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": 7.896956593866752,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.1817,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 8.250890619374346,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.2341,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 8.161570874063477,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.2421,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 8.14138655736581,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.2129,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 7.6400542283563135,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.1968,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 7.366307458534047,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.2071,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 5.906559386801082,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 1.1148,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 5.645414715937756,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 1.1308,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 3.230356448544544,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 1.0682,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.7503732386247943,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 1.0641,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.4424156293748056,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 1.009,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 4.304573195668639,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 1.0096,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.455665086224842,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9854,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.429432174129503,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9983,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 4.162399756010438,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9912,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.9184681788117923,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.9427,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.982359319576255,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.9343,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.7613764146088684,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.9206,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 2.2919096534798946,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8715,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 2.093097399092634,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8567,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.8251750793663921,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.8893,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.8613192675489114,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8737,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.8907206955955638,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.9066,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.7626663527091586,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.8515,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.5390011540538677,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.8632,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.3649686000456573,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.8623,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.3550378514328105,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.8552,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 1.1164679232012884,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.8176,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 1.028790892089515,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.8067,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 1.0555091738398206,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7917,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.8477262398410288,
+      "learning_rate": 1e-05,
+      "loss": 0.7879,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 1.0431428668592304,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.8274,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 1.0067342571333378,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.8173,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.9453301547643268,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7928,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.8269252113517598,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.8067,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.8464035774380969,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7965,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.9614275427541185,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.8014,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.7584668239489472,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7835,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.7783325320635575,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7835,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.7799771097150662,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7797,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.8677795533720064,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.772,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.9002757522145695,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.7943,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.7694272241410326,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7683,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 1.038402953922992,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.7702,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.8961175986675348,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7863,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.8057272153682119,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.7796,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.7771578248857729,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.7586,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.8772294556721161,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.7754,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.8500989755892019,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.7863,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.7815542837006726,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.7701,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.9159987118797778,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.7634,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.7858338464693339,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.7615,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.8754500670539732,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.7839,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.8315694885288852,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.7684,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.9719554306235046,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.7955,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 1.0155389372571137,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.7835,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 1.0645274823812731,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.7731,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 1.0224840117364242,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.7655,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.8511744101272428,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.7566,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.9679398301357474,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.7692,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.8488103345383574,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.7793,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.9701815624098458,
+      "learning_rate": 2e-05,
+      "loss": 0.7456,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.8521202985399893,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.7566,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.8979050575745416,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.7526,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.9752394739693592,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.7514,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.9629065314553338,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.7681,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 1.1194919436414723,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.7453,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 1.0904012043510785,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.77,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 1.0542119496070057,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.7489,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 1.0926670049306186,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.7661,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.7952443613694119,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.7294,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.9226126008750115,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.7752,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.9420385653518116,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.7425,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.8727062075127135,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.7671,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 1.1606585810413057,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.7486,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.8205444268781921,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.7599,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.9982002393594918,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.7714,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.757916124459772,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.7427,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.8393138948588185,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.721,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.8470749124914952,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.7656,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.9190020324158813,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.7657,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.989018622269614,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.7466,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.9078623162848197,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.7324,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.8048574386886347,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.7446,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.7650092088351168,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.7188,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.9944521057525681,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.7737,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 1.0042955916256968,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.768,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.8279205317703516,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.737,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.0033302340573467,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.7343,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.8105962329092502,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.7589,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.8620154794632287,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.7744,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 1.0049361361989435,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.7555,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.820998895182546,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.736,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.085767612239963,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.7126,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.8519887626225435,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.7431,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.9770204093520869,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.7615,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.890776067879005,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.7676,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.9454357628364443,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.7696,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.254559494730336,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.7509,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.846643461717218,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.7713,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.136355583653842,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.7673,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.0650534120184716,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.7304,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.8312516303374767,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.7379,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.151777395306924,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.7434,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.8491077014225029,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.7405,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.0073813977486343,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.7453,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.9205339004518173,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.763,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.8233864076172686,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.7218,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.8903624003528929,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.7564,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.9506675655752435,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.7403,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.8350896695621072,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.7012,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 1.10053085379262,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.7404,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.9948518863476845,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.7555,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.9096254812842482,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.7612,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.8466526744109061,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.7432,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.1122314328097833,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.7536,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.8898628370902996,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.7504,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.9665970881698415,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.7267,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 4.521263740732135,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.7371,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.51334018080246,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.7495,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.8305252177250677,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.7237,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.3787250499579446,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.7366,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.0704184656420916,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.7646,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.091772175517905,
+      "learning_rate": 4e-05,
+      "loss": 0.7488,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.9966851038278123,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.7492,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.804559014365656,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.7334,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 1.0647955371323579,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.7419,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.7041453866734583,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.7323,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.0808462451486691,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.7341,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.8346921663103315,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.7359,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.9871109910571816,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.7246,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.9217196113155944,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.7279,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.0792086104263792,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.7479,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.8773304557428783,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.721,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.1233774714116695,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.7273,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.9521303058675346,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.7184,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.8167018363758924,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.7499,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.8677017305323712,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.755,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.6578197791820202,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.69,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.9404010509286478,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.7541,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.7674518162763317,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.7524,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.7760327965119508,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.7387,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.9465568395510758,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.7459,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.8302297151895294,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.7358,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.9134188658874217,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.7029,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.9563311027616868,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.7488,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.8740932905889962,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.751,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.8169439997059696,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.766,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.801933079253408,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.7323,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.8136271205189634,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.7727,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 1.1542172895006149,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.7412,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.8373261487954003,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.7454,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 1.0537328486839306,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.742,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.8585957159033233,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.7159,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.9188286274443758,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.7206,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.8715311863602735,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.7394,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.8701222215940531,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.7118,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.8777495481346238,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.7371,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.8989486103648016,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.7526,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.8347979818867436,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.757,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.9545399014004454,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.7881,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.8092721018923198,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.7024,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.9021640721752451,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.7196,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.7994342222592613,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.7641,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.923044882014875,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.7081,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.748481925699889,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.7495,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.8731156643356355,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.7635,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.7501331557933156,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.7268,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.7365188838012106,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.7386,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.6815097202486469,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.7445,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6944886438531291,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.7187,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.7282747402859536,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.7333,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.6558739961496217,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.7397,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.7850419293577147,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.7455,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.7359816056868546,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.7232,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.7398496758490406,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.7396,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.6668274085705151,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.7207,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.6962069114775007,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.7042,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.7361365700034131,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.724,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.684327079473779,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.7408,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.7280600931261505,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.7318,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.6375424942770173,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.7469,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.6996677581966146,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.7163,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.6733774274718524,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.7276,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.7253073145283652,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.7396,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.689144539706276,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.7088,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.6975328711377113,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.7163,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.7225318017525385,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.7372,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.5906263915112032,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.6834,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.7942390003133437,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.7291,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.727917892144924,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.7085,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.6947964183405534,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.7263,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.6707403193544363,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.7105,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.731800619423467,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.7205,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.7691615980232677,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.7397,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6756055333102378,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.7077,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.6895627071893281,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.7618,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.755479529822667,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.7263,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.6975907124905093,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.7394,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.7579550646190952,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.6979,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.6108939515327224,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.7138,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.7829301319326165,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.7026,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.6819566169772593,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.7304,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.7842196180076135,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.7388,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.7150555992783668,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.7388,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.7287402574907651,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.7116,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.8480167537080249,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.752,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.6587570767598163,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.7085,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.8162187280641727,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.7362,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.7167585579551728,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.7183,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.7303453267442132,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.709,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.7658307565503085,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.7087,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.604638768493775,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.6937,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.7586505537480442,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.7355,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.6894078496063702,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.754,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.6966270031577029,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.7335,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.6020172631437388,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.7204,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.702148502363764,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.7249,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.6205202481966265,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.7369,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.6805728235496064,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.6874,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5750671294434272,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.7113,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.6452002918441831,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.7118,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.6266162112957526,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.704,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.6218633685694466,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.7157,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.6983437330437375,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.7406,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.5972454925945231,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.7191,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.647250363343804,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.6985,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.6341300257653193,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.7356,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.6492626706396889,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.7204,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.6306880550219593,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.7229,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.6370960873425177,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.6889,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.6587509402756125,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.7267,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.6697308593769057,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.7246,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.6545624188417978,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.7296,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.6975725539692176,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.7147,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.6524961005505715,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.7342,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.6490338389860988,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.7734,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.7055159801630626,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.7305,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.6451434316629133,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.6986,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.6324686040438535,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.7071,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.6545326758903276,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.7141,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.7924889707574986,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.6923,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.6556930522149884,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.7136,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.7471678682613927,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.7181,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.667033761981706,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.7117,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.6465000308168121,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.7291,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5979998228585848,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.7068,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.7918806698007783,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5412,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.645839275390746,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5268,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.7475116244201249,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5306,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.8560203430842198,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5447,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.7419067741358536,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5459,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.8535741184990855,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5278,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.7030974081255914,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5518,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.6899216077837134,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5068,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.704930464094611,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5265,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.7638741781265855,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5465,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.6836191156526904,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5486,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.6594794600111621,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5443,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.7057489690653247,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5471,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.7209952920963361,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5453,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.7717548025798928,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5293,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.6487163421376152,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.534,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.7083663709883242,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5509,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.6433086381016127,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5562,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.7358545968389828,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.525,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.7083722477982212,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5211,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.8301583467130484,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5265,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.729813562782329,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5033,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.752200069229649,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5501,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.7491843649848936,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5177,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.6834415836379277,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5334,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.8617374241115866,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5569,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.6624226049447375,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5375,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.7536829400395826,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.522,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.6541731353958475,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.533,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.7366287334337182,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5291,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.6210478638147476,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5305,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.6721453238832759,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5492,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.6143018746969647,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5369,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.6590949369458804,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5596,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.6596485121414294,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5426,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.6368927983022917,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.547,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.6794872970703435,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5491,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.7207386066170471,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5457,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.5793934879441103,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5131,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.7871727952411967,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5579,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.670229355968719,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5424,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.6997968312916134,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.563,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.6646858171111037,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.537,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.7247560450158957,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5268,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.6803512582658766,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5473,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.691104912889742,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.543,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5971538144241855,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5183,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.688919303057941,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5012,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.729623679194009,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5617,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.718098658016614,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5543,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.7324660621322207,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.4951,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.6559616214343055,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5253,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.7143569148242955,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5537,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.6116229639726337,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.526,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.6192788655357062,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5432,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.5870901853982465,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5423,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.6355427875394826,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5623,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.6010685340972887,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5506,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.6105342982647867,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5527,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.6601026567825459,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5338,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.5876353233939494,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5518,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.6178035659258662,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5503,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.6067045969172247,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.54,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.5970482525467229,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.557,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.6168495786518905,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5295,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.593043129906852,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5414,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.6708317288228063,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5346,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.6216137277537424,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5481,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.6139562793273009,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5591,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.6886057705171386,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.533,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.6452678654563264,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5241,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.7416525678274916,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5482,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.6712654607000791,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5421,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.7041065626866309,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.533,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.6699445936161539,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5362,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.7198819186462806,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5347,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5882485756232412,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5398,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.7333170799028271,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5076,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.625179795147293,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5381,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.6378058563993624,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.5446,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.6298135575613442,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5358,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.6144008882434406,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.539,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.6770679872358158,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5499,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.6453408259358712,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5396,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.6797875277716332,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5414,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.5914568275183317,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5411,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.7472575378926508,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5551,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.6140369477641603,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5289,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.6096186287300118,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5376,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.7668642543873226,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5337,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.626485779968028,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5686,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.7051197130315752,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5068,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.6501128573734031,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5206,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.607287860558443,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5189,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.6913712686003229,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5245,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.6275572033098413,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5384,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.6525606185111263,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.575,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.6416282662815204,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5268,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.6910353756685812,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5417,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.6880055929006236,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5262,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.5968587104135665,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5268,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.7929662986431356,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5358,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.6061493417240312,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5484,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.7208919666811039,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5444,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.59520219114923,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5334,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.6340274125819809,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5286,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.6502370228472837,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5348,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.6706983768793204,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5417,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.6203715032190583,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5349,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.6387190094327716,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5434,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.624572974087524,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5547,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.5673921804341926,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5344,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.6546423890720546,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5444,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.585456877828548,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5735,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.5346874619905109,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5428,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.6083675315287471,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5608,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.5593975772015052,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5095,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.5861574488388085,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5236,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.5614172172917737,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.535,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.6716322531952549,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.558,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.5752826797645595,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5413,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.5981573143560327,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5343,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.630996072980425,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.538,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.6206347871499921,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5616,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.5817826744658627,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5674,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.6070834708059147,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5354,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.5435100050131151,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5278,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.620001200837295,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5514,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.5773108907246014,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.547,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.6468647228440233,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5391,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.6222593929436858,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5611,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.5830159571390442,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5335,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.6789153011316809,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5498,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.5475314507255191,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5218,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.6505239501857234,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5521,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.5926874783717488,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5458,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.633837598797931,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5396,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.6808099613883282,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5615,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.5754816932266656,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5361,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.6222561626910301,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5502,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.6317656746352714,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5207,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.5721588495925056,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5651,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.5374242801899645,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5397,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.6057063826076984,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5314,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.5487621035587309,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5514,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.5521862263176056,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5528,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.5705950448905123,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5413,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.5882027623857716,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5334,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.6128022691301136,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5548,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.5814891018172945,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5597,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.6570193342040848,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5431,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.6282600488547089,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5368,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.6699516254514705,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5365,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.6172632411604175,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5326,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.6151918700964317,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.576,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.6203476736745196,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5277,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.6014763422314395,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5321,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.6341415604711222,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5542,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.5803239470764502,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5222,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.640358293690612,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.545,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.5904888706381538,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.535,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.5298073949432225,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5287,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.6275177686744439,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5567,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.5924752473064901,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5445,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.5759341155719487,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5341,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.6367959173887359,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5479,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.6275355459890076,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5571,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.6570109028101975,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5523,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.5708420419002955,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.5553,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.5851427625856627,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5341,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.5818191505588025,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5526,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.6173282290512913,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.537,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.588843111982373,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5538,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.5796385472721386,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.52,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.6072385560727243,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5313,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.5489524116118614,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5503,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.5985197749618895,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5185,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.6183523797454198,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5376,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.589359863100689,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5555,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.6280410379127301,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.538,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.6184619399271417,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5656,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.6168762748241193,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5401,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.613606462362541,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5487,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.5658380122604596,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5392,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.6207896189925389,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.538,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.5833712811784336,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5464,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.5254610503525575,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5425,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.5864151148960993,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5243,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.5731156299501053,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5088,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.531748035784143,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5296,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.5771970324844771,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5274,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.5833559341396615,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.542,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.5599897062637607,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5287,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.599476140308303,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5143,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.5798436898550795,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5487,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.585349850836082,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5413,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.5907333337984962,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5107,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.5702038810279063,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.544,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.5829463417170855,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5437,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.5567704543503441,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5417,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.5429793422191448,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.557,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.5559676868846265,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5204,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.5623367771011216,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5441,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.5656635312805552,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5535,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.5922400173729401,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5476,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.6617000708584949,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5127,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.5371541130180763,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5492,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.7193618618605213,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5322,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.5330481233399853,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5622,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.6608288009755023,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5136,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.7107230373305515,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5489,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.5768603034644065,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5319,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.6433147318355988,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5386,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.5646639097916635,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5338,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.5862792442190161,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.519,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.5702943662320945,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5588,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.5649025643201037,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5352,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.5923784275953722,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.563,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.5680123630409147,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5352,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.5406247779879336,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5431,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.5949549583880022,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5301,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.5460248814422107,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5465,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.6039297907783509,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5416,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.5918467760708311,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5384,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.5611520715265647,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.5388,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.5988009150916238,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5385,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.5996417735733256,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.569,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.5722382149191179,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5403,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.5626435192165211,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5391,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.5579459844937822,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5238,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.6219358436112798,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5904,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.5838330774472333,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5322,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.5888407322025457,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5357,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.6018112387709398,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5336,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.5943325875918147,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.553,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.615803256907434,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.539,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.6001867690364487,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5433,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.550852967727887,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5244,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.6258382169369585,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.5585,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.6308830496788315,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5125,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.6453263565623456,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5287,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.588277214257248,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.551,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.7286044483846403,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.542,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.6498406491837265,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.548,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.6059589376360205,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5563,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.5875786800628212,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5192,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6210357247914922,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.541,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.9561376460183231,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.3573,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.6251233596010198,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3493,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 1.1775548424544031,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3457,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.6755629154473264,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3324,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.6749669496315537,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3169,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.6058439758395823,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.3222,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.6839015998307398,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3249,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.5536924531680131,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3643,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.6623445837437189,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3327,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.578909646153525,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3207,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.6313439706538881,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3366,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.5878024665353784,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3294,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.5385270007409896,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3368,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.612440797228422,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3652,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.5596638420760371,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3145,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.5182112387997669,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3133,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.5938777082622732,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3273,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.5351777102627749,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3172,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.5567094796011324,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.336,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.5216325313717205,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3307,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.5187768670944479,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3313,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.5248533479434337,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3411,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.4961345465925558,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3426,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.5593835542083903,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3455,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.5185134599579961,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3339,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.5293912183037898,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.3316,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.5770150569758211,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3517,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.5259566339914634,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3276,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.5019520397414248,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.333,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.5132816106268442,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3155,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.5289369361506726,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3546,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.5281118489553415,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3405,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.5206274102619982,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3166,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.5079435906580224,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.3135,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.5040165898557063,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3263,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.5227566229978546,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3369,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.5211772422304101,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3077,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.5099540730196319,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.338,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.5284446081268709,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.336,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.5176741268867192,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3498,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.5439841308983797,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3292,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.5227679867917473,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.3463,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.49366837172978373,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3306,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.5040311275693086,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3175,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.5112599990825986,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3079,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.48512058337452396,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3518,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.5008919720767387,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3295,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.4980304148571713,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3111,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.5197823751808713,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3101,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.5459204523194214,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3221,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.502353771807027,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3298,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.5134152211881277,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3235,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.4929381179437613,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3248,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.49822472967232245,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3315,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.5400412406768654,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3516,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.5278632776606373,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3668,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.5561721908541211,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.3197,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.518816142687651,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3256,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.5396073854045892,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3225,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.5823509937369001,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3322,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.5021593033586006,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3599,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.5602786897811859,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3418,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.4886477965844604,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3383,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.5227906660977483,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3171,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.5478924325672999,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3489,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.5197944942536591,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.31,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.5318424022526256,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3189,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.5055013377604511,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3488,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.4909088684693141,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3319,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.5251337550808545,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3365,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.520644295175242,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.3496,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.49632330909144634,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3195,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.5092208095628089,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3283,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.5091152889601727,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.3451,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.5395206378819085,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.322,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.5094753963405843,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3478,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.5329741123811217,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3268,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.5588879721280001,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3392,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.5254709892700182,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3497,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.5292224363048875,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3383,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.5303114118899298,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3322,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.5101847646399678,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3353,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.5345652690481326,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3201,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.5725050671651839,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3369,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.5615077531433403,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3331,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.620742806486385,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3292,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.5309899064721884,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3269,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.5879518054963494,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3428,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.501236954798775,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.3266,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.5595858250493975,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3195,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.5556759587614016,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3295,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.5132959547291918,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3139,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.5175689176082969,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3267,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.5605331356245039,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3509,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.4899759085366955,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3394,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.5371772878274168,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3462,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.5485973866778019,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3199,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.49468370224744423,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3376,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.5101967442450068,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3467,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.5286599232759308,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3306,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.5543939580447448,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3249,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.5273169642395282,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3439,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.5364553756700644,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3436,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.5458692504497876,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3393,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.537004739363394,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3455,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.5048138826151596,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3298,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.5544971947493254,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.304,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.5299594886576667,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.2977,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.5281626380509807,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3343,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.523099951401833,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3269,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.513209574545583,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3497,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.5031972957607549,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3321,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.5416147585674184,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3322,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.4958827602046742,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3317,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.4819965336834454,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3487,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.5266985842727613,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3442,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.5082975677433722,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3297,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.5161478132445707,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3228,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.5092159535656366,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3131,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.5008229977818367,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.3154,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.5067637242840125,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3255,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.5001137789574943,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3532,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.49227965841851223,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.333,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.5496993809427412,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3582,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.5391343759308996,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3517,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.5235992803517996,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.337,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.5235059983184671,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.3198,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.5147401474149721,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.332,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.48749541216225756,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.336,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.4970282656529527,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.3475,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.5249311953698331,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.3218,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.4850838427586397,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3337,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.5368001376196136,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3311,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.5319275398677763,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.328,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.5236222963219404,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3137,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.5237149776122466,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3468,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.4923971068113294,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3327,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.49699578267631433,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3197,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.4824021864653325,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3473,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.5115443785419962,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3399,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.4804154571888371,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3393,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.4898004492986149,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3241,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.5186784271917372,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3273,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.47596581251406206,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.334,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.4903958391649802,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.318,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.5036646811015992,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3396,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.48075872736134373,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3275,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.5285652872265865,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3083,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.5135754067962908,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.3266,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.5246590165434307,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3229,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.5241078845773992,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3053,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.5159856149019909,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3443,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.4943985129322829,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3495,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.5428222905342281,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3364,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.5109965969707837,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.3442,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.5356678527750582,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3217,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.4907151314205694,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3402,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.5408609965562604,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3473,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.5390384898021859,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3325,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.48648498303322485,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3345,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.5337391586003477,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3499,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.524130575531712,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.3306,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.4602360751514776,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3401,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.4943584310321846,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.348,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.48115191386940753,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3488,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.49576867336339064,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3365,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.4822065022076074,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3566,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.5019027277572762,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3483,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.48936783948940277,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3405,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.4861411240450023,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.326,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.5191817671506175,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3296,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.5154708117389282,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3352,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.4966448878433611,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3216,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.4984713152010231,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3312,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.48039919863103636,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3172,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.5020034320663027,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3361,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.5299409399916055,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3392,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.4928593915477008,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3337,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.5134897557559529,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3421,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.4759343010224339,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.344,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.4876588486823063,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3267,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.5155279890881398,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3267,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.4671270956845811,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.33,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.5003311934822238,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3289,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.5284564123839222,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3302,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.4825876135469532,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3342,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.512075081077978,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.3159,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.4849875332355992,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.3281,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.5203649916625274,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3344,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.5435921252301606,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3317,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.45974662289741375,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3423,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.5088960555483448,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3279,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.5226143059238112,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3456,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.46102103225606955,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3494,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.5258134572745363,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3163,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.48277489999491785,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3369,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.4974404940682969,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3364,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.4756347500844287,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3582,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.4626383310738306,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3275,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.48113356787762557,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3084,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.4491601162308387,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3408,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.476127019328394,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3351,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.4790381698761587,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.3606,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.4615640433600368,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3616,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.4915518394987355,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3273,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.4884623610085759,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3038,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.4795417702282911,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3236,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.4949678358030222,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.33,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.5012964382854123,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3293,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.4807582475386661,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3396,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.5107152445374297,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3339,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.5050618521112045,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3162,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.4745220692630866,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3429,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.5203929902967701,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.2985,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.47670960800192846,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3184,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.49725547021772537,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3306,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.5014647239002694,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3251,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.5009095643416256,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3399,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.49365430549247674,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3177,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.5548005896434409,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.334,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.4848410671359143,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3467,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.49053890804449585,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3286,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.511889645775235,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3446,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.4831283407540862,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3569,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.48245000483704636,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.318,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.4920599793867877,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3206,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.4689769500496085,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3297,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.5048793285469493,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.3191,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.4864997220867531,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3428,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.4684524513455118,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3501,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.436534900316035,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3368,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.47811089333202833,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3291,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.4817014438864516,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3164,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.4778055331766441,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3557,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.4481982680553162,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3542,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.45031308363644657,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3462,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.46551186813486545,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3335,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.4869387932728957,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3124,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.4520215184454977,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.3378,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.44639311738174037,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.3168,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.4730754434234949,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3432,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.46073723478445183,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3315,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.4590908231775504,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3369,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.4886358659458799,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.332,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.4582575466396626,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3029,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.46511372313505556,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.318,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.44023593606305456,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3174,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.5974434537479153,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.1772,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.45276777699053844,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.1861,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.4197849742672843,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2207,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.6079291605108523,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2207,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.6228117262585126,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.203,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4565517256089258,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2242,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.560481959329884,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.1863,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.5493476807727913,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.1981,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.45986577622945174,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2099,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.4639176680646764,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.1851,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.45884545039064006,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.1951,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.47043214582699266,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.1976,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.44950662898996846,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.227,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.41327537082025895,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.197,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.434480771656148,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.1889,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.4047815207550183,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.1893,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.4030471476900367,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.1911,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.4376984919102052,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.1857,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.40461773604390366,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.1882,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.38734917145893966,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2008,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.4107024967409707,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.1977,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.4099809091276119,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.191,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.40118315259067516,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.1862,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.37585104633862043,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.1642,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.41532156832472944,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.1772,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.43321264712986957,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.1875,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.40213255836255707,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2149,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.40292226815299714,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2037,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.4021456932425278,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.1975,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.396152665032875,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.1944,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.4003289163669197,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.1975,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.3950253272807709,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.1973,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.40784054866654995,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.19,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.40698099035882945,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2046,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.3865165164325973,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2155,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.4163083311662694,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2081,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.41027452103452317,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.39890285751160337,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2208,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.36739960545889033,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2155,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.3939694428921529,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.1896,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.40332311382498803,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2239,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.3843861119587302,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.1989,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.41223277098575517,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.1966,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.38538729751764045,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.1744,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.37689070514294987,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.1863,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.39893681292779426,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.193,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.40674200642320746,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2026,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.37743165802459283,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.1946,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.3836583470859372,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.1846,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.3896169067943886,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.224,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.3790738322986921,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.1869,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.41009326265224394,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.18,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.3868308964169996,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2275,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.38319473297755957,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.1866,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.4092853729685683,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.1744,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.412738718615427,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2037,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.3936366310271036,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.192,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.39865845675766987,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2056,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.4215051253540589,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.1974,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.39724138186637475,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2001,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.3896211909934466,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.173,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.37712581089123276,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.186,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.403079915835177,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.1893,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.37651077870564814,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.1798,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.40609299621508704,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.1923,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.3916075757590921,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.1775,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.360940453425393,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2111,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.3773491545618289,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2006,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.3907125253028253,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.1804,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.3838086434899943,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.1767,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.37905908530379734,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.1877,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.3987195695545708,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.1904,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.37589419908096006,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2068,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.4002074416043618,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.1984,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.38408382218263865,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.1736,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.3942198463279369,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.1913,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.38410834970140584,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.196,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.38260872339744184,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.1938,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.4060821957550745,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.1915,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.3835739684356381,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2148,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.3782595685652323,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2008,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.39755859731706994,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2001,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.37891202118772754,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.1831,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.3880119339468308,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.1675,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.3827510095461639,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.1958,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.3992949791647187,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.1827,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.3730771615854695,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.1946,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.39857387073298195,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.1958,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.38182189957239315,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.1948,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.3716885103301511,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.175,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.38323332072015637,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.1876,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.3682870010317586,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.1858,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3717864238292802,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2056,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.3628717230714638,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2113,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.38321256293466577,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.1978,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.373590415717196,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.1934,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.38575777395132754,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.1794,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.37589729968714714,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.1868,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.36766241594188176,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2038,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.3843495984071599,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.1876,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.3623728160981848,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.1736,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.4012266401855185,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.1797,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.3830947926937361,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.1775,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.3785941151575758,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.1907,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.3819968688836211,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2077,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.3658835200557448,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2073,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.388088162357279,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.1977,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.38816224709650654,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.1937,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.3841241588215632,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.1794,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.3801781752761419,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2024,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.3640582098593209,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.1958,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.3736063060162092,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.1919,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.3971293478317797,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.1976,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.38687310726493407,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.1765,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.3883996285256185,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.1933,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.3745119439860919,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.1645,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.37127845398682197,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.1829,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.39780023854657964,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.1907,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.3819978880554131,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.173,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.39575213836371204,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.1732,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.4086561624545444,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2074,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.3818649387469906,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.1671,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.3751856625620452,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.1875,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.3992616595953391,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2091,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.36989297388846276,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.1812,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.36969030467185404,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2275,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.3654882887965206,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.1983,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.38035691387093706,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.1872,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.3761084766561415,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2008,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.36186218410151816,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.1737,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.3826627575447781,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.1801,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.38089009581056144,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.1787,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.36594046751611065,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.185,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.3721341253398479,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2032,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.37511751380541714,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.1755,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.37579934668742737,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.1864,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.3775052238935721,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2001,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.3662379840380862,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.1979,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.37295535061142326,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.1864,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.3709712699527873,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.1742,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.3764647430933775,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.1942,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.38073707974283044,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.1838,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.36752364852930314,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.1795,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.37288842106098413,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2016,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.35687361442607934,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.1656,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.36764082507186113,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.1857,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.3704884723137268,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.1831,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.3844867308614277,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.1786,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.38096706605157155,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.1836,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.36958754149832207,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.1949,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.38386455779058115,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.1773,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.37630635461242884,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.198,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.3944185614372966,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2024,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.37054182707539807,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.1871,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.3659553219422058,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.1765,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.35723884112509324,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.201,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.3850464735228209,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.1968,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.3773316159845961,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.1724,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.3678860738188148,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.1868,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.37382927274073857,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.197,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.38582938106052916,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.1675,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.36930280116694425,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.1719,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.3685438474248349,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.1569,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.3776421156066127,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2053,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.37757081765315736,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2076,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.3680243071482207,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2178,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.3703029569012227,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2033,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.35120114123727525,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.184,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.36377437536447677,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2046,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.3668602153188336,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.1778,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.3948247759132509,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.1905,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.3881195504408917,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2077,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.35375618589505115,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.185,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.35695130246809564,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2146,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.3916464550682213,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.1635,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.38149847248634133,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.1906,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.3762598966739746,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.1763,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.362658151221331,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.1775,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.36822682867425416,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2034,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.358662663413512,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.1839,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.3588478861899968,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.1828,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.379799497913373,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.1914,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.37918977203789206,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2155,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.3649806878655295,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.1941,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.3607029576812349,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.1946,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.36595897156567875,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.1855,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.3691503805782982,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.1869,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.36898604718266587,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2074,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.3746953487810768,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.1892,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.379332267265396,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.1855,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.34827561315601735,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.1784,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.3718339005647526,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.1712,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.37815555370122933,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.1889,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.36288264716922464,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.199,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.3669937698310996,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.1964,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.3658956574042133,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.186,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.3699764142513695,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.1808,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.3643245482250208,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.1793,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.37187056763163767,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.1744,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.37725114754257477,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.1808,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.35083042453523033,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.1909,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.3632387394843996,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.1718,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.3640124167103497,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.1891,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.3626842681258393,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.1923,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.38257359738242097,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.1805,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.3689473637558465,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.1839,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.36839001891250894,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.1759,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.35005784095346765,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.1714,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.3563813705867102,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.1945,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.387514725081635,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.1806,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.3701368439342887,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.1976,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.35142361510322817,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.1757,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.3797687121278205,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.1878,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.359765193526977,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2004,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.3629297620684388,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.1856,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.36668841178924844,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.1763,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.3479644983117742,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.1952,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.3581358593852166,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.187,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.35222624331979785,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.1609,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.37211427754915466,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.1943,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.34510626523682136,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2136,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.36730568939693353,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.1727,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.36612402102307606,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.1809,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.36118769815333,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2017,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.3572025412488459,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.36064742089306256,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.1907,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.3583937921270184,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.1841,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.3434792182545754,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.1685,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.3463672346431188,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.1794,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.36363035414025136,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.191,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.36631129729207007,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.1818,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.359639737954924,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2073,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.3489308561509024,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.1836,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.3565092302876537,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2009,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.3605916820227822,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.1669,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.3615816557668373,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.1766,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.3457107092137401,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2195,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.36261718236777096,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.1782,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.359435135615048,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.1825,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.38455280642819173,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.1819,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.37830854955103316,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2008,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.3632669354388954,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2155,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.3639350930297768,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.1762,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.35773934168435173,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2092,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.35660423641609434,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.1871,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.36561856124741454,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.1859,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.33814817705408007,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.1848,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.4164305692711213,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1262,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.38094466689855794,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1364,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.33020636115225593,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1324,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.27743721996269266,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1388,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.261574509018901,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1278,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.3130811834064452,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1347,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3244487200498987,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1413,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.31471793657895614,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1225,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.3729962205784512,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1534,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.32426714078412705,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1127,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.31315535373538705,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1315,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.3133284467824177,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1272,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.2950978575946418,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1392,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2960122760385938,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.132,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.30950246796345443,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1198,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.2944833173608919,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1333,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2868645961146282,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1236,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.2766708201352724,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1397,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.275073086472039,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1295,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.3010784663851177,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.1289,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.28858773238606267,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1539,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.26326211679258393,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1243,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.287675674720111,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1516,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.28131807600939757,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.1538,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.2550509119271561,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1123,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.26357372909734744,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1236,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.2637499217959702,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1106,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.2793785997070372,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.1272,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.2818967009186503,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1443,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.2682547153487216,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1216,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.25957907400852,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1206,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.26981361504866574,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1314,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.25662659271294325,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.124,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.2783609418222511,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.119,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.26691203755747744,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1178,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.27485508553793825,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1229,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.26077045860680537,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1307,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.27231583586845665,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1352,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2650627889376335,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.1554,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.2606514261596702,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1333,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.26046818555300244,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1313,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.2704294341448038,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1177,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.2619154552381053,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1439,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.24873950305967094,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.1389,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.2521844150241456,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1109,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.27566504803120384,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1441,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.2684171659976439,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1302,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.2667899917873379,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.1333,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.2729350623131414,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1219,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.263128117330335,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1277,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.27345086060566665,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1266,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.2536440846411039,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.111,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.25500476494954954,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1249,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.2680769768588839,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1063,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.26733998600349546,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1208,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.28466625950906416,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1223,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.2679818847109725,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1262,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.2713824103777235,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1333,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.2681996898806442,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1318,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.2671473958575293,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1303,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.27078117997955764,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1208,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.2776592596135665,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1208,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.25477345549435765,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.1096,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2673307108883562,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.1262,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.25758588005501365,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1277,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.2582347906645367,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1149,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.2715276672561504,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1335,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2597056831255997,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.132,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.2492765554499706,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.1604,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.2655613194448304,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.127,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.26236797441554555,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.117,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.267058412516969,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1238,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.27479320823700404,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1283,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.27829570571089185,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1502,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.26341996698165193,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.125,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.2633205493803472,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1054,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.26914928828227125,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.117,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.2653611839868599,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1159,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.2588930703777674,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1175,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.25552960284855747,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1447,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.26708647953036624,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1463,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.26780080158204334,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1338,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.2886363323979974,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1319,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.2523796355148789,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.1416,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.26443114358941205,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.0973,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.2597698181122491,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1347,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2623916244348604,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1253,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.25517103361397575,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1491,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.26295822686606407,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1026,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.26557557557878153,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1135,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2691970314916236,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.1204,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.2577946358350458,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.1156,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.2622720891028971,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1462,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.2755863397533489,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1403,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.25905375495036537,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1279,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2727740740373861,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1326,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.26533541048453413,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1128,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.2563928056211843,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1296,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.2682097409032537,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1148,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.2495323489702948,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1116,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.27999488238071824,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1156,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.2539493083772409,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1243,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.254841956392742,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1153,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.26502010023470607,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1274,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.26537360202216637,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.127,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.28943351732067035,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1324,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.2590928234827473,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1145,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.24171679252464748,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1194,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.27037696985659293,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.131,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.25659904037715403,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1345,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.2719748953878259,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1405,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.26285714616291334,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.1219,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.25696037738784483,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1218,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.2634512692429073,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1245,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.25652203237338517,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1232,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.25490348134342683,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1298,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.25398902076692187,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.1235,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.2701106792596946,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1564,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.26515755067410096,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1311,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.26809055983628227,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1116,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.27154413541902206,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1289,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.24826909037984043,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1302,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.25124949789506,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1216,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.26251676879271246,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1422,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.2535488347341743,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1186,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.26304837560545535,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.1187,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.24733630027470774,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1125,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.26907792472723496,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1396,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.26580290444832355,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1467,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.24680230331064257,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1136,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.2503348356280754,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1408,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.2656585064529949,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1127,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.2554392087013873,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1226,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.25891480791367916,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.0999,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.2655770809021952,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.099,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.2745838734245946,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.1274,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.2736636736645475,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1365,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.2611682591053911,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1426,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2498922167541896,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1609,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.277700031777722,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1399,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.27751550301164013,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1494,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.25300672645761796,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1328,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.24863437453369136,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1275,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.27182878618527195,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1378,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.2571416700877065,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1237,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2686258030830597,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1225,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.26925951479369564,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1225,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.26912798369696334,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1278,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.2698471535249445,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1317,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.25943124176062315,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.1286,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.24204926258816617,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1132,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.2537292958317066,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1276,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.24342951623263348,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1323,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.2695457568943355,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1349,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.26159631555188895,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1164,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.2671060349925584,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1255,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.2563010646170726,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1201,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.2659930033944024,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1189,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.26194011815387924,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1133,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2695030636970598,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1035,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.2718560790160448,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1297,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.2664344703749486,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.1164,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.25852935338631017,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.138,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.26034735633311973,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1168,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.2831830414051803,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1353,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.2516479713411473,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1294,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.26115656924873476,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1375,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.2642884324838008,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1216,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.2690736947993892,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.1279,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.2676546445807836,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.135,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.2703958436527685,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.1295,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.2524515556162816,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1618,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.26965472506032406,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1348,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.26467492490097455,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1582,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.24467783775501056,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1172,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.26740226983779675,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1006,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.2721629970736541,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1115,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.279525490099223,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1487,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.29144812365612904,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1176,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.2722674219242803,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1158,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2627337576805755,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.1174,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.2795500533142634,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1234,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.2743195737433811,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1222,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.2628830684842733,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1254,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.2841215208496794,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1368,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.24990666237609419,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1158,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2781576990330721,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.1209,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.2507563950454011,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1155,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.25823798816990995,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1331,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.27571064470389495,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1322,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.2630858264865323,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1271,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.2596058118182548,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1537,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.2688354190379071,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.1299,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.2675503679488366,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1202,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2540947446668114,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1099,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.2490256158207312,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1186,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.26349664820347524,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1518,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.26190043313133166,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1187,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.27245592673552854,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1437,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.2563688003976948,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.0934,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.25859024257724855,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1541,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.25592038699953473,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.1066,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.26097551026899896,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1236,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.25753085503779166,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1088,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.2685799295523616,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.119,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.29049584778307136,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1478,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.26160736415518593,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1484,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.262463356057707,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1384,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.2735801405635864,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1524,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.26253537114522457,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1146,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.2699708658201266,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1341,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.270874686256745,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1409,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.24807135139956463,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1444,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.25932344266299784,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1443,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.2502530907043193,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.1153,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.2489686085590865,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1123,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.250090150172712,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1157,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.277803045535119,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1436,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.2728159212168563,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1395,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.2654117805446476,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1183,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2681227622359158,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1355,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.2755335161137832,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1422,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.250994537380382,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1204,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.27594940993767925,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.0958,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.2520368463393784,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.1285,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.25531708157259997,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1542,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.24775480879336056,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.1306,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.256604835641186,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1021,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.25303940575608663,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1125,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.2652339393916005,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1194,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.25998577954621754,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1086,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.25038613558395156,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1334,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.25794688914037345,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1202,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.27741489230931343,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1535,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.2528516629059354,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.1224,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.2502032764738309,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1222,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.2579914824196696,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1401,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.25932046889209504,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1183,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.27258909791999303,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1499,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.25479043616284336,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.117,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.2607983518839157,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1263,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2568083629736199,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1177,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.2616649267096724,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1167,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.2593892264389287,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.1246,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.2647491368387386,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1245,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.269557086443595,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1276,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2501788702377557,
+      "learning_rate": 0.0,
+      "loss": 0.103,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 1.1719477303695114e+18,
+      "train_loss": 0.39265167106862975,
+      "train_runtime": 15780.8805,
+      "train_samples_per_second": 10.012,
+      "train_steps_per_second": 0.078
+    }
+  ],
+  "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": 1.1719477303695114e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}