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": 6.942972601071445,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.1412,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 7.274370492384997,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.1589,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 7.1020087230512186,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.1232,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 7.263503625040815,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.1795,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.864001703205238,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.1576,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.43408609509781,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.1284,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 5.091777759508549,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 1.0763,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.7469857766489625,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 1.0637,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.9470197843377615,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 1.0339,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.721098659943853,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 1.0385,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.5882800098726566,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 1.0302,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.5181546634914795,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9872,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 3.8013969136384334,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9835,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.433757721518224,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9622,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.0239563963805423,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9707,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.240022848630295,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.9201,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.3273255168028792,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8789,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.113306161617231,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.919,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.68736462565708,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.9035,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.3767746723147263,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8626,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.1198893287691045,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.8461,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.423667650562201,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8499,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.4999422984650983,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.8423,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.264872363977058,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.802,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.902631656171295,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.8404,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.0658802795525926,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.8089,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.0204861170986583,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7974,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.8918372688162807,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7802,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.8476251166579827,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7817,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.924276810330819,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.803,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7966660804867749,
+      "learning_rate": 1e-05,
+      "loss": 0.7916,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7516440584182685,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7826,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.753752669678862,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7635,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.8048052931178297,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7648,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.7447793180946454,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.7648,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.7211976997906194,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7662,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.7984424905003054,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7596,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.7122970427960141,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7713,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.7349324751668059,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7547,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.8275199861901612,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.77,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.6748422856669954,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.7709,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.7593989211214066,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.7365,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.8658862252460563,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.758,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.673414857238853,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.752,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.8898220844829617,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7458,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.7140888351059241,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.7543,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.8730073960801225,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.7526,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.7467842898185023,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.7436,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.9263038157266031,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.7241,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.6853848621245014,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.7537,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.7812854795860303,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.7179,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.6522148306786013,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.7269,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.786281945934443,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.7291,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.6458813920573292,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.7355,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.650713876705963,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.7371,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.7209426511490539,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.7385,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.721012291496249,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.7432,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.6839674953291137,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.7339,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.722531271585373,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.7598,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.6647583106844287,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.7096,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.7349781502027213,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.7187,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.687792872291443,
+      "learning_rate": 2e-05,
+      "loss": 0.7231,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.6738263084649144,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.7337,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.9460160058121421,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.7177,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.686157619524909,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.7228,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 1.2409435922508296,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.7081,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.6926946431408979,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.7012,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.8001529370112573,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.7229,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.7569781072532157,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.729,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.7209439880691869,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.7256,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.7538474356922459,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.7338,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.7226709707920258,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.7,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.7651751007725693,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6993,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.6921476296394198,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.7222,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.8550270461275882,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.712,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.607610099512061,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.7197,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.8628200902777112,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.7329,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.8672639512738568,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.7037,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 1.0953765299454246,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.7068,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.7846421778284934,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.7044,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.9005008934518463,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.7153,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.8699981850192329,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.7398,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.747562134370576,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.7168,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.8075498243597028,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.7321,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.7556162844508504,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.7114,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.8003892616656831,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6914,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.8066419344053487,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.7147,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.9012063352518128,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.7127,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.806885883149076,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.7217,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.693730466704314,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.7202,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.8637814357741358,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.7136,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.8423876133681261,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6982,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.7473934079116322,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.7087,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.1179286036404779,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6911,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.9957626841798636,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.694,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.8041473659607422,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.7051,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.8192675264856624,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.7139,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.7083720317282798,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6831,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.6587850951061014,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.709,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.0686660592820016,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.7101,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.91396538558362,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6986,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.0992570516081237,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.715,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.006898062697234,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6967,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.979283254749083,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.7284,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.0106876915170604,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.7168,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.0110342614123942,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.708,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.0403903495133706,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.7237,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.9157809274127848,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.706,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.0887644991461911,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.684,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.9935112137614743,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.7106,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.9245036779053515,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.7179,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.9306034501796548,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.7066,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.7299663554539398,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6891,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.0314333911682174,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.728,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.725563661346128,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.7222,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.8026871546217916,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.7155,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.9969573432069145,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.7354,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.8739742654641134,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6865,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.9181839503770259,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6954,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.8723376862617166,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6942,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.9067018120643638,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.7038,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.3427600038861043,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.7243,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.8328831152304988,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.7038,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.425027152281263,
+      "learning_rate": 4e-05,
+      "loss": 0.7268,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.0600374212011863,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.7336,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.322589552053183,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.7167,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 1.0751129195051716,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.7101,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.0084353394156995,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6917,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.06929110324764,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.718,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.0416659567835245,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.7354,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.8215800411898269,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.7112,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.9557796923545172,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.6975,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.7782618001621247,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6962,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.8617870877142595,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6998,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.6988547154850117,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.7059,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.7796400090372853,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6816,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.7394622544020901,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.7162,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.8276019425290009,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.6995,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.690455314660011,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.6969,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.8727963668677908,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6994,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.7478989978026196,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6829,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.8628743406124969,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.7256,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.7236700982613496,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.7004,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.7079452772643655,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.7244,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.7541017874533913,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.6877,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.6898163223415911,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.721,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.6449293350354117,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.7045,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.7234247575303413,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.7014,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6753197175719816,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6885,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.6374253180550165,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.7072,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.8009810744968906,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.7177,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.9588159693043375,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.711,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.6462502860589621,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6851,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.8251565748182984,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.697,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.6145067821573502,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.712,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.6464688750049639,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.6756,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.8233899034314994,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.7025,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.7448840845940736,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.711,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.6582388219448405,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.7079,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 1.1140880110487328,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6734,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.6433763352080922,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6931,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.7850720334680423,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6951,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.8615643372242394,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.7189,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.6739990149273377,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6953,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.8573195094755125,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.7048,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.8147304513941195,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.7027,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.5846910211557799,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.7041,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.7274410546057029,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.7062,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.6548575674407744,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.7019,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.6332408783218992,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6983,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5270542751021902,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.6883,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.6154585516669865,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.6876,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.6091550888330678,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.7005,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.7499099452373762,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.7174,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.5544488631833826,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.7094,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.6791321884659147,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.7034,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5327242843886634,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.7051,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.5919521863942397,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.7062,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.570686314363627,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6981,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.5898936143272694,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.6811,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.6472576923479176,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.715,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.674078171133154,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.726,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.5959226171129522,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.6941,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.5740421558344734,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.708,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.773619867065025,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.7206,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.5755769830521705,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.6763,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.6505331221586954,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.693,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.7282567323442279,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.7128,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.5785822518920648,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.7023,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.6845393817067426,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.7066,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.7320629502329586,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.674,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.6673004383471355,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.6934,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.6944279514411141,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6699,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.6291241044776039,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.691,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.6021160404919286,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6906,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.5987605795965831,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.7128,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.59657092492049,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.7154,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.5777438152764647,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.7041,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.6375070230151364,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.7009,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.6901181112996669,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.7114,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.7828832089068205,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.7266,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.8878375342405622,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.6962,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.5543872196858022,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.7066,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.8328166471135329,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.6821,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.5940246162107751,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.6867,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.7869704501561257,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6973,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.5658867798964767,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.6631,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.6341732194407964,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6588,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.7332694898367749,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6726,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.5998351509210339,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6961,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.6523256834848006,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.6858,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5366664552082545,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6636,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.6924387916213781,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.6818,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.6157687592746773,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.6827,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.5250269733793669,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6938,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.6257776360479353,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.6755,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.6334685205390918,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.7085,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5788987933975644,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.6879,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.7913978539047446,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.7002,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.6368547451210717,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.6831,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.7179132405211135,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.6768,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.6568811824503954,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6865,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.6947517902774524,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.6799,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.719372329277766,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.685,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.5542753622599904,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.6847,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.7590901742936617,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.658,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.49897602000149277,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.6923,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.6299026079585931,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6786,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5571658362153324,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.691,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.6288318042397765,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.6702,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.6499560201654135,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.7024,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5254213120919178,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.6868,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.5783369174905522,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.6639,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.6401850796560458,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.7039,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.5779172728788429,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.679,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.8136089136396493,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.6975,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.5173424484423915,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.6735,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.7336077212647041,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.6812,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.6004484408596527,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.6718,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.7062720833731685,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.6957,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.7910134774401758,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.6889,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.5738923579550766,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.673,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.8094404398743662,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.6768,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.588104954774358,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.6589,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.8643266209500539,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.6818,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.591595016225481,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.672,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5663054328111914,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.6726,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.6570701924000998,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5694,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.5450768130037024,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5815,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.6793651298496692,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5443,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.7452308245123384,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5525,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.7274606520994371,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5385,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.6497891973916592,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.559,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.8481499787509749,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5697,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5715199914690972,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5643,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.8093862890330905,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5728,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.6039387368089019,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5656,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.7138129473801568,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5571,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.6542033058873279,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5705,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.7235667610297482,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5618,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.5628241593709418,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5634,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.7560106795874016,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5605,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5880140599329021,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5574,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.6842038425105134,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5582,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.582531052495127,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5681,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.5615496724846258,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5548,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.5828795299842768,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5722,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.5955465705395669,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5517,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5795102264045158,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5391,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.5303310368063463,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5612,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.6628962889971413,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5392,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.6413487975738116,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5635,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.6809239634188292,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5341,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.6053779974131289,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5622,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.7561312548325491,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5547,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.5135639116236025,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.553,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.684931725287035,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5407,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.6121827555304616,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5294,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.6327571630266581,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5604,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.6015507338598143,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5414,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.6014763650436604,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5568,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.5603996095556509,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5509,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.5851196369161478,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.5573,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5682915125031825,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.556,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5868574735421211,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5482,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.6327424443838472,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5714,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.7169224358722898,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.548,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.516560040841256,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5635,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.6961671282393811,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5446,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.56985843386344,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5606,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5748515889405462,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5468,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.585869111437449,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5489,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.615362491555497,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.568,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5230489981850811,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5547,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.58220690458362,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5568,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.5657186024723249,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5373,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.6141616980650735,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5602,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.4984593191063502,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.5366,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.5712336803938952,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5724,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.532387832362116,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5641,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.5644043560746359,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5606,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.4993173095704987,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5603,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.5544037879781516,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5607,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.5419128357408739,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5701,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.5141904300277415,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.542,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5241412550641447,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5583,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.5958672747120616,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5509,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.6193352581000144,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5739,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.5341001944902151,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5504,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.6229290575572415,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5643,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.5607395496281096,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.5441,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.551426197304066,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5531,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.6630373505137549,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5693,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.5065957252870706,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5664,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5249030358524007,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5568,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.6831267733928549,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5572,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.7698261993838464,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.5759,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.5635372249548691,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5609,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.7780736855443777,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5586,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.6651453487800393,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5616,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.6190634892033566,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.5584,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.6302323172867258,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5553,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.6000749230810436,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5381,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.6194680744518758,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5495,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.6415800789528868,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5511,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.6311903856618839,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.554,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.7574375470349596,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.5764,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.6114806844465063,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5829,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.7805021167241296,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5622,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.576516790705326,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5574,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.7141174185249417,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5476,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.6924409644708011,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5663,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.664057845601002,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5767,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.5512596112679503,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5439,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.6064724115772029,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5487,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.5022239755736536,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5897,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.6731548087089059,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5613,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.4869335159890864,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5506,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.6054559777818486,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5469,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.6136321962382456,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5676,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.5829638719102298,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5579,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.5577824599304676,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5651,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.5554359737308044,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5656,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.6150262282753375,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5755,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.6697754063138541,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5459,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.5651017602497931,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5495,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.6533466108694358,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5872,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.6060045987294435,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5495,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5599237523181013,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5529,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.6208191985117438,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5724,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5221037990780857,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5762,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5839489626317362,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5598,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.5739928838416055,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.557,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.4983548362247336,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5576,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.5956435668823638,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5465,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.49706367199041795,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5658,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.5471928723851542,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5259,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.5810447342543532,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5491,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.5373056945035952,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5429,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.581737223191348,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5385,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.6060637226979688,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5523,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.6223415722278556,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5699,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.561777926863728,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5841,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.549137482905935,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5475,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.4919406163816352,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5423,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.5782414453892972,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5741,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.49072451014056656,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.564,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.5911114167850081,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5698,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.5274026906420255,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5721,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.5313893573943679,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5738,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.5095794289209861,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5299,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.5560882610583895,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5544,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.520762230157012,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5569,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.6116130906768851,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5651,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.5464316538633777,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5663,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.5692682583877192,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5447,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.546958175979348,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5643,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.6138010848394085,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5613,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.5430968206713975,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5546,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.599797597292036,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5674,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.5827312710989697,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5572,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.5471115414828701,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5497,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.5371325710076547,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5347,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.5704004613756388,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5633,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.5436785847400979,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5583,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.5934598032225769,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5455,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.5531274612866255,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5481,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.5394548195114188,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5515,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.5294746928193378,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5593,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.5099136089678604,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5449,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.5073135832689362,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5469,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.47343158388989365,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.549,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.49677960497030194,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5602,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.5415737523390325,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5597,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.5414095455423743,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.557,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.5434188743739381,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5745,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.6042125870168678,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5497,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.5611326726606312,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5648,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.6303524652249016,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5694,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4488798547947485,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5335,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.6604925957371425,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5802,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.4705451703656483,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.5722,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.5863125723930181,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5649,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.49674937846742795,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5602,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.5165998474090634,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.56,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.5119503857148233,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5392,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.47056801855351565,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5595,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.5228425203004609,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5529,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.4863838316142981,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5417,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.46588179114231243,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.541,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.5131124731786317,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5703,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.48563499926133863,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.552,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.47916936654976133,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5593,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.4815162536399678,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.547,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.5568440866036163,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5561,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.5472828437629763,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.5665,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.5823093172170815,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5662,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.5589871353313982,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5803,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.6578611652732539,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5647,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.6019819088308884,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5584,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.5581484831108845,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5463,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.6005543492709602,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5509,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.5585047470916519,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5487,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.5682335580372245,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5659,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.5600993825196519,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5577,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.5461912635496476,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5509,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.5926260177506153,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5644,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.5959021244437845,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5592,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.46270016264933966,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.548,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.584734135015607,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5549,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.5732133502567226,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5665,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.534352463317189,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5364,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.5635585345796398,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5522,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.5225394559726037,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5575,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.5307479160317833,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.554,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.5910421586073352,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5614,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.5365283071163274,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5785,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.5806500505477514,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5779,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.5194083839649574,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5698,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.5601707994594164,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5526,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.5183214313528204,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5725,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.5206018408888748,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5583,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.4959494692915475,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5453,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.5241168429039557,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5446,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.5122145689367384,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.5749,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.4601494673560328,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5649,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.4392217441216441,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5598,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.49623745667255476,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.553,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.4372682851562209,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5594,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.4884054020910858,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5709,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.446548067552475,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5459,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.5317485412174123,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5882,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.4603440975200663,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5611,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.6074875340326577,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5471,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.5166083450234906,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5417,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.6514374045344168,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5855,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.47836858206023325,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5615,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.6181881894731301,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5628,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.5035626333192696,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5411,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.5283476503203801,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5585,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.49127897104199386,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5652,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.5515895796578943,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.5548,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.5051392725814681,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5835,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.5698140063827895,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5508,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.4768952531903233,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5553,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.4996205755886404,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5604,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.4780992281182001,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5409,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.48061718719833796,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5615,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.5701099674631144,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5432,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.48041755683173437,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5401,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.5701187148784184,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5412,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.4950547300164685,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.5539,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.5412975617935949,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5616,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.5223012874661932,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.5641,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.540990516507999,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5665,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.47851755861389866,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5448,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.49838511848807915,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.558,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.5111056014258318,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5674,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.5195742995991699,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5486,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.4802221786400986,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5536,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.4926312013844188,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5569,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.5271582299419536,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.5421,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.4977219833642667,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5595,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.5071067670879992,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5434,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.4433339143125335,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5351,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.5074643140264908,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.57,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.4562669398491736,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.544,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.46999408473845355,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5598,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.4728500606143059,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5706,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.4607057535382554,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5558,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.47435132446344164,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5749,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.5082866128456671,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5725,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.4084911126514184,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5667,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5020018693895874,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.561,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.714891118751107,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.4202,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.847599364959623,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.4479,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.4238068866523268,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.4172,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.6680509624739339,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.4283,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.45551944783844966,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.4312,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.5401553501950365,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.4141,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.46224406713280164,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3993,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.5103968133246267,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3936,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.4534023900698407,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.42,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.5878264964470541,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.4016,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.45046162100888637,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.4218,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.5882558196837897,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.4258,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.5188329630233126,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.4375,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.45377728834252906,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.4048,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.50279082943128,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.4198,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.5178435442367886,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.4245,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.46343680625013534,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.4168,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.5119553093213111,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.4103,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.5460398740577684,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.4134,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.489912606853807,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3987,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.5404832659102159,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.4102,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.4778938985095347,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.4162,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.5027206147611618,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.4192,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.458886870963197,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.4256,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.48759080128044563,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.4263,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.4611618910509574,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.4052,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.46446472139649164,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.4112,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.44195930833645625,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.4084,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.5058838505533448,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.4059,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.46840056774971955,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.4124,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.44390618925490016,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.423,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.4915768643423526,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.4264,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.44601587937497117,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.4056,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.52392947521619,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.4426,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.4595933258176781,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3945,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.527984118711373,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.4109,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.497508019260246,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.4303,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.46861596466758093,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.4134,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.4923300675812259,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.4095,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.4302665458299563,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.4059,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.552204056524918,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.4002,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.4529565213930843,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.4192,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.49761990053654603,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.422,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.539224086468645,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.4287,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.5056168669701598,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.4232,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.5188971285905014,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.4165,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.5344614988542594,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.4162,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.47608970153779395,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3995,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.4776410086053797,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.4204,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.5263753839342112,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.4241,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.45679837191004374,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3999,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.4764766700525995,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.411,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.4735988563448093,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.4108,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.4326601154890732,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.4294,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.4632678572904072,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3925,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.4537915961240924,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.4202,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.4552257700299121,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.4208,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.4807085465855315,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3771,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.43622188727775013,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.4451,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.40536458781364404,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.4203,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.43448485388826374,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.4249,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.4770768942723972,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.4076,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.43683176487991,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.4203,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.4719172775243783,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.4178,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.46492948158388697,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3942,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.4585091572582227,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.4033,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.47431181400777606,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.4088,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.4525186256254847,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.4248,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.5238614445037011,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.4018,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.4530262999488542,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.4001,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.5773872187766409,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.4198,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.4191784928458021,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.4069,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.5216313146961177,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.413,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.4645213482414846,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.3854,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.5200519407456169,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3789,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.45513246584198336,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.411,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.5179090970715635,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4169,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.4205595422022868,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.4429,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.5333641901248988,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.4092,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.4671339237227354,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3914,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.5726420935893702,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.4217,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.47258581795606436,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.4481,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.5053717499286753,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.4274,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.5031325351831356,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.4236,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.4341526073003402,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3903,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.4672312728673047,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.4111,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.4275626509461323,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3757,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.4912315173812233,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.4103,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.43984776501376344,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.4159,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.4331569883592562,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.4249,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.4330981544812645,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.4034,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.4662383525575362,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.4145,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.43758800056920544,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.433,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.45335228030372865,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.4034,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.40969320270682336,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.4105,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.4565529667257317,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.4179,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.44310860236407107,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.4177,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.4290493608427889,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.4309,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.4712316642172006,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.4156,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.4456587904009671,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.4127,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.4570275987463864,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3956,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.4301225118412888,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.4039,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.47660087758727665,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.4204,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.4052705176144776,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3996,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.5105165652506455,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.4111,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.47000472910147734,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.4284,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.4287380339302124,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.4019,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.5019226214365751,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.4338,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.43304366696337065,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.4147,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.43333783738259407,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3986,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.4458055465765878,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.4285,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.43511745785719275,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.394,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.44725309286614023,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.4161,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.4688437781032993,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.4126,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.4240582097783621,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3985,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.4597335144103167,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3956,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.42185501737502573,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.4012,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.4270505899302832,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.4209,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.46955005647853276,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.4197,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.4439259655004338,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.4159,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.4220982279682082,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.4402,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.45684772650087546,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.4298,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.4247327421623316,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.4444,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.46780302156874815,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3875,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.45855617768781876,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3954,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.43983328858188886,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.4009,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.4475821028058819,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.417,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.4390378956677425,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.4116,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.4421497894763745,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.428,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.4223722331392806,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.3916,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.4383467171613578,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.4104,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.4194301751632662,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.4045,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.4265214418127582,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.4082,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.4523437741279549,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.4258,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.4078386927174807,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3944,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.421201080971416,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.4187,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.4598456142475718,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3878,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.4384760834361649,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.4192,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.4213018645195888,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.4231,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.4933140971012986,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.399,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.41935675177890613,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.424,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.4467438356049115,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.408,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.40672170186667456,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.398,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.4450269536171574,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.4196,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.4494945244752626,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.4148,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.44586079624198094,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.4231,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.4564406536739291,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3835,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.4101836363740444,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.4298,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.4415805773377833,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.4014,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.4401401328652939,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.4121,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.4073142834689562,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.4181,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.4487547140765669,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.4055,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.37910746935008727,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.4184,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.43858318896096893,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.4091,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.42246757777150695,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.4251,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.403564468831465,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.4398,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.4315040261035724,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.4184,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.38793127527376714,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.4287,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.4431094712057941,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.4231,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.4002468186453386,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.4215,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.44361315604943774,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4083,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.44085406058947824,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.4147,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.42908704398730657,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.4146,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.45031384536797453,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.4164,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.4214796551472105,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.4255,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.414008325329202,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.4213,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.430394454865695,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.4102,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.41528067241402317,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.4185,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.4255419904459261,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.4061,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.42194508350405635,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3787,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.412460798776737,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3998,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.4140039198446458,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.4247,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.43071674576582497,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3999,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.41919882795114205,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.4127,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.42703815483844,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.4122,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.42984616480753535,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3863,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.38313050219807604,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.4133,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.4156589306416081,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.4272,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.41193222410496816,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.407,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.43959973776608857,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.4099,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.4374383752340829,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3924,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.44439903903025374,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.4069,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.4356574192058686,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.4342,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.4004522407008378,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.4257,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.42019737246888145,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.4105,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.404439611807079,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.4056,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.420814587928824,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.4092,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.42398149482702546,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.4142,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.4302343303427169,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.4308,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.42614554412596495,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.4248,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.44738614463314247,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.4034,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.40662950130308434,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.4241,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.43646383910763703,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.4228,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.41035015367283784,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.4078,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.40644928492107457,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.4295,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.41371936687977323,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.4182,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.4454269135212288,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3765,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.4024836652418892,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.4166,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.4571976696507955,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3906,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.4103012578488923,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.393,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.467135586054299,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.4065,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.4412327689180016,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.4055,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.4502103628815858,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.4015,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.43366601073527916,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3898,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.4072425794443396,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.4182,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.38688739926475163,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3986,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.38931492147653013,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.4136,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.39593953403200277,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3873,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.4028976686271683,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3943,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.410669599604932,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.4053,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.42465454517610685,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.4371,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.4222157429717235,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.4023,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.41417174890225433,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3768,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.4944221028292575,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.4137,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.4111530781224235,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.406,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.42266704162283947,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3831,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.44659508886925814,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3999,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.3898345002061509,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.4222,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.4515588957321067,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.4008,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.4214978862601101,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3931,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.44836571833961,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.4154,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.41088539230924914,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.4307,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.41019783808410676,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.4134,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.4207586922628982,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.4193,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.4044155574014975,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3883,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.43587773544500774,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.4335,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.3992275433015573,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.4237,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.40961935202551,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.4282,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.4127500809717396,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.4133,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.4135263837791416,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3976,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.4482863768103365,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.4122,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.392072558750952,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.4027,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.452256957727583,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3846,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.4456474397685214,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.418,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.4036698936378597,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.4172,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.4203626125027886,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3868,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.4135373835322303,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.4095,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.402346884715839,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.4058,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.39535219454420384,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.4207,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.40496418414171986,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.4019,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.4118508448145776,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.4048,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.452296993178111,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.4023,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.4281064231097284,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.4112,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.44397997738275014,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3991,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.48628957168153975,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.4082,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.4851402662262113,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.4155,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.4041784217943458,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.4131,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.6532461194724538,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.3203,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.412284516847881,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.3148,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.756958466333922,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.3105,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.5809113794615819,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.3385,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.538087353183133,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.3236,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.615121417164802,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.3016,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.4460807196080808,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.3251,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.4944462181049625,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.3169,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.560704659598515,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.3162,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.5020808694149362,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.308,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.4303453807240522,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2765,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.4247204333399636,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.3005,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.451569711374309,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.297,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.4187627361144555,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.3182,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3794397545086729,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2927,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.47662603871932563,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.3026,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.42959800498928946,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2922,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.3673332961599993,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.3238,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.4191140069036359,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.3052,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.385171288337976,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.3032,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.3803527782958374,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2867,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.37934511826345096,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.316,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.3970904313040757,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.305,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.4078215994700007,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.3083,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.3648163965933911,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.296,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.43086024840143816,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2943,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.42044617549207036,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.3113,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.37749624776827934,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.3095,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.40226318208269346,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2954,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.39853677003733917,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2996,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.3492226172294328,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.3439,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.35696270839870814,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.3127,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.35814945595980974,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2689,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.3502138404483637,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.3363,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.3470592677561302,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.3125,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.35731895538827246,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2893,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.3660700876530698,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.3054,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.38978183512032755,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.3257,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.3911743212192198,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.3323,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.3748878956597432,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2891,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.3640120269920037,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.3199,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.33221559826304,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.3017,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.35312420096368496,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2918,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.37259301880738066,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2735,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.3425814112662897,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.3185,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.3624309674537822,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.3079,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.3673989649593521,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2822,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.3462092105824052,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2793,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.361406927733927,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.3161,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.3683415529309823,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2961,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.3549556661317242,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2972,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.371587821994228,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.3197,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.3546448501505361,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.3002,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.3366747283195333,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2828,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.36667366414738445,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.3053,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.37151516560308484,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2917,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.34213140988944174,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.302,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.3569751593816248,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.3484,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.3548951763771542,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.3145,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.3500692403974437,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2887,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.35332254569392874,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.3147,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.37553295147430754,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2905,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.3494388315687768,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.3308,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.36042920605432355,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.3023,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.37999755839793914,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2896,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.3444555619832787,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.3105,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.3472596777378627,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2944,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.3532390255090748,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.3103,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.33489694221059907,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2896,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.33832400490048015,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.3047,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.34876111094077145,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2966,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.3617499341271051,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.3021,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.3625942623757372,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.3125,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.3597927688689315,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.3269,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.35838490674378265,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.3012,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.38034971136068224,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.3163,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.3666494470822015,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2915,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.3219135508257872,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.3092,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.3426589224099514,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2831,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.3388244541809436,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.3167,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.3604856547720706,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.3175,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.34407081327720695,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2825,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.334150014074755,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.3326,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.3547142475987456,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.3117,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.3602007021942534,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.3079,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.37280865501133065,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.3071,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.3419356866749891,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2817,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.3647005952961701,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2714,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.37435725572064815,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2772,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.3702169176305095,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2812,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.3560708106577693,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2982,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.37222516702281516,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2977,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3527168807663789,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2793,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.3645198912283778,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.3237,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.3636854641948233,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2872,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.3608256446063911,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2881,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.33906251087625755,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2865,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.3506194526158018,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.3068,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.3784315433307159,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.3088,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.37579758758290954,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.33,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.3435736217428535,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.299,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.34184424023250104,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.3036,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.36007174995717633,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.3088,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.3354016686538351,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.3154,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.3459584617452908,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.285,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.3428879631868324,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2963,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.3534559238013964,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2789,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.33761913838462304,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.3166,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.3463678610671779,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.282,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.3409177551253938,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.3087,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.3278529138618826,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2866,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.3388586909399508,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.279,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.34769115015539104,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.3038,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.31876854585469877,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2727,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.3395452690645815,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2813,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.3319393003198318,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2687,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.3260853330722799,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.3287,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.33740605643128513,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.3003,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.33459114497652764,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.3079,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.3422804197972526,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2954,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.34605821042458973,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2752,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.35142985413231526,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.3085,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.32544849592778663,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2799,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.3584082876418853,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2955,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.35612574735484037,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2854,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.33507473184226155,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2684,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.3418621098107702,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.3085,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.3633438675804184,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2961,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.3577053993252952,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2566,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.3372404258938528,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2969,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.3232455788696414,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.3165,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.3466477447147838,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2934,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.3598659651985966,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2994,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.35654884470422443,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2721,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.3280082168550394,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2917,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.3249812996967332,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2704,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.35112180655566344,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2977,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.3572798305939122,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2984,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.3372133137652759,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2812,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.350226633480729,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2904,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.3366545318410615,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.3364,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.32559584289149807,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2937,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.33618514617594303,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2796,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.3632128350362819,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2822,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.33168901711618815,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.3022,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.33916734123490094,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2637,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.3463136923372992,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2985,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.33398922477999143,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.297,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.3270136916229444,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.3143,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.33096896684469007,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2936,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.33172588510900053,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2916,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.3205819229745957,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.3008,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.34570042348655855,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2959,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.33249862359144006,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2785,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.32647618376420157,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2924,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.3388189426538843,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.3035,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.33080442247764724,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.3005,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.344880744835281,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2951,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.34734531589555306,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2917,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.3469314981804123,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2947,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.35073812360940704,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2811,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.3557024038059759,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2904,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.3411925259003655,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.3226,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.33558304787229415,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2969,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.3252551679269038,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2977,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.34168437328699125,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.3061,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.34107499059975993,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.3057,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.3209479964886134,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.3003,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.32628756472744885,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.3193,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.3347069766741303,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.3002,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.33851570657993885,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2835,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.3396308492433413,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2786,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.3360268446862623,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.321,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.30674349748130025,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.3278,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.32810384967526984,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2738,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.34384351243705763,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.3023,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.3234295017406861,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.288,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.3320137622447016,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.3153,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.3432815562891715,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2811,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.3356360429144249,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.3243,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.3404150460000038,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2952,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.3402038131432844,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.3284,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.36185159514407805,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2932,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.3377880881658572,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2953,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.3420291610615689,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.3044,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.3298626648955638,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2894,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.34305513070003474,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.293,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.3279683977756324,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.3082,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.3108653489897831,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2879,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.3178790593634154,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.3113,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.3242173209379642,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.328,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.3487750486570586,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2878,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.33016630950494824,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2724,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.3467525231319372,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2838,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.34485200855680814,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2984,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.3336340045865642,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2935,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.31925791420234373,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.3032,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.334536394956521,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2726,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.32348305657357995,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2953,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.34389739085701826,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2532,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.3474652611829293,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.3012,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.3288865224428118,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2986,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.32405981046769083,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.3116,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.33899081192056874,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2633,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.3276214182551796,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2872,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.32131135807887834,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2914,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.3156118681013975,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2631,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.33437392167948776,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2846,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.3392512212810045,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2826,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.34405959672111397,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2902,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.32780779022422535,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2923,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.3378291752567787,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2673,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.3289530730085772,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2994,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.3333047911607234,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.3202,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.32006691591952086,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.3136,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.3421618664411335,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.3119,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.3395261805986684,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.3029,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.32063680308126885,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2928,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.31440137954294334,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.3089,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.3123914899054246,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.3066,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.33545003753765845,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.281,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.3360906545829926,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2914,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.31139040496952974,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.3097,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.3257360372369606,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2967,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.350696446539407,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2927,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.34167859474411727,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2747,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.345346031799756,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.3185,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.3170287785846727,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2665,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.32884411876198916,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2938,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.3369925897705977,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2897,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.31919589018200556,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.3069,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.32687743243118417,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2735,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.3363726347359758,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.3133,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.3251888637967272,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2551,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.3417245547912569,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.308,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.3190133483865621,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.3015,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.34426998420142046,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2939,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.3169106870838058,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2983,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.3219095420899402,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2851,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.3173217032189768,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.3379,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.32082762321543096,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2803,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.32431141951761583,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2704,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.3177583009183231,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.3019,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.34360291803778686,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2827,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.35017086100480177,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2885,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.3103892134818647,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2958,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2886845768466698,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.3049,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.5430668967492195,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.2151,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.40939167200355137,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.2469,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.3098677026545855,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.2231,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.33271704892395587,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.2182,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.5444561020610956,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.2497,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.5185482039860664,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.2753,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3969345972719522,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.261,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.4061694750889794,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.2415,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.3819340410647206,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.258,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.4126309210128106,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.2205,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.40255582108166776,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.2102,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.33012294457950525,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.2663,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.2853726310542805,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.2288,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.3284168756080659,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.238,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.39388913323312813,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.2603,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.4036433610874266,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.2599,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.34296396527696754,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.2212,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.29403767598349123,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.2585,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.316573098735475,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.2369,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.32300082587246154,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.2228,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.33306018628228495,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.2154,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.33163660827751307,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.2387,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.2718604495263681,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.2285,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.27818108417987486,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.2416,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.30354609376584313,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.2343,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.2934574296527086,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.2474,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.3061018152202143,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.2324,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.2997532784035326,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.2321,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.2780143679084382,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.2444,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.27160753671308924,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.2852,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.28520705708929694,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.2475,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2874288689426395,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.2317,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.2822137717409348,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.2478,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.28088298233647585,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.235,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.2760902853564496,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.2407,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2936504682870651,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.2034,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.2781479045261909,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.2103,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.26909580206786043,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.2323,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2938040373429214,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.2373,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.2789341388237214,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.2459,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.2837270868807281,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.2372,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.2798917740030386,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.207,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.27439513659965326,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.2234,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.27277119128065735,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.237,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.28655203977763294,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.2278,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.26625793126531777,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.2442,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.2722201068285373,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.2422,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.2706654489860444,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.2275,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.25136891895543706,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.2576,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.2845703386489974,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.2545,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.275314433940237,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.2218,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.2750439096663465,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.2543,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.2807991369003968,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.2418,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.28862214099645017,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.2575,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.2708556436951992,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.2292,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.2674870229948473,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.2287,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.29784371842854973,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.2383,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.26333840395137487,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.2194,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.2772858265318135,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.227,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.2824242062807,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.2165,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.2651246289720427,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.2176,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.27720588685385084,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.2043,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.2619144388372195,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.2449,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.26308598456654203,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.226,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.27254891919734875,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.2139,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.2749525207677551,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.2369,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.2712652819657269,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.2417,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.26553348951474687,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.2047,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.2711153124371305,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.2446,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.2635550669481188,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.2323,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.27217895957524524,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.2176,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.25676360238199925,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.2559,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.26296063732626135,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.2109,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.26442043042701696,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.2234,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.26972799446270895,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.2457,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.2576582721193562,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.2493,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.2704179835753245,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.2214,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.2605068956209222,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.2226,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.27979506542422744,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.2267,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.2756756108179579,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.2258,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.2588038367643494,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.2125,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.2759683715173921,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.2424,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.25911791441668935,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.2421,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.26774859556395875,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.262,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.25898162295235483,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.2197,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.2492270268812741,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.254,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.26101833004179037,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.2249,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.2704010406294356,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.2344,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.27674396977153154,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.2736,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2723719507891954,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.2571,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2716034993521182,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.2377,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.27226093478910185,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.2366,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.25649126034192365,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.231,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.27713874643558967,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.2522,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.26870907233178243,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.2472,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.26137182425897176,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.2373,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.27191969456238074,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.2298,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.25969420151083994,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.2451,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.27162431039310353,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.2216,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.2715612511582366,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.2223,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.2801265392762274,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.2192,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.27174619703740444,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.231,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.24272892967581,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.2132,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.25664941287055804,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.2234,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.25774770819457854,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.2441,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.26760363029033735,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.2392,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.26793550823024515,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.2287,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.2636738762927424,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.2156,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.2631385512537731,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.2539,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.26706944312830494,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.2194,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.28469151373226587,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.2218,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.2850712365604702,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.2569,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2699823857706972,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.2557,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.27707809215806484,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.2265,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.27031390789707865,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.2332,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.2742086935632072,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.2505,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.2656265847982999,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.2309,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.26817090625020007,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.2132,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.259218586320286,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.2595,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.2656374123087416,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.2547,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.28069372410035676,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.2599,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.270362150920403,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.2282,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.25019309168068893,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.2303,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.25670094597476534,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.2074,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.2647939170504991,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.2343,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.2912394045437335,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.244,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.25918482747399596,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.2085,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.2691740054402849,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.2238,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.2826925148242043,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.256,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2663898484092447,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.2406,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.2531483428823242,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.2568,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.27524908046713076,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.2453,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.2842184772690936,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.2337,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.2772634697540167,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.22,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.25898569301833313,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.2167,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.26340333798607307,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.2422,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.2804355306732828,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.2601,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.26014748564642776,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.2,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2692332020480921,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.2298,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.27459228231969135,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.2206,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.26775794366412,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.2419,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.25604763370980965,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.2268,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.25011507405984706,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.2308,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.2729391579396118,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.2429,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.2500063433078464,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.2534,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2607893212428444,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.2341,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.2750062293621644,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.2374,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.26824695691581973,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.2376,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.2752773992217112,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.2325,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.267473010843177,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.2396,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.2740887279124805,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.2104,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.26214910525101454,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.2283,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.2614827305825287,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.2481,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.27857596273004154,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.2179,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.25689765397675113,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.2326,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.2616983853476394,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.2629,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.2622046611602647,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.2677,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.2649085337459911,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.2063,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.2565829495514667,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.2542,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.274432576234703,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.2424,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.2526724726504924,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.2319,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.2762945044328773,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.2164,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.27203441210559237,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.2484,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.26860978721241724,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.2296,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.26908447326377827,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.2331,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.251571996723459,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.2191,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.27290661461933474,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.2391,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.26565391845778946,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.2212,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.25133218391538104,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.2316,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.27944802569516564,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.2613,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.26323004284113627,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.21,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.27425103215629526,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.2685,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.2567220464192621,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.2368,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.26291194114430766,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.2363,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.25830892657362303,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.2119,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.26511045685936546,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.2453,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.272593177718769,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.2249,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.262358802493824,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.2264,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.2639589881036655,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.2386,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.2628980557653983,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.2182,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2638066464372134,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.2247,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.2631359563683872,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.2318,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.25753616436936594,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.242,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.25893467764241934,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.2429,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.2750187058004876,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.2328,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.25960964706581113,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.2499,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2665569502000977,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.2528,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.27206676758962,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.2011,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.27465824838065833,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.2225,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.25462590468828084,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.2366,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.2816669628641986,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.2229,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.2681561366600725,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.2592,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.25905087113926933,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.2333,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.2836343274990789,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.2579,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2568626754817778,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.2484,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.2601404006392459,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.2169,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.26916258521786035,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.2415,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.28292742392054754,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.2316,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.26348656143523275,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.2371,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.27091974364854576,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.2384,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.25761720793776494,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.266,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.27279529619644416,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.2351,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.2743344979491243,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.2382,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.2647739713998883,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.2444,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.25927821170731963,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.2439,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.27092758290348445,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.2258,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.26557751827244025,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.2357,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.25971160780094515,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.2262,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.25735573553239727,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.2333,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.2616559111158278,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.2563,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.25944016464785974,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.2253,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.26498886239477587,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.2203,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.26240477627332137,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.2367,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.26382193213078764,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.2581,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.2633795854934592,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.2347,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.2648357570245524,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.2396,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.26714824378268653,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.2628,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.26726842177957594,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.2433,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.26741104563859436,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.225,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.2649138189009745,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.2394,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.27038658272778016,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.2347,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.2729275556503985,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.2128,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.28025009495424386,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.2543,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.256551140719159,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.213,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.2807931663492663,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.2303,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.27088595178392183,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.2438,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.2728142080212622,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.2345,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.25815492066688256,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.2604,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.26173398681685733,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.2489,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.26362796233475105,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.2567,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.27364398763912373,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.2648,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.2539233302622435,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.2244,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.254989881579746,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.2328,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.2618882450975687,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.2377,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.27296055275273384,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.2388,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.27184132853530013,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.2444,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.26285781800030084,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.2578,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.2525292240152352,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.2187,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.27043697786468557,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.2063,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.2618886352978046,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.2165,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.26784253457205237,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.2225,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2656397125319535,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.2469,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.26447667382028056,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.2488,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.26065775156893367,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.2405,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.25057529082292945,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.2205,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.2715594487750537,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.2446,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2553871772916396,
+      "learning_rate": 0.0,
+      "loss": 0.2477,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 2.1632960746246636e+18,
+      "train_loss": 0.4484364619380549,
+      "train_runtime": 24245.7929,
+      "train_samples_per_second": 6.517,
+      "train_steps_per_second": 0.051
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.1632960746246636e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}