diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 6.459665465324552,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0524,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.471427441754463,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0451,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.372776075871749,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0591,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.324941191648752,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0462,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.054142670599006,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.028,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.809413679027237,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0359,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.535710297173685,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 1.0039,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.173756090498895,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9918,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.7817270446850877,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9631,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.613369122927364,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9616,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.537286595373564,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.972,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.6448933037869127,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9341,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 3.9939114842405035,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9493,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.9371649523315053,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9365,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.4446382933776185,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9014,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.732008724137745,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8924,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.5077847592074067,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8734,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.011303386691361,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8527,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.4884428962403327,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8401,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.1210643249878753,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8361,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.0938728596060001,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.813,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.154633449816931,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8265,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.0359515760425966,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7855,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9131428541303851,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.8001,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.9205972763034083,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7889,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.8550913931176193,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7897,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.832741628899,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7736,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.7980307202231869,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.8036,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.760303641695837,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7675,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.8740434518029085,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.756,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.6987352477201874,
+      "learning_rate": 1e-05,
+      "loss": 0.7721,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.6123363890529894,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7562,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.6007552001523746,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7507,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.6223685334469656,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7333,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6294996218079126,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.7342,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.568854818978871,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7403,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6570047588986183,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7484,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6133252559424294,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7246,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.508730838019348,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7366,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6644680578804396,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7173,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.5584935371041069,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.7439,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5426003893937492,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.6975,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.576644427668356,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7153,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.4751436841331954,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.7154,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.4440907289823238,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7203,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.5441092414859566,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.7131,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5465953055854101,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.7225,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.5233359622988442,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.716,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.4917765222875724,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.7239,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.5143061530222288,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.701,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6643722743628421,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.7015,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.5764605209070669,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.7106,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.6377819040573797,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.7071,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.5325470319297528,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.711,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6215050447585039,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.7053,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.47014782791341425,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.7095,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.776832788854264,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6986,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.8346690412168706,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.7113,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.5420747438903296,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.7018,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.7105042034015364,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.7035,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 1.0194548836762647,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6873,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 1.061213162836727,
+      "learning_rate": 2e-05,
+      "loss": 0.6745,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.500265008299495,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6909,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.7579669772706773,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6779,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 1.1059480917747941,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6757,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.6718267484148415,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6749,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.6889065317688966,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6744,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 1.0906908935970876,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6786,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.8865053318753872,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6756,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.6772160169670091,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.6893,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.7060292180866043,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6817,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.6498321417689452,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6958,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.8231190398603231,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6861,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 1.3365150035193878,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6862,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.6985067828000411,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.675,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.6461752118181774,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6858,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 1.169822844602762,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.6778,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.1600759918236914,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6694,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.5688958126336329,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6848,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.7029589620768897,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6896,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 1.6116095739273482,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6859,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.501600282161531,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6593,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 1.382319671005502,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6616,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.170249545329363,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6752,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.759201679822918,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.6543,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 1.2792289325295112,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6691,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 1.1257870996947978,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6947,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.9213241795627908,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6586,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.346602812342591,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6597,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.7321996116839679,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6632,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.3445209458826555,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6657,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.8369341930715465,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6814,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.198582707849498,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6815,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.9922170354117397,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6579,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.1345430123344087,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6703,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.0556796298193054,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6696,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.521176008910272,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6775,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 1.0244458279686801,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6788,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.438274100130431,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6923,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.3375067485772074,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6706,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.269275553963895,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6708,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.526826586152908,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6645,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.0809558635141259,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6814,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.5298495378300774,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6735,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.0601493552301249,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6742,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.4931318995588678,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.6816,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.3117864661360066,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6777,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.0967720408750508,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.672,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.4211387701644143,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6689,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.0013546260050405,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6574,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.3967393513495623,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6652,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 1.0532583363466255,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6718,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.330156396894285,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6796,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.1202734869814452,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.648,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.2459260532304832,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6659,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.082198287576687,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6567,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.959222823922708,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6718,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 1.0524783096148418,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6582,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.0811277895222442,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6602,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.052040730153533,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6614,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.9677108822436438,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6708,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.2502569356446345,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6808,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.0466701599241446,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6728,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.902807533074889,
+      "learning_rate": 4e-05,
+      "loss": 0.6702,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.5602918371246075,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6705,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.2566669223357327,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6583,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.8234944826362134,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.6623,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.953266891036256,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6519,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.0419130443992124,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.6689,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 2.7076356866693208,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6778,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 2.602029595250716,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6817,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.2548396012962875,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.6679,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.5873687869593935,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6667,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.2742208211259172,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6671,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.245616407127005,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.647,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 1.1130606257051756,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6588,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 1.1111392975674002,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6466,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.981115348274201,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.6535,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 1.0145201428015687,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.6605,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.9975480541658427,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6694,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.8520471000050983,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6577,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.8355964904259912,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.6486,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.7414893013945396,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6475,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.9114636575560996,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.644,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.7946473888402759,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.6536,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.9153110643551885,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6556,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.9459624672202098,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6582,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 1.2229154969228304,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6468,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.8517056667193024,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6546,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.7839190387363419,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.646,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.9357224415834496,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6532,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 1.023889629658584,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6647,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.7841526917970647,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6355,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.524326894247506,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6571,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.6265676970698151,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6439,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.9305092412114612,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.6564,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.9861461646258585,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.642,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.7806180922857873,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.645,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.5595965812560422,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.6565,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.7130098519610254,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6487,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 1.0410924147659828,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6547,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.7136806537365763,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6446,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.5821100271252931,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6405,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.5560372991556043,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6312,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.5286784993832532,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.6462,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6313860923805222,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.6321,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.5399168795646403,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6478,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.5124549709616008,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6319,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.5209874937572953,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6511,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.4496263223800329,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6294,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5442265048685908,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.6355,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.49969222489796417,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.6505,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.5404878548313056,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6499,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.5315284235665768,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.6452,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.6062905315775704,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6252,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.6181757405784966,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.6523,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.639419488417689,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.6566,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.6337139955416552,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.64,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.4937785744339398,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6354,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.5320103872036227,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.6402,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.5228852144190683,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.6434,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5179765534196565,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6365,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.48866894128533905,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.6481,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.4461556398310811,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.642,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.6351686950303617,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.6366,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.6775358803798571,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.6401,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.7027187155346797,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.639,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.4469551228940771,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.6464,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.704006865467612,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.6451,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.798292809949734,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.6491,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.7577097819802122,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.6422,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.6176965682967446,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.6372,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5783822570104997,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6456,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.6203720399044865,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.6364,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.6040012134859962,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6323,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6933393609797239,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.6188,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.6757434065915104,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.625,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.700695603944701,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.6452,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.5255619342566007,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.6281,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.7936178120149665,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.6235,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5329637488477694,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.6339,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.47177084744635406,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.6531,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.5217904885928017,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.6414,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.6815591978693452,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.6368,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.5962036518086017,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.6313,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.5122558371160887,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6307,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.5067215835766831,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.6388,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5341835412036363,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6398,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.5336729210642319,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6456,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.6093516455012424,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6502,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.45948460236778704,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.6381,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.4483854371747767,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6437,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5494607332585906,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.6293,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.543565028672638,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.6316,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.47500825540803576,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6394,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.46737885717456135,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.6278,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.4741210676404677,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.6434,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5006588118179628,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.6252,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.4825859197246936,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.6171,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.41698115288070475,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.6357,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.46081830243647853,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.62,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.4822935005599183,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6498,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.5127377034697886,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.6404,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5471221391902577,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.6434,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.42455995092230797,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.6186,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.5182577417319736,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.6188,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.5176801715374499,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.6361,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.47853385682297755,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6183,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.4922888505195442,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.6228,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.5152980751382746,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.6349,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.5769373099794932,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.632,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5946663792365409,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.6237,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.49323185733848857,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.6322,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.4429063454094144,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.6314,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.5521749066711108,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.632,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.6222946968200259,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.6482,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.5691509143879261,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.6235,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.4911000245832107,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.6372,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.546700676051417,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.6214,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.6084507890267942,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.6275,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.5624956583070048,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.6409,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.4758040778233731,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.639,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.5401998848785035,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.6376,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.6316096946959638,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.6216,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.6746642389118835,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.6253,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.5318363100693085,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.6327,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4258866919744573,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.6268,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.47469135414126395,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5824,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.5141528733690494,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5888,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.5939435963835401,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5857,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.7629538483691288,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5872,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.6418802365691467,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5722,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.5333740006359746,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.563,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.45325121679818314,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.6069,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.7473866884901172,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5947,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.5500844930362414,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.571,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.6140979724274733,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5869,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.7636783805047836,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5755,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.5683392414084004,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.57,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.4302509114606557,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.577,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.458458090354974,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5745,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.4616116607651978,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5722,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5982620990558437,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5766,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.7979169091451579,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5824,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.758959009423279,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.572,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.5939876556314869,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5695,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.4727162938233323,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.57,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.6090242797055386,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5941,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5334046325297247,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5676,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.4842845426331719,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5837,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.4473930274169014,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5691,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.4971265961959341,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5746,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.5663430022343765,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5683,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.45044326503472865,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5688,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.46416715489396154,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.581,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.5640033416933319,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.5744,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.554480183033315,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5859,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.618111418348891,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5876,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.5298857605397558,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5647,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.39394552025316054,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5781,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.422682834405,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5846,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.4714724703894466,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.585,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.42498724629560153,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.585,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5339960253855923,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5696,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5231442061474477,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5677,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.41137883618462573,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5735,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.551181444178857,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5694,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.5626598541711041,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5833,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.6343682865861435,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.579,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.44927586868384484,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5818,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5158760918783195,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.57,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.5842082605528931,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5786,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.46236052718052484,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.5795,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.4092345245779309,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5964,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.40979081186747607,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5512,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.45700989126632985,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.581,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.48147963791626147,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5827,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.44657092797902925,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.585,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.40987093909570826,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5732,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.4473318956473719,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5607,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.46614479598383823,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.578,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.38545556567821143,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5815,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.47997992752963314,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5696,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.3932945468169103,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5839,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.4693536076929455,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5709,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.4619569056209716,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5634,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.4798890447806358,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5755,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.5774194498065013,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5672,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.5599887049261802,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5687,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.5089715960296369,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5682,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.4017729942385334,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.5666,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.61803523580949,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5693,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.4874307254373677,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5663,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.5499624564666089,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5721,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.423215287621879,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5725,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.3964984390976774,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5725,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.4887500342433081,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.5652,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.43922738785805127,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5848,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.3991442831548997,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5647,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.5201893849576074,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5615,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.5304432804394799,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.5837,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.4664994160256728,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5769,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.3288959865640498,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5723,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.4629286041168874,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5694,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.35981602974803567,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5813,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.449619842615558,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5778,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.4299074250933994,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.5699,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.34889451653673814,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5767,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.35748890480923656,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5759,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.4153754811591562,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5571,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.4037580816312372,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.569,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.4450032963403848,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5873,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.5257376742760217,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.579,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.4632164750837019,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5624,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.42488196387405314,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5658,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.39653819631767556,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5758,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.41666444398249175,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5806,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.40097751053924235,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5867,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.38268244157642,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5738,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.4900230847086495,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.565,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.42222767206527834,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5718,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.4219009701066636,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.588,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.36321998239812814,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5714,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.4531680311466365,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5607,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.5122908816990849,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5735,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.4350548526133703,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5792,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.4195992767810922,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5708,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.4050521699248782,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5828,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.4430900281808868,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5659,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.41911954769346177,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5907,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.3612436691382918,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5805,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.34859456497891644,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5705,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.4467304800081078,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5625,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.47892117306413956,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.561,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.4905032682073677,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5661,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.5645230163127484,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5753,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.389471046471546,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5769,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.4961793698004379,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5623,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.6356574450649256,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5727,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.5552289375906598,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.58,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.38191020106757184,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5702,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.5252165199139697,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.575,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.5869891741318807,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.576,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.4415030930927111,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5823,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.5535589327356933,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.563,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.5764487538257506,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5749,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.492885289659622,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.587,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.49194666462582326,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5792,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.4141267209192759,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5742,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.5178448330451588,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5556,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.439670475033425,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5824,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.4267574293149541,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5725,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.39364308839423945,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5669,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.38175397529423505,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5772,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.3143213593864884,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5743,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.4187621405543284,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5675,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.43264056589160965,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.575,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.4078863674763763,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5628,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.4054685846962102,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5539,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.4626685447586439,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5745,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.46818647240697325,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5842,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.4843049793453663,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5662,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.4551426840854115,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5797,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.44285242659826435,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5909,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.35910274825289995,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5675,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.3636586214294395,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5709,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.40584609927708837,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5635,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.4611184237026507,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5722,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.3947351480234868,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5633,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.428709025095758,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5787,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.5908843501756174,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5769,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.4946349015829665,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5849,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.4609663942884448,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5765,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.6128809063629149,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5703,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.43386268668223127,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5622,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.47637708406480955,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5785,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.5729276061878527,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5782,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.4689965662345007,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5684,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.4113919441023652,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5696,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.6070392665121848,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5755,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.49591226101629843,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5719,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.34571770318256706,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.5739,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.5769628869907851,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5701,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.4620413568427658,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5776,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.44295735509429063,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5715,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.5021182567173607,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5601,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.38586079418398006,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.572,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.3545021221350271,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5694,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.4222740824175228,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5692,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.4082001306622224,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5531,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.3185812544580993,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5763,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.6959744879999296,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5655,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.6231088133815548,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5696,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.3717245691943316,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5692,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.5146648752108882,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5739,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.39330615338051755,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.5667,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.4112144373247347,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5524,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.48293430008760646,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5854,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.3960169150816137,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.569,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.4584067865320419,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5719,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.4450903830028189,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5749,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.4755164367978987,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5691,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.4710651537552274,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5683,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.43304445186491003,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5638,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.4675599392762509,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5796,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.5762296798340942,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5769,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.5850575160054865,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5748,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.39211100241385066,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5663,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.4526698263905519,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5724,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.48366867703239225,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5708,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.42446910190578896,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5736,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.3657523874581048,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5865,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.4774433199378076,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5543,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.35499695540163306,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5603,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.3735659058001478,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5664,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.4238983432112332,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5691,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.4244807042504971,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5724,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.4511069458751411,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5738,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.40283126526733165,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5669,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.46288005454301706,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5772,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.3355055045193123,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.568,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.4339994783284762,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5582,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.337538539868481,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5687,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.36809976982652454,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5634,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.3816336510277478,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.561,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.3587500122515907,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5652,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.32980983459612845,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5744,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.3963861917501752,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.5801,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.3475712211964662,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5702,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.327128887624614,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5627,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.3244014467217673,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5531,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.32597018964530977,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5625,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.34353583936407295,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5681,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.345506045053071,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5627,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.4192661386278193,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5646,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.43108487169151927,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5758,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.31707633945098873,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5653,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.45163172209655517,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5817,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.3338721602124623,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5642,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 1.415837316262739,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5584,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.3219310392339379,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5653,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.3331470889713053,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.573,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.2977021357912867,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5712,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.339774481690552,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5675,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.3287594517679952,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5708,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.2972411404394,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5664,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.34939476569209194,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5749,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.345609271639419,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5599,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.3672244481452618,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5684,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.3840369963102914,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5669,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.38167470182103364,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5679,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.3684408851143079,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.5679,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.3265366311795297,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5569,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.3393194666037429,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.5803,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.3874967891283752,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5661,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.3628754736362243,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5698,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.3856844755120131,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5842,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.41175791367255005,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5814,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.35843999659200115,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5741,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.31196449478798954,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5627,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.42311542252438167,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5648,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.3276601543372154,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.5706,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.33216494609864916,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5619,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.31970203345371073,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5511,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.36913157687178555,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5762,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.3159099710591718,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.56,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.4159646697313424,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5503,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.3387965024800779,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5606,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.4148466045249964,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.573,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.383063388203945,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5657,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.3730844270382058,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5709,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.4144634394027677,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5651,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.33174696761532335,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5607,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4181345673901539,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.5629,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5282189705460727,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.5001,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.7478232940131694,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.5086,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.6503520867529902,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.5035,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.5180497316839383,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.4953,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.5180425080390696,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.4779,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.5289860365393836,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.4981,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.49507423502793096,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.4966,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.4955871631349802,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.4907,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.445736544003744,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.499,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.3633637539622773,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.4993,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.42968193156266515,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.4975,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.9492211677970467,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.4854,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.37359867204373925,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.485,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.4543741208047001,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.4951,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.4752439575371558,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.4978,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.41595112089471736,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.4898,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.41974855624871277,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.4923,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.49045158417357243,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.4979,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.3949550356873287,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.4955,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.45723316199601577,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.4907,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.3859294968477481,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.4898,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.4006510454521406,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.4884,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.43618155730367175,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.498,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.3407203080583359,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.4873,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.47083060282546696,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.4996,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.3307989243344453,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.4982,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.3910625145568562,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.5005,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.3840881917508381,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.4996,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.3198393760722777,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.4911,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.3887759958113042,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.4762,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.47193601866966217,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.5039,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.4136848441991303,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.4986,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.43063589217229215,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.4919,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.4356356928340353,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.4868,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.37258727831150834,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.4944,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.4353157877377001,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.5071,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.40920578687870285,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.5052,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.44589934932082886,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.5055,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.47548181185585986,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.4922,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.537445492891941,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.4845,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.4430127454938254,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.4995,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.4318349827409834,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.4887,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.4725754532749146,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.493,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.3404773648045616,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.493,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.3963365824871784,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.4892,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.453700947707705,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.5001,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.7700888212399075,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.5013,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.40334556937825616,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.491,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.4143663135941732,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.4924,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.31420887067939984,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.5017,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.44859268208868674,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.4829,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.31702946286915923,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.4902,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.4464914842025124,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.4817,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.40041315250749665,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.4817,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.4045557452526137,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.5032,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.4413079005170389,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.4895,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.40884458217735326,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.4914,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.349516777849422,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.49,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.3432294180091726,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.5131,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.3441807456935691,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.5118,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.32457779870458026,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.4989,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.31646758689226556,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.4838,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.3254607629257324,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.4999,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.31285065608384743,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.5103,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.30588737136014604,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.4944,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.33742622579563797,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.5061,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.31450028500865485,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.4943,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.37744899002091703,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.4823,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.308634713828327,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.5026,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.34562364222699743,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.4966,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.29507301775751044,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.504,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.2993596252763041,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.5021,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.29905602068177856,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.5081,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.3260276218952712,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.5118,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3362739420018949,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.4931,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.30224895283837583,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.4909,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3505237449496007,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4788,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.32726499839041684,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.494,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.36244591738166404,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.5131,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.31381699273849806,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.4918,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.3573719500186489,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.4944,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.30157362603162235,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.501,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.3806960143674884,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.4895,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3210901270854166,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.5008,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.32756315512550493,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.478,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.3401888335058829,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.4842,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.33154670393480845,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.5085,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.33666156527411734,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.4886,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.3682928014196928,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.4861,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.3374043326093659,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.4923,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.3897457029545706,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.4846,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.32026910363458566,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.4971,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.5279530799131006,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.4989,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.34246764240474736,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.4912,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.3887603973229571,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.5,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3010006392561998,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.5077,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.4062661455299412,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.5026,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3582288631387254,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.5105,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3276554853369119,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.493,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.37823149643178583,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.5033,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.31216258171996514,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.5059,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.33452256271267167,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.4919,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.37013569390003187,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.4974,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3278634057117822,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.4995,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3190353755636182,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.5018,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3097989069876717,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.4967,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.36388169170304885,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.4997,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.28686078633770595,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.4918,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.4282955794613573,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.5036,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3140938593492917,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.5099,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.42977494555627077,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.499,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.38014994105222866,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.4995,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.3998378335929432,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.5039,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.42571027995865235,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.4943,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.31447710347492236,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.5071,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.43399540215754906,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.5052,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.30743402127138914,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.5061,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.4211339742132858,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.5023,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.3270294039592575,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.4974,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.44995219433231315,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.4908,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.3822589485237031,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.4977,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.33463411092001577,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.5017,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.387916202703603,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.4806,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.31769300692587793,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.4916,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.38293045741818155,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.4972,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.45516069415611393,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.5131,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.3704692946878139,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.4914,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.32374418291664653,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.4925,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.3202591286619051,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.4959,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.36058090386674085,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.5075,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.2926188455698945,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.4871,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3593972402316347,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.5109,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.3112066558137283,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.5007,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.449842192652982,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.4959,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.30261092147066077,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.4636,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 2.2014567771010265,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.5219,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 4.036697118703728,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.5265,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3755938021554953,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.5086,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.3215781526516941,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.4959,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3788397580729665,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.5123,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.3426219621834552,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.5061,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.40967353596258,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.5023,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.37221761750759397,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.5094,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.36633215406409886,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.5081,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.41439276027939215,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.5016,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.3200054031740638,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.4962,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.40872254056846014,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.5036,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.2971685799643418,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.4963,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.32083013719226017,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.489,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.29480665245951976,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.4996,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3224552283182997,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.4997,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.31649356911070553,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.5107,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.3279239114657478,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.5038,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.3336488114953472,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.5013,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.3448655833798366,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.4876,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.34651832853462955,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.508,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.36212659779688683,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.4937,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.36265411471030684,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.5223,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.396725943610999,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.4953,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.31620477189748714,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.5046,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.43356638956825994,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4863,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.3100392688389492,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.4975,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.37174879823287876,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.5044,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.36875400082828297,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.4978,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.30157223267916244,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.5067,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.35294132839129727,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.4832,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.3461813642529951,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.5241,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.31167213736440347,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.4963,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.4158108244876341,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.4992,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.31266850970402676,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.5013,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.31503210032343704,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.5086,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3280688226496025,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.4934,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.2826461182855877,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.5017,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.2947178976594695,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.4972,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.29788749085756777,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.4939,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.29113225453028574,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.4918,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.26716579962315695,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.4994,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.3158527315236615,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.5008,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.27368385195180833,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.4983,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.30517677188913206,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.497,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.27883719038408217,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.5011,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.29966788999854704,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.4947,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.32205824695834395,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.4941,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3130350693328434,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.4954,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.3173541811333573,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.5101,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.28004778847583034,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.502,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.30741494723671053,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.5046,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.9485034450366308,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.5494,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.2978788379799312,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.509,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.3251787729160983,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.5118,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3156523002952909,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.5082,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.28342949230999603,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.5074,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.301459355476137,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.5068,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.29139113626649477,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.4878,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.30059833171624245,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.4769,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.3032687480695521,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.5126,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.3365250804096517,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.4915,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.2846047016823376,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.491,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.34162936715978387,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.4828,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.25388163601291297,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.5059,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.33284019170796403,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.493,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.2691855040282454,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.4966,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.27659695230775294,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.511,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.28863521302262013,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.5008,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.32438296416500095,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.5021,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.2584483829291308,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.4948,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.30403718862340745,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.4867,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.2903583005565281,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.5039,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.3185961015203256,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.5005,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.262337917673509,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.4954,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.32565142646934025,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.4837,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.27396835466737707,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.4951,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.2672715695889588,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.4851,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.2928497975391583,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.4984,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.2751591230440774,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.4991,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.24954984622825802,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.489,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.2892315822733234,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.5019,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.2760981716963271,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.5187,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.26651903861040666,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.4889,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.28983986415761553,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.4967,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.2733904058104156,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.4909,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.2646969346107782,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.4973,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.36617626441155404,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.4858,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.2849878973922972,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.4971,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.29857699080661104,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.4986,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 1.439999490860653,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.5391,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.31772284600642436,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.5017,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.27564297106999536,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.5037,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3936521540211617,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.5023,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.2714998144269784,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.5016,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.2766180191813574,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.4948,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.26276144165681203,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.4963,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.272894770270464,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.5015,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.25293268792813306,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.499,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3035551251556596,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.4976,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.2691820187644683,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.4928,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.295958728398659,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.5185,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.2650229351876123,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.4928,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.2913982877464376,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.4876,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.26394269883416516,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.4895,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.27560106168784354,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.4952,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.24589088649237006,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.4909,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.3146925909548399,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.4889,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.2740878539396498,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.4902,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.36044616176147953,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.4868,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.2846100263010349,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.5118,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.30655204685751275,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.4988,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.4504113039992735,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.4188,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3914368284653486,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.427,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.5521381766317431,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.4247,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.5240602288178258,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.4458,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.411314141992625,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.4397,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4626858557009983,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.4326,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.3921378521933744,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.427,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.45682889286167716,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.429,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.6011860020316524,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.4358,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.4136935593835205,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.4345,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.4011335929276381,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.4397,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3664884651093178,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.4101,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.39589794689275043,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.4126,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.38784392382996574,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.4116,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3859223805763991,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.4148,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.4255637784035202,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.4073,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.41469826766714063,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.4395,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.9587968413751474,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.4737,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3723991137587226,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.4315,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.3495321518610353,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.4364,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.32817725092514066,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.4197,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.3321196251275955,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.4227,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.3257859889450846,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.4207,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.3069188554798484,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.4317,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.3189315311707027,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.4322,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3057411895507506,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.4337,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.29008900934269477,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.4288,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.3052219557061328,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.4172,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3131277857034955,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.4394,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.2796689242816851,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.4276,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.30078154304278276,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.4388,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.30117894892062835,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.4248,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.2792812587121499,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.4305,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.2981635945121314,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.4495,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.2809843347468263,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.4369,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.3119614180888181,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.4187,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.27648948177887717,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.4311,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.3301477987845189,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.4299,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.3534678275382521,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.4376,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.3060276933049569,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.413,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.403151416948464,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.4405,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.28619218753810965,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.4132,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.39591471498325653,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.4487,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.2779925155412773,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.4263,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.3552055700188044,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.4394,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.26256964161229435,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.4403,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.30732860009976787,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.4156,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.27280978621534707,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.4218,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.2982046259664555,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.4342,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.2747145345142786,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.4158,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.3114103248052139,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.4415,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.28005844883509584,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.436,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.3053144633834214,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.4405,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.26606695948028364,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.4406,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.3009950430282216,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.451,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 2.0458291913167184,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.461,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.3233605375514536,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.4317,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.3180245744491582,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.4328,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.28549286544327873,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.426,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.3061346451814859,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.4339,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.24605090732636745,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.4221,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.2675617794800944,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.415,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.2561093971585491,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.4377,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.2557291388062938,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.4229,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.24591451973215073,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.4292,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.351809850274213,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.4116,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.23876979124530395,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.4305,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.2736043779809072,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.4401,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.25128347081454594,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.433,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.2850741813547461,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.4148,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.2523176478155811,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.4411,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.2800472026717812,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.4294,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.25910656620417266,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.415,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.2439763463662115,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.4232,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.24494641002563056,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.4347,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.25379220230869787,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.4485,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.2347054754933408,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.4297,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.2904049324848912,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.4318,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.24624789563393298,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.4282,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.28604402426539016,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.439,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.22904413342224858,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.4392,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.24212075207821196,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.4373,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.2411800335508837,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.4226,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.2620986208551137,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.4322,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.24922579986464521,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.4284,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.2473947553148224,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.4361,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2512393123193338,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.4492,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.2406911462406621,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.4296,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.24996220758193788,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.4314,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.256609913018724,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.4349,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.2461302102420052,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.4383,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.2737199203004341,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.4481,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.2502368925882613,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.4344,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.24597354905333949,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.4332,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2693677776477121,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.4315,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.23892094204090686,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.4266,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.23999555423620433,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.4286,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.24137293812492705,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.4423,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.2566258174761299,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.4345,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.26517847784421655,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.4252,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.232262662395166,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.4305,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.25980050826773377,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.4419,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.2965960738449794,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.4299,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.2533936637671041,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.4477,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.2959662282420935,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.4231,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.25294786792320356,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.4269,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.24061790035743752,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.4294,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.28453641099298194,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.4282,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.257473480451623,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.4062,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.24322134384301483,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.4368,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.2668820168536047,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.4262,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.2679475171172626,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.4327,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.2523362662151896,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.43,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.2587266178866944,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.4229,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.27694719062502604,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.4157,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.2579054143509735,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.4295,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.25576892250385236,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.4397,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.2887486169569586,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.4301,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.2323816631662225,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.4459,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.259489872064675,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.4298,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.2876464516461242,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.4365,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.24915386810653445,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.4347,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.25828088049360715,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.4434,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.2596288163094018,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.4425,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.2544592669539469,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.4275,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.23826195485855098,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.4376,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.26319646112573697,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.4228,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.2623147652742952,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.4443,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.25737979941371647,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.4245,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.28212921578393535,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.4274,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.3121260690480352,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.4465,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.2401018045922969,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.414,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2874321600997058,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.4337,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.2680281669187174,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.4471,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.2227962108306868,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.4259,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.3216922496008065,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.4293,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.23753643418796055,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.4224,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.2324540171981671,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.422,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.2740012028297674,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.4153,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.23177648071523577,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.446,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.2331092867366247,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.4397,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.2304541912291757,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.4343,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.2577435592841916,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.4249,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2359854891249285,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.4177,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.22081740482660575,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.4182,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.2461363367520829,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.4343,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.24227044286588442,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.4264,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.24296818788391594,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.4161,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.2580005733989028,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.428,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.24678224145282338,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.4202,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.24639729754952752,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.4297,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.24440606352935748,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.437,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.23386370229134215,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.4304,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.24625669913199696,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.4361,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.5191351508654154,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.4227,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.25434665682582036,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.4192,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.25612066399809497,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.4347,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.2722304951635677,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.4184,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.24410105040838126,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.423,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.2488514311345245,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.4374,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.24685297024183578,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.4313,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.23705640264959701,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.4294,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.24551805382475728,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.4494,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.23690461524224535,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.4352,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.23020062655274842,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.436,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.23312622241550246,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.4231,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.23079029102528958,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.4253,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.21880971667486257,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.4263,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.2350220702061284,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.4315,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.22833505957169326,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.4319,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.21451089136390028,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.4375,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.2706441670698682,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.4271,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.2352895571222755,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.4215,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2374477761290762,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.4345,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.22535104673231432,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.4322,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.21601922768282497,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.4317,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.2445161391479925,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.4297,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.22717030924094242,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.4407,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.21422149969343027,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.4297,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.21224406091191672,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.4389,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.22957633872401606,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.4228,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.23757633112780702,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.4381,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.22399467397207942,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.4245,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.21814109591035308,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.4194,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.22203952148122544,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.4312,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.22656242916687802,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.4206,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.2236019440265143,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.4155,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.2390787966961733,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.4327,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.2247970774218248,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.4297,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.24095062041305879,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.4205,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.21687993522491422,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.4365,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.21764798341667752,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.4306,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.22605953754407718,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.4428,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.2244208717978109,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.4329,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.21591095118787038,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.4222,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.2305624306920765,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.4107,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.2268924770528969,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.4103,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.21407127730040768,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.4424,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.23060609426830422,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.414,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.23175324656918636,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.4079,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.21972617378310116,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.4359,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.20598006302565278,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.4372,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.22153649452465937,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.4304,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.23072635726944463,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.4323,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.2153429755908241,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.4224,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.22091002539768467,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.4253,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.21766005071482164,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.4191,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2118083773317974,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.4261,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.22696167517943638,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.4327,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.22246874088260282,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.4264,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2228831901711511,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.4218,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.22368410469029948,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.4296,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.2234638787495273,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.4437,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2368087518550631,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.4206,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.21563316107303915,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.4451,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.2228861639009254,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.4328,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.22671985305871847,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.4363,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.22583331896852127,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.4328,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.2544581343319045,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.4318,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.22964714769400071,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.4331,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.22459433703209886,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.413,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.21671128891789984,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.4061,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.21866390457729598,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.4158,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2166487919708008,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.4484,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2342377204663719,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.4342,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.21823489701277296,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.4192,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.2164785087184789,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.4327,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.2108600237698421,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.4297,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.2569943706444706,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.4094,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.232524603192592,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.412,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.2320096652851375,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.4262,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.22318439579168262,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.4355,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.21911529311783817,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.4325,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.2204886607493766,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.4305,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.2154367823485719,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.43,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.22915973906577558,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.4276,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.2170306458439218,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.4281,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.2227078144802261,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.421,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.22967293224952698,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.4113,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.23420247566764618,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.4312,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.22883690088991804,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.4359,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.2062203739455029,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.4192,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2225337054836429,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.4512,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.24930268344510956,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.4273,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.2217083387667037,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.4303,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.20993992277851284,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.4342,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2177549119378194,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.4283,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3922132646931689,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.3824,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.3108919377888719,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.3752,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.2770111611700889,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.3881,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.41264631227677184,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.3817,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3582174801411419,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.37,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.28688678173204385,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.3783,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3675383317252137,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.3783,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.37097334483782946,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.3823,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.32758756709649084,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.3775,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.2541057718526196,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.3794,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.33691214003565795,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.3935,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.2966990997693529,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.3767,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.2603015373004761,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.3742,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2593777055101183,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.3842,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.2767488369523831,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.3761,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.2691526841000442,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.3875,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.24160631009275413,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.3866,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.24656793183268869,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.382,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.2708249491586843,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.3953,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.2460317175166156,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.3968,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2330107324947574,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.3875,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.22576117733799328,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.3854,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.23259419010263518,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.3939,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.23208848429369425,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.3788,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.23374627737566936,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.3921,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.22474762370168408,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.3729,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.24022279744627406,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.387,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.21907612030258958,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.3972,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.21713817671124383,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.3551,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.22198973482374781,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.3873,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.24292434022551687,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.374,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.223698570985009,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.3936,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.22494308240750818,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.3652,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.21283177048800178,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.3844,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.2592231463335273,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.3908,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2252713896758508,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.3768,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.22777553884915727,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.384,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.21483792951429553,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.3773,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2342189751313108,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.3775,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.2333025980001946,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.3882,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.21391945765246767,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.3874,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.21364667257977785,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.3991,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.22258573991126474,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.3866,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.21482721789370876,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.3862,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.21277380513835895,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.3843,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.2103849106569351,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.3938,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.20494121795267595,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.3781,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.2159541070843611,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.3804,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 3.2148005420428944,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.4137,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.22666656459284149,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.3826,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.21287804932119428,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.3819,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.20463003812283861,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.3841,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.21095225598499082,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.3797,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.21506696106288822,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.3733,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.2090532072793429,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.3899,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.20294453366285822,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.3981,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.20770030829869462,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.3877,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.1993187462391782,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.3976,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.22031678685739373,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.3781,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.21283336555112578,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.3754,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.20752169760329994,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.3666,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.21929312447090085,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.3818,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.20328109036782213,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.3848,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.22096364235206356,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.3833,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.2075127391047829,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.397,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.21290060909992553,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.3919,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.20456038942313418,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.3959,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.21145451899633838,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.3763,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.2078745705918575,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.3918,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.2139146500309758,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.3955,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.2088693841200297,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.3852,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.21263572549581944,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.3935,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.22390700906789004,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.3749,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.4666032647522683,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.3756,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.2038178826935211,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.3792,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.2148625207465219,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.3822,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.22352892647766162,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.3886,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.20804179674200043,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.3824,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.21096374195932596,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.3954,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.21283273304734757,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.3747,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.20993712155300845,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.3768,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.21661588513487054,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.3867,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.20450363816050554,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.3858,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.20855445561881572,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.3867,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.21371456778257203,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.3743,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.21167042512596912,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.3945,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.21405766981687682,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.3875,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.1994857624864544,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.3804,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.2052652027713752,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.3873,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2138778549605151,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.3704,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2025612686088349,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.3786,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.2051793348330142,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.3703,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.19891307016342585,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.387,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.21777483434426062,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.3669,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.20915718728437002,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.379,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.20354510016944688,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.3814,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.20138074037162085,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.3891,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.21215865602494052,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.3938,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.19989277496436259,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.3955,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.20470850708337968,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.3861,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.205709442482576,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.3697,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.20504180634562988,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.3793,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.2032863516188537,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.383,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.2113077132625886,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.3888,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.20981029438327703,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.3936,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.20449871423479782,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.3632,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.19706945446914678,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.3872,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.20721257093645004,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.3772,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.20029055668091167,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.3793,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.20062976294880755,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.3849,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.2085543655137471,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.3865,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.21195870122606808,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.3784,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2022845636783801,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.3833,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.19824704135587573,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.3738,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.1962980216993788,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.3719,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.20208058453182207,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.3907,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.19289974200711166,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.3729,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.20175352939799077,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.3752,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.20993192650829817,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.374,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.2037172453141576,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.381,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.20860194829388423,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.3768,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.20243584370028359,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.3881,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.20228595712343317,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.3843,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.19946001630619134,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.3932,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.20551036306593504,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.3704,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.1964825672199204,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.397,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.2073919111825547,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.3947,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.22161165797843083,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.4015,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.2125104861980743,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.3588,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.20091401413179222,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.3951,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.2003998852645089,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.3722,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.21283122580127925,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.3733,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.20396852112449265,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.3878,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.20292065772311468,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.3763,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.1987973039524154,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.3898,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.20610455882922524,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.3873,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.19736905558335358,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.3821,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.20469399773020436,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.3957,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.20327507500815964,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.3901,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.20604010200492126,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.3819,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.5681236090490008,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.3795,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.19983459822624391,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.3812,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.19967077414508694,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.3886,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.2136951694224824,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.3644,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.2031887680960763,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.3778,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.20318199678087778,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.3705,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.19695592461395206,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.3918,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.20741790369790836,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.393,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.20785561582284373,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.3799,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.20046971577975706,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.3947,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.2022883686326143,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.3858,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.19878617877070173,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.3842,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.20585241284451603,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.3788,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.2044651318296073,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.3867,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.19491777919156025,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.3918,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.20535196177876208,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.3842,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.20416286406499654,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.3954,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.19522004777633578,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.3785,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.19584841456827806,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.3828,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.19801338511773925,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.3855,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.19214703444505515,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.395,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.19826381731533402,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.3956,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.2051664360702941,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.3809,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.20132722186272223,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.4028,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.19969463546750746,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.3767,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.19409260392703853,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.3889,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.19556832375259692,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.3781,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.2011683982938109,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.3838,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.19914680101391166,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.3794,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.193819718298843,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.3858,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.20914898039315688,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.3594,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.2027485014705548,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.3854,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.19690662783657178,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.3786,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.20504184227726993,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.3771,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.194070576288239,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.3913,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.20523393786567404,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.3862,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.20200135478409284,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.3745,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.2059599734777962,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.3833,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.19247642626012124,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.3899,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.19873403387517935,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.3864,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.19736990382268008,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.4001,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.19057555775378018,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.3913,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.19157941978675694,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.3863,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.192398566934559,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.3939,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.20933348147108186,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.3852,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.19533014124614848,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.3992,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.20103947710953188,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.3651,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.19739460938349915,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.3877,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.1960431393289219,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.3895,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.19608173628052616,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.3743,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.20369953095215398,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.3673,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.19321024178782686,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.3887,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.1892281097919001,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.4013,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.19205789377416022,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.3961,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.19975738415113278,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.3833,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.1957496423631938,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.3763,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.19205487966511664,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.3855,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.1998165771915459,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.3826,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.20673334568307625,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.3929,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.19562536727887436,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.387,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.19954815073022206,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.3653,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.19828520094032534,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.3994,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.19216268881211812,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.4003,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.19662331595704202,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.385,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.1957421281560983,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.3878,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.19502176512710917,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.3831,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.19085685910698477,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.3977,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.20272135273221878,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.381,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.1987972410488697,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.384,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.196999652829902,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.3829,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.1972581298087618,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.3845,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.20844545951160576,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.3577,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.20146422116979898,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.381,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.19413792949056155,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.3859,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.19258234961881887,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.3881,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.19210524453282854,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.388,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.22483049988577278,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.3763,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.1985214246882749,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.3754,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.19664651738220393,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.3776,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.19722974376713287,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.3714,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.20146155951595238,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.3899,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.19665216737403954,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.3946,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.2016422837326845,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.3858,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.19694879501338983,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.3757,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.22389882918438517,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.3814,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.19958123851929432,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.3916,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.19586110091722314,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.3874,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.19516235049449984,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.3992,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.197231254665183,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.3806,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.1934141145424699,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.3989,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.20361724496270805,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.3787,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.20365274517570295,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.379,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.20319091871938993,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.3891,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.20208604033127225,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.3703,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.19705627010207388,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.3944,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.19444265619137377,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.3862,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.1987684821466451,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.3688,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.1931114753099553,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.3973,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.19368341511020362,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.3792,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.19366137808007788,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.3822,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.19223559840501597,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.3997,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.20039766911441861,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.3797,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.2010754275057287,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.3716,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.19236197661266896,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.4022,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.19525950014046842,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.3721,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.20164023797282968,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.376,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2030788705376874,
+      "learning_rate": 0.0,
+      "loss": 0.3793,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 5.149708790995616e+18,
+      "train_loss": 0.5151235838650692,
+      "train_runtime": 29707.6144,
+      "train_samples_per_second": 5.319,
+      "train_steps_per_second": 0.042
+    }
+  ],
+  "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": 5.149708790995616e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}