diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 7.114226976307344,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0268,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.823243252425845,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0074,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.93550360590846,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0042,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.843720268311807,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0108,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.5681210634936855,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.002,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.27335459746003,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.9782,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.761329461643682,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.9519,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.421261307865699,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9544,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.948318244576834,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9195,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.808252057239988,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9215,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.6632350101566096,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9039,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.463488126754135,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.8661,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 3.669496087588936,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.8709,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.312072095426366,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.8559,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 2.893706048239516,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.8491,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.2545544356772025,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8102,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.246849550454957,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.7859,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 1.751883737915054,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.7627,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.2953233988349793,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.766,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.1584419109679023,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.7422,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.19948585321261,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.736,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.150439582446643,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.718,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.0639970611610798,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7321,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9820003062371879,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7003,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.9011953279906983,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.6935,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.7962694039191154,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.6948,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.8393685095759259,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.6747,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.6985665100676669,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.6641,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7386708152794241,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.691,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.6899694779465942,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.6737,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.6746680843229063,
+      "learning_rate": 1e-05,
+      "loss": 0.6624,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.658077586360844,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.6584,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.6210412060328028,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.6515,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.5624162937960794,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.6424,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.5239430813800069,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.6362,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.5515333962279187,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.6399,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.519620801601042,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.6308,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6592287723888879,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.6299,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5811365498651043,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.6294,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.7672532018565443,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.6259,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.5840280757208299,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.6238,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.6160889988752574,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.6212,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.7275955989372745,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.6117,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.5520121928907699,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.6292,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.6823682194498206,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.6168,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.5908613201343594,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.6184,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5148119868588893,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.6193,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.5340052316028103,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6258,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.538994685016397,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.6095,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.6423403569905928,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.5946,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.526956219193085,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.6132,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.7469826585849013,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6093,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.671506817237587,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.5891,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.7757361843010703,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.5826,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6724011851661684,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.5923,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.7924362768394398,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.593,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.7145765033108378,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6003,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.6073808427658355,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.5856,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.8579912340307408,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.587,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.7020931646910384,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.5929,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.733981680975229,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.5843,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.7909671217388238,
+      "learning_rate": 2e-05,
+      "loss": 0.5732,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.6559768926708804,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.5905,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.9710270448398912,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.5874,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 1.1550341472854284,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.5885,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.647616671288994,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.5833,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.6721617036455116,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.5784,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.9995249143233349,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.5836,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 1.4354056287709995,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.5819,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.7702045983540574,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.5851,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 1.6665497000921767,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.5806,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.9476075162522167,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.566,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 1.574891578059407,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.5801,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 1.1850367690336496,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.5854,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 1.6751938830536084,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.5697,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 1.2103551809870419,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.5616,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 1.4810585704013224,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.5676,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.324542247259397,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.5677,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 1.5587261291103458,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.568,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 1.322856179711789,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.5673,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 1.6217685638033166,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.5574,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.3184499657050164,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.582,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 1.2714681688954323,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.5675,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.1711992019002215,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.5649,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 1.2026658545812778,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.5683,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.8697933039016337,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.5577,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 1.106883141521429,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.5682,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.9917765883061842,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.5602,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.5359561273467894,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.5684,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 1.021032149030094,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.5659,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.2676987726332511,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.551,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 1.4630830497648293,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.5653,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.8723048799814457,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5541,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.9067975623237527,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.5422,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.0759604433270291,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.543,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.8256245150198112,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.5773,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.5205567333813557,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.5596,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 1.557498826947946,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.5425,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.3836048750592889,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.5528,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.2253880571003868,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.553,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.3954808564080559,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.5517,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.9884958816954522,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.5534,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.7086779098755804,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.5555,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.4038364601472366,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.5592,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.5189107839949154,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.556,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.4243047793889483,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.5369,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.2528107668489161,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.543,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.6005395456370943,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.5249,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.2410923408270853,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.5411,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.3649693713428568,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.5393,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.3244687979064798,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.544,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.9565985448080758,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.5401,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.3464865319861343,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.5316,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.1318965076638179,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.5345,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.9403899544665524,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.533,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.3551697986166693,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.5463,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.772986439269267,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.5298,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 1.302655294443736,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.5522,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.9812954910127641,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.5292,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.1982831942661312,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.5446,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.8527830470866056,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.5438,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.2598351235728356,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.5362,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.9224354056153569,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.5386,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.8577517458092544,
+      "learning_rate": 4e-05,
+      "loss": 0.5292,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.020003884520917,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.5411,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.2109954565743017,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.5306,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.8746902957567236,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.5328,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.975903732515012,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.538,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.3035629393293395,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.5175,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.8776502620304968,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.5251,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.3886878550381356,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.5228,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.1824658541766833,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.5379,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.8448210034039814,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.5067,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.006663579895604,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.5331,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.1388405183196237,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.5142,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.706343868506072,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.5079,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 1.068319664595277,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.5289,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 1.092989877125497,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.5114,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.9062169653279493,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.5161,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 1.0231139982791657,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.5054,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.7392135753722859,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.5253,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 1.0554620900437193,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.5281,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 1.0794755425211011,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.5217,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.850262465208569,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.5082,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.7641732486937696,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.5042,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 1.3015014326243848,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.5241,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.6809519546085684,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.5018,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 1.0465862400154935,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.5115,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.8765782980525759,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.5161,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.9223585001640474,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.507,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.7116643234347547,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.4906,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.7843481957403297,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.4979,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.8365137290708132,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.5114,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 1.028680359067452,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.4999,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.5170372448584647,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.5033,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.9378429424634904,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.5178,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.58208609942589,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.4955,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.7120540849016272,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.4932,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.7297327854114466,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.4881,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.6831260281588274,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.4902,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.7645417829719593,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.4876,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.6867540650360291,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.5013,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.7817373084603998,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.5038,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.6284370384153564,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.4962,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.5738235210570166,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.4948,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6099430530869066,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4893,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.6082159738006128,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.4809,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.6443047084941422,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.4903,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.6386693445591629,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.4812,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.7899340397469603,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.473,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6307436749646044,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.4913,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5361227086660292,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.5,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.5696739393791692,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.488,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.6790320827056937,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.4642,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.596046709301768,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4934,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.5663493674891897,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.4823,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5518691689290341,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.4617,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.819692155335645,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.4863,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.7676350478235296,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.4752,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.6411257369581546,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.475,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.6538677609343855,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4648,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.766216414782569,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.4694,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.6846483302038855,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.4815,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.6205456235032429,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.4698,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.9007422203704472,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.4732,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.8010313526445725,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.4535,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.6060490928505674,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.4723,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.8905837736601526,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.4701,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.7488531970376175,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4915,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.6712929575936696,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.4607,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.7343568864572043,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.4892,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.6227749081802506,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.5023,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.7275156339838091,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.4684,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.5996392126980136,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.4609,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.7213434107842805,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4597,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6112516246637231,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4629,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.736397196712402,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.4733,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.8218944084129113,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.4499,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.5701509604760082,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.4607,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.7127900270453104,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.4592,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.8497835144469892,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.4609,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.5394972561813497,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.4431,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.7194657121076055,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.4638,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.7171452173513965,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.432,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.6760291090493512,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.4448,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.8308682455416495,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4432,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.47861434872262987,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.4629,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.7792256672319722,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.4633,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.6069676684215779,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.4423,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.6030522958236488,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4415,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.6745066593124447,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.4501,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5743662448721891,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.4516,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.6214999287774493,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.4467,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.6318245875557705,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.4177,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.7312069332981894,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.4226,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.6965257361777851,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.4341,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.6105040725492799,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.4443,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5561574512438544,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.4331,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.6906509889972667,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.4272,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.5848007708754247,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.4589,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.7188535825081847,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.4197,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.5007951350613536,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.4415,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.699534703128445,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4353,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5067010396061123,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.4484,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.6568899847014354,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.4422,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.5557745042842687,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.409,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.9299582088348566,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.4156,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.5783315163510357,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.4157,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.7227613924590878,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.4406,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.6255824158614407,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4335,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.5056910043550916,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.4313,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5444043491927443,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.438,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.5098201033353602,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.425,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.4922711652344555,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.4313,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.6920620279978467,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.4173,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.5437203837524964,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.4193,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.6444870940062666,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.4261,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.6548081594973988,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.424,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.5288313383967759,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4237,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.49835770508955785,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4179,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.5142495339910189,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.4192,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.5620710706482116,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.4355,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.6981438183426263,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4319,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.5454001954639964,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.4214,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.7696218850782683,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.4218,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.6207474017516889,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.432,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7682251898259739,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.4182,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.6757476467424128,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.3506,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.6192726622265251,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.3582,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.7647115347660998,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.3607,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.6413940942565369,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.3509,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.570334729178281,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.3429,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.8583584372486077,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.3408,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.9033627700916018,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.3646,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.7200201376571489,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.3426,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.5938265226157662,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.3414,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.7015854952824675,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.3469,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.662950638194443,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.3407,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.5519645840901345,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.3628,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5392336854373473,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.3566,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.5796011483370286,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.3625,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.5070024528575735,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.3525,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5874335681340777,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.3624,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.5192432773222412,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.3522,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.4755528027145419,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.347,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.6087344390533755,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.3532,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.5379869053822307,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.3334,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.5641887788405713,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.3244,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.7120956013784201,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.331,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.654894376306451,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.3464,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.6169878723663599,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.3296,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.6604380370338732,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.349,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.7145862642833642,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3594,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.6883251136057624,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.3489,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.7981050399206261,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.3393,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.5993747917832551,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3439,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.6191498689073499,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.3321,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.49121566458827065,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.3155,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.5440030285520485,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.3231,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.45527830051069346,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.3115,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.5127036775817909,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.3397,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.5012674111027076,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.3225,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.5891766884716853,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.3387,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.6057364721635156,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.3242,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5747251546148945,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.3408,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.6215786984764327,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.302,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.5326553368978214,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.3334,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.511205011848968,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.3626,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.5668350324392251,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.3246,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.4624022243696653,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.318,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5686974633110563,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.3169,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.6175141297652048,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.3205,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.41172964457538996,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.3068,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.712658886515756,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.325,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.6306542096952428,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.3323,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.5863855520141114,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.3042,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.6130433112716517,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.3284,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.5865294386601786,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.3169,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.6232704657283656,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.2888,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.5275808456363389,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.3192,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.6897315583045173,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.3478,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.5023313861899064,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.3277,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.579683431309866,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.3274,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.5366422048896198,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.3302,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.5229097905193483,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.3173,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.61931815536183,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.3062,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.4634027791684282,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.3162,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.541456767692388,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.3088,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.5726217211609801,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.3141,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.7303280892902416,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.2965,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.4710089509209088,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.3028,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.5556228136788505,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.3032,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.5273296236368723,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.3294,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.5160202639071401,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.3291,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5066055972127445,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.3246,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.5253763387565633,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.3156,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.6065849676431206,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.2967,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.5063130119244295,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.2938,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.5463985839834996,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.302,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.49087329314414296,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.3225,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.47977924495944985,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.3066,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.4769738484405003,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.2927,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.47342789639316685,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.3225,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.3876745519596092,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.2846,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.5051560653337944,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.297,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.44297009524755354,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.2908,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.5025210241718157,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.2845,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.5209087409530332,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.2825,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.5058295446933186,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.2717,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.5889459649708597,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.3111,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.4667810233832782,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.3001,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.608338541363304,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.3039,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.5418132922168261,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.3119,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.5614340210499108,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.2899,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.3831858980046882,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.2834,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.5584672116240562,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.2805,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.4490964545987542,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.2905,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.5424385691923251,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.298,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.45135662876260085,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.2841,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.5431152390155998,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.2762,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.5810772435434395,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.2753,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.38543144498595455,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.2674,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.5429967068455626,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.2849,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.4531447807999564,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.2683,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.5998460965240968,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.2894,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.526252861055792,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.3051,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.4812231292410862,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.3029,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.4404853594502609,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.2901,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5157160970126969,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.2917,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.4054611234470367,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.2752,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5425230807552287,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.2811,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5095885364767981,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.2717,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.3917973270620053,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.2857,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.4796345969781351,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.2794,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.4097103211324211,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.2804,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.4693781852676285,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.2815,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.46569143373144145,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.2835,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.4663287522468997,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.2819,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4477546327643259,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.2761,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.46612807832613706,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.2733,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.4198164029965005,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.271,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.4390543143413172,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.2701,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.4424993169464221,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.2708,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.5001309263441134,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.2764,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.4025118986925197,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.2647,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.5344820558384235,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.2798,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.44456306721274086,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.2327,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.46184990060248937,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.261,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.49996215101557856,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.2665,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.5328333398520886,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.2695,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.5412561248356805,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.2741,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.5952730479043927,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.2588,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.5216513085084413,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.2712,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.5641482098517344,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.266,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.482184258845709,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.2699,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.4136921477235899,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.2612,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.5364069806905752,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.2498,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.4979804911273512,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.2511,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.4828686276560079,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.2762,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.5241286161738753,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.2523,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.4299437859971156,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.2541,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.4368515039960552,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.2618,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.49277860053294026,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.2662,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.5229215831117131,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.2633,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.5068085127981562,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.2768,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.5590901649145359,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.2674,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.5570545977094467,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.2582,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.4876295337913524,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.2476,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.5439532478978393,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.2293,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.4995059884449875,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.2702,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.4233762568758728,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.2295,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.46757193942853364,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.2375,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.4593119441950135,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.2318,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.44103267751487135,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.2549,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.522326038673655,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.2667,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.443704850935825,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.2756,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.4717688631207691,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.2579,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.46990625861713947,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.2634,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.413932199403329,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.2437,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4501549472888603,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.2636,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.44670593949266135,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.2478,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.49886620495994655,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.2346,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.39884388093883416,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.2475,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.44012556390355617,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.2496,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.4519591880323006,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.2516,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.43164510040448706,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.2396,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.46863644580850383,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.2382,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.40373189450018865,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.2494,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.46624995722499263,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.2246,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.4157909731708204,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.2289,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.4379445170171008,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.2459,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.4083558505606154,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.2298,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.43691318435851545,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.2462,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.4147928610791711,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.2111,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.5051364813022657,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.238,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.39874385049299704,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.2132,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.4282696134930731,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.2239,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.43692109825923375,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.236,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.3941934803000771,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.2176,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.424255784112043,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.2197,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.3917499739917114,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.2108,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.40529611852405384,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.2192,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.4079991476512434,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.2215,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.40680636875712134,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.2269,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.4086322154428503,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.2409,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.4338109585678527,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.1945,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.41252102493001436,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.243,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.4197001301956176,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.2053,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.4274446597118927,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.2314,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.39388607387840013,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.2324,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.4076561748208201,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.2,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.4145244385338656,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.1969,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.38256841700149824,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.2192,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.4220268053325287,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.2115,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.36510681835174513,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.2116,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.402410745123727,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.2102,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.41314909871534494,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.2158,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.4048565641777879,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.2125,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.395135435214346,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.2005,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.456937102541613,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.2452,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.38309938792439147,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.2146,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.4196808738124701,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.2058,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.38111691568884126,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.1926,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.42372204202172115,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.1963,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.38534459631221596,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.2015,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.4205960588622298,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.2122,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.36597166808516024,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.2079,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.41649404458002515,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.2132,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.43636732874001116,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.2105,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.3946601455120183,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.1812,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.45122784036755986,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.1925,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4017796271537724,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.1982,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.4791137733187544,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.2012,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.4667791635434564,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.1967,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.37705022019118706,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.1773,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.43164889601131773,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.1709,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.36680239324332387,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.2,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.3887785116759851,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.1944,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.43787661610102874,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.2013,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.39640577025510165,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.2001,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.38094278633296097,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.1825,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.3948830944145819,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.1958,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.39442565200435264,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.2079,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.3827699605019993,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.195,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.41033195729094285,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.1887,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.3916418345544464,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.2024,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.3868124892663438,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.1722,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.43025588786610836,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.2029,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.403407032649107,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.1996,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.3802039448415926,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.1905,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.39931768629259595,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.1721,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.38270855366689566,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.1747,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.37901183642762165,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.1821,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.43784704919474543,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.1979,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.37959588483405504,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.1753,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.36867588129039525,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.1899,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.3948701855822849,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.172,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.349543344245847,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.1709,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.40145293217980926,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.1756,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.37289636496455425,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.1714,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.3673694955856495,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.1774,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.38729204116940014,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.1731,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.3714130737296065,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.1804,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.3636080024274385,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.1904,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.3641183986492469,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.1601,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.3985934173737231,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.1757,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.3652039314712971,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.1712,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.3821997037990993,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.1644,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.4045431951400469,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.1629,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.36133172604898756,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.1714,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.41646722992521573,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.1737,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.36498916268069237,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.1688,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.38800847394726357,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.1641,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.3887611189469426,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.1705,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.3952124189790676,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.116,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.3241696587519347,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.1134,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.42180592345514933,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.1056,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.35539405621392195,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.1135,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.34150354892172513,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.1103,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.3827199490930834,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.1047,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.36903353065843664,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.1195,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.35903348730314577,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.1208,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.3892181020202926,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.1047,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.35508542189786735,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.105,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.4058138905240334,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.1181,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.3416260485281971,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.1055,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.33984983324938645,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.11,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.3618563598144018,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.1026,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.3730555240194491,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.1145,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.38729211166000815,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.1041,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.393236627545051,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.111,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.3539037624102464,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.1052,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.36909680054552174,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.0844,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.40060677994613286,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.0982,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.36441658074215694,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.1142,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.36734236537693904,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.1003,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.37142658800397194,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.0994,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.40743443420406295,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.1195,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.3922263018625202,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.1148,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.37643246123328455,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.1185,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.3761566508289387,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.1033,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.33963929431175977,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.1126,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.37790922685950795,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.1076,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.3564804998540527,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.0978,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.3292813282938648,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.1042,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.365520259177511,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.0991,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.36113534555694266,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.0997,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.34456762470662133,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.0973,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.36589987150415465,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.0927,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.3620926112614718,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.0931,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.37468979004135844,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.1049,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.3471711781948294,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.0935,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.3537859957617662,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.1078,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.40125566325443135,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.1109,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.3574488953813267,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.109,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.36063671070033343,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.1032,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.41137237689792633,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.1128,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.36294980875537874,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.0957,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.3448057786869242,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.0963,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.37026667627364274,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.1134,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3193770390092366,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.0983,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.3904364416806637,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.1021,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.3504990237115117,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.0923,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.3456402594171997,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.0828,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.3651861963259521,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.0997,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3273058467797983,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.1143,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.352798326062349,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.0908,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.31962058305238183,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.0915,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.33310720342953776,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.0931,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.3294897455959873,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.0807,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.3745238287725312,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.0914,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.31837711397151736,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.0913,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.351985839314606,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.0956,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.30918141263875004,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.0913,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.37062453489482,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.1035,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.3220857617964363,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.1014,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.3613386305490762,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.1068,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.33166560253768623,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.1054,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.3680294933362338,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.0945,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.31376510029041443,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.0915,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.31861683944405383,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.0809,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.35009134888859883,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.1024,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.347766289786633,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.0919,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.32670057057038343,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.0892,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.32003648597112055,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.095,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.32277936700552273,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.1032,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.31871867817026195,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.0844,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.33583794851067933,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.0867,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3164292913977101,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.0969,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.30309145364967477,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.0857,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3228924663618951,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.0835,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.33744682701267076,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.1098,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.31835334506450713,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.0993,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.28822873754161954,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.0877,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.32678145724253865,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.087,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.2868506864370932,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.0786,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.3593876046490686,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.0831,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.31049921924119644,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.0916,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.33521225974997393,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.0779,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.31997302472736855,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.0912,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.339686286374968,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.0854,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.3286261838727106,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.0892,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.33165430152733816,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.0817,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.2853349548601994,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.0701,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.35044120273338364,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.0924,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.33371563570824203,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.0876,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.33267203490696434,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.0893,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.3427945731617254,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.0869,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.29951114177205257,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.0789,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3460799448381564,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.0931,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.3228975488622833,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.0968,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3569770346590971,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.0895,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.33467428857552306,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.0791,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.3106661366680819,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.0839,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.2970921072367564,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.0825,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3028745451345777,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.0741,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.28783160648586376,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.0713,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.2937462647664457,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.0661,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3062581212850377,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.0824,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3093447064336362,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.0808,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.2883886543935193,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.0708,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.3022318774010405,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.0556,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.29538185995080823,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.071,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3173641983273913,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.07,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.3101021793938913,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.0832,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.30283086750616706,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.0622,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.2984869491648959,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.0711,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.3192571006834449,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.0866,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.3016149140544656,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.0681,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.3246478352761254,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.0845,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.2954248372762179,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.0686,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.30559518136984426,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.0655,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.2901548700448351,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.0612,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.31523179195214523,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.082,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.30750697450079684,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.0718,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.30634644012855455,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.0753,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.3230740456264466,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.0792,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.3000771284006283,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.0784,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.29345397453024735,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.071,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.29030241071599183,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.0656,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.2851102191629578,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.0698,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.2944197259468657,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.0708,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.2887165746515513,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.0609,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3074861223113701,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.067,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.27402711408407315,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.0578,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.30589823580945746,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.0591,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.29596360163221175,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.0721,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.30927083511641645,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.0663,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3053735970528671,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.0658,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.29282847761283437,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.0704,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.2964384981312799,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.0623,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.32594727508339866,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.0589,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.29499903325349686,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.0746,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.30189314628924535,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.0639,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.3175006154094282,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.0669,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.2892804391362886,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.0657,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.2966798634180088,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.0716,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.2910781084556746,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.0662,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.295168470851888,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.0646,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.3103757354898736,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.0625,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.2978176049321637,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.0569,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.29138467372509347,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.0633,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.3403119882921393,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.0664,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.2899736488454072,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.0623,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3075989057382275,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.0666,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.3250610638896431,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.0643,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.2818783723636903,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.0637,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.31330925238626733,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.0557,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.2785495939546398,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.0603,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.30248080307360786,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.0561,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.31155193985016844,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.0711,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.28760015709169456,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.0597,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.287700495695342,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.0622,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.30735016150951544,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.0492,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.29270040884109333,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.0525,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.29362337730480353,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.0556,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.2894475391362164,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.0561,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.2827665693166563,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.0549,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.2853717366397669,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.0573,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.2682674415473706,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.0633,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.27278526160005484,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.0512,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.27974569626862283,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.0624,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.2826719563451568,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.0461,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.2812005588886208,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.05,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.27632320170816954,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.0616,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.2840324580582358,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.055,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.3000895525187686,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.0557,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.29758548181259653,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.0549,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.28231249284718524,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.06,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.2766233657370569,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.054,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.2897094773790861,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.0509,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.302855694860025,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.0519,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.2517095924821794,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.0461,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.32292647564820065,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.0692,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.2663456233924653,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.0464,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.27835666620802557,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.0566,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.29949087830290216,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.0464,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.2631066673589524,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.0505,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.27987411912951693,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.0634,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.29256169344101013,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.0547,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.2934681865867812,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.0576,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.2576999121673978,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.0515,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.26475952340118064,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.0455,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.2477520610467691,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.038,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.2629951584454191,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.0531,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.26487118057473014,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.0504,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.27037157687598384,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.0554,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.26885562461509205,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.0527,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.24629712651371335,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.048,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.25862983165273057,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.0411,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.25934796081359,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.0446,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.2549879550260934,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.0462,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.2683736967655737,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.045,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3652181891457886,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.0506,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.2541210678284303,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.0384,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.2640790948829442,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.0466,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.2537350570032988,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.0451,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.26337773108893303,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.0427,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.27830182413772475,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.0411,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.25818610965386796,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.0455,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.24833952587313618,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.0396,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.33728491314831144,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.0481,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.2745495420752194,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.0402,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.2698103826666054,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.0492,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.25205673320875793,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.0454,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.25158600519888646,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.0421,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.25624582046391253,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.0487,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.26270400260246457,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.0466,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.2602539355908384,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.0395,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.24778699843972443,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.04,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.2704562603415123,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.0451,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.22808638579397236,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.0339,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.2462375597935742,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.0403,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.25880027635483827,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.0456,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.25599474818999607,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.0428,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.2449158745149275,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.04,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.24416842923951962,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.0438,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.23675978013073154,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.0386,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.2551204729686027,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.0414,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.24363072828617746,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.0388,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.25513973452508715,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.0356,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.2405989403185756,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.0388,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.2381011436184982,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.0352,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.25306555068365993,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.038,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.23696524517247844,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.0363,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.24916365112816224,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.036,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.23166830614112932,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.0326,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.24589656764251955,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.0344,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.2406853345852265,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.0377,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.24859047148540928,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.0344,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.24088935874395612,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.036,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.2403514320419359,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.0343,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.2505180906407298,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.0463,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.2296718186656429,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.0322,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.2376295594309593,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.0309,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.22860678935025194,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.0312,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.24438274423961331,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.0414,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.233209511061214,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.0285,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.25708821844523905,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.0366,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.23056147910592747,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.0344,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.25944046442985036,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.0314,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.20541720106884642,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.0181,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.1874915736895638,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.0204,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.19082586256295334,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.0153,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.21736722231825378,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.0201,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.22463002857947367,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.0166,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.22093999155217492,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.0205,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.19734065232694128,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.02,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.20471806262643952,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.0168,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.1979253519226936,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.0174,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.20158650205817535,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.0222,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.19172372661597434,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.0175,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.20340546014763236,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.0206,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.20284307701563775,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.0165,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.201813066089805,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.0187,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.20525952384645144,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.0233,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.18554935196813777,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.0171,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.1861759055796356,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.0181,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.18417919408176583,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.0126,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.18641551522402575,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.0157,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.18390943875092172,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.0175,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.8204241439525604,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.0203,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.18635578154643315,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.0154,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.20231611603230013,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.0202,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.19299412411445352,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.0196,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.16431923624749,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.0126,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.19435857488264346,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.0185,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.17966783966882796,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.0143,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.18469700874485748,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.0155,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.17466127173656593,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.0136,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.1826905591738935,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.0167,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.20524434487007434,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.0185,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.1959902764051585,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.0158,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.19510902408133157,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.0153,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.1961273759619965,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.0204,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.17575560699061296,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.0124,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.16969086833435676,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.0118,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.19741009197812825,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.0138,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.19374275184608897,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.0147,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.1791062341894546,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.0136,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.17639171050454994,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.0148,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.23791071337114741,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.0155,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.17688024682039608,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.0191,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.19166068871455344,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.0163,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.17749860284149233,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.0139,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.18027981476605198,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.0143,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.1666277574355885,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.0114,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.17455542206982738,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.0131,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.18465120468821397,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.0158,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.1866287769633372,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.0143,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.16557207352338543,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.0127,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.18379579579646277,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.015,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.18559021333530712,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.0139,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.18377646707514542,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.0153,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.17048387558715242,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.0123,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.18181950707571834,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.0161,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.1703751004390071,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.0177,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.18142793799867132,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.0159,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.1691999304200095,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.0119,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.18184524969830543,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.0128,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.1774105310996861,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.0146,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.17319208667851183,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.0142,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.1775183268455283,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.0147,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.16495329784474672,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.0138,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.18491645969061338,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.0149,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.17148479034656244,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.0173,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.18337691917014678,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.0164,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.17902734578647087,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.0145,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.15890637783446931,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.0111,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.16052352839969514,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.0119,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.15624641385195934,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.0105,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.18521672235293274,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.0144,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.17521287984879513,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.0126,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.167710677762983,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.0123,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.17189222491675554,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.0128,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.16989850836590212,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.0116,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.17071994499537765,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.0135,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.16299985388564267,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.0125,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.1727770395071183,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.0142,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.1636296726091575,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.0135,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.17028181559727848,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.0121,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.1661008710636989,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.0099,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.163829994870747,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.0122,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.18180254536321855,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.0161,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.16596524860586478,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.0107,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.16040797513544575,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.0097,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.16466472111754973,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.0119,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.1540333387516671,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.0108,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.15124183237479688,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.0093,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.16701535437610113,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.0147,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.15098212255677187,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.0107,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.16713493694631018,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.0134,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.16894365479644607,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.0133,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.1746515335370343,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.0118,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.15667635871223878,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.0117,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.1444736385852286,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.0086,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.17749436519159156,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.0118,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.1748116345676669,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.0128,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.15683240091532896,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.0123,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.15288229419511515,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.0132,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.13962227493214396,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.0057,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.16169058838535835,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.0111,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.15828537261838552,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.0115,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.1665846193183141,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.0097,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.16470252798919066,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.0103,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.158540761541256,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.0107,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.15884576719202173,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.0121,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.15185110335480456,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.0098,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.1593913888133789,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.0124,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.15624235227758673,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.0113,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.1574788135352717,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.0123,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.15201837014862982,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.0105,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.1576061794548123,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.0097,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.15731150132488764,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.0108,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.1500051053584008,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.0106,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.1532375550712184,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.0077,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.159494762572442,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.0119,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.1540465027541923,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.0109,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.15668428698912867,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.0108,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.15727064750934963,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.0094,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.15213783924430885,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.0102,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.15090495538995116,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.0124,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.14660718974961626,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.0095,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.14049895208582142,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.0068,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.14125967405773518,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.009,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.1544360592807957,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.0111,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.15610533669705137,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.0082,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.1490015038216058,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.0134,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.14929717251895375,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.0111,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.1233061207785064,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.0055,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.14230086129144062,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.0084,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.14455781630248446,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.0082,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.15208167325639851,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.0092,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.1465468856766422,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.0108,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.147667365337004,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.0088,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.14101728945760172,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.0102,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.14441096280067506,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.009,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.1413680533793005,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.0095,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.13234916126224888,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.0059,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.17032467064431386,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.0124,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.1414403117396058,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.0086,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.14326997520060672,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.009,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.14667165011645114,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.0092,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.1585418274779253,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.0119,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.14656936219538133,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.0095,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.12829025890171505,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.0073,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.1397156072246295,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.0082,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.14901094295425626,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.0107,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.14993569150525063,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.0078,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.13466186229710642,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.0081,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.12395639194245477,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.0052,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.13234773023489677,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.0064,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.12847149385386814,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.0063,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.15920794709405744,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.0121,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.13701333868738236,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.0086,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.11628425496723656,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.0048,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.11870988855658256,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.006,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.11894385368079538,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.0059,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.13064294565023396,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.008,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.13941134721879073,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.0081,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.15957586295102816,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.0096,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.1297142119310439,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.0063,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.12327662782149743,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.007,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.12029315173076262,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.0064,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.13032043978497956,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.0062,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.12751666025788555,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.0056,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.12458357847577003,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.0067,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.12848056873677116,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.0075,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.12169823002637754,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.0056,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.11390521249288205,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.0053,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.12820193920233866,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.0083,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.11677132451464077,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.006,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.11699310242913259,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.0057,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.12529987346188193,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.0069,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.1261008001251447,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.0059,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.12754525719239856,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.0063,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.11440413388620226,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.0056,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.1353941863907901,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.0086,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.11992101001087817,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.0066,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.11433795864648098,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.0075,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.12387431239637017,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.006,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.13077223679665037,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.0063,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.12026799504189024,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.004,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.11232202394910713,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.0062,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.10946139415683633,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.0049,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.10540108736494803,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.0051,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.13802896422812902,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.0103,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.11809346470819365,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.0048,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.11003920075259967,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.0053,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.11585305769581744,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.005,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.11731019253446567,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.0055,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.10827519905741839,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.0048,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.11391053405315292,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.0045,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.10707374509261255,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.0043,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.11413042861187131,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.0051,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.15502753671651753,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.0069,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.10546762022485931,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.0041,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.11211670289366113,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.0055,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.1032832660807849,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.005,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.10506293712500865,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.005,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.11476362755932457,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.0053,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.10943980097974326,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.0048,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.11285527467713659,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.0048,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.10699166721103276,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.0053,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.1097419377294878,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.0057,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.11115870052950595,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.0063,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.10512127418572167,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.0052,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.10352273335117232,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.0041,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.10570196402647787,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.005,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.09814077907548227,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.0038,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.11559929313178484,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.0073,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.09673374846874727,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.0042,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.11249116810591006,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.0057,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.09231356969492527,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.0041,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.08691400138292399,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.003,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.09784615360429627,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.0056,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.09364600909645075,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.0039,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.10056744913811692,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.0035,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.08962786001746821,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.0042,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.08635208470675917,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.0035,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.1215800179443063,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.007,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.08459021227324545,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.0041,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.09787863426204857,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.0047,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.10668127605525157,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.0041,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.09662544909071354,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.0038,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.09770178292591185,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.0053,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.09614921051731116,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.0036,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.0830453824555553,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.0037,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.09282327580949633,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.0051,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.09121777031854007,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.0035,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.09644805653841083,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.0045,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.10934980753448524,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.0058,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.10008177001814868,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.0035,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.08067843396765995,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.0031,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.09693975384084563,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.0033,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.0900711417595501,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.0042,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.08917829028654524,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.0033,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.09583734719927275,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.0028,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.08099308408417666,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.0032,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.07339817411009371,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.003,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.09856135718889049,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.0042,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.08482278055094322,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.0036,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.0979953700352469,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.0036,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.07705126365152701,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.0028,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.0848041027536405,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.0035,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.07528118269969317,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.0023,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.07849380413537242,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.0033,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.07298989088714364,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.0026,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.05204296700286365,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.0016,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.05773893784371216,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.0018,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.057007410747105,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.0017,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.054034773199336514,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.0021,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.054594649915388385,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.0022,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.04925073380419735,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.0018,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.05532935791334816,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.0021,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.05835347057149607,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.0012,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.05290169731863809,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.0015,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.053140952466009694,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.0012,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.06234242808384632,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.0015,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.06098215584699549,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.0021,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.062063044421459684,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.0019,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.050626485707721665,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.0014,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.06376778832791562,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.0017,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.0599791808291001,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.0016,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.06199576350216603,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.0017,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.05990688571157757,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.0017,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.05363429701445285,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.0015,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.05087987756148815,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.0014,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.060258859258813635,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.0017,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.05398800344906819,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.0016,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.043552753377658154,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.0012,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.0620517896301348,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.0016,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.046614088425438197,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.0012,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.05426932429128966,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.0016,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.052482175666439056,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.0017,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.048085448327705574,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.0017,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.06319224455366214,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.0012,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.04901424854247344,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.0016,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.04723193866461885,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.0013,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.048101577525817375,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.0014,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.04407561492070258,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.0011,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.06756345487789561,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.0023,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.03759896955672313,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.0011,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.044849592148029645,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.0013,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.04956712854739809,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.0014,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.049998342609820876,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.0011,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.04729081172561275,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.0015,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.04715945087042008,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.0013,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.04906701528574314,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.001,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.06076666964084672,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.0019,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.04655621457882325,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.0012,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.04416252372864719,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.0016,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.055302010421376765,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.0013,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.04124301850460134,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.0015,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.0402102917819068,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.0013,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.05176781586645517,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.0017,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.05846527616526151,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.0017,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.04275673146003062,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.0012,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.05608738779848541,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.0012,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.056441068068771985,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.0019,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.05437838558411408,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.0016,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.04332061230458002,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.0015,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.04921892772700717,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.0014,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.04172884825740209,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.0008,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.051982159261104596,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.0014,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.041337717972453815,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.0014,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.042248252736725694,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.0009,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.04828359529674988,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.0013,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.044656721612912495,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.0015,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.06991274469496156,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.0019,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.04041843860665941,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.0012,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.04666190002563977,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.0014,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.042446957822767556,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.001,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.04623485047734316,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.0016,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.0508327942338693,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.0014,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.045262587709247556,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.0014,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.04876300942754564,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.0015,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.04492298489605455,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.0013,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.047546180944110825,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.0011,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.04550049063843597,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.0011,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.044540620742745604,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.0013,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.04178895698356649,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.0015,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.04368847575220791,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.0012,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.0424623147098431,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.0009,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.04626435897043272,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.0014,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.04296246625530364,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.0012,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.04356640846308451,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.0015,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.041040836248546576,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.0011,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.041990222547551234,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.0012,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.03930827507826577,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.001,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.04305249546895658,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.0015,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.03829144712276813,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.001,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.05002546782086568,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.0014,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.05499189105197184,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.0012,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.05795109891896602,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.0015,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.05161297951548109,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.0014,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.051322561636900646,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.0017,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.05378245663566002,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.0012,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.0499153589518134,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.0017,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.04351275954875249,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.0015,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.03981195278441172,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.0012,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.06306974448580276,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.0023,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.024140198562791683,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.0005,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.04677331003823121,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.0013,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.045372411404951185,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.0014,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.039266395968343144,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.0014,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.03447532317235575,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.0007,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.04815517596851518,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.0014,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.03271349939014719,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.001,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.04473128290324987,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.0012,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.049951103006735234,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.0012,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.049157761904328516,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.0013,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.03814650033045412,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.0015,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.040727055882241965,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.001,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.04248364014923922,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.0014,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.035475153418756365,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.0011,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.041979164329127364,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.0013,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.0471334720866644,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.0013,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.04911350135472233,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.0015,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.03382677792163763,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.0009,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.03741801155189026,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.0012,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.038474713921293494,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.0013,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.032118872894658985,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.0009,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.036772624439523206,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.0014,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.04346509500003399,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.0011,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.05457920972930003,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.0013,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.0433250641342956,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.0012,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.05059532675609269,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.0015,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.03263502776829861,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.001,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.046141449909282074,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.0014,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.02194179295633395,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.0006,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.023853277823948894,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.0006,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.04017932067517861,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.0013,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.04509294271207957,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.0012,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.033273683788581386,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.001,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.030155747940848476,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.001,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.033518867217713556,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.0011,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.03182650165581313,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.0008,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.030484357469737575,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.0008,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.03599200723501593,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.001,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.032779900233816925,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.0008,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.03137829557495043,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.001,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.028491328684328812,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.001,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.03926545118500328,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.0011,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.04504333451554328,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.0015,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.031478098887310166,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.0009,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.02766867421889181,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.0006,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.035536215227495276,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.0012,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.030894409867039913,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.001,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.03764038387248957,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.001,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.041314427594097015,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.0015,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.036564490088593735,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.0009,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.03057142111752281,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.0009,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.03609222527916196,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.0011,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.03661791259237522,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.0012,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.042693781340119405,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.0012,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.037569495534210365,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.0014,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.026071237805805607,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.0008,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.03763843693080899,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.001,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.043587885314857994,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.0011,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.03247303481029576,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.0011,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.0255708530936397,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.0009,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.020281074853450765,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.0006,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.03449280445073408,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.0011,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.03487655716449958,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.0013,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.034694111160568336,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.0011,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.03806408767934333,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.0017,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.02351699368585384,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.0007,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.03459974395296212,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.0013,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.02947344697122565,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.001,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.03077108831836505,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.0011,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.04198908918718181,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.0009,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.0330481164925262,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.0011,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.03298355303799536,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.0013,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.031441310851894076,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.0009,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.030843718525785217,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.0011,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.030115446129065256,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.0011,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.017375184814033714,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.0006,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.04037026592843756,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.001,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.04773283294205063,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.0015,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.03293112460652502,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.0013,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.033940376125370174,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.0011,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.02818393971440681,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.001,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.040215430730727086,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.0016,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.033101499556175216,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.0011,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.039186695253359424,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.0011,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.026516903439140295,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.0009,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.025960117430873713,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.0006,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.02846881061042262,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.001,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.029169384818429817,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.001,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.03079259564733529,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.0011,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.03874619648534124,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.0014,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.03370730395631837,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.0014,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.028348110067004738,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.0008,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.04120988678451968,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.0009,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.02844050281478281,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.0008,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.03132955901680712,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.001,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.03789727049662089,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.0016,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.027871652832050856,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.0009,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.03763496038668608,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.0016,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.028750392714171166,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.0012,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.02218648555200923,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.0008,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.023678654126213044,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.0008,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.023695695309771884,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.0008,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.02620019794202227,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.0009,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.03095385594009785,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.001,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.022805333061317844,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.0009,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.03142699214912273,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.001,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.030811480836259515,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.0011,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.027893750769134473,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.0007,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.03911121296762184,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.0012,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.03108293891421958,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.001,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.03285757041610076,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.0009,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.03842412540600895,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.0013,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.03848191620312672,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.0013,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.031522081044749364,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.0011,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.023552715108211723,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.0008,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.03980532895850969,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.001,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.03266930787295375,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.0012,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.03405515189601465,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.0011,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.022442822238414463,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.0006,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.024636597889818725,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.001,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.028141545101971227,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.001,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.03095926700537304,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.0011,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.030775415761536223,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.0008,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.03237229196669056,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.0012,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.033491740337787464,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.0013,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.03194629083360365,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.0013,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.03444569858351394,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.0011,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.022002345865948226,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.0007,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.021393060271189136,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.0007,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.028577350121914324,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.001,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.026191333966386888,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.0009,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.02478755209650598,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.0009,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.023829054928724107,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.0007,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.03724409306816833,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.0011,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.026804037017212783,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.0008,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.029808646478268346,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.0012,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.02983474000732463,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.0009,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.03462251900595818,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.0012,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.027051179205484718,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.001,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.027110753818631223,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.0009,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.034229620002890475,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.0012,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.03405431413440894,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.0011,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.04716617435361974,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.0015,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.02350667499968646,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.0007,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.02437900133400284,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.0007,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.021928928000783072,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.0006,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.029876708285934483,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.0008,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.028327985003880816,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.0009,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.030784777363009132,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.001,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.02418063454488776,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.0009,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.01972577010142427,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.0006,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.028178949257879976,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.001,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.02717233278618788,
+      "learning_rate": 0.0,
+      "loss": 0.0011,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 2.846545834932175e+18,
+      "train_loss": 0.18064658682569976,
+      "train_runtime": 25002.5539,
+      "train_samples_per_second": 6.319,
+      "train_steps_per_second": 0.049
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.846545834932175e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}