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": 6.4576805277374,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0467,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.3633678760919015,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0187,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.474482842709658,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0407,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.37188963656656,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0498,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.025639575881523,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.0341,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.868542177036071,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0317,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.616118107056145,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.9956,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.238359019734606,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9867,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.7944074454859567,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9646,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.5967949061477102,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9749,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.496858484999185,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9549,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.954208574996178,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9149,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.163531615470529,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9346,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.078319491283546,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9204,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.801013908923328,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9194,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.768011046809733,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8839,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.4784518690099464,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8623,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.174847875135467,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8674,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.6372765076048046,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8484,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.1859194702516667,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8224,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.060647948615501,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.7927,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.1621009539850669,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.7973,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.1367645526333463,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.8126,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9607886519801782,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7687,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.9073620446406971,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7632,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.9557064200452087,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7679,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.8324775463617341,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7513,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.8313768448349383,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7574,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7644876659825101,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7493,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.795594990117005,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7366,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.8516934733192878,
+      "learning_rate": 1e-05,
+      "loss": 0.7396,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7958761290128898,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7281,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.7461843677983585,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7197,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.9550868845043304,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.714,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6082385067117011,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.7009,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.7163777915487319,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7113,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6259943631170258,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7058,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.5936230955456848,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7105,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5618737271950303,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.6991,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.7802312246581131,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7034,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.5798752147806977,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.6882,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.616318626071167,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.6864,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.6030213640054157,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7004,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.780264923186977,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.6854,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.695979641821267,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.6734,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.7647662261396853,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.6857,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5879380246741798,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.6789,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.7607338589488528,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6805,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.5485980816630737,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.6693,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.685916991220263,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.6758,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6439528533911836,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.6668,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.631227940250043,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6634,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.7725370134823678,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.6684,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.7075926620662965,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.6629,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.5058275288479809,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6728,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.755245735339481,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.6652,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.613488137806361,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.668,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.6685508136065996,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6591,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.765262133540382,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.6579,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.685139164116857,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6566,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.6952487916441916,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6623,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.715716808632893,
+      "learning_rate": 2e-05,
+      "loss": 0.6653,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.8750160295108307,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6625,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.7227431487824444,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6522,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.7471140886408016,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6469,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.8873027321836182,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6524,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.9069638051363405,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6597,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.71913153144016,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6389,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.8093794839286521,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6551,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.8048124606359653,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.6625,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.8592994694896612,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6448,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.6283236843412697,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6479,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.8236887446911357,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6452,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 1.2001514148141346,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6427,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 1.0552937796428103,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6457,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.8815552584852745,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6305,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 1.3280590349071824,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.6503,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.9098443031905042,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6418,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 1.2897366777124972,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.66,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 1.205470762934797,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6544,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 1.1089231518724751,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6455,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.711134017857066,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6555,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.9422043585284016,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6495,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 2.0034568776306236,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6415,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 1.7116824847289107,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.6389,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 1.6806551676569736,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6501,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 1.58061274240446,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6471,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 1.494504432245134,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6483,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.4815193585702942,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6566,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 1.5516350750733627,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6484,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.5067832239685792,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6376,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 1.0962456899058293,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6339,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.4509046266359258,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6215,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.1581232749755557,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6454,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.3728272949812264,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6393,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.3341068278103818,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6409,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.0255616262177563,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6256,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 1.5616904386383346,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6346,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.0628665658397332,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6443,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.4616509938943625,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6376,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.4244619135886751,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6029,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.9754347785287966,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6455,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.3865695599738441,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6331,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.0196537418242142,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6413,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.0338644707267417,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6409,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.6995012507005631,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.6283,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.7858014550987024,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6306,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 2.1058555492343682,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.6193,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.24838487891165,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6316,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 2.0703071077453608,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6346,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.5365572256594244,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6536,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 1.832171049618836,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.635,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.2340279462569648,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6335,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.9253615629708136,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.6363,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.6781488265129398,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6477,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.5973056429832255,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6131,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.3430991831532064,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6409,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 1.4311206414787536,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6357,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.2940965248414038,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6241,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.2991799819460441,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6175,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 1.035667157369696,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6178,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.3401517785096304,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6431,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.0669273106058113,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6265,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.3865941540587932,
+      "learning_rate": 4e-05,
+      "loss": 0.6227,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.267104030107667,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6179,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.130882581750431,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6169,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.8839568574619707,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.6168,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.0632704615757336,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6069,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.1474089700305976,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.6188,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.052618298256343,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.622,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.013397561772301,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6163,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.5500288587282889,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.619,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.9237337166897704,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6312,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.2005039443402952,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6196,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.9645955893005146,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.6081,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.886709263637031,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6339,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.7109988300560216,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6113,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.7960176087302295,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.635,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.864553635580836,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.625,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.86139294464879,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6141,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.6146596256585393,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6198,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.7647459653529677,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.608,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.7048214694381668,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6052,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.7339939340493472,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.6029,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.8816657252566782,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.6047,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.9830840143506725,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6171,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.9148655092481464,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6065,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.8775039656205047,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6207,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.7853986003516787,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6035,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.5351698328623787,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.6148,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.6923761222041092,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6245,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.8248575864424534,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6093,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.5884426195473208,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.5941,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.6257427394767463,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6142,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.8006951563960394,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6061,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.618531822586279,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.5941,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.5693806512276458,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.6122,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.6202141840919696,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.6101,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.4865657068109101,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.603,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.4716555133520285,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6093,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.5814595006089668,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.614,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.6659935910116571,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.5988,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 1.1158971681392873,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6122,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.672266254769799,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6141,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.7260372562424515,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.607,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6498361207725679,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.6138,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.6912457666946383,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6117,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.6170333845704532,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6036,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.5411952363741989,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6021,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.5552604270170359,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6092,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5767289015089132,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.6003,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.6349738811219758,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.6131,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.727242794354254,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6086,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.518029789302584,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.5989,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.6501054183819525,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6102,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.8768726091640601,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.5994,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.8810550933448285,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.6117,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.5407812673367098,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.6136,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.4373329371052941,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6208,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.534455815873109,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.5869,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.5766099660388782,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.5988,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5728964384461903,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6048,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.4634068817762017,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.5859,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.510468281233829,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.5947,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.5328989113464621,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.5972,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.7023841622384055,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.5997,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5911678041287016,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.6203,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5412302257426849,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.6218,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.5843108687055252,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.5966,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.9128508665413902,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.6119,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.8288266916222455,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.6162,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.5161321946946388,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.614,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.48969669319093967,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6111,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.5531524941671261,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.5849,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5811485930728056,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6083,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.519652692353517,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.6024,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.5170433196630216,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.6089,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.5725727849313519,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.6052,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.6210343042736366,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.5988,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.6819592491834878,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.5904,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5743326715078187,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.6044,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.6187141119939527,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.5919,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.5395296114408239,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.5881,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.7040660760651569,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.5966,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.7897096257901146,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.594,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.6109315746891444,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6016,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.6510242240208545,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.617,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5845105070281634,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.5975,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.6975232570896907,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.5943,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.6072576675938474,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6038,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.5181532262107797,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.5954,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.6662125318188135,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.5847,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5412282160801553,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.5929,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 3.0432470040999453,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.5977,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 8.689816335063275,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6234,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 148.67033544295282,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.7618,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 1.2175595851025047,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.6175,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 1.1114109055683665,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.604,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.9525993938031869,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.5848,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.9875921830948513,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.5953,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.854668425840757,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.6173,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.7775974949350375,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6034,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.7317191372177227,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.5969,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.8832110528112876,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.5916,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 1.002505992950982,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.5769,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.7223052870296279,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.5938,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.879860300538462,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.5956,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.539767688508127,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.5801,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5944710972460049,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.5996,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 3.9126519262997723,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.5922,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 1.1937385861348178,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.5907,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.8415090459462536,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.6008,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.6802267200631734,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.6003,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.9393388020652677,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.5947,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.539969688035524,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.5832,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.7633782720810772,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.5922,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.728009619403396,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.5962,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.7016326620764834,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.5903,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.5582350484793566,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.5942,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.7007788925324399,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.5806,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.5480320489930586,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.5832,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.6435240139368261,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.5801,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.5108322710489129,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.5921,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.45206263367007593,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.5923,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.5767688739215794,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.5908,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.45735439721713267,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.5864,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.43937276040289636,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.5931,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.5296214468920195,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5392,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.4889461231248204,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5468,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.5585740098319993,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.539,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.5828053002936635,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5469,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.4235749243248414,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5486,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.6542691427928425,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5473,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5542739287754719,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5381,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5764976370750128,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5491,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.5106992974923714,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5238,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.5047944002940753,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5455,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.5410184275643929,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5284,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.592132991479928,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5532,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5014508086786136,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5457,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.47474254286401607,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5335,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.5072735352892433,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5401,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5658199157670931,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5479,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.7096145059909459,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.533,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.5694895745443466,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5369,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.555100225512485,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5436,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 1.9306394996991945,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.543,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.7765967707731842,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5409,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.9522386278604992,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5438,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.4969197423798699,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5445,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5795853977474442,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5454,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.48897741395363864,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5468,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.5628979386431644,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5411,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.4743280934831043,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5389,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.49346514914995965,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5339,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.681944654871647,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.5419,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.4495294938729085,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5379,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5408501886531236,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5408,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.557954059082625,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5461,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.4193153980458834,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5433,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.4890221977843074,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5439,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.49705765828084786,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5328,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.42435283364301024,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.5458,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.430829513218629,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5451,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.41912502000527885,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.548,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.43328362160030626,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5412,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.3859264724463287,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5438,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.43678340809769234,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5404,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.433391966476135,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5484,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.3685077445965236,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5379,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.44362366197669245,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5408,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.36888788428308467,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5411,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.4540218951482447,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.5604,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.45767555785243524,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5449,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.4369159876768135,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5413,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.3717552959171639,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5469,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.47294911232838865,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5509,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.42404324419828676,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.5391,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.4723623681163445,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5395,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.5820467896489243,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5387,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.40904665874314644,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5453,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.43906792161113395,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5445,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.343989475112032,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5308,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.4435019642371615,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5407,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.4314979841453714,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5526,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.39133581280588176,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5417,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.4196640454546816,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5531,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.41574941531076237,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5419,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.36239177683704776,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5471,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.37284782878967243,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5401,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.3916440826081326,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.5278,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.3385513242576728,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5411,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.4741047877894491,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5305,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.37523141546410643,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5476,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.47111148089573723,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5529,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.41700653250760844,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5485,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.44286938307676443,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.5477,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.44216163881042325,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5458,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.4335085541887388,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5376,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.45434543342885053,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5536,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.5086817469104363,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.5528,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.3570382651399442,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5447,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.4834170324206242,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5417,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.46897700807509324,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5405,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.5044126551561389,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5438,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.41489689493786425,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5328,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.5362794101119884,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.538,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.528710818300108,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5382,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.41668233576319075,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5495,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.3778318710532727,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5378,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.6171105599193623,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5364,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.49158805834479,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5314,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.3927045023856944,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5377,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.5461535210599626,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5395,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.4797410783359126,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5442,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.9513770903270331,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5286,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.511200173397715,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5219,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.4426137031432017,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5421,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.4127283958298289,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5221,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.49773604575237507,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5463,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.3955434142889222,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5433,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.34627659440153535,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5287,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.5713366119678243,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5514,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 3.899278914635574,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5633,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.7397310332439267,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5476,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 1.1719078273371506,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5516,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.565560166295888,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5448,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.9913797350203428,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5473,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.8301419536059896,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5419,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.6181781436544843,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.551,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.8698209088754812,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5521,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.51569288724247,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5358,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.641701601475714,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5422,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.5160278500540039,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5458,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.5010332916843963,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5313,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.527689490350092,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5355,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.5329735337765056,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5424,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.48250199531819216,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5337,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.43824402117530264,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5482,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.4025494837775103,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.549,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.5283944492151974,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5386,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.4344071710252463,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5402,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.5082361298128638,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5397,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.3758569775509337,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5295,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.4681183881182206,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.542,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.4109037133006743,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5317,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.43757690335906874,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.542,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.46615263380689265,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5362,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.3714523920426941,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5359,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.42475540554427443,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5337,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.4449779238827899,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5426,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.45494549443124344,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5351,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.3779201642953057,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5444,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.42521087155308185,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5359,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.4221247381194866,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5357,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.38213296230482524,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5416,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.3753633369766499,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5403,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.33804665713162924,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5316,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.4035932924269497,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5375,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.32761192413114315,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5467,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.4103966195741819,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.535,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.34927849228802954,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5429,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.37483784547491117,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5474,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.3411843108926859,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5339,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.3831887336032112,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5459,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.3436333658106659,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5433,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.3209838712638273,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5394,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.4606835209145216,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5391,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.388525623989058,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5423,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.41557642313394816,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5408,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.40131628980141504,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5396,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.4257468828868746,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.546,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.4020740294795907,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.527,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.36961200412818995,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5258,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.4055839207354904,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5269,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.3592484575197053,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.551,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.3806632981668532,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5537,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.3338123066852918,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5331,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.37499220636297476,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5334,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.36163669879231264,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5396,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.47431378950598985,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5343,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.3751269177427494,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.5252,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.4714985640390288,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5361,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.4608345142744209,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5263,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.44964227845146076,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.532,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.49298785309300214,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5308,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.4446307552259452,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.542,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.33082299886897987,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5373,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.3675790581108493,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5187,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.36061157408061506,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5468,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.3483507947956696,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5258,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.36967786198032127,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5401,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.32142468346690994,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5237,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.3407601308326797,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5178,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.3782803490045996,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.528,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.37050271069275487,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.5499,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.4375927220288063,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5413,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.43282436282656434,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5381,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.3496722900671478,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5359,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.34968805622608523,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5423,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.3410263051130017,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5351,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.3165590872279204,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5446,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.37138024541899606,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5447,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.28093316829542125,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5249,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.4264480083654586,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5289,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.32810618599942515,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5376,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 3.302374181080447,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5824,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.41954853328174857,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5173,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.42757707033554726,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5193,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.4088310810558018,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5372,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.41520781650424987,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5448,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.45351765443576236,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5471,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.38260022445670056,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5291,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.42703247010945505,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5309,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.4467438865299205,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5397,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.5505394895889614,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5442,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.4159829641557025,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5332,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.3924230959674851,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5399,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 5.000052806861568,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5359,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.6529237059497968,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5423,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.5473030973951157,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5324,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.4919263790695153,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5246,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.5858468281575583,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5344,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.44972380994820677,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5417,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.4817519640804834,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.535,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.4009320544838859,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5276,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.44658184984591515,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5482,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.42435506529789596,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.5343,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.4786549195208785,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5317,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.41244112864517624,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5426,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.38797045234228733,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5347,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4331892029294097,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5321,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.3930495874980896,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5281,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.3870553776496449,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5513,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.3883022768965384,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5354,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.3771518889798142,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5467,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.3821406717739271,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5417,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.38123756329715824,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5188,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.4080270269828664,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5329,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.31985267858075844,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5374,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.3701044512769983,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.535,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.3604320410302114,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.5204,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.33149269241045515,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5399,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.3385492421486355,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5246,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.3426180561969019,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5272,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.3373469454950969,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5298,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.328958690298476,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5287,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.35796238883762965,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5284,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.4245626710118605,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5325,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.3078175084914415,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5345,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.46742893380908374,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5283,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.3291624701258338,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.5401,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.4233293255039732,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5338,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.37911499928861847,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.5438,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.32325367888968604,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5411,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.38220993879848614,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5354,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.380914363742081,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5324,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.33444961947388513,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5249,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.36970418701768376,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5261,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.3354641906537629,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5401,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.48183031308836527,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5198,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.29194434043709644,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.5276,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.42327649325711164,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5245,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.30903489644506743,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5336,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.40825578127664447,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5382,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.34518945727922634,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.5443,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.3624324972682968,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5282,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.3872769152052105,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5351,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.36951703196736946,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5371,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.38547582107276596,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5346,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.336348434956659,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5338,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.43260634702009654,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5401,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.3027738877441493,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.514,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.3941856154848139,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.528,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.45394635686609,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.4782,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.3497601436374178,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.4692,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5079594623910333,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.4772,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.4871879533884612,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.4719,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.43430977951899136,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.4727,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.42765223914508366,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.4648,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.4599219136330044,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.4669,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.4732627224790653,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.4621,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 8.362849817713629,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.5306,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.726890818944512,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.4621,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.7015278468112518,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.4685,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.46304714943603603,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.4678,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.56876868158564,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.4755,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.519731344501471,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.4599,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.4288025263236545,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.4726,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 1.639467030856446,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.4684,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.506941300290542,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.4841,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.5970213836774897,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.4725,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 1.868756982862492,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.4859,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.7498584574361769,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.4596,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.847451829959669,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.4702,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.638189697091603,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.474,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.6946846031102561,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.4801,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.5417475753675832,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.485,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.6132979383371503,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.4724,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.6410788626228738,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.4731,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.4947128602207681,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.4633,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.5897071553390205,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.4644,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.4167397807683381,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.4765,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.4944596627227177,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.4839,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.3961293698562305,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.4783,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.4547934726632149,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.4755,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.35417980061620347,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.4781,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 1.79271350277976,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.4575,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.5300454129157421,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.4847,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 1.504341232411586,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.4789,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 5.34509605378001,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.5449,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.5974592719033606,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.469,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.6877346023864425,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.4885,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.49870448966632785,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.4737,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.7692239847427462,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.4754,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.45473313984194424,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.4728,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.5761355029762897,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.468,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.4563395250056329,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.482,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.7406740666209599,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.4866,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.3992323031736619,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.4737,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.507682370772295,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.474,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.3469018342597289,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.4726,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.46694284405080944,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.4714,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.312732633544916,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.471,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.4622285320043684,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.4743,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.33360699535183636,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.4772,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.4353497499101615,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.4719,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.32481756694783664,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.47,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 13.476843419038941,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.5373,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 7.505217872194464,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.699,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.760589226199742,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.4751,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 1.1189623138002254,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.4834,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.6936983863021657,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.4742,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.7330084344330785,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.4932,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.5832495545567773,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.4752,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.6607371707846781,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.4817,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.5686119118581259,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.4835,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.5085717168746307,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.4794,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.6414233161030176,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.4769,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.43614849477108003,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.4759,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.5933923146069829,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.4823,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.40303289496506406,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.4749,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.4872178254442704,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.4784,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.42847573668634475,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.4807,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.4996432008474541,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.4792,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.3868446783684828,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.4795,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.4844516332548,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.4779,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.3420924665863662,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.4676,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.43316033721578,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.4716,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.33855104298204114,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.4622,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.4058260305838945,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4779,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.3989978098779538,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.4769,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.4089385473638304,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.4829,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.41024823783091807,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.4691,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.35800811271246247,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.4712,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.4422064002414372,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.4757,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.37083700335590414,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.4834,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.4227012655231,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.4794,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.31065190415347127,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.4677,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.43012289324920844,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.4792,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.3165092105182924,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.4836,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.3631284871604974,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.4664,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.3066932601544059,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.4642,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.3849489074161355,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.47,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.31569946960462614,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.4726,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.366299919515113,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.483,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.29305694786459924,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.4731,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.3324428976138318,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.4684,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.31523624576420656,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.4665,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.32961796388661263,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.4624,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.315632212393381,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.4665,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.2951646430733345,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.4632,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3058303796688548,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.477,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.3177803556506064,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.4706,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.3103735385291091,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.4654,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.367802875897843,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.4837,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.2893244012500115,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.4694,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.32337428720863737,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.4749,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3146209749438074,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.4807,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.30263434436626635,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.4625,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.29204781484201525,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.4671,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.27341280756888725,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.474,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.2938899240033719,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.4734,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3165081712595855,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.4633,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.26272248779292207,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.4767,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.2791072953380711,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.4727,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.3115084364719211,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.4657,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.2615433801607689,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.4861,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 1.0478610554482388,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.4702,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.2696706273970768,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.4627,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.2604335178473818,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.484,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.2954660741949583,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.4818,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.2835225515631131,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.466,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.2840126565177403,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.4738,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.35531610167118005,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.4618,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.2923329666268743,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.4728,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.33934024312540784,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.478,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.33258774875889996,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.4721,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.29786076698025865,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.479,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.28761922963075964,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.4771,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.2987323841337894,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.4771,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.27669440532592837,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.4634,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.31115046496951987,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.4588,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.37738676782666636,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.4736,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.2947339056025845,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.4732,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3307945127494029,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.4739,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.27176253661973815,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.4767,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.28995749554274186,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.4716,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.27496188460472903,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.4737,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3071457448309229,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.4704,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.5334085326872272,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.4856,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.28876789506132994,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.4675,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.2708437164741343,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.4777,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3071823022873115,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.4851,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.26908688465028574,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.4789,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.28177382547366286,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.4672,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.2912076863023911,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.4799,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.28545664605187476,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.4742,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.2821251579707922,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.4714,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.2951627796463692,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.4748,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.29151492603498885,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.4727,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.29091695190579964,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.4533,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.2998757916145576,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.4596,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.2831559366425363,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.4664,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.2830873847087029,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.4639,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.27392329560616135,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.4671,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.27948873812488356,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.48,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.3086613796166838,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.4689,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.2766942069918444,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.4926,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3008771493909261,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.4784,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.289690692012994,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.4742,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.27069554748103947,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.4713,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.31891639227680557,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.4669,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.2743868309905548,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.4741,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.31181318618275494,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4762,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.2804209043590127,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.4683,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.2694987665260241,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.475,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.28213607426659126,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.4682,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.29392359171968124,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.4749,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.26723234462990236,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.4867,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.2806137326271767,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.4582,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.28000232439881373,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.4681,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.2624403243038678,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.4723,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.28339210554390465,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.4709,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.2688490252142536,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.466,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.2462182470247603,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.4763,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.26721537971075937,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.4776,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.26380677693097115,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.4773,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.25513121842323216,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.4652,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.2531560699062049,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.4667,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.27572572551872326,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.4605,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 1.0445904795525864,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.4715,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.25430718200312946,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.4727,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.2793972625068528,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.4568,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.2697030148752019,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.4707,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.2540618160267754,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.4711,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.2758680962217576,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.4613,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.25363990846337114,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.4698,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.2685493241443898,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.4793,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.28661154543946943,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.4792,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.26928240230307515,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.4658,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.25920126610294303,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.4655,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.27372983853020083,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.4811,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.26223844592304535,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.4794,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.26414212019978855,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.4534,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.27815685416786295,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.4798,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.2569398238517087,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.4849,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.2788707615917957,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.4683,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.26244064355063135,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.4824,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.29291627178184054,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.4684,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.26896468330223455,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.4669,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.24816418362049752,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.458,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.2957604266146241,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.4857,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.2770979775539658,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.469,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.27291118211957616,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.4755,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.7524229851893677,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.4667,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.25735798498758544,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.4843,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.26644417990632924,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.4744,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.28854555730679626,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.4713,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.25218897237342575,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.4624,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.28433915517848407,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.4655,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.3117333733038494,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.4779,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.2832149810959309,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.478,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.2673953092253229,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.4693,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.2589749745049599,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.4612,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.24712661401600075,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.4629,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.2776207303694512,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.4679,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.2744351652863762,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.4682,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.29921681021514246,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.4766,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.25787061017313007,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.479,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.28344791103258005,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.4654,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.2759013158992343,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.4638,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.26961193975523023,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.467,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.31253112888727663,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.4676,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.2645685779720682,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.4719,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.27698805983480185,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.4696,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.28673880835706766,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.4643,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.26089819826504324,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.4709,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.301845856023087,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.467,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.2516289895938231,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.4766,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.29825360886135643,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.4822,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.2758511572780391,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.4754,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.30342603435931337,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.4772,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.2826843552292578,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.4637,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.3038515582363174,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.4756,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.2923007364424485,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.4866,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.30655526558645424,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.4675,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.2842227661841636,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.4799,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.2674662206647013,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.4656,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.6575087967686535,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.4646,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.2577862858797889,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.4693,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.2840673351909301,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.4748,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.3071834540794957,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.4697,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.274891355430367,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.471,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.30807781914077625,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.4705,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.28076710730121396,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.4659,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.2826073010130978,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.4691,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.6123673795579665,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.4746,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.27540279945878654,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.4755,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.3224755276509576,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.4718,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2696589522623033,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.4579,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.43637132679783797,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.4188,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.29930241981666394,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.4097,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.5029282365672337,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.4127,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.3487926355340343,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.4113,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.368544422020649,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.405,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.3658239754975543,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.4205,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.331581055391099,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.4157,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.36240150608776894,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.4091,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.3478729756855832,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.4101,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.38741386749439816,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.4131,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.33312992360513843,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.4158,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3225674039476713,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.4165,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3563545184452297,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.4054,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3165629038114712,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.4099,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3200372498084527,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.4066,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.32454069847672523,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.4,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.3200106273176754,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.4038,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.3049905911885024,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.4134,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3284029011380285,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.4133,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.3524769579279301,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.4124,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.2868269355234788,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.4099,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.31951168583350975,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.4109,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.28639361755295445,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.4137,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.30757782633831415,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.412,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.32114806903911053,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.4147,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.30717671560679444,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.4063,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.2912303635244877,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.4012,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.27491086814603705,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.4172,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.7448649302739736,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.4176,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.2879982008236426,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.4167,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.2674186440472874,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.4051,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.28508496266831573,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.41,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.2544502947470696,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.4187,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.2649705252513182,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.4223,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.25032814516541646,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.4053,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.26489066475069767,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.409,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.4131094226896746,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.4036,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.27525355808231705,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.42,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.2594142636170966,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.4142,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.25436210499598155,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.4204,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.26721862501957133,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.4175,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.2714861850563534,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.4112,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.2734426735863882,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.4091,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.2708649886232534,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.4198,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.25305713519675405,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.4027,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.28781470809218845,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.4307,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2522827423196585,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.4137,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.2854832177139244,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.4079,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.29426661811823873,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.4052,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.27672281589002967,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.4089,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.28260682112354024,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.4135,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.29619482811717984,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.4157,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.252190919097644,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.4042,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.24978346590564143,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.4127,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.279790845558154,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.4039,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.2747402662613917,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.4172,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.25320480982972443,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.4016,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.26245850916535607,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.4216,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.24648127548637289,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.4108,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.26160674419674773,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.4172,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.24278997802901506,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.4115,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.2618699638454409,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.4162,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.2506335476887826,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.4163,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.2530435081119448,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.4032,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.2716847072532106,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.4165,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.2475517605365274,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.4104,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.2596471307270437,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.4113,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.24626203780614542,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.4174,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.25172800646634363,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.419,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.2516959993610152,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.4065,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.25317378391309114,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.4091,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.5270910737718609,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.4244,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.25617657737555927,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.4062,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.2487649122174302,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.4197,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.2645699951461412,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.4101,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.24881300763757933,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.4139,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.288726426822403,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.4154,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.2624376208470002,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.4106,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.3408908161458269,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.4218,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.25852093240296636,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.4169,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.26263918961427046,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.4147,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.26595409686298843,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.4048,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.25505246931024445,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.4158,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.25426761364636113,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.413,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.2752824213968177,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.4047,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.25468388005781945,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.4103,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2827190999591702,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.4058,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.2592298711485064,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.4137,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.2499972476329392,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.4078,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.25967888742847894,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.4167,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.25259706771222384,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.418,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.2821508132711538,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.4129,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.24796163259994264,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.408,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.23179997140856481,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.4201,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2813147437213595,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.4103,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.2677894478743601,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.4135,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.26128010405340774,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.412,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.287822235187298,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.4226,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.27140698867435864,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.3987,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.26313781406610126,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.4294,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.2871498276569352,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.4123,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.2913583446871985,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.4167,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.254668241707422,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.4176,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.26632999840999255,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.4144,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.29467780863559595,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.4185,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.24330058408060173,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.4058,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.27717198552899713,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.4132,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.27009781083407636,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.4183,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.2300992670825822,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.4123,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.26462458780234205,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.4117,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.25368411177731526,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.4172,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.2362187672034192,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.4191,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.247969435424048,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.419,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.2649562503247566,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.4148,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.2600992051046922,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.4163,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.24382701318401623,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.4174,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.2588760550291378,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.4138,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.26619261397374505,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.4129,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.2554269737433302,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.4188,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.25311457786048214,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.4029,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.2518450605627866,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.4012,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.24610562203725575,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.4002,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.24844783709650697,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.4203,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 9.612554075997874,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.417,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.29923434027160395,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.4094,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.26074484675515935,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.4286,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.24960827337485225,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.4049,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.2778963027502794,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.4208,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.23302681941837047,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.4053,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.25987282170483933,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.4153,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.2904784667355861,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.4074,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.24502939178876176,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.4151,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2533218564118443,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.4083,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.26211447310760694,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.4143,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.2391928480846179,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.4134,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.26370416823826315,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.4144,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.24822262530796782,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.4164,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.2512318628872042,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.4102,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.23863323774449102,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.4119,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.24382297693883329,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.4069,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.25195415878701993,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.414,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.255289436222328,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.4019,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.24398412143083384,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.4176,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.8231620522692135,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.4402,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.24106639004737565,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.402,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.2324299447871331,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.4257,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 2.44120439897893,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.4138,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.2626905768300182,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.406,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.24335756650376367,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.4068,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2455912111837562,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.414,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.24626273593689738,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.4212,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 1.914607981932772,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.4444,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.24267274824409768,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.3969,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 4.07525605039324,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.4602,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.2463639233822839,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.4098,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.2439662232557943,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.4075,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.23499001562484328,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.4087,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.23353990786039655,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.4164,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.2327253632463121,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.4187,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.4272691596580049,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.4323,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.2599547311153755,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.4084,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.24728816157249447,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.4159,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.23636933338263175,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.4126,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.24815975394957399,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.4109,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.3176711907978759,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.4145,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.23081608805390288,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.4181,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.24440887544890477,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.4051,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.2554988099568679,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.4066,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.23923716225829778,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.4042,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.22601941824005814,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.41,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.5765973289802934,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.4039,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.24908458934046196,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.4072,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.2311877229406325,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.4123,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 1.7635925688456702,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.4468,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.2441246972260112,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.4099,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.21321121089734724,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.4004,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.23257400564020114,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.4052,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.22603176612108922,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.4199,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.2205061924963481,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.4081,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.23141316353546929,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.4166,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.2279760950909608,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.4114,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.23551548002185888,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.418,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.23236496672610255,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.4083,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.21353400343024923,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.4144,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.28709491244768204,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.4145,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.27071477465414273,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.4055,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.22151075819568225,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.4109,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.22537706721224265,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.418,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.2457639118715228,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.4134,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.22457157882444365,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.4151,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.2152786380474088,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.4064,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.24957898451187402,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.4198,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.23361178189343482,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.4126,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.22802852609812055,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.4098,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.24151484660286543,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.4143,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.22969510968970694,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.4104,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.233430774591082,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.4162,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.23148986646966768,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.42,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.22083872287090017,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.4183,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.23293288215224667,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.4142,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.24636738983016127,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.4113,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.2228469820341981,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.4246,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.23047013531339114,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.4113,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.2491437997697633,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.4103,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 1.0693141569042974,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.4105,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.21847135816563842,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.4114,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.2263805378530708,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.4094,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2335290266882657,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.4024,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.2330180668262211,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.4114,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.23352541727567028,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.4223,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.24494433111156252,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.4017,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.23135975536004622,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.4126,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.22950989418071968,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.4173,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2334735471790285,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.4183,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.22465725557666086,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.4158,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.2245629502269219,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.4198,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.2594169363061784,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.3976,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.24218046463916612,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.4178,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.22299892788247777,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.403,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.22820852361788482,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.4204,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.21492295061457103,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.4164,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 4.7305774009366495,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.427,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.2517510621821203,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.3959,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2347487389165283,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.4118,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2367132510295154,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.4132,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.27216307420308544,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.4047,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.22815741332216052,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.4111,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.21640941811011502,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.4081,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.22098116518337987,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.4086,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.23519716865147408,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.4089,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.23974537098247622,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.4061,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.2233894804147601,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.4145,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.22580688689714254,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.4034,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.23110775356929597,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.4104,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 1.255747464170372,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.4388,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.21492617703358904,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.4112,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.2262426236763954,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.4193,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.21525787782317057,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.4192,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.20655254887541327,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.4233,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.22824854888118265,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.4234,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.22398998898261596,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.4112,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.2202696033466498,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.4014,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.22948816434658562,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.404,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.2172966548743071,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.4102,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.22543221914011372,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.4099,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 1.027614699967583,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.4047,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2160071308214485,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.4073,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.37321553031507443,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.3754,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.3081357489632571,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.3808,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.24667067069822785,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.3835,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.3011021240744058,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.365,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.35146714684172353,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.3637,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.31191642149865956,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.3629,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3235415026152105,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.3772,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.29964012062118545,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.3781,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.3110051665409754,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.3726,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.27990234474158254,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.378,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.24989983748228325,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.3652,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.25572467866322607,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.362,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.2711263026142407,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.3692,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2825807900316572,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.3652,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.25356906658044404,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.3752,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.2591549541950435,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.3787,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.24676929996016614,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.3716,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.2551060841176538,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.3698,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.2505963783682618,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.3657,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.23045943202796665,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.359,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.24486889588298844,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.3635,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.26157285298363475,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.3784,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.24996161374426587,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.3691,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.24327102021857663,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.3694,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.23793417767504088,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.3784,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.2567528043901597,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.363,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.23537708506863794,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.3769,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.2461454937335168,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.3714,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.23360792050479826,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.3668,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.22608233603967043,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.3628,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.23901480227254354,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.3775,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.22688743211386658,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.3791,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.23142304795726903,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.3648,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.23012800805596764,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.3586,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.22721343238815692,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.3661,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2307588608722202,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.368,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.7993880396501319,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.3913,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.23261937856646903,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.3791,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2215386362935551,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.374,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.23836502902696274,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.3835,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.22990039133009194,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.367,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.22168174801896087,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.3625,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.22603481352629562,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.3717,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.21621143983011634,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.373,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.2267785432220509,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.3756,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.9453143402880974,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.3832,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.23639145818364907,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.3797,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.2192084824500539,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.3722,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.21553355941461372,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.3669,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.21796454021580694,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.3776,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.22696385219853799,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.3743,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.23350366273902645,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.3583,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.22581650771864287,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.3722,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.23535710367689883,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.3664,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.23574539940939498,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.3669,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.23695375420118667,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.3644,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.2339544991984321,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.3628,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.22354391505961838,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.3719,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.22151334161252528,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.3857,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.2338039187090513,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.3617,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.2356214449291888,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.3589,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.22003702337902975,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.3689,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.21305832565240873,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.364,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.235226748517246,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.3708,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.22856574434590918,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.3673,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.21366257655659232,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.3767,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.22800334400793604,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.3706,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2351626150018153,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.3575,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.21065437174093543,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.3696,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.22174398548276336,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.3719,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.3015398906094147,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.3623,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.22568295487300138,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.3642,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.23207781271818712,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.3742,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.23717565750168582,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.3719,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.22305719463798904,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.3652,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.222945989834579,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.3631,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.21644914939852633,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.3609,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.22506652238562064,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.3584,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.2380199163244916,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.3719,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.2228759198838121,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.3706,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.21974635884797658,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.3625,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.22133236256571656,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.3704,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.22916674659924088,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.3649,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.22074440171136098,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.3705,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.2910932283635072,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.3743,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.22945439413997742,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.3636,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2175742140108454,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.3635,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.21863682970719026,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.3677,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.20984055393972578,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.3715,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2240975531347134,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.3668,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.21089027383800654,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.3749,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.22963516408904064,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.3711,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.22588409994875328,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.3631,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.21502948681367803,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.3758,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.24048330064677823,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.3659,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.22375173542958499,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.3715,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.2206471703037627,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.3603,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.22180909245189967,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.3653,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.21632801451488956,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.3761,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.21816161627073072,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.3809,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.2184844917949915,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.3687,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.2192110934232643,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.3625,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.22415525802980377,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.3782,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.22390153628025627,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.3623,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.22345367006627032,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.3683,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.23335962028484913,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.3707,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.21764921047321947,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.373,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.22035045094590264,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.3617,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.3958117284390404,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.4114,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.21564381548786427,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.3775,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.21762473260352933,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.3759,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.22396992996404982,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.3677,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2076096526398509,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.3624,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.21000873581699056,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.3611,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.21187112554644466,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.3699,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.21195018799701712,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.3627,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.2158726701462431,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.3732,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.21730238749660358,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.3775,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.22862881544836197,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.3645,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.19866229858637613,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.3725,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 1.0364615815839568,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.3773,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.21898648115341318,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.3656,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.21082949217425564,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.3599,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.21676028562518138,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.3632,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.22003036783638125,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.3658,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.20409657631521416,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.3734,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 4.50653595821962,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.3751,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.21376118875125352,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.3839,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.2179016329214353,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.3716,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2269380607378638,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.3613,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.2217898543076479,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.3709,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.22038184283441759,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.3764,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.21599694355455304,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.3822,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.2056934587211255,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.3703,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.21441368448845335,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.359,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.2088935151558965,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.3772,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.21107351724500684,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.3648,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.2210990269864139,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.3587,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.22136300030235437,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.3627,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.21788565913022992,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.3675,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2213799827702262,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.3567,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.22222107595149357,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.3727,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 1.0082440086465838,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.3814,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.21081858400379053,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.3765,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.2222613199674564,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.3765,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.21477275806441476,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.3655,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.21107431789081268,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.3779,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.21324807329095025,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.3584,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.20938214375650624,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.3708,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.2244021265370967,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.3761,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.21093987784241283,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.3707,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.20699424023734914,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.3728,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.207451774206176,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.3735,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.25051593799884375,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.3732,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.21966860370418362,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.3682,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.20195655146966954,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.3819,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.22366224154710948,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.3717,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.3736202207516026,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.3755,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.26545057104872216,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.3733,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.21516942654868018,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.3622,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.26088135769356063,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.371,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.20790545072153266,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.379,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.2162359015236616,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.3672,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.21521605383085043,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.3683,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.20631027045420247,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.3745,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.21228671144262148,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.3616,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.21816208428421455,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.3673,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.2107502706457402,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.3675,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.2064686331927141,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.374,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.20567289437504185,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.3763,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.21627406432758467,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.3616,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.19885443850986173,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.3704,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.22710711979441625,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.3495,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.21594897534729293,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.3614,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.21773755944644818,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.3566,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.21808076800950651,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.3596,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.20486812526959647,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.3612,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.21622951767279908,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.3675,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.21494311395173019,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.3669,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.2180178463339316,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.359,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2211230755658558,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.3769,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.21658251699170122,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.3744,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.2148410110287674,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.3541,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.2189051502969592,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.3691,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.21832312204640256,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.3656,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.20418923162011493,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.3764,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.21807246165898325,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.3846,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.21988967655414773,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.3511,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.21348400248566682,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.3704,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.21595945897844238,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.3595,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.22011737080731825,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.3638,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.21399818486818087,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.3622,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.2179851957049784,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.3623,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.22058537967918748,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.3705,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.21868237645286617,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.3701,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.20848667668284737,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.3789,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.2001103729669512,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.3699,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.2157174414877935,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.3587,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 4.249682276228918,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.4603,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.21479041542495328,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.3701,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.21705859452856432,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.3653,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.21237344425073007,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.3743,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.2075979496777426,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.3658,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.2087004309465195,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.3672,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.22061937202880189,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.3678,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.2047747614572636,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.3739,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.20877704992714047,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.382,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2089377183694615,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.3797,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.206797316940726,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.3788,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.21415566122832683,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.3684,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.2224899039904018,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.3772,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.20093921573836504,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.3765,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.21390894506721894,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.3769,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.2028037361980611,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.393,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.21282283233143456,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.362,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.2167687796243451,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.3687,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.21606177343835378,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.3701,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.20477337937293544,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.3612,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.2293406434834965,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.3715,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.20939315824048801,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.3661,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.21345867931421356,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.3646,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.19941439560254662,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.365,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.23625833620225278,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.3527,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.20846123530134344,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.3623,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.20500893289994523,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.3733,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.21782786819043917,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.3618,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 1.3583270038617354,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.3842,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 1.2969579563044564,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.3848,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.21901713504589676,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.3631,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.21512376657004412,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.3714,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.20009224319877575,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.3699,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.21852864666817748,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.3693,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.2225904658791034,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.3625,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.20566783681180908,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.3759,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.2071264348448113,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.3748,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.21966588277318014,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.359,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.20738862786477388,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.3687,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.21194130358388497,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.3724,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.20015195290491472,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.3703,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.21849157925786233,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.3727,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.20315227222952445,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.3817,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.20519865979177,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.3771,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.20758050660888655,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.3585,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.20730520072634412,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.3645,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.21133858314215384,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.3738,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.20860144409219658,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.3702,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2192109903333941,
+      "learning_rate": 0.0,
+      "loss": 0.3676,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 3.9653987591515013e+18,
+      "train_loss": 0.49109085100865074,
+      "train_runtime": 32140.9544,
+      "train_samples_per_second": 4.916,
+      "train_steps_per_second": 0.038
+    }
+  ],
+  "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": 3.9653987591515013e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}