diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6342 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5002086521073863,
+  "eval_steps": 899,
+  "global_step": 899,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0005564056196967589,
+      "grad_norm": 1.3203125,
+      "learning_rate": 2e-05,
+      "loss": 1.7806,
+      "step": 1
+    },
+    {
+      "epoch": 0.0005564056196967589,
+      "eval_loss": 1.7791756391525269,
+      "eval_runtime": 1750.0366,
+      "eval_samples_per_second": 2.681,
+      "eval_steps_per_second": 1.341,
+      "step": 1
+    },
+    {
+      "epoch": 0.0011128112393935178,
+      "grad_norm": 1.1875,
+      "learning_rate": 4e-05,
+      "loss": 1.7467,
+      "step": 2
+    },
+    {
+      "epoch": 0.0016692168590902768,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6e-05,
+      "loss": 1.6933,
+      "step": 3
+    },
+    {
+      "epoch": 0.0022256224787870356,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8e-05,
+      "loss": 1.6916,
+      "step": 4
+    },
+    {
+      "epoch": 0.0027820280984837947,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001,
+      "loss": 1.6973,
+      "step": 5
+    },
+    {
+      "epoch": 0.0033384337181805537,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.00012,
+      "loss": 1.6518,
+      "step": 6
+    },
+    {
+      "epoch": 0.0038948393378773127,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00014,
+      "loss": 1.6383,
+      "step": 7
+    },
+    {
+      "epoch": 0.004451244957574071,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00016,
+      "loss": 1.5714,
+      "step": 8
+    },
+    {
+      "epoch": 0.005007650577270831,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00018,
+      "loss": 1.553,
+      "step": 9
+    },
+    {
+      "epoch": 0.005564056196967589,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0002,
+      "loss": 1.5321,
+      "step": 10
+    },
+    {
+      "epoch": 0.006120461816664348,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0001999999829570899,
+      "loss": 1.5589,
+      "step": 11
+    },
+    {
+      "epoch": 0.006676867436361107,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0001999999318283654,
+      "loss": 1.4519,
+      "step": 12
+    },
+    {
+      "epoch": 0.007233273056057866,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0001999998466138439,
+      "loss": 1.4035,
+      "step": 13
+    },
+    {
+      "epoch": 0.007789678675754625,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001999997273135545,
+      "loss": 1.4071,
+      "step": 14
+    },
+    {
+      "epoch": 0.008346084295451384,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0001999995739275378,
+      "loss": 1.2862,
+      "step": 15
+    },
+    {
+      "epoch": 0.008902489915148143,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019999938645584616,
+      "loss": 1.2801,
+      "step": 16
+    },
+    {
+      "epoch": 0.009458895534844901,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001999991648985434,
+      "loss": 1.2358,
+      "step": 17
+    },
+    {
+      "epoch": 0.010015301154541661,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0001999989092557051,
+      "loss": 1.2122,
+      "step": 18
+    },
+    {
+      "epoch": 0.01057170677423842,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019999861952741839,
+      "loss": 1.2294,
+      "step": 19
+    },
+    {
+      "epoch": 0.011128112393935179,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019999829571378197,
+      "loss": 1.2641,
+      "step": 20
+    },
+    {
+      "epoch": 0.011684518013631937,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001999979378149063,
+      "loss": 1.2373,
+      "step": 21
+    },
+    {
+      "epoch": 0.012240923633328696,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.0001999975458309133,
+      "loss": 1.2389,
+      "step": 22
+    },
+    {
+      "epoch": 0.012797329253025456,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019999711976193663,
+      "loss": 1.2716,
+      "step": 23
+    },
+    {
+      "epoch": 0.013353734872722215,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0001999966596081215,
+      "loss": 1.2038,
+      "step": 24
+    },
+    {
+      "epoch": 0.013910140492418973,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00019999616536962473,
+      "loss": 1.2588,
+      "step": 25
+    },
+    {
+      "epoch": 0.014466546112115732,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00019999563704661483,
+      "loss": 1.198,
+      "step": 26
+    },
+    {
+      "epoch": 0.01502295173181249,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00019999507463927188,
+      "loss": 1.2047,
+      "step": 27
+    },
+    {
+      "epoch": 0.01557935735150925,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00019999447814778755,
+      "loss": 1.1884,
+      "step": 28
+    },
+    {
+      "epoch": 0.01613576297120601,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.0001999938475723652,
+      "loss": 1.2156,
+      "step": 29
+    },
+    {
+      "epoch": 0.016692168590902768,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00019999318291321968,
+      "loss": 1.1636,
+      "step": 30
+    },
+    {
+      "epoch": 0.017248574210599527,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00019999248417057767,
+      "loss": 1.1857,
+      "step": 31
+    },
+    {
+      "epoch": 0.017804979830296285,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019999175134467726,
+      "loss": 1.1862,
+      "step": 32
+    },
+    {
+      "epoch": 0.018361385449993044,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00019999098443576827,
+      "loss": 1.2155,
+      "step": 33
+    },
+    {
+      "epoch": 0.018917791069689802,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00019999018344411207,
+      "loss": 1.1009,
+      "step": 34
+    },
+    {
+      "epoch": 0.019474196689386564,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00019998934836998173,
+      "loss": 1.1714,
+      "step": 35
+    },
+    {
+      "epoch": 0.020030602309083323,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00019998847921366187,
+      "loss": 1.0824,
+      "step": 36
+    },
+    {
+      "epoch": 0.02058700792878008,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00019998757597544875,
+      "loss": 1.0703,
+      "step": 37
+    },
+    {
+      "epoch": 0.02114341354847684,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00019998663865565023,
+      "loss": 1.1183,
+      "step": 38
+    },
+    {
+      "epoch": 0.0216998191681736,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00019998566725458586,
+      "loss": 1.1328,
+      "step": 39
+    },
+    {
+      "epoch": 0.022256224787870357,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.0001999846617725867,
+      "loss": 1.1076,
+      "step": 40
+    },
+    {
+      "epoch": 0.022812630407567116,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00019998362220999548,
+      "loss": 1.1143,
+      "step": 41
+    },
+    {
+      "epoch": 0.023369036027263874,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00019998254856716657,
+      "loss": 1.1768,
+      "step": 42
+    },
+    {
+      "epoch": 0.023925441646960633,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0001999814408444659,
+      "loss": 1.0872,
+      "step": 43
+    },
+    {
+      "epoch": 0.02448184726665739,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00019998029904227103,
+      "loss": 1.122,
+      "step": 44
+    },
+    {
+      "epoch": 0.025038252886354154,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0001999791231609712,
+      "loss": 1.0025,
+      "step": 45
+    },
+    {
+      "epoch": 0.025594658506050912,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00019997791320096718,
+      "loss": 1.1672,
+      "step": 46
+    },
+    {
+      "epoch": 0.02615106412574767,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00019997666916267144,
+      "loss": 1.1096,
+      "step": 47
+    },
+    {
+      "epoch": 0.02670746974544443,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.000199975391046508,
+      "loss": 1.1446,
+      "step": 48
+    },
+    {
+      "epoch": 0.027263875365141188,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0001999740788529125,
+      "loss": 1.1254,
+      "step": 49
+    },
+    {
+      "epoch": 0.027820280984837947,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00019997273258233223,
+      "loss": 1.1143,
+      "step": 50
+    },
+    {
+      "epoch": 0.028376686604534705,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00019997135223522605,
+      "loss": 1.1123,
+      "step": 51
+    },
+    {
+      "epoch": 0.028933092224231464,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00019996993781206448,
+      "loss": 1.0469,
+      "step": 52
+    },
+    {
+      "epoch": 0.029489497843928222,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00019996848931332965,
+      "loss": 1.1145,
+      "step": 53
+    },
+    {
+      "epoch": 0.03004590346362498,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0001999670067395153,
+      "loss": 1.0489,
+      "step": 54
+    },
+    {
+      "epoch": 0.030602309083321743,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00019996549009112675,
+      "loss": 1.0671,
+      "step": 55
+    },
+    {
+      "epoch": 0.0311587147030185,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019996393936868098,
+      "loss": 1.0341,
+      "step": 56
+    },
+    {
+      "epoch": 0.03171512032271526,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019996235457270654,
+      "loss": 1.0725,
+      "step": 57
+    },
+    {
+      "epoch": 0.03227152594241202,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00019996073570374367,
+      "loss": 1.1031,
+      "step": 58
+    },
+    {
+      "epoch": 0.032827931562108774,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001999590827623441,
+      "loss": 0.9974,
+      "step": 59
+    },
+    {
+      "epoch": 0.033384337181805536,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00019995739574907133,
+      "loss": 1.1398,
+      "step": 60
+    },
+    {
+      "epoch": 0.0339407428015023,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00019995567466450035,
+      "loss": 1.0728,
+      "step": 61
+    },
+    {
+      "epoch": 0.03449714842119905,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00019995391950921782,
+      "loss": 0.9649,
+      "step": 62
+    },
+    {
+      "epoch": 0.035053554040895815,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.000199952130283822,
+      "loss": 1.1536,
+      "step": 63
+    },
+    {
+      "epoch": 0.03560995966059257,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001999503069889227,
+      "loss": 1.0135,
+      "step": 64
+    },
+    {
+      "epoch": 0.03616636528028933,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0001999484496251415,
+      "loss": 1.0599,
+      "step": 65
+    },
+    {
+      "epoch": 0.03672277089998609,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00019994655819311144,
+      "loss": 1.0521,
+      "step": 66
+    },
+    {
+      "epoch": 0.03727917651968285,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00019994463269347725,
+      "loss": 1.0737,
+      "step": 67
+    },
+    {
+      "epoch": 0.037835582139379605,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00019994267312689525,
+      "loss": 1.0856,
+      "step": 68
+    },
+    {
+      "epoch": 0.03839198775907637,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019994067949403337,
+      "loss": 0.9674,
+      "step": 69
+    },
+    {
+      "epoch": 0.03894839337877313,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00019993865179557117,
+      "loss": 1.1069,
+      "step": 70
+    },
+    {
+      "epoch": 0.039504798998469884,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019993659003219978,
+      "loss": 1.0198,
+      "step": 71
+    },
+    {
+      "epoch": 0.040061204618166646,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00019993449420462197,
+      "loss": 1.0864,
+      "step": 72
+    },
+    {
+      "epoch": 0.0406176102378634,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00019993236431355217,
+      "loss": 1.037,
+      "step": 73
+    },
+    {
+      "epoch": 0.04117401585756016,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00019993020035971633,
+      "loss": 1.0724,
+      "step": 74
+    },
+    {
+      "epoch": 0.04173042147725692,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019992800234385201,
+      "loss": 0.9854,
+      "step": 75
+    },
+    {
+      "epoch": 0.04228682709695368,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019992577026670854,
+      "loss": 1.0655,
+      "step": 76
+    },
+    {
+      "epoch": 0.042843232716650435,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00019992350412904663,
+      "loss": 1.0487,
+      "step": 77
+    },
+    {
+      "epoch": 0.0433996383363472,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019992120393163876,
+      "loss": 0.9593,
+      "step": 78
+    },
+    {
+      "epoch": 0.04395604395604396,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00019991886967526897,
+      "loss": 1.1132,
+      "step": 79
+    },
+    {
+      "epoch": 0.044512449575740715,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019991650136073287,
+      "loss": 0.9343,
+      "step": 80
+    },
+    {
+      "epoch": 0.04506885519543748,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0001999140989888378,
+      "loss": 1.0228,
+      "step": 81
+    },
+    {
+      "epoch": 0.04562526081513423,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00019991166256040256,
+      "loss": 1.0239,
+      "step": 82
+    },
+    {
+      "epoch": 0.046181666434830994,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00019990919207625764,
+      "loss": 1.0706,
+      "step": 83
+    },
+    {
+      "epoch": 0.04673807205452775,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00019990668753724515,
+      "loss": 1.0069,
+      "step": 84
+    },
+    {
+      "epoch": 0.04729447767422451,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00019990414894421876,
+      "loss": 1.0511,
+      "step": 85
+    },
+    {
+      "epoch": 0.047850883293921266,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0001999015762980438,
+      "loss": 0.9398,
+      "step": 86
+    },
+    {
+      "epoch": 0.04840728891361803,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00019989896959959712,
+      "loss": 1.1496,
+      "step": 87
+    },
+    {
+      "epoch": 0.04896369453331478,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019989632884976727,
+      "loss": 1.0488,
+      "step": 88
+    },
+    {
+      "epoch": 0.049520100153011545,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00019989365404945436,
+      "loss": 1.0843,
+      "step": 89
+    },
+    {
+      "epoch": 0.05007650577270831,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00019989094519957015,
+      "loss": 1.0822,
+      "step": 90
+    },
+    {
+      "epoch": 0.05063291139240506,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019988820230103796,
+      "loss": 1.0301,
+      "step": 91
+    },
+    {
+      "epoch": 0.051189317012101825,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001998854253547927,
+      "loss": 1.0098,
+      "step": 92
+    },
+    {
+      "epoch": 0.05174572263179858,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019988261436178093,
+      "loss": 0.9847,
+      "step": 93
+    },
+    {
+      "epoch": 0.05230212825149534,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0001998797693229608,
+      "loss": 1.0233,
+      "step": 94
+    },
+    {
+      "epoch": 0.0528585338711921,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.0001998768902393021,
+      "loss": 1.0747,
+      "step": 95
+    },
+    {
+      "epoch": 0.05341493949088886,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019987397711178614,
+      "loss": 0.9823,
+      "step": 96
+    },
+    {
+      "epoch": 0.053971345110585614,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.0001998710299414059,
+      "loss": 1.0238,
+      "step": 97
+    },
+    {
+      "epoch": 0.054527750730282376,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00019986804872916593,
+      "loss": 0.9832,
+      "step": 98
+    },
+    {
+      "epoch": 0.05508415634997914,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019986503347608245,
+      "loss": 0.9833,
+      "step": 99
+    },
+    {
+      "epoch": 0.05564056196967589,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0001998619841831832,
+      "loss": 1.0971,
+      "step": 100
+    },
+    {
+      "epoch": 0.056196967589372655,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001998589008515076,
+      "loss": 1.0077,
+      "step": 101
+    },
+    {
+      "epoch": 0.05675337320906941,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.0001998557834821065,
+      "loss": 1.0066,
+      "step": 102
+    },
+    {
+      "epoch": 0.05730977882876617,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00019985263207604264,
+      "loss": 1.0267,
+      "step": 103
+    },
+    {
+      "epoch": 0.05786618444846293,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001998494466343901,
+      "loss": 0.975,
+      "step": 104
+    },
+    {
+      "epoch": 0.05842259006815969,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019984622715823474,
+      "loss": 1.0134,
+      "step": 105
+    },
+    {
+      "epoch": 0.058978995687856445,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00019984297364867385,
+      "loss": 1.1247,
+      "step": 106
+    },
+    {
+      "epoch": 0.05953540130755321,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019983968610681648,
+      "loss": 0.974,
+      "step": 107
+    },
+    {
+      "epoch": 0.06009180692724996,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001998363645337832,
+      "loss": 1.0099,
+      "step": 108
+    },
+    {
+      "epoch": 0.060648212546946724,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019983300893070623,
+      "loss": 0.8479,
+      "step": 109
+    },
+    {
+      "epoch": 0.061204618166643486,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00019982961929872927,
+      "loss": 1.0034,
+      "step": 110
+    },
+    {
+      "epoch": 0.06176102378634024,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00019982619563900782,
+      "loss": 1.0165,
+      "step": 111
+    },
+    {
+      "epoch": 0.062317429406037,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00019982273795270876,
+      "loss": 1.0338,
+      "step": 112
+    },
+    {
+      "epoch": 0.06287383502573377,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00019981924624101071,
+      "loss": 1.0465,
+      "step": 113
+    },
+    {
+      "epoch": 0.06343024064543051,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019981572050510387,
+      "loss": 0.9284,
+      "step": 114
+    },
+    {
+      "epoch": 0.06398664626512728,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00019981216074619,
+      "loss": 1.074,
+      "step": 115
+    },
+    {
+      "epoch": 0.06454305188482404,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00019980856696548242,
+      "loss": 0.8981,
+      "step": 116
+    },
+    {
+      "epoch": 0.0650994575045208,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00019980493916420616,
+      "loss": 1.0605,
+      "step": 117
+    },
+    {
+      "epoch": 0.06565586312421755,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019980127734359782,
+      "loss": 1.0433,
+      "step": 118
+    },
+    {
+      "epoch": 0.06621226874391431,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00019979758150490546,
+      "loss": 1.0483,
+      "step": 119
+    },
+    {
+      "epoch": 0.06676867436361107,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0001997938516493889,
+      "loss": 1.0543,
+      "step": 120
+    },
+    {
+      "epoch": 0.06732507998330783,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001997900877783195,
+      "loss": 1.0162,
+      "step": 121
+    },
+    {
+      "epoch": 0.0678814856030046,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0001997862898929802,
+      "loss": 0.8468,
+      "step": 122
+    },
+    {
+      "epoch": 0.06843789122270134,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019978245799466554,
+      "loss": 1.0417,
+      "step": 123
+    },
+    {
+      "epoch": 0.0689942968423981,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019977859208468161,
+      "loss": 0.9901,
+      "step": 124
+    },
+    {
+      "epoch": 0.06955070246209487,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00019977469216434619,
+      "loss": 1.0117,
+      "step": 125
+    },
+    {
+      "epoch": 0.07010710808179163,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0001997707582349886,
+      "loss": 0.9719,
+      "step": 126
+    },
+    {
+      "epoch": 0.07066351370148838,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019976679029794968,
+      "loss": 0.9176,
+      "step": 127
+    },
+    {
+      "epoch": 0.07121991932118514,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00019976278835458203,
+      "loss": 1.0633,
+      "step": 128
+    },
+    {
+      "epoch": 0.0717763249408819,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00019975875240624968,
+      "loss": 0.9845,
+      "step": 129
+    },
+    {
+      "epoch": 0.07233273056057866,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019975468245432833,
+      "loss": 0.9352,
+      "step": 130
+    },
+    {
+      "epoch": 0.07288913618027543,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.0001997505785002053,
+      "loss": 1.0966,
+      "step": 131
+    },
+    {
+      "epoch": 0.07344554179997217,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001997464405452794,
+      "loss": 0.9278,
+      "step": 132
+    },
+    {
+      "epoch": 0.07400194741966894,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001997422685909611,
+      "loss": 0.9244,
+      "step": 133
+    },
+    {
+      "epoch": 0.0745583530393657,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019973806263867247,
+      "loss": 0.8941,
+      "step": 134
+    },
+    {
+      "epoch": 0.07511475865906246,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00019973382268984713,
+      "loss": 0.9702,
+      "step": 135
+    },
+    {
+      "epoch": 0.07567116427875921,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019972954874593026,
+      "loss": 0.9809,
+      "step": 136
+    },
+    {
+      "epoch": 0.07622756989845597,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019972524080837873,
+      "loss": 1.0604,
+      "step": 137
+    },
+    {
+      "epoch": 0.07678397551815273,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019972089887866092,
+      "loss": 0.9702,
+      "step": 138
+    },
+    {
+      "epoch": 0.0773403811378495,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019971652295825681,
+      "loss": 0.9903,
+      "step": 139
+    },
+    {
+      "epoch": 0.07789678675754626,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019971211304865795,
+      "loss": 0.979,
+      "step": 140
+    },
+    {
+      "epoch": 0.078453192377243,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00019970766915136746,
+      "loss": 1.1067,
+      "step": 141
+    },
+    {
+      "epoch": 0.07900959799693977,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001997031912679002,
+      "loss": 0.966,
+      "step": 142
+    },
+    {
+      "epoch": 0.07956600361663653,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00019969867939978236,
+      "loss": 1.0207,
+      "step": 143
+    },
+    {
+      "epoch": 0.08012240923633329,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019969413354855195,
+      "loss": 0.9127,
+      "step": 144
+    },
+    {
+      "epoch": 0.08067881485603004,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0001996895537157584,
+      "loss": 1.0944,
+      "step": 145
+    },
+    {
+      "epoch": 0.0812352204757268,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019968493990296277,
+      "loss": 0.948,
+      "step": 146
+    },
+    {
+      "epoch": 0.08179162609542356,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019968029211173778,
+      "loss": 0.9824,
+      "step": 147
+    },
+    {
+      "epoch": 0.08234803171512033,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019967561034366763,
+      "loss": 0.9328,
+      "step": 148
+    },
+    {
+      "epoch": 0.08290443733481709,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019967089460034814,
+      "loss": 1.0162,
+      "step": 149
+    },
+    {
+      "epoch": 0.08346084295451384,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001996661448833867,
+      "loss": 0.7696,
+      "step": 150
+    },
+    {
+      "epoch": 0.0840172485742106,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019966136119440233,
+      "loss": 0.9133,
+      "step": 151
+    },
+    {
+      "epoch": 0.08457365419390736,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00019965654353502554,
+      "loss": 1.0071,
+      "step": 152
+    },
+    {
+      "epoch": 0.08513005981360412,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0001996516919068985,
+      "loss": 0.9797,
+      "step": 153
+    },
+    {
+      "epoch": 0.08568646543330087,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00019964680631167492,
+      "loss": 0.9521,
+      "step": 154
+    },
+    {
+      "epoch": 0.08624287105299763,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0001996418867510201,
+      "loss": 1.034,
+      "step": 155
+    },
+    {
+      "epoch": 0.0867992766726944,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019963693322661087,
+      "loss": 0.9971,
+      "step": 156
+    },
+    {
+      "epoch": 0.08735568229239116,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019963194574013573,
+      "loss": 1.0262,
+      "step": 157
+    },
+    {
+      "epoch": 0.08791208791208792,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001996269242932947,
+      "loss": 0.9913,
+      "step": 158
+    },
+    {
+      "epoch": 0.08846849353178467,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00019962186888779936,
+      "loss": 1.0822,
+      "step": 159
+    },
+    {
+      "epoch": 0.08902489915148143,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019961677952537292,
+      "loss": 0.8287,
+      "step": 160
+    },
+    {
+      "epoch": 0.08958130477117819,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0001996116562077501,
+      "loss": 0.9663,
+      "step": 161
+    },
+    {
+      "epoch": 0.09013771039087495,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019960649893667722,
+      "loss": 0.7657,
+      "step": 162
+    },
+    {
+      "epoch": 0.0906941160105717,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001996013077139122,
+      "loss": 0.8837,
+      "step": 163
+    },
+    {
+      "epoch": 0.09125052163026846,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001995960825412245,
+      "loss": 0.9727,
+      "step": 164
+    },
+    {
+      "epoch": 0.09180692724996523,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001995908234203952,
+      "loss": 0.9438,
+      "step": 165
+    },
+    {
+      "epoch": 0.09236333286966199,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019958553035321686,
+      "loss": 0.9364,
+      "step": 166
+    },
+    {
+      "epoch": 0.09291973848935874,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019958020334149366,
+      "loss": 0.9435,
+      "step": 167
+    },
+    {
+      "epoch": 0.0934761441090555,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00019957484238704139,
+      "loss": 1.034,
+      "step": 168
+    },
+    {
+      "epoch": 0.09403254972875226,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019956944749168737,
+      "loss": 0.9662,
+      "step": 169
+    },
+    {
+      "epoch": 0.09458895534844902,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00019956401865727053,
+      "loss": 1.0057,
+      "step": 170
+    },
+    {
+      "epoch": 0.09514536096814578,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019955855588564125,
+      "loss": 0.9569,
+      "step": 171
+    },
+    {
+      "epoch": 0.09570176658784253,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019955305917866164,
+      "loss": 0.9651,
+      "step": 172
+    },
+    {
+      "epoch": 0.0962581722075393,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019954752853820523,
+      "loss": 0.8911,
+      "step": 173
+    },
+    {
+      "epoch": 0.09681457782723606,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019954196396615721,
+      "loss": 0.9656,
+      "step": 174
+    },
+    {
+      "epoch": 0.09737098344693282,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019953636546441431,
+      "loss": 0.8851,
+      "step": 175
+    },
+    {
+      "epoch": 0.09792738906662957,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019953073303488486,
+      "loss": 0.91,
+      "step": 176
+    },
+    {
+      "epoch": 0.09848379468632633,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019952506667948868,
+      "loss": 1.0532,
+      "step": 177
+    },
+    {
+      "epoch": 0.09904020030602309,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00019951936640015723,
+      "loss": 1.052,
+      "step": 178
+    },
+    {
+      "epoch": 0.09959660592571985,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019951363219883344,
+      "loss": 0.9436,
+      "step": 179
+    },
+    {
+      "epoch": 0.10015301154541661,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.0001995078640774719,
+      "loss": 1.0067,
+      "step": 180
+    },
+    {
+      "epoch": 0.10070941716511336,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019950206203803874,
+      "loss": 0.8923,
+      "step": 181
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019949622608251156,
+      "loss": 0.9681,
+      "step": 182
+    },
+    {
+      "epoch": 0.10182222840450689,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019949035621287967,
+      "loss": 0.9994,
+      "step": 183
+    },
+    {
+      "epoch": 0.10237863402420365,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019948445243114384,
+      "loss": 1.0139,
+      "step": 184
+    },
+    {
+      "epoch": 0.1029350396439004,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001994785147393164,
+      "loss": 0.9698,
+      "step": 185
+    },
+    {
+      "epoch": 0.10349144526359716,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001994725431394213,
+      "loss": 0.9727,
+      "step": 186
+    },
+    {
+      "epoch": 0.10404785088329392,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019946653763349394,
+      "loss": 0.9109,
+      "step": 187
+    },
+    {
+      "epoch": 0.10460425650299068,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001994604982235814,
+      "loss": 0.9881,
+      "step": 188
+    },
+    {
+      "epoch": 0.10516066212268745,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019945442491174227,
+      "loss": 1.0062,
+      "step": 189
+    },
+    {
+      "epoch": 0.1057170677423842,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00019944831770004665,
+      "loss": 0.9374,
+      "step": 190
+    },
+    {
+      "epoch": 0.10627347336208096,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019944217659057626,
+      "loss": 0.907,
+      "step": 191
+    },
+    {
+      "epoch": 0.10682987898177772,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019943600158542436,
+      "loss": 0.9287,
+      "step": 192
+    },
+    {
+      "epoch": 0.10738628460147448,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019942979268669573,
+      "loss": 0.9948,
+      "step": 193
+    },
+    {
+      "epoch": 0.10794269022117123,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001994235498965067,
+      "loss": 0.9651,
+      "step": 194
+    },
+    {
+      "epoch": 0.10849909584086799,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019941727321698521,
+      "loss": 0.9267,
+      "step": 195
+    },
+    {
+      "epoch": 0.10905550146056475,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019941096265027074,
+      "loss": 0.8551,
+      "step": 196
+    },
+    {
+      "epoch": 0.10961190708026151,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019940461819851425,
+      "loss": 0.9683,
+      "step": 197
+    },
+    {
+      "epoch": 0.11016831269995828,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019939823986387834,
+      "loss": 0.9127,
+      "step": 198
+    },
+    {
+      "epoch": 0.11072471831965502,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019939182764853707,
+      "loss": 1.0557,
+      "step": 199
+    },
+    {
+      "epoch": 0.11128112393935179,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019938538155467615,
+      "loss": 0.9967,
+      "step": 200
+    },
+    {
+      "epoch": 0.11183752955904855,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019937890158449272,
+      "loss": 0.9391,
+      "step": 201
+    },
+    {
+      "epoch": 0.11239393517874531,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00019937238774019557,
+      "loss": 1.9551,
+      "step": 202
+    },
+    {
+      "epoch": 0.11295034079844206,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.000199365840024005,
+      "loss": 0.9682,
+      "step": 203
+    },
+    {
+      "epoch": 0.11350674641813882,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019935925843815285,
+      "loss": 0.961,
+      "step": 204
+    },
+    {
+      "epoch": 0.11406315203783558,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0001993526429848825,
+      "loss": 1.0718,
+      "step": 205
+    },
+    {
+      "epoch": 0.11461955765753234,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0001993459936664489,
+      "loss": 0.9232,
+      "step": 206
+    },
+    {
+      "epoch": 0.1151759632772291,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.0001993393104851185,
+      "loss": 1.0022,
+      "step": 207
+    },
+    {
+      "epoch": 0.11573236889692586,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019933259344316934,
+      "loss": 0.9549,
+      "step": 208
+    },
+    {
+      "epoch": 0.11628877451662262,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019932584254289096,
+      "loss": 0.9472,
+      "step": 209
+    },
+    {
+      "epoch": 0.11684518013631938,
+      "grad_norm": 2.859375,
+      "learning_rate": 0.00019931905778658443,
+      "loss": 1.9077,
+      "step": 210
+    },
+    {
+      "epoch": 0.11740158575601614,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00019931223917656247,
+      "loss": 1.0227,
+      "step": 211
+    },
+    {
+      "epoch": 0.11795799137571289,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00019930538671514923,
+      "loss": 1.0234,
+      "step": 212
+    },
+    {
+      "epoch": 0.11851439699540965,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019929850040468035,
+      "loss": 0.9573,
+      "step": 213
+    },
+    {
+      "epoch": 0.11907080261510641,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001992915802475032,
+      "loss": 0.9576,
+      "step": 214
+    },
+    {
+      "epoch": 0.11962720823480318,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001992846262459765,
+      "loss": 0.9715,
+      "step": 215
+    },
+    {
+      "epoch": 0.12018361385449992,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001992776384024706,
+      "loss": 0.8743,
+      "step": 216
+    },
+    {
+      "epoch": 0.12074001947419669,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019927061671936737,
+      "loss": 0.8957,
+      "step": 217
+    },
+    {
+      "epoch": 0.12129642509389345,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0001992635611990602,
+      "loss": 0.9718,
+      "step": 218
+    },
+    {
+      "epoch": 0.12185283071359021,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019925647184395402,
+      "loss": 0.8916,
+      "step": 219
+    },
+    {
+      "epoch": 0.12240923633328697,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001992493486564653,
+      "loss": 0.9546,
+      "step": 220
+    },
+    {
+      "epoch": 0.12296564195298372,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019924219163902204,
+      "loss": 0.9285,
+      "step": 221
+    },
+    {
+      "epoch": 0.12352204757268048,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019923500079406373,
+      "loss": 0.9095,
+      "step": 222
+    },
+    {
+      "epoch": 0.12407845319237724,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00019922777612404148,
+      "loss": 0.9921,
+      "step": 223
+    },
+    {
+      "epoch": 0.124634858812074,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019922051763141788,
+      "loss": 0.9756,
+      "step": 224
+    },
+    {
+      "epoch": 0.12519126443177075,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019921322531866702,
+      "loss": 0.832,
+      "step": 225
+    },
+    {
+      "epoch": 0.12574767005146753,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00019920589918827455,
+      "loss": 0.741,
+      "step": 226
+    },
+    {
+      "epoch": 0.12630407567116428,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001991985392427376,
+      "loss": 0.8839,
+      "step": 227
+    },
+    {
+      "epoch": 0.12686048129086103,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019919114548456497,
+      "loss": 1.0329,
+      "step": 228
+    },
+    {
+      "epoch": 0.1274168869105578,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0001991837179162768,
+      "loss": 0.9777,
+      "step": 229
+    },
+    {
+      "epoch": 0.12797329253025455,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019917625654040485,
+      "loss": 0.9311,
+      "step": 230
+    },
+    {
+      "epoch": 0.12852969814995133,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0001991687613594924,
+      "loss": 0.9973,
+      "step": 231
+    },
+    {
+      "epoch": 0.12908610376964808,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00019916123237609428,
+      "loss": 1.054,
+      "step": 232
+    },
+    {
+      "epoch": 0.12964250938934482,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019915366959277674,
+      "loss": 0.886,
+      "step": 233
+    },
+    {
+      "epoch": 0.1301989150090416,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019914607301211765,
+      "loss": 0.9879,
+      "step": 234
+    },
+    {
+      "epoch": 0.13075532062873835,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019913844263670638,
+      "loss": 0.8272,
+      "step": 235
+    },
+    {
+      "epoch": 0.1313117262484351,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019913077846914377,
+      "loss": 0.8881,
+      "step": 236
+    },
+    {
+      "epoch": 0.13186813186813187,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019912308051204226,
+      "loss": 0.9737,
+      "step": 237
+    },
+    {
+      "epoch": 0.13242453748782862,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00019911534876802572,
+      "loss": 1.1547,
+      "step": 238
+    },
+    {
+      "epoch": 0.1329809431075254,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019910758323972961,
+      "loss": 0.9975,
+      "step": 239
+    },
+    {
+      "epoch": 0.13353734872722214,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019909978392980087,
+      "loss": 0.9454,
+      "step": 240
+    },
+    {
+      "epoch": 0.1340937543469189,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019909195084089792,
+      "loss": 0.969,
+      "step": 241
+    },
+    {
+      "epoch": 0.13465015996661567,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019908408397569075,
+      "loss": 1.0012,
+      "step": 242
+    },
+    {
+      "epoch": 0.13520656558631242,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001990761833368609,
+      "loss": 0.9292,
+      "step": 243
+    },
+    {
+      "epoch": 0.1357629712060092,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001990682489271013,
+      "loss": 0.8401,
+      "step": 244
+    },
+    {
+      "epoch": 0.13631937682570594,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019906028074911645,
+      "loss": 0.9614,
+      "step": 245
+    },
+    {
+      "epoch": 0.1368757824454027,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00019905227880562246,
+      "loss": 0.9454,
+      "step": 246
+    },
+    {
+      "epoch": 0.13743218806509946,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001990442430993468,
+      "loss": 0.8745,
+      "step": 247
+    },
+    {
+      "epoch": 0.1379885936847962,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019903617363302846,
+      "loss": 0.9185,
+      "step": 248
+    },
+    {
+      "epoch": 0.138544999304493,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019902807040941806,
+      "loss": 0.9842,
+      "step": 249
+    },
+    {
+      "epoch": 0.13910140492418974,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019901993343127763,
+      "loss": 1.0057,
+      "step": 250
+    },
+    {
+      "epoch": 0.13965781054388648,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001990117627013807,
+      "loss": 0.8891,
+      "step": 251
+    },
+    {
+      "epoch": 0.14021421616358326,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019900355822251233,
+      "loss": 0.9715,
+      "step": 252
+    },
+    {
+      "epoch": 0.14077062178328,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001989953199974691,
+      "loss": 0.8475,
+      "step": 253
+    },
+    {
+      "epoch": 0.14132702740297676,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001989870480290591,
+      "loss": 0.9162,
+      "step": 254
+    },
+    {
+      "epoch": 0.14188343302267353,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019897874232010186,
+      "loss": 0.8788,
+      "step": 255
+    },
+    {
+      "epoch": 0.14243983864237028,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0001989704028734285,
+      "loss": 1.8645,
+      "step": 256
+    },
+    {
+      "epoch": 0.14299624426206706,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019896202969188148,
+      "loss": 1.0641,
+      "step": 257
+    },
+    {
+      "epoch": 0.1435526498817638,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.000198953622778315,
+      "loss": 1.0014,
+      "step": 258
+    },
+    {
+      "epoch": 0.14410905550146055,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00019894518213559457,
+      "loss": 1.0273,
+      "step": 259
+    },
+    {
+      "epoch": 0.14466546112115733,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001989367077665972,
+      "loss": 0.9224,
+      "step": 260
+    },
+    {
+      "epoch": 0.14522186674085408,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019892819967421154,
+      "loss": 0.9585,
+      "step": 261
+    },
+    {
+      "epoch": 0.14577827236055085,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.0001989196578613376,
+      "loss": 0.9612,
+      "step": 262
+    },
+    {
+      "epoch": 0.1463346779802476,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001989110823308869,
+      "loss": 0.8635,
+      "step": 263
+    },
+    {
+      "epoch": 0.14689108359994435,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0001989024730857825,
+      "loss": 0.9478,
+      "step": 264
+    },
+    {
+      "epoch": 0.14744748921964113,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019889383012895896,
+      "loss": 1.0103,
+      "step": 265
+    },
+    {
+      "epoch": 0.14800389483933787,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019888515346336226,
+      "loss": 0.8712,
+      "step": 266
+    },
+    {
+      "epoch": 0.14856030045903465,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019887644309195,
+      "loss": 0.9338,
+      "step": 267
+    },
+    {
+      "epoch": 0.1491167060787314,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019886769901769104,
+      "loss": 0.9435,
+      "step": 268
+    },
+    {
+      "epoch": 0.14967311169842815,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019885892124356597,
+      "loss": 1.0054,
+      "step": 269
+    },
+    {
+      "epoch": 0.15022951731812492,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00019885010977256675,
+      "loss": 0.9563,
+      "step": 270
+    },
+    {
+      "epoch": 0.15078592293782167,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0001988412646076968,
+      "loss": 0.889,
+      "step": 271
+    },
+    {
+      "epoch": 0.15134232855751842,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019883238575197113,
+      "loss": 0.8797,
+      "step": 272
+    },
+    {
+      "epoch": 0.1518987341772152,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019882347320841615,
+      "loss": 0.9045,
+      "step": 273
+    },
+    {
+      "epoch": 0.15245513979691194,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019881452698006973,
+      "loss": 0.8376,
+      "step": 274
+    },
+    {
+      "epoch": 0.15301154541660872,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001988055470699813,
+      "loss": 0.9139,
+      "step": 275
+    },
+    {
+      "epoch": 0.15356795103630547,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019879653348121176,
+      "loss": 0.9675,
+      "step": 276
+    },
+    {
+      "epoch": 0.15412435665600221,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019878748621683344,
+      "loss": 0.8359,
+      "step": 277
+    },
+    {
+      "epoch": 0.154680762275699,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019877840527993018,
+      "loss": 0.9296,
+      "step": 278
+    },
+    {
+      "epoch": 0.15523716789539574,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019876929067359726,
+      "loss": 0.9716,
+      "step": 279
+    },
+    {
+      "epoch": 0.15579357351509251,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019876014240094154,
+      "loss": 0.9313,
+      "step": 280
+    },
+    {
+      "epoch": 0.15634997913478926,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001987509604650812,
+      "loss": 0.9064,
+      "step": 281
+    },
+    {
+      "epoch": 0.156906384754486,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019874174486914602,
+      "loss": 1.0199,
+      "step": 282
+    },
+    {
+      "epoch": 0.1574627903741828,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019873249561627723,
+      "loss": 0.9184,
+      "step": 283
+    },
+    {
+      "epoch": 0.15801919599387954,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001987232127096275,
+      "loss": 0.8171,
+      "step": 284
+    },
+    {
+      "epoch": 0.15857560161357628,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019871389615236094,
+      "loss": 0.9977,
+      "step": 285
+    },
+    {
+      "epoch": 0.15913200723327306,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019870454594765323,
+      "loss": 0.9063,
+      "step": 286
+    },
+    {
+      "epoch": 0.1596884128529698,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019869516209869144,
+      "loss": 1.0768,
+      "step": 287
+    },
+    {
+      "epoch": 0.16024481847266658,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019868574460867416,
+      "loss": 0.9082,
+      "step": 288
+    },
+    {
+      "epoch": 0.16080122409236333,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019867629348081138,
+      "loss": 0.9355,
+      "step": 289
+    },
+    {
+      "epoch": 0.16135762971206008,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001986668087183246,
+      "loss": 0.8966,
+      "step": 290
+    },
+    {
+      "epoch": 0.16191403533175686,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001986572903244468,
+      "loss": 0.8964,
+      "step": 291
+    },
+    {
+      "epoch": 0.1624704409514536,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019864773830242244,
+      "loss": 0.9721,
+      "step": 292
+    },
+    {
+      "epoch": 0.16302684657115038,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0001986381526555073,
+      "loss": 0.9809,
+      "step": 293
+    },
+    {
+      "epoch": 0.16358325219084713,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001986285333869688,
+      "loss": 0.9851,
+      "step": 294
+    },
+    {
+      "epoch": 0.16413965781054388,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019861888050008575,
+      "loss": 0.8559,
+      "step": 295
+    },
+    {
+      "epoch": 0.16469606343024065,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019860919399814837,
+      "loss": 0.9342,
+      "step": 296
+    },
+    {
+      "epoch": 0.1652524690499374,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0001985994738844584,
+      "loss": 0.9079,
+      "step": 297
+    },
+    {
+      "epoch": 0.16580887466963418,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019858972016232907,
+      "loss": 0.8145,
+      "step": 298
+    },
+    {
+      "epoch": 0.16636528028933092,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019857993283508497,
+      "loss": 0.9047,
+      "step": 299
+    },
+    {
+      "epoch": 0.16692168590902767,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019857011190606215,
+      "loss": 0.9684,
+      "step": 300
+    },
+    {
+      "epoch": 0.16747809152872445,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00019856025737860828,
+      "loss": 1.0272,
+      "step": 301
+    },
+    {
+      "epoch": 0.1680344971484212,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001985503692560822,
+      "loss": 0.9,
+      "step": 302
+    },
+    {
+      "epoch": 0.16859090276811795,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001985404475418545,
+      "loss": 1.0117,
+      "step": 303
+    },
+    {
+      "epoch": 0.16914730838781472,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019853049223930697,
+      "loss": 0.8626,
+      "step": 304
+    },
+    {
+      "epoch": 0.16970371400751147,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019852050335183305,
+      "loss": 0.9383,
+      "step": 305
+    },
+    {
+      "epoch": 0.17026011962720825,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019851048088283742,
+      "loss": 0.9119,
+      "step": 306
+    },
+    {
+      "epoch": 0.170816525246905,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019850042483573644,
+      "loss": 0.9732,
+      "step": 307
+    },
+    {
+      "epoch": 0.17137293086660174,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019849033521395773,
+      "loss": 0.9586,
+      "step": 308
+    },
+    {
+      "epoch": 0.17192933648629852,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019848021202094042,
+      "loss": 0.943,
+      "step": 309
+    },
+    {
+      "epoch": 0.17248574210599527,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001984700552601351,
+      "loss": 0.9244,
+      "step": 310
+    },
+    {
+      "epoch": 0.17304214772569204,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019845986493500378,
+      "loss": 1.0124,
+      "step": 311
+    },
+    {
+      "epoch": 0.1735985533453888,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019844964104901992,
+      "loss": 1.0024,
+      "step": 312
+    },
+    {
+      "epoch": 0.17415495896508554,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019843938360566843,
+      "loss": 0.856,
+      "step": 313
+    },
+    {
+      "epoch": 0.1747113645847823,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0001984290926084456,
+      "loss": 0.9451,
+      "step": 314
+    },
+    {
+      "epoch": 0.17526777020447906,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019841876806085923,
+      "loss": 0.9995,
+      "step": 315
+    },
+    {
+      "epoch": 0.17582417582417584,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019840840996642852,
+      "loss": 0.9539,
+      "step": 316
+    },
+    {
+      "epoch": 0.17638058144387259,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019839801832868413,
+      "loss": 0.9055,
+      "step": 317
+    },
+    {
+      "epoch": 0.17693698706356933,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001983875931511681,
+      "loss": 0.8727,
+      "step": 318
+    },
+    {
+      "epoch": 0.1774933926832661,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019837713443743397,
+      "loss": 0.899,
+      "step": 319
+    },
+    {
+      "epoch": 0.17804979830296286,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019836664219104666,
+      "loss": 0.8607,
+      "step": 320
+    },
+    {
+      "epoch": 0.1786062039226596,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019835611641558253,
+      "loss": 1.0124,
+      "step": 321
+    },
+    {
+      "epoch": 0.17916260954235638,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0001983455571146294,
+      "loss": 1.0013,
+      "step": 322
+    },
+    {
+      "epoch": 0.17971901516205313,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019833496429178652,
+      "loss": 0.8477,
+      "step": 323
+    },
+    {
+      "epoch": 0.1802754207817499,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019832433795066446,
+      "loss": 0.889,
+      "step": 324
+    },
+    {
+      "epoch": 0.18083182640144665,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019831367809488537,
+      "loss": 0.789,
+      "step": 325
+    },
+    {
+      "epoch": 0.1813882320211434,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001983029847280827,
+      "loss": 0.8853,
+      "step": 326
+    },
+    {
+      "epoch": 0.18194463764084018,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001982922578539014,
+      "loss": 0.9427,
+      "step": 327
+    },
+    {
+      "epoch": 0.18250104326053693,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001982814974759978,
+      "loss": 0.8221,
+      "step": 328
+    },
+    {
+      "epoch": 0.1830574488802337,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00019827070359803968,
+      "loss": 0.9266,
+      "step": 329
+    },
+    {
+      "epoch": 0.18361385449993045,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0001982598762237062,
+      "loss": 1.0117,
+      "step": 330
+    },
+    {
+      "epoch": 0.1841702601196272,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00019824901535668796,
+      "loss": 0.9188,
+      "step": 331
+    },
+    {
+      "epoch": 0.18472666573932398,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.000198238121000687,
+      "loss": 0.952,
+      "step": 332
+    },
+    {
+      "epoch": 0.18528307135902072,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019822719315941672,
+      "loss": 0.9938,
+      "step": 333
+    },
+    {
+      "epoch": 0.18583947697871747,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.000198216231836602,
+      "loss": 0.8886,
+      "step": 334
+    },
+    {
+      "epoch": 0.18639588259841425,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019820523703597908,
+      "loss": 0.8996,
+      "step": 335
+    },
+    {
+      "epoch": 0.186952288218111,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001981942087612956,
+      "loss": 0.8635,
+      "step": 336
+    },
+    {
+      "epoch": 0.18750869383780777,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0001981831470163107,
+      "loss": 1.029,
+      "step": 337
+    },
+    {
+      "epoch": 0.18806509945750452,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001981720518047948,
+      "loss": 0.9771,
+      "step": 338
+    },
+    {
+      "epoch": 0.18862150507720127,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0001981609231305298,
+      "loss": 0.919,
+      "step": 339
+    },
+    {
+      "epoch": 0.18917791069689804,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019814976099730904,
+      "loss": 0.9555,
+      "step": 340
+    },
+    {
+      "epoch": 0.1897343163165948,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019813856540893722,
+      "loss": 1.0713,
+      "step": 341
+    },
+    {
+      "epoch": 0.19029072193629157,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019812733636923044,
+      "loss": 0.9679,
+      "step": 342
+    },
+    {
+      "epoch": 0.19084712755598832,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001981160738820162,
+      "loss": 0.9542,
+      "step": 343
+    },
+    {
+      "epoch": 0.19140353317568506,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0001981047779511334,
+      "loss": 0.991,
+      "step": 344
+    },
+    {
+      "epoch": 0.19195993879538184,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0001980934485804324,
+      "loss": 0.9589,
+      "step": 345
+    },
+    {
+      "epoch": 0.1925163444150786,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019808208577377486,
+      "loss": 0.9242,
+      "step": 346
+    },
+    {
+      "epoch": 0.19307275003477536,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0001980706895350339,
+      "loss": 0.9109,
+      "step": 347
+    },
+    {
+      "epoch": 0.1936291556544721,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019805925986809403,
+      "loss": 1.0326,
+      "step": 348
+    },
+    {
+      "epoch": 0.19418556127416886,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019804779677685112,
+      "loss": 0.9841,
+      "step": 349
+    },
+    {
+      "epoch": 0.19474196689386564,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001980363002652125,
+      "loss": 0.9203,
+      "step": 350
+    },
+    {
+      "epoch": 0.19529837251356238,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001980247703370968,
+      "loss": 0.8824,
+      "step": 351
+    },
+    {
+      "epoch": 0.19585477813325913,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00019801320699643416,
+      "loss": 0.7956,
+      "step": 352
+    },
+    {
+      "epoch": 0.1964111837529559,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019800161024716598,
+      "loss": 0.922,
+      "step": 353
+    },
+    {
+      "epoch": 0.19696758937265266,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019798998009324512,
+      "loss": 0.9232,
+      "step": 354
+    },
+    {
+      "epoch": 0.19752399499234943,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019797831653863582,
+      "loss": 0.8868,
+      "step": 355
+    },
+    {
+      "epoch": 0.19808040061204618,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001979666195873137,
+      "loss": 0.9103,
+      "step": 356
+    },
+    {
+      "epoch": 0.19863680623174293,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019795488924326574,
+      "loss": 0.8998,
+      "step": 357
+    },
+    {
+      "epoch": 0.1991932118514397,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019794312551049039,
+      "loss": 0.8973,
+      "step": 358
+    },
+    {
+      "epoch": 0.19974961747113645,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019793132839299732,
+      "loss": 0.9458,
+      "step": 359
+    },
+    {
+      "epoch": 0.20030602309083323,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019791949789480775,
+      "loss": 1.0083,
+      "step": 360
+    },
+    {
+      "epoch": 0.20086242871052998,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019790763401995415,
+      "loss": 0.9661,
+      "step": 361
+    },
+    {
+      "epoch": 0.20141883433022673,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019789573677248047,
+      "loss": 0.8475,
+      "step": 362
+    },
+    {
+      "epoch": 0.2019752399499235,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019788380615644196,
+      "loss": 0.8632,
+      "step": 363
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019787184217590523,
+      "loss": 0.9522,
+      "step": 364
+    },
+    {
+      "epoch": 0.20308805118931703,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019785984483494838,
+      "loss": 0.8749,
+      "step": 365
+    },
+    {
+      "epoch": 0.20364445680901377,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019784781413766073,
+      "loss": 0.8714,
+      "step": 366
+    },
+    {
+      "epoch": 0.20420086242871052,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001978357500881431,
+      "loss": 0.9581,
+      "step": 367
+    },
+    {
+      "epoch": 0.2047572680484073,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019782365269050756,
+      "loss": 1.0496,
+      "step": 368
+    },
+    {
+      "epoch": 0.20531367366810405,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019781152194887764,
+      "loss": 0.8804,
+      "step": 369
+    },
+    {
+      "epoch": 0.2058700792878008,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019779935786738822,
+      "loss": 0.8474,
+      "step": 370
+    },
+    {
+      "epoch": 0.20642648490749757,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001977871604501855,
+      "loss": 0.8413,
+      "step": 371
+    },
+    {
+      "epoch": 0.20698289052719432,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019777492970142707,
+      "loss": 0.8636,
+      "step": 372
+    },
+    {
+      "epoch": 0.2075392961468911,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019776266562528192,
+      "loss": 0.8351,
+      "step": 373
+    },
+    {
+      "epoch": 0.20809570176658784,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001977503682259303,
+      "loss": 0.8751,
+      "step": 374
+    },
+    {
+      "epoch": 0.2086521073862846,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019773803750756393,
+      "loss": 0.9113,
+      "step": 375
+    },
+    {
+      "epoch": 0.20920851300598137,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001977256734743858,
+      "loss": 0.9433,
+      "step": 376
+    },
+    {
+      "epoch": 0.20976491862567812,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001977132761306103,
+      "loss": 0.8863,
+      "step": 377
+    },
+    {
+      "epoch": 0.2103213242453749,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019770084548046318,
+      "loss": 0.8073,
+      "step": 378
+    },
+    {
+      "epoch": 0.21087772986507164,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019768838152818153,
+      "loss": 0.9198,
+      "step": 379
+    },
+    {
+      "epoch": 0.2114341354847684,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001976758842780138,
+      "loss": 0.9204,
+      "step": 380
+    },
+    {
+      "epoch": 0.21199054110446516,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019766335373421973,
+      "loss": 0.9714,
+      "step": 381
+    },
+    {
+      "epoch": 0.2125469467241619,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019765078990107048,
+      "loss": 1.0368,
+      "step": 382
+    },
+    {
+      "epoch": 0.21310335234385866,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001976381927828486,
+      "loss": 0.8881,
+      "step": 383
+    },
+    {
+      "epoch": 0.21365975796355544,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019762556238384784,
+      "loss": 1.0236,
+      "step": 384
+    },
+    {
+      "epoch": 0.21421616358325218,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0001976128987083734,
+      "loss": 0.9628,
+      "step": 385
+    },
+    {
+      "epoch": 0.21477256920294896,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001976002017607418,
+      "loss": 0.8861,
+      "step": 386
+    },
+    {
+      "epoch": 0.2153289748226457,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00019758747154528092,
+      "loss": 1.0946,
+      "step": 387
+    },
+    {
+      "epoch": 0.21588538044234246,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019757470806632994,
+      "loss": 1.0047,
+      "step": 388
+    },
+    {
+      "epoch": 0.21644178606203923,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019756191132823937,
+      "loss": 0.8831,
+      "step": 389
+    },
+    {
+      "epoch": 0.21699819168173598,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019754908133537113,
+      "loss": 0.8999,
+      "step": 390
+    },
+    {
+      "epoch": 0.21755459730143276,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019753621809209842,
+      "loss": 0.8368,
+      "step": 391
+    },
+    {
+      "epoch": 0.2181110029211295,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019752332160280576,
+      "loss": 0.9262,
+      "step": 392
+    },
+    {
+      "epoch": 0.21866740854082625,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.000197510391871889,
+      "loss": 0.9496,
+      "step": 393
+    },
+    {
+      "epoch": 0.21922381416052303,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001974974289037554,
+      "loss": 0.9303,
+      "step": 394
+    },
+    {
+      "epoch": 0.21978021978021978,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0001974844327028235,
+      "loss": 0.924,
+      "step": 395
+    },
+    {
+      "epoch": 0.22033662539991655,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019747140327352306,
+      "loss": 0.9763,
+      "step": 396
+    },
+    {
+      "epoch": 0.2208930310196133,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001974583406202954,
+      "loss": 0.7589,
+      "step": 397
+    },
+    {
+      "epoch": 0.22144943663931005,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019744524474759296,
+      "loss": 0.9391,
+      "step": 398
+    },
+    {
+      "epoch": 0.22200584225900682,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001974321156598796,
+      "loss": 0.9961,
+      "step": 399
+    },
+    {
+      "epoch": 0.22256224787870357,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019741895336163046,
+      "loss": 0.9214,
+      "step": 400
+    },
+    {
+      "epoch": 0.22311865349840032,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.000197405757857332,
+      "loss": 0.8915,
+      "step": 401
+    },
+    {
+      "epoch": 0.2236750591180971,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019739252915148208,
+      "loss": 0.8334,
+      "step": 402
+    },
+    {
+      "epoch": 0.22423146473779385,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019737926724858976,
+      "loss": 0.9007,
+      "step": 403
+    },
+    {
+      "epoch": 0.22478787035749062,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001973659721531755,
+      "loss": 0.8694,
+      "step": 404
+    },
+    {
+      "epoch": 0.22534427597718737,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019735264386977098,
+      "loss": 0.9033,
+      "step": 405
+    },
+    {
+      "epoch": 0.22590068159688412,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019733928240291932,
+      "loss": 0.8603,
+      "step": 406
+    },
+    {
+      "epoch": 0.2264570872165809,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019732588775717486,
+      "loss": 1.024,
+      "step": 407
+    },
+    {
+      "epoch": 0.22701349283627764,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019731245993710327,
+      "loss": 0.8826,
+      "step": 408
+    },
+    {
+      "epoch": 0.22756989845597442,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019729899894728158,
+      "loss": 0.8364,
+      "step": 409
+    },
+    {
+      "epoch": 0.22812630407567117,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.000197285504792298,
+      "loss": 1.0239,
+      "step": 410
+    },
+    {
+      "epoch": 0.22868270969536791,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019727197747675218,
+      "loss": 0.9286,
+      "step": 411
+    },
+    {
+      "epoch": 0.2292391153150647,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.000197258417005255,
+      "loss": 0.8623,
+      "step": 412
+    },
+    {
+      "epoch": 0.22979552093476144,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019724482338242866,
+      "loss": 0.8571,
+      "step": 413
+    },
+    {
+      "epoch": 0.2303519265544582,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019723119661290664,
+      "loss": 1.0023,
+      "step": 414
+    },
+    {
+      "epoch": 0.23090833217415496,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019721753670133376,
+      "loss": 0.8866,
+      "step": 415
+    },
+    {
+      "epoch": 0.2314647377938517,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001972038436523661,
+      "loss": 0.7956,
+      "step": 416
+    },
+    {
+      "epoch": 0.2320211434135485,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019719011747067108,
+      "loss": 0.9645,
+      "step": 417
+    },
+    {
+      "epoch": 0.23257754903324523,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019717635816092732,
+      "loss": 0.9324,
+      "step": 418
+    },
+    {
+      "epoch": 0.23313395465294198,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019716256572782482,
+      "loss": 0.9626,
+      "step": 419
+    },
+    {
+      "epoch": 0.23369036027263876,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001971487401760649,
+      "loss": 0.8419,
+      "step": 420
+    },
+    {
+      "epoch": 0.2342467658923355,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019713488151036003,
+      "loss": 0.8471,
+      "step": 421
+    },
+    {
+      "epoch": 0.23480317151203228,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019712098973543408,
+      "loss": 1.0177,
+      "step": 422
+    },
+    {
+      "epoch": 0.23535957713172903,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001971070648560222,
+      "loss": 0.944,
+      "step": 423
+    },
+    {
+      "epoch": 0.23591598275142578,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001970931068768708,
+      "loss": 0.8007,
+      "step": 424
+    },
+    {
+      "epoch": 0.23647238837112255,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001970791158027375,
+      "loss": 0.9658,
+      "step": 425
+    },
+    {
+      "epoch": 0.2370287939908193,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019706509163839136,
+      "loss": 0.8966,
+      "step": 426
+    },
+    {
+      "epoch": 0.23758519961051608,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001970510343886126,
+      "loss": 1.0448,
+      "step": 427
+    },
+    {
+      "epoch": 0.23814160523021283,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0001970369440581927,
+      "loss": 0.937,
+      "step": 428
+    },
+    {
+      "epoch": 0.23869801084990958,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019702282065193455,
+      "loss": 0.9181,
+      "step": 429
+    },
+    {
+      "epoch": 0.23925441646960635,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019700866417465216,
+      "loss": 0.9101,
+      "step": 430
+    },
+    {
+      "epoch": 0.2398108220893031,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019699447463117093,
+      "loss": 0.9438,
+      "step": 431
+    },
+    {
+      "epoch": 0.24036722770899985,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019698025202632743,
+      "loss": 0.8315,
+      "step": 432
+    },
+    {
+      "epoch": 0.24092363332869662,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019696599636496962,
+      "loss": 0.8385,
+      "step": 433
+    },
+    {
+      "epoch": 0.24148003894839337,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001969517076519566,
+      "loss": 0.8667,
+      "step": 434
+    },
+    {
+      "epoch": 0.24203644456809015,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019693738589215883,
+      "loss": 0.881,
+      "step": 435
+    },
+    {
+      "epoch": 0.2425928501877869,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019692303109045795,
+      "loss": 0.9509,
+      "step": 436
+    },
+    {
+      "epoch": 0.24314925580748364,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019690864325174699,
+      "loss": 0.9065,
+      "step": 437
+    },
+    {
+      "epoch": 0.24370566142718042,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001968942223809301,
+      "loss": 1.072,
+      "step": 438
+    },
+    {
+      "epoch": 0.24426206704687717,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001968797684829228,
+      "loss": 0.9068,
+      "step": 439
+    },
+    {
+      "epoch": 0.24481847266657394,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019686528156265175,
+      "loss": 0.8168,
+      "step": 440
+    },
+    {
+      "epoch": 0.2453748782862707,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.000196850761625055,
+      "loss": 0.8748,
+      "step": 441
+    },
+    {
+      "epoch": 0.24593128390596744,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019683620867508176,
+      "loss": 0.7686,
+      "step": 442
+    },
+    {
+      "epoch": 0.24648768952566422,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019682162271769256,
+      "loss": 0.9036,
+      "step": 443
+    },
+    {
+      "epoch": 0.24704409514536096,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001968070037578591,
+      "loss": 0.9472,
+      "step": 444
+    },
+    {
+      "epoch": 0.24760050076505774,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019679235180056437,
+      "loss": 0.8882,
+      "step": 445
+    },
+    {
+      "epoch": 0.2481569063847545,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019677766685080264,
+      "loss": 0.9693,
+      "step": 446
+    },
+    {
+      "epoch": 0.24871331200445124,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019676294891357937,
+      "loss": 0.8764,
+      "step": 447
+    },
+    {
+      "epoch": 0.249269717624148,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00019674819799391129,
+      "loss": 0.9671,
+      "step": 448
+    },
+    {
+      "epoch": 0.24982612324384476,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001967334140968264,
+      "loss": 0.9398,
+      "step": 449
+    },
+    {
+      "epoch": 0.2503825288635415,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019671859722736389,
+      "loss": 0.9572,
+      "step": 450
+    },
+    {
+      "epoch": 0.2509389344832383,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001967037473905742,
+      "loss": 0.958,
+      "step": 451
+    },
+    {
+      "epoch": 0.25149534010293506,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019668886459151908,
+      "loss": 0.8959,
+      "step": 452
+    },
+    {
+      "epoch": 0.2520517457226318,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019667394883527139,
+      "loss": 0.9529,
+      "step": 453
+    },
+    {
+      "epoch": 0.25260815134232856,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019665900012691529,
+      "loss": 0.9213,
+      "step": 454
+    },
+    {
+      "epoch": 0.25316455696202533,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001966440184715462,
+      "loss": 0.952,
+      "step": 455
+    },
+    {
+      "epoch": 0.25372096258172205,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019662900387427075,
+      "loss": 0.9548,
+      "step": 456
+    },
+    {
+      "epoch": 0.25427736820141883,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019661395634020673,
+      "loss": 0.8359,
+      "step": 457
+    },
+    {
+      "epoch": 0.2548337738211156,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019659887587448327,
+      "loss": 0.9204,
+      "step": 458
+    },
+    {
+      "epoch": 0.2553901794408123,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019658376248224063,
+      "loss": 0.9216,
+      "step": 459
+    },
+    {
+      "epoch": 0.2559465850605091,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019656861616863038,
+      "loss": 0.9383,
+      "step": 460
+    },
+    {
+      "epoch": 0.2565029906802059,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019655343693881526,
+      "loss": 0.9428,
+      "step": 461
+    },
+    {
+      "epoch": 0.25705939629990265,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001965382247979692,
+      "loss": 0.9357,
+      "step": 462
+    },
+    {
+      "epoch": 0.2576158019195994,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019652297975127737,
+      "loss": 0.967,
+      "step": 463
+    },
+    {
+      "epoch": 0.25817220753929615,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001965077018039362,
+      "loss": 0.8993,
+      "step": 464
+    },
+    {
+      "epoch": 0.2587286131589929,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019649239096115336,
+      "loss": 1.0113,
+      "step": 465
+    },
+    {
+      "epoch": 0.25928501877868965,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019647704722814755,
+      "loss": 0.9117,
+      "step": 466
+    },
+    {
+      "epoch": 0.2598414243983864,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019646167061014892,
+      "loss": 0.9318,
+      "step": 467
+    },
+    {
+      "epoch": 0.2603978300180832,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019644626111239863,
+      "loss": 0.9237,
+      "step": 468
+    },
+    {
+      "epoch": 0.2609542356377799,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0001964308187401492,
+      "loss": 0.9926,
+      "step": 469
+    },
+    {
+      "epoch": 0.2615106412574767,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001964153434986643,
+      "loss": 0.8647,
+      "step": 470
+    },
+    {
+      "epoch": 0.26206704687717347,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001963998353932187,
+      "loss": 0.7973,
+      "step": 471
+    },
+    {
+      "epoch": 0.2626234524968702,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019638429442909853,
+      "loss": 0.8922,
+      "step": 472
+    },
+    {
+      "epoch": 0.26317985811656697,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019636872061160109,
+      "loss": 0.9267,
+      "step": 473
+    },
+    {
+      "epoch": 0.26373626373626374,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019635311394603476,
+      "loss": 0.8743,
+      "step": 474
+    },
+    {
+      "epoch": 0.2642926693559605,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019633747443771924,
+      "loss": 0.7997,
+      "step": 475
+    },
+    {
+      "epoch": 0.26484907497565724,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019632180209198542,
+      "loss": 0.9695,
+      "step": 476
+    },
+    {
+      "epoch": 0.265405480595354,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019630609691417527,
+      "loss": 0.7992,
+      "step": 477
+    },
+    {
+      "epoch": 0.2659618862150508,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00019629035890964208,
+      "loss": 0.7188,
+      "step": 478
+    },
+    {
+      "epoch": 0.2665182918347475,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0001962745880837503,
+      "loss": 0.93,
+      "step": 479
+    },
+    {
+      "epoch": 0.2670746974544443,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019625878444187551,
+      "loss": 0.8121,
+      "step": 480
+    },
+    {
+      "epoch": 0.26763110307414106,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019624294798940451,
+      "loss": 0.9366,
+      "step": 481
+    },
+    {
+      "epoch": 0.2681875086938378,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001962270787317353,
+      "loss": 0.94,
+      "step": 482
+    },
+    {
+      "epoch": 0.26874391431353456,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.000196211176674277,
+      "loss": 0.9317,
+      "step": 483
+    },
+    {
+      "epoch": 0.26930031993323134,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019619524182245,
+      "loss": 0.8751,
+      "step": 484
+    },
+    {
+      "epoch": 0.2698567255529281,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019617927418168587,
+      "loss": 0.9093,
+      "step": 485
+    },
+    {
+      "epoch": 0.27041313117262483,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019616327375742722,
+      "loss": 0.8399,
+      "step": 486
+    },
+    {
+      "epoch": 0.2709695367923216,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019614724055512799,
+      "loss": 0.9025,
+      "step": 487
+    },
+    {
+      "epoch": 0.2715259424120184,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001961311745802532,
+      "loss": 0.8661,
+      "step": 488
+    },
+    {
+      "epoch": 0.2720823480317151,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019611507583827903,
+      "loss": 0.8429,
+      "step": 489
+    },
+    {
+      "epoch": 0.2726387536514119,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019609894433469295,
+      "loss": 0.9272,
+      "step": 490
+    },
+    {
+      "epoch": 0.27319515927110866,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001960827800749935,
+      "loss": 0.87,
+      "step": 491
+    },
+    {
+      "epoch": 0.2737515648908054,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019606658306469034,
+      "loss": 0.865,
+      "step": 492
+    },
+    {
+      "epoch": 0.27430797051050215,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019605035330930442,
+      "loss": 0.8939,
+      "step": 493
+    },
+    {
+      "epoch": 0.27486437613019893,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019603409081436773,
+      "loss": 0.9009,
+      "step": 494
+    },
+    {
+      "epoch": 0.27542078174989565,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001960177955854235,
+      "loss": 0.8828,
+      "step": 495
+    },
+    {
+      "epoch": 0.2759771873695924,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019600146762802613,
+      "loss": 0.9509,
+      "step": 496
+    },
+    {
+      "epoch": 0.2765335929892892,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019598510694774108,
+      "loss": 0.7998,
+      "step": 497
+    },
+    {
+      "epoch": 0.277089998608986,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019596871355014506,
+      "loss": 0.8909,
+      "step": 498
+    },
+    {
+      "epoch": 0.2776464042286827,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019595228744082586,
+      "loss": 0.9643,
+      "step": 499
+    },
+    {
+      "epoch": 0.2782028098483795,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019593582862538248,
+      "loss": 0.9311,
+      "step": 500
+    },
+    {
+      "epoch": 0.27875921546807625,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019591933710942503,
+      "loss": 0.8344,
+      "step": 501
+    },
+    {
+      "epoch": 0.27931562108777297,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019590281289857478,
+      "loss": 0.9109,
+      "step": 502
+    },
+    {
+      "epoch": 0.27987202670746975,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019588625599846417,
+      "loss": 0.994,
+      "step": 503
+    },
+    {
+      "epoch": 0.2804284323271665,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019586966641473673,
+      "loss": 0.8917,
+      "step": 504
+    },
+    {
+      "epoch": 0.28098483794686324,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019585304415304711,
+      "loss": 0.8417,
+      "step": 505
+    },
+    {
+      "epoch": 0.28154124356656,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019583638921906124,
+      "loss": 0.9425,
+      "step": 506
+    },
+    {
+      "epoch": 0.2820976491862568,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019581970161845603,
+      "loss": 0.8792,
+      "step": 507
+    },
+    {
+      "epoch": 0.2826540548059535,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001958029813569196,
+      "loss": 0.9791,
+      "step": 508
+    },
+    {
+      "epoch": 0.2832104604256503,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019578622844015117,
+      "loss": 0.8546,
+      "step": 509
+    },
+    {
+      "epoch": 0.28376686604534707,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019576944287386113,
+      "loss": 0.8836,
+      "step": 510
+    },
+    {
+      "epoch": 0.28432327166504384,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.000195752624663771,
+      "loss": 0.8764,
+      "step": 511
+    },
+    {
+      "epoch": 0.28487967728474056,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019573577381561336,
+      "loss": 0.9411,
+      "step": 512
+    },
+    {
+      "epoch": 0.28543608290443734,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019571889033513198,
+      "loss": 0.9222,
+      "step": 513
+    },
+    {
+      "epoch": 0.2859924885241341,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019570197422808173,
+      "loss": 0.9914,
+      "step": 514
+    },
+    {
+      "epoch": 0.28654889414383083,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019568502550022858,
+      "loss": 0.8788,
+      "step": 515
+    },
+    {
+      "epoch": 0.2871052997635276,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001956680441573497,
+      "loss": 0.811,
+      "step": 516
+    },
+    {
+      "epoch": 0.2876617053832244,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019565103020523328,
+      "loss": 0.9191,
+      "step": 517
+    },
+    {
+      "epoch": 0.2882181110029211,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019563398364967868,
+      "loss": 0.9129,
+      "step": 518
+    },
+    {
+      "epoch": 0.2887745166226179,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019561690449649636,
+      "loss": 0.9591,
+      "step": 519
+    },
+    {
+      "epoch": 0.28933092224231466,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00019559979275150783,
+      "loss": 0.8129,
+      "step": 520
+    },
+    {
+      "epoch": 0.2898873278620114,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019558264842054585,
+      "loss": 0.8881,
+      "step": 521
+    },
+    {
+      "epoch": 0.29044373348170816,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00019556547150945418,
+      "loss": 0.8941,
+      "step": 522
+    },
+    {
+      "epoch": 0.29100013910140493,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001955482620240877,
+      "loss": 0.8872,
+      "step": 523
+    },
+    {
+      "epoch": 0.2915565447211017,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001955310199703124,
+      "loss": 0.7887,
+      "step": 524
+    },
+    {
+      "epoch": 0.2921129503407984,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019551374535400543,
+      "loss": 0.8684,
+      "step": 525
+    },
+    {
+      "epoch": 0.2926693559604952,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001954964381810549,
+      "loss": 0.7739,
+      "step": 526
+    },
+    {
+      "epoch": 0.293225761580192,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019547909845736013,
+      "loss": 0.8568,
+      "step": 527
+    },
+    {
+      "epoch": 0.2937821671998887,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019546172618883155,
+      "loss": 0.9188,
+      "step": 528
+    },
+    {
+      "epoch": 0.2943385728195855,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019544432138139062,
+      "loss": 0.8119,
+      "step": 529
+    },
+    {
+      "epoch": 0.29489497843928225,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019542688404096986,
+      "loss": 0.9192,
+      "step": 530
+    },
+    {
+      "epoch": 0.29545138405897897,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.000195409414173513,
+      "loss": 0.8616,
+      "step": 531
+    },
+    {
+      "epoch": 0.29600778967867575,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019539191178497472,
+      "loss": 0.9191,
+      "step": 532
+    },
+    {
+      "epoch": 0.2965641952983725,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00019537437688132094,
+      "loss": 0.7765,
+      "step": 533
+    },
+    {
+      "epoch": 0.2971206009180693,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019535680946852852,
+      "loss": 0.9185,
+      "step": 534
+    },
+    {
+      "epoch": 0.297677006537766,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019533920955258546,
+      "loss": 0.8514,
+      "step": 535
+    },
+    {
+      "epoch": 0.2982334121574628,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00019532157713949083,
+      "loss": 1.8071,
+      "step": 536
+    },
+    {
+      "epoch": 0.29878981777715957,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019530391223525478,
+      "loss": 0.9006,
+      "step": 537
+    },
+    {
+      "epoch": 0.2993462233968563,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019528621484589859,
+      "loss": 0.9186,
+      "step": 538
+    },
+    {
+      "epoch": 0.29990262901655307,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019526848497745448,
+      "loss": 0.8292,
+      "step": 539
+    },
+    {
+      "epoch": 0.30045903463624984,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019525072263596585,
+      "loss": 0.7303,
+      "step": 540
+    },
+    {
+      "epoch": 0.30101544025594656,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001952329278274872,
+      "loss": 0.9214,
+      "step": 541
+    },
+    {
+      "epoch": 0.30157184587564334,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019521510055808396,
+      "loss": 0.9173,
+      "step": 542
+    },
+    {
+      "epoch": 0.3021282514953401,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019519724083383273,
+      "loss": 0.8349,
+      "step": 543
+    },
+    {
+      "epoch": 0.30268465711503684,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019517934866082116,
+      "loss": 0.9247,
+      "step": 544
+    },
+    {
+      "epoch": 0.3032410627347336,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001951614240451479,
+      "loss": 0.8788,
+      "step": 545
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019514346699292274,
+      "loss": 0.8521,
+      "step": 546
+    },
+    {
+      "epoch": 0.30435387397412716,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019512547751026648,
+      "loss": 0.8488,
+      "step": 547
+    },
+    {
+      "epoch": 0.3049102795938239,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019510745560331097,
+      "loss": 0.9655,
+      "step": 548
+    },
+    {
+      "epoch": 0.30546668521352066,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00019508940127819913,
+      "loss": 1.0282,
+      "step": 549
+    },
+    {
+      "epoch": 0.30602309083321744,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019507131454108497,
+      "loss": 1.0152,
+      "step": 550
+    },
+    {
+      "epoch": 0.30657949645291416,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001950531953981334,
+      "loss": 0.8472,
+      "step": 551
+    },
+    {
+      "epoch": 0.30713590207261093,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001950350438555206,
+      "loss": 0.8322,
+      "step": 552
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019501685991943357,
+      "loss": 0.9437,
+      "step": 553
+    },
+    {
+      "epoch": 0.30824871331200443,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001949986435960705,
+      "loss": 0.817,
+      "step": 554
+    },
+    {
+      "epoch": 0.3088051189317012,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019498039489164058,
+      "loss": 0.9954,
+      "step": 555
+    },
+    {
+      "epoch": 0.309361524551398,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019496211381236401,
+      "loss": 0.8284,
+      "step": 556
+    },
+    {
+      "epoch": 0.3099179301710947,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001949438003644721,
+      "loss": 0.8699,
+      "step": 557
+    },
+    {
+      "epoch": 0.3104743357907915,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019492545455420703,
+      "loss": 0.8738,
+      "step": 558
+    },
+    {
+      "epoch": 0.31103074141048825,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001949070763878222,
+      "loss": 0.894,
+      "step": 559
+    },
+    {
+      "epoch": 0.31158714703018503,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019488866587158196,
+      "loss": 0.8258,
+      "step": 560
+    },
+    {
+      "epoch": 0.31214355264988175,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019487022301176168,
+      "loss": 0.9347,
+      "step": 561
+    },
+    {
+      "epoch": 0.3126999582695785,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019485174781464773,
+      "loss": 0.9083,
+      "step": 562
+    },
+    {
+      "epoch": 0.3132563638892753,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019483324028653754,
+      "loss": 0.784,
+      "step": 563
+    },
+    {
+      "epoch": 0.313812769508972,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001948147004337396,
+      "loss": 0.8954,
+      "step": 564
+    },
+    {
+      "epoch": 0.3143691751286688,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001947961282625733,
+      "loss": 0.8873,
+      "step": 565
+    },
+    {
+      "epoch": 0.3149255807483656,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019477752377936917,
+      "loss": 0.9064,
+      "step": 566
+    },
+    {
+      "epoch": 0.3154819863680623,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019475888699046866,
+      "loss": 0.8424,
+      "step": 567
+    },
+    {
+      "epoch": 0.31603839198775907,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0001947402179022243,
+      "loss": 0.9908,
+      "step": 568
+    },
+    {
+      "epoch": 0.31659479760745585,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019472151652099963,
+      "loss": 0.9116,
+      "step": 569
+    },
+    {
+      "epoch": 0.31715120322715257,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001947027828531691,
+      "loss": 0.9074,
+      "step": 570
+    },
+    {
+      "epoch": 0.31770760884684934,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019468401690511832,
+      "loss": 0.9643,
+      "step": 571
+    },
+    {
+      "epoch": 0.3182640144665461,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019466521868324375,
+      "loss": 0.9002,
+      "step": 572
+    },
+    {
+      "epoch": 0.3188204200862429,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001946463881939529,
+      "loss": 0.9181,
+      "step": 573
+    },
+    {
+      "epoch": 0.3193768257059396,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019462752544366436,
+      "loss": 0.9275,
+      "step": 574
+    },
+    {
+      "epoch": 0.3199332313256364,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001946086304388076,
+      "loss": 0.8565,
+      "step": 575
+    },
+    {
+      "epoch": 0.32048963694533317,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001945897031858232,
+      "loss": 0.9334,
+      "step": 576
+    },
+    {
+      "epoch": 0.3210460425650299,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001945707436911626,
+      "loss": 0.9068,
+      "step": 577
+    },
+    {
+      "epoch": 0.32160244818472666,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019455175196128838,
+      "loss": 0.8463,
+      "step": 578
+    },
+    {
+      "epoch": 0.32215885380442344,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019453272800267393,
+      "loss": 0.8119,
+      "step": 579
+    },
+    {
+      "epoch": 0.32271525942412016,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019451367182180378,
+      "loss": 0.9386,
+      "step": 580
+    },
+    {
+      "epoch": 0.32327166504381694,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019449458342517338,
+      "loss": 0.9527,
+      "step": 581
+    },
+    {
+      "epoch": 0.3238280706635137,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019447546281928916,
+      "loss": 0.9448,
+      "step": 582
+    },
+    {
+      "epoch": 0.3243844762832105,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019445631001066853,
+      "loss": 0.8528,
+      "step": 583
+    },
+    {
+      "epoch": 0.3249408819029072,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019443712500583993,
+      "loss": 0.9443,
+      "step": 584
+    },
+    {
+      "epoch": 0.325497287522604,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019441790781134265,
+      "loss": 0.9369,
+      "step": 585
+    },
+    {
+      "epoch": 0.32605369314230076,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001943986584337271,
+      "loss": 0.8631,
+      "step": 586
+    },
+    {
+      "epoch": 0.3266100987619975,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001943793768795545,
+      "loss": 0.921,
+      "step": 587
+    },
+    {
+      "epoch": 0.32716650438169426,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001943600631553972,
+      "loss": 0.8827,
+      "step": 588
+    },
+    {
+      "epoch": 0.32772291000139103,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019434071726783843,
+      "loss": 0.9897,
+      "step": 589
+    },
+    {
+      "epoch": 0.32827931562108775,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001943213392234724,
+      "loss": 0.7868,
+      "step": 590
+    },
+    {
+      "epoch": 0.32883572124078453,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019430192902890423,
+      "loss": 0.8787,
+      "step": 591
+    },
+    {
+      "epoch": 0.3293921268604813,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001942824866907501,
+      "loss": 0.861,
+      "step": 592
+    },
+    {
+      "epoch": 0.329948532480178,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019426301221563702,
+      "loss": 0.8415,
+      "step": 593
+    },
+    {
+      "epoch": 0.3305049380998748,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019424350561020312,
+      "loss": 0.9665,
+      "step": 594
+    },
+    {
+      "epoch": 0.3310613437195716,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001942239668810973,
+      "loss": 0.934,
+      "step": 595
+    },
+    {
+      "epoch": 0.33161774933926835,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019420439603497956,
+      "loss": 0.9518,
+      "step": 596
+    },
+    {
+      "epoch": 0.3321741549589651,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019418479307852074,
+      "loss": 0.8112,
+      "step": 597
+    },
+    {
+      "epoch": 0.33273056057866185,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001941651580184027,
+      "loss": 0.7649,
+      "step": 598
+    },
+    {
+      "epoch": 0.3332869661983586,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019414549086131815,
+      "loss": 0.9321,
+      "step": 599
+    },
+    {
+      "epoch": 0.33384337181805535,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001941257916139709,
+      "loss": 0.9537,
+      "step": 600
+    },
+    {
+      "epoch": 0.3343997774377521,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001941060602830755,
+      "loss": 0.961,
+      "step": 601
+    },
+    {
+      "epoch": 0.3349561830574489,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019408629687535763,
+      "loss": 0.9001,
+      "step": 602
+    },
+    {
+      "epoch": 0.3355125886771456,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00019406650139755375,
+      "loss": 0.8772,
+      "step": 603
+    },
+    {
+      "epoch": 0.3360689942968424,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019404667385641128,
+      "loss": 0.9444,
+      "step": 604
+    },
+    {
+      "epoch": 0.33662539991653917,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001940268142586887,
+      "loss": 0.8729,
+      "step": 605
+    },
+    {
+      "epoch": 0.3371818055362359,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001940069226111552,
+      "loss": 0.8825,
+      "step": 606
+    },
+    {
+      "epoch": 0.33773821115593267,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019398699892059112,
+      "loss": 0.9166,
+      "step": 607
+    },
+    {
+      "epoch": 0.33829461677562944,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019396704319378754,
+      "loss": 0.9447,
+      "step": 608
+    },
+    {
+      "epoch": 0.3388510223953262,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001939470554375466,
+      "loss": 0.932,
+      "step": 609
+    },
+    {
+      "epoch": 0.33940742801502294,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001939270356586812,
+      "loss": 0.8981,
+      "step": 610
+    },
+    {
+      "epoch": 0.3399638336347197,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019390698386401533,
+      "loss": 0.8561,
+      "step": 611
+    },
+    {
+      "epoch": 0.3405202392544165,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019388690006038375,
+      "loss": 0.8992,
+      "step": 612
+    },
+    {
+      "epoch": 0.3410766448741132,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019386678425463221,
+      "loss": 0.952,
+      "step": 613
+    },
+    {
+      "epoch": 0.34163305049381,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019384663645361736,
+      "loss": 0.8244,
+      "step": 614
+    },
+    {
+      "epoch": 0.34218945611350676,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019382645666420674,
+      "loss": 0.9347,
+      "step": 615
+    },
+    {
+      "epoch": 0.3427458617332035,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019380624489327876,
+      "loss": 0.9098,
+      "step": 616
+    },
+    {
+      "epoch": 0.34330226735290026,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00019378600114772283,
+      "loss": 1.9243,
+      "step": 617
+    },
+    {
+      "epoch": 0.34385867297259703,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00019376572543443915,
+      "loss": 0.8078,
+      "step": 618
+    },
+    {
+      "epoch": 0.34441507859229376,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019374541776033888,
+      "loss": 0.8494,
+      "step": 619
+    },
+    {
+      "epoch": 0.34497148421199053,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019372507813234405,
+      "loss": 0.8841,
+      "step": 620
+    },
+    {
+      "epoch": 0.3455278898316873,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001937047065573876,
+      "loss": 0.8424,
+      "step": 621
+    },
+    {
+      "epoch": 0.3460842954513841,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001936843030424133,
+      "loss": 0.8028,
+      "step": 622
+    },
+    {
+      "epoch": 0.3466407010710808,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019366386759437596,
+      "loss": 0.8883,
+      "step": 623
+    },
+    {
+      "epoch": 0.3471971066907776,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001936434002202411,
+      "loss": 0.7923,
+      "step": 624
+    },
+    {
+      "epoch": 0.34775351231047436,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019362290092698515,
+      "loss": 0.8398,
+      "step": 625
+    },
+    {
+      "epoch": 0.3483099179301711,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001936023697215956,
+      "loss": 0.866,
+      "step": 626
+    },
+    {
+      "epoch": 0.34886632354986785,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019358180661107051,
+      "loss": 0.8925,
+      "step": 627
+    },
+    {
+      "epoch": 0.3494227291695646,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001935612116024191,
+      "loss": 0.8689,
+      "step": 628
+    },
+    {
+      "epoch": 0.34997913478926135,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001935405847026613,
+      "loss": 0.8397,
+      "step": 629
+    },
+    {
+      "epoch": 0.3505355404089581,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019351992591882798,
+      "loss": 0.8053,
+      "step": 630
+    },
+    {
+      "epoch": 0.3510919460286549,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019349923525796084,
+      "loss": 0.8547,
+      "step": 631
+    },
+    {
+      "epoch": 0.3516483516483517,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019347851272711247,
+      "loss": 0.944,
+      "step": 632
+    },
+    {
+      "epoch": 0.3522047572680484,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019345775833334633,
+      "loss": 0.8962,
+      "step": 633
+    },
+    {
+      "epoch": 0.35276116288774517,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001934369720837367,
+      "loss": 0.8379,
+      "step": 634
+    },
+    {
+      "epoch": 0.35331756850744195,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019341615398536877,
+      "loss": 0.9129,
+      "step": 635
+    },
+    {
+      "epoch": 0.35387397412713867,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00019339530404533854,
+      "loss": 0.8244,
+      "step": 636
+    },
+    {
+      "epoch": 0.35443037974683544,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019337442227075285,
+      "loss": 0.8405,
+      "step": 637
+    },
+    {
+      "epoch": 0.3549867853665322,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001933535086687295,
+      "loss": 0.906,
+      "step": 638
+    },
+    {
+      "epoch": 0.35554319098622894,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.000193332563246397,
+      "loss": 0.9176,
+      "step": 639
+    },
+    {
+      "epoch": 0.3560995966059257,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001933115860108948,
+      "loss": 0.8164,
+      "step": 640
+    },
+    {
+      "epoch": 0.3566560022256225,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019329057696937317,
+      "loss": 0.8414,
+      "step": 641
+    },
+    {
+      "epoch": 0.3572124078453192,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019326953612899318,
+      "loss": 0.9027,
+      "step": 642
+    },
+    {
+      "epoch": 0.357768813465016,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001932484634969268,
+      "loss": 0.8986,
+      "step": 643
+    },
+    {
+      "epoch": 0.35832521908471276,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019322735908035678,
+      "loss": 0.891,
+      "step": 644
+    },
+    {
+      "epoch": 0.35888162470440954,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001932062228864768,
+      "loss": 0.9787,
+      "step": 645
+    },
+    {
+      "epoch": 0.35943803032410626,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001931850549224912,
+      "loss": 0.8145,
+      "step": 646
+    },
+    {
+      "epoch": 0.35999443594380304,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019316385519561538,
+      "loss": 0.8913,
+      "step": 647
+    },
+    {
+      "epoch": 0.3605508415634998,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019314262371307532,
+      "loss": 0.7981,
+      "step": 648
+    },
+    {
+      "epoch": 0.36110724718319653,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019312136048210801,
+      "loss": 0.8772,
+      "step": 649
+    },
+    {
+      "epoch": 0.3616636528028933,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0001931000655099612,
+      "loss": 0.8837,
+      "step": 650
+    },
+    {
+      "epoch": 0.3622200584225901,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019307873880389345,
+      "loss": 0.923,
+      "step": 651
+    },
+    {
+      "epoch": 0.3627764640422868,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019305738037117412,
+      "loss": 0.8609,
+      "step": 652
+    },
+    {
+      "epoch": 0.3633328696619836,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019303599021908342,
+      "loss": 0.8668,
+      "step": 653
+    },
+    {
+      "epoch": 0.36388927528168036,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019301456835491237,
+      "loss": 0.9394,
+      "step": 654
+    },
+    {
+      "epoch": 0.3644456809013771,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001929931147859628,
+      "loss": 0.7789,
+      "step": 655
+    },
+    {
+      "epoch": 0.36500208652107385,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019297162951954728,
+      "loss": 0.9009,
+      "step": 656
+    },
+    {
+      "epoch": 0.36555849214077063,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019295011256298929,
+      "loss": 0.7917,
+      "step": 657
+    },
+    {
+      "epoch": 0.3661148977604674,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019292856392362303,
+      "loss": 0.9266,
+      "step": 658
+    },
+    {
+      "epoch": 0.3666713033801641,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019290698360879355,
+      "loss": 0.849,
+      "step": 659
+    },
+    {
+      "epoch": 0.3672277089998609,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001928853716258567,
+      "loss": 0.8509,
+      "step": 660
+    },
+    {
+      "epoch": 0.3677841146195577,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00019286372798217905,
+      "loss": 0.8732,
+      "step": 661
+    },
+    {
+      "epoch": 0.3683405202392544,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019284205268513803,
+      "loss": 0.8759,
+      "step": 662
+    },
+    {
+      "epoch": 0.3688969258589512,
+      "grad_norm": 1.8046875,
+      "learning_rate": 0.00019282034574212185,
+      "loss": 1.8908,
+      "step": 663
+    },
+    {
+      "epoch": 0.36945333147864795,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019279860716052947,
+      "loss": 0.8435,
+      "step": 664
+    },
+    {
+      "epoch": 0.37000973709834467,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019277683694777074,
+      "loss": 0.8151,
+      "step": 665
+    },
+    {
+      "epoch": 0.37056614271804145,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019275503511126618,
+      "loss": 0.9638,
+      "step": 666
+    },
+    {
+      "epoch": 0.3711225483377382,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019273320165844707,
+      "loss": 0.9098,
+      "step": 667
+    },
+    {
+      "epoch": 0.37167895395743494,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019271133659675563,
+      "loss": 0.9032,
+      "step": 668
+    },
+    {
+      "epoch": 0.3722353595771317,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019268943993364462,
+      "loss": 0.7979,
+      "step": 669
+    },
+    {
+      "epoch": 0.3727917651968285,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019266751167657778,
+      "loss": 0.9499,
+      "step": 670
+    },
+    {
+      "epoch": 0.37334817081652527,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001926455518330295,
+      "loss": 0.8741,
+      "step": 671
+    },
+    {
+      "epoch": 0.373904576436222,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.000192623560410485,
+      "loss": 0.9162,
+      "step": 672
+    },
+    {
+      "epoch": 0.37446098205591877,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019260153741644021,
+      "loss": 0.8135,
+      "step": 673
+    },
+    {
+      "epoch": 0.37501738767561554,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00019257948285840188,
+      "loss": 0.6664,
+      "step": 674
+    },
+    {
+      "epoch": 0.37557379329531226,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019255739674388746,
+      "loss": 0.9409,
+      "step": 675
+    },
+    {
+      "epoch": 0.37613019891500904,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019253527908042522,
+      "loss": 0.9325,
+      "step": 676
+    },
+    {
+      "epoch": 0.3766866045347058,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019251312987555408,
+      "loss": 0.8681,
+      "step": 677
+    },
+    {
+      "epoch": 0.37724301015440254,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019249094913682382,
+      "loss": 0.9045,
+      "step": 678
+    },
+    {
+      "epoch": 0.3777994157740993,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019246873687179495,
+      "loss": 0.8807,
+      "step": 679
+    },
+    {
+      "epoch": 0.3783558213937961,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019244649308803867,
+      "loss": 0.9102,
+      "step": 680
+    },
+    {
+      "epoch": 0.37891222701349286,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019242421779313692,
+      "loss": 0.8124,
+      "step": 681
+    },
+    {
+      "epoch": 0.3794686326331896,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019240191099468253,
+      "loss": 0.9337,
+      "step": 682
+    },
+    {
+      "epoch": 0.38002503825288636,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00019237957270027884,
+      "loss": 1.8244,
+      "step": 683
+    },
+    {
+      "epoch": 0.38058144387258314,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001923572029175401,
+      "loss": 0.8975,
+      "step": 684
+    },
+    {
+      "epoch": 0.38113784949227986,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001923348016540912,
+      "loss": 0.8755,
+      "step": 685
+    },
+    {
+      "epoch": 0.38169425511197663,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019231236891756787,
+      "loss": 0.8125,
+      "step": 686
+    },
+    {
+      "epoch": 0.3822506607316734,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019228990471561636,
+      "loss": 0.8673,
+      "step": 687
+    },
+    {
+      "epoch": 0.38280706635137013,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001922674090558939,
+      "loss": 0.8462,
+      "step": 688
+    },
+    {
+      "epoch": 0.3833634719710669,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019224488194606826,
+      "loss": 1.0229,
+      "step": 689
+    },
+    {
+      "epoch": 0.3839198775907637,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019222232339381802,
+      "loss": 0.8654,
+      "step": 690
+    },
+    {
+      "epoch": 0.3844762832104604,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019219973340683244,
+      "loss": 0.877,
+      "step": 691
+    },
+    {
+      "epoch": 0.3850326888301572,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019217711199281148,
+      "loss": 0.9084,
+      "step": 692
+    },
+    {
+      "epoch": 0.38558909444985395,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019215445915946584,
+      "loss": 0.9667,
+      "step": 693
+    },
+    {
+      "epoch": 0.38614550006955073,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00019213177491451692,
+      "loss": 0.8251,
+      "step": 694
+    },
+    {
+      "epoch": 0.38670190568924745,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019210905926569688,
+      "loss": 0.8764,
+      "step": 695
+    },
+    {
+      "epoch": 0.3872583113089442,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0001920863122207485,
+      "loss": 0.8484,
+      "step": 696
+    },
+    {
+      "epoch": 0.387814716928641,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00019206353378742528,
+      "loss": 1.0047,
+      "step": 697
+    },
+    {
+      "epoch": 0.3883711225483377,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019204072397349144,
+      "loss": 0.8996,
+      "step": 698
+    },
+    {
+      "epoch": 0.3889275281680345,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019201788278672193,
+      "loss": 0.9355,
+      "step": 699
+    },
+    {
+      "epoch": 0.3894839337877313,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019199501023490232,
+      "loss": 0.8971,
+      "step": 700
+    },
+    {
+      "epoch": 0.390040339407428,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001919721063258289,
+      "loss": 0.8437,
+      "step": 701
+    },
+    {
+      "epoch": 0.39059674502712477,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001919491710673087,
+      "loss": 0.9162,
+      "step": 702
+    },
+    {
+      "epoch": 0.39115315064682155,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019192620446715933,
+      "loss": 0.9021,
+      "step": 703
+    },
+    {
+      "epoch": 0.39170955626651827,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00019190320653320918,
+      "loss": 0.8099,
+      "step": 704
+    },
+    {
+      "epoch": 0.39226596188621504,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001918801772732973,
+      "loss": 0.9569,
+      "step": 705
+    },
+    {
+      "epoch": 0.3928223675059118,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001918571166952734,
+      "loss": 0.9673,
+      "step": 706
+    },
+    {
+      "epoch": 0.3933787731256086,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001918340248069978,
+      "loss": 0.8443,
+      "step": 707
+    },
+    {
+      "epoch": 0.3939351787453053,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019181090161634167,
+      "loss": 0.9339,
+      "step": 708
+    },
+    {
+      "epoch": 0.3944915843650021,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019178774713118663,
+      "loss": 0.834,
+      "step": 709
+    },
+    {
+      "epoch": 0.39504798998469887,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019176456135942515,
+      "loss": 0.9933,
+      "step": 710
+    },
+    {
+      "epoch": 0.3956043956043956,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019174134430896027,
+      "loss": 0.924,
+      "step": 711
+    },
+    {
+      "epoch": 0.39616080122409236,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001917180959877057,
+      "loss": 0.8976,
+      "step": 712
+    },
+    {
+      "epoch": 0.39671720684378914,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019169481640358583,
+      "loss": 0.8488,
+      "step": 713
+    },
+    {
+      "epoch": 0.39727361246348586,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019167150556453569,
+      "loss": 0.8426,
+      "step": 714
+    },
+    {
+      "epoch": 0.39783001808318263,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019164816347850096,
+      "loss": 1.0289,
+      "step": 715
+    },
+    {
+      "epoch": 0.3983864237028794,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019162479015343803,
+      "loss": 0.8953,
+      "step": 716
+    },
+    {
+      "epoch": 0.39894282932257613,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019160138559731386,
+      "loss": 0.9065,
+      "step": 717
+    },
+    {
+      "epoch": 0.3994992349422729,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019157794981810607,
+      "loss": 0.908,
+      "step": 718
+    },
+    {
+      "epoch": 0.4000556405619697,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019155448282380294,
+      "loss": 0.796,
+      "step": 719
+    },
+    {
+      "epoch": 0.40061204618166646,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019153098462240342,
+      "loss": 0.9603,
+      "step": 720
+    },
+    {
+      "epoch": 0.4011684518013632,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019150745522191701,
+      "loss": 0.9096,
+      "step": 721
+    },
+    {
+      "epoch": 0.40172485742105996,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019148389463036395,
+      "loss": 1.0184,
+      "step": 722
+    },
+    {
+      "epoch": 0.40228126304075673,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019146030285577506,
+      "loss": 0.8274,
+      "step": 723
+    },
+    {
+      "epoch": 0.40283766866045345,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019143667990619177,
+      "loss": 0.9454,
+      "step": 724
+    },
+    {
+      "epoch": 0.4033940742801502,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019141302578966614,
+      "loss": 0.8516,
+      "step": 725
+    },
+    {
+      "epoch": 0.403950479899847,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019138934051426092,
+      "loss": 0.9234,
+      "step": 726
+    },
+    {
+      "epoch": 0.4045068855195437,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019136562408804934,
+      "loss": 0.8836,
+      "step": 727
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019134187651911546,
+      "loss": 0.9283,
+      "step": 728
+    },
+    {
+      "epoch": 0.4056196967589373,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019131809781555373,
+      "loss": 0.9731,
+      "step": 729
+    },
+    {
+      "epoch": 0.40617610237863405,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00019129428798546938,
+      "loss": 0.7996,
+      "step": 730
+    },
+    {
+      "epoch": 0.4067325079983308,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019127044703697816,
+      "loss": 0.8759,
+      "step": 731
+    },
+    {
+      "epoch": 0.40728891361802755,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019124657497820644,
+      "loss": 0.9137,
+      "step": 732
+    },
+    {
+      "epoch": 0.4078453192377243,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019122267181729123,
+      "loss": 0.8381,
+      "step": 733
+    },
+    {
+      "epoch": 0.40840172485742104,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001911987375623801,
+      "loss": 0.9209,
+      "step": 734
+    },
+    {
+      "epoch": 0.4089581304771178,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019117477222163127,
+      "loss": 0.8937,
+      "step": 735
+    },
+    {
+      "epoch": 0.4095145360968146,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001911507758032135,
+      "loss": 0.9494,
+      "step": 736
+    },
+    {
+      "epoch": 0.4100709417165113,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019112674831530616,
+      "loss": 0.9375,
+      "step": 737
+    },
+    {
+      "epoch": 0.4106273473362081,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019110268976609923,
+      "loss": 0.8973,
+      "step": 738
+    },
+    {
+      "epoch": 0.41118375295590487,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00019107860016379324,
+      "loss": 0.9579,
+      "step": 739
+    },
+    {
+      "epoch": 0.4117401585756016,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019105447951659935,
+      "loss": 0.7406,
+      "step": 740
+    },
+    {
+      "epoch": 0.41229656419529837,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001910303278327393,
+      "loss": 0.9315,
+      "step": 741
+    },
+    {
+      "epoch": 0.41285296981499514,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019100614512044536,
+      "loss": 0.8731,
+      "step": 742
+    },
+    {
+      "epoch": 0.4134093754346919,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001909819313879604,
+      "loss": 0.8679,
+      "step": 743
+    },
+    {
+      "epoch": 0.41396578105438864,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019095768664353785,
+      "loss": 0.8881,
+      "step": 744
+    },
+    {
+      "epoch": 0.4145221866740854,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001909334108954418,
+      "loss": 0.8616,
+      "step": 745
+    },
+    {
+      "epoch": 0.4150785922937822,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019090910415194682,
+      "loss": 0.8466,
+      "step": 746
+    },
+    {
+      "epoch": 0.4156349979134789,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.000190884766421338,
+      "loss": 0.963,
+      "step": 747
+    },
+    {
+      "epoch": 0.4161914035331757,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019086039771191112,
+      "loss": 0.8879,
+      "step": 748
+    },
+    {
+      "epoch": 0.41674780915287246,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001908359980319724,
+      "loss": 0.9345,
+      "step": 749
+    },
+    {
+      "epoch": 0.4173042147725692,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019081156738983874,
+      "loss": 0.9356,
+      "step": 750
+    },
+    {
+      "epoch": 0.41786062039226596,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019078710579383746,
+      "loss": 0.8295,
+      "step": 751
+    },
+    {
+      "epoch": 0.41841702601196273,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019076261325230655,
+      "loss": 0.9092,
+      "step": 752
+    },
+    {
+      "epoch": 0.41897343163165945,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00019073808977359446,
+      "loss": 0.83,
+      "step": 753
+    },
+    {
+      "epoch": 0.41952983725135623,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0001907135353660602,
+      "loss": 0.7945,
+      "step": 754
+    },
+    {
+      "epoch": 0.420086242871053,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019068895003807339,
+      "loss": 0.959,
+      "step": 755
+    },
+    {
+      "epoch": 0.4206426484907498,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001906643337980141,
+      "loss": 0.8404,
+      "step": 756
+    },
+    {
+      "epoch": 0.4211990541104465,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.000190639686654273,
+      "loss": 0.811,
+      "step": 757
+    },
+    {
+      "epoch": 0.4217554597301433,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019061500861525127,
+      "loss": 0.9065,
+      "step": 758
+    },
+    {
+      "epoch": 0.42231186534984005,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019059029968936062,
+      "loss": 0.8662,
+      "step": 759
+    },
+    {
+      "epoch": 0.4228682709695368,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001905655598850233,
+      "loss": 0.7609,
+      "step": 760
+    },
+    {
+      "epoch": 0.42342467658923355,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019054078921067205,
+      "loss": 0.8826,
+      "step": 761
+    },
+    {
+      "epoch": 0.4239810822089303,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019051598767475016,
+      "loss": 0.8253,
+      "step": 762
+    },
+    {
+      "epoch": 0.42453748782862705,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00019049115528571147,
+      "loss": 0.7537,
+      "step": 763
+    },
+    {
+      "epoch": 0.4250938934483238,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019046629205202027,
+      "loss": 0.9363,
+      "step": 764
+    },
+    {
+      "epoch": 0.4256502990680206,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019044139798215143,
+      "loss": 0.8248,
+      "step": 765
+    },
+    {
+      "epoch": 0.4262067046877173,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00019041647308459022,
+      "loss": 0.7364,
+      "step": 766
+    },
+    {
+      "epoch": 0.4267631103074141,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019039151736783262,
+      "loss": 0.9075,
+      "step": 767
+    },
+    {
+      "epoch": 0.42731951592711087,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019036653084038486,
+      "loss": 0.966,
+      "step": 768
+    },
+    {
+      "epoch": 0.42787592154680765,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019034151351076387,
+      "loss": 0.8645,
+      "step": 769
+    },
+    {
+      "epoch": 0.42843232716650437,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019031646538749707,
+      "loss": 0.824,
+      "step": 770
+    },
+    {
+      "epoch": 0.42898873278620114,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019029138647912218,
+      "loss": 0.8858,
+      "step": 771
+    },
+    {
+      "epoch": 0.4295451384058979,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019026627679418767,
+      "loss": 0.8066,
+      "step": 772
+    },
+    {
+      "epoch": 0.43010154402559464,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00019024113634125234,
+      "loss": 0.8256,
+      "step": 773
+    },
+    {
+      "epoch": 0.4306579496452914,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001902159651288855,
+      "loss": 0.8785,
+      "step": 774
+    },
+    {
+      "epoch": 0.4312143552649882,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.000190190763165667,
+      "loss": 0.8643,
+      "step": 775
+    },
+    {
+      "epoch": 0.4317707608846849,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019016553046018712,
+      "loss": 0.9032,
+      "step": 776
+    },
+    {
+      "epoch": 0.4323271665043817,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019014026702104662,
+      "loss": 0.8785,
+      "step": 777
+    },
+    {
+      "epoch": 0.43288357212407846,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019011497285685678,
+      "loss": 0.8607,
+      "step": 778
+    },
+    {
+      "epoch": 0.4334399777437752,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001900896479762393,
+      "loss": 0.8355,
+      "step": 779
+    },
+    {
+      "epoch": 0.43399638336347196,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019006429238782637,
+      "loss": 0.9727,
+      "step": 780
+    },
+    {
+      "epoch": 0.43455278898316874,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019003890610026067,
+      "loss": 0.8807,
+      "step": 781
+    },
+    {
+      "epoch": 0.4351091946028655,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019001348912219533,
+      "loss": 0.9325,
+      "step": 782
+    },
+    {
+      "epoch": 0.43566560022256223,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018998804146229387,
+      "loss": 0.8559,
+      "step": 783
+    },
+    {
+      "epoch": 0.436222005842259,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018996256312923043,
+      "loss": 0.8304,
+      "step": 784
+    },
+    {
+      "epoch": 0.4367784114619558,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00018993705413168944,
+      "loss": 0.9767,
+      "step": 785
+    },
+    {
+      "epoch": 0.4373348170816525,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001899115144783659,
+      "loss": 0.8711,
+      "step": 786
+    },
+    {
+      "epoch": 0.4378912227013493,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018988594417796516,
+      "loss": 0.8525,
+      "step": 787
+    },
+    {
+      "epoch": 0.43844762832104606,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018986034323920312,
+      "loss": 0.83,
+      "step": 788
+    },
+    {
+      "epoch": 0.4390040339407428,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.000189834711670806,
+      "loss": 0.8088,
+      "step": 789
+    },
+    {
+      "epoch": 0.43956043956043955,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00018980904948151062,
+      "loss": 0.784,
+      "step": 790
+    },
+    {
+      "epoch": 0.44011684518013633,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018978335668006407,
+      "loss": 0.9129,
+      "step": 791
+    },
+    {
+      "epoch": 0.4406732507998331,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.000189757633275224,
+      "loss": 1.0234,
+      "step": 792
+    },
+    {
+      "epoch": 0.4412296564195298,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001897318792757584,
+      "loss": 0.8711,
+      "step": 793
+    },
+    {
+      "epoch": 0.4417860620392266,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018970609469044577,
+      "loss": 0.9124,
+      "step": 794
+    },
+    {
+      "epoch": 0.4423424676589234,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00018968027952807494,
+      "loss": 0.7081,
+      "step": 795
+    },
+    {
+      "epoch": 0.4428988732786201,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00018965443379744535,
+      "loss": 0.7776,
+      "step": 796
+    },
+    {
+      "epoch": 0.4434552788983169,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00018962855750736656,
+      "loss": 0.8216,
+      "step": 797
+    },
+    {
+      "epoch": 0.44401168451801365,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018960265066665882,
+      "loss": 0.8406,
+      "step": 798
+    },
+    {
+      "epoch": 0.44456809013771037,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001895767132841527,
+      "loss": 0.9074,
+      "step": 799
+    },
+    {
+      "epoch": 0.44512449575740715,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00018955074536868913,
+      "loss": 0.88,
+      "step": 800
+    },
+    {
+      "epoch": 0.4456809013771039,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001895247469291195,
+      "loss": 0.9454,
+      "step": 801
+    },
+    {
+      "epoch": 0.44623730699680064,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00018949871797430557,
+      "loss": 0.7475,
+      "step": 802
+    },
+    {
+      "epoch": 0.4467937126164974,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00018947265851311954,
+      "loss": 0.9466,
+      "step": 803
+    },
+    {
+      "epoch": 0.4473501182361942,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.000189446568554444,
+      "loss": 0.791,
+      "step": 804
+    },
+    {
+      "epoch": 0.44790652385589097,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018942044810717188,
+      "loss": 0.898,
+      "step": 805
+    },
+    {
+      "epoch": 0.4484629294755877,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00018939429718020661,
+      "loss": 0.7666,
+      "step": 806
+    },
+    {
+      "epoch": 0.44901933509528447,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001893681157824619,
+      "loss": 0.9343,
+      "step": 807
+    },
+    {
+      "epoch": 0.44957574071498124,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018934190392286198,
+      "loss": 0.8074,
+      "step": 808
+    },
+    {
+      "epoch": 0.45013214633467796,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00018931566161034126,
+      "loss": 0.8394,
+      "step": 809
+    },
+    {
+      "epoch": 0.45068855195437474,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018928938885384472,
+      "loss": 0.8655,
+      "step": 810
+    },
+    {
+      "epoch": 0.4512449575740715,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00018926308566232763,
+      "loss": 0.792,
+      "step": 811
+    },
+    {
+      "epoch": 0.45180136319376824,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018923675204475564,
+      "loss": 0.8638,
+      "step": 812
+    },
+    {
+      "epoch": 0.452357768813465,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018921038801010477,
+      "loss": 0.8151,
+      "step": 813
+    },
+    {
+      "epoch": 0.4529141744331618,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018918399356736146,
+      "loss": 0.8198,
+      "step": 814
+    },
+    {
+      "epoch": 0.4534705800528585,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00018915756872552242,
+      "loss": 0.6904,
+      "step": 815
+    },
+    {
+      "epoch": 0.4540269856725553,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001891311134935948,
+      "loss": 0.8762,
+      "step": 816
+    },
+    {
+      "epoch": 0.45458339129225206,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0001891046278805961,
+      "loss": 0.7527,
+      "step": 817
+    },
+    {
+      "epoch": 0.45513979691194884,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018907811189555412,
+      "loss": 0.7777,
+      "step": 818
+    },
+    {
+      "epoch": 0.45569620253164556,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018905156554750708,
+      "loss": 0.8556,
+      "step": 819
+    },
+    {
+      "epoch": 0.45625260815134233,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001890249888455035,
+      "loss": 0.8273,
+      "step": 820
+    },
+    {
+      "epoch": 0.4568090137710391,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001889983817986023,
+      "loss": 0.8405,
+      "step": 821
+    },
+    {
+      "epoch": 0.45736541939073583,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018897174441587266,
+      "loss": 0.9006,
+      "step": 822
+    },
+    {
+      "epoch": 0.4579218250104326,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018894507670639418,
+      "loss": 0.9191,
+      "step": 823
+    },
+    {
+      "epoch": 0.4584782306301294,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00018891837867925678,
+      "loss": 0.9777,
+      "step": 824
+    },
+    {
+      "epoch": 0.4590346362498261,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018889165034356072,
+      "loss": 0.892,
+      "step": 825
+    },
+    {
+      "epoch": 0.4595910418695229,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00018886489170841648,
+      "loss": 0.8615,
+      "step": 826
+    },
+    {
+      "epoch": 0.46014744748921965,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018883810278294503,
+      "loss": 0.8765,
+      "step": 827
+    },
+    {
+      "epoch": 0.4607038531089164,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018881128357627763,
+      "loss": 0.8803,
+      "step": 828
+    },
+    {
+      "epoch": 0.46126025872861315,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018878443409755575,
+      "loss": 0.9017,
+      "step": 829
+    },
+    {
+      "epoch": 0.4618166643483099,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00018875755435593129,
+      "loss": 0.8139,
+      "step": 830
+    },
+    {
+      "epoch": 0.4623730699680067,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001887306443605664,
+      "loss": 0.9611,
+      "step": 831
+    },
+    {
+      "epoch": 0.4629294755877034,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018870370412063365,
+      "loss": 0.9311,
+      "step": 832
+    },
+    {
+      "epoch": 0.4634858812074002,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00018867673364531578,
+      "loss": 0.9108,
+      "step": 833
+    },
+    {
+      "epoch": 0.464042286827097,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00018864973294380587,
+      "loss": 1.0142,
+      "step": 834
+    },
+    {
+      "epoch": 0.4645986924467937,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018862270202530738,
+      "loss": 0.744,
+      "step": 835
+    },
+    {
+      "epoch": 0.46515509806649047,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018859564089903404,
+      "loss": 0.8419,
+      "step": 836
+    },
+    {
+      "epoch": 0.46571150368618724,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00018856854957420983,
+      "loss": 0.7806,
+      "step": 837
+    },
+    {
+      "epoch": 0.46626790930588397,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00018854142806006902,
+      "loss": 0.9135,
+      "step": 838
+    },
+    {
+      "epoch": 0.46682431492558074,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00018851427636585623,
+      "loss": 0.836,
+      "step": 839
+    },
+    {
+      "epoch": 0.4673807205452775,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018848709450082637,
+      "loss": 0.8926,
+      "step": 840
+    },
+    {
+      "epoch": 0.4679371261649743,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001884598824742445,
+      "loss": 0.8962,
+      "step": 841
+    },
+    {
+      "epoch": 0.468493531784671,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001884326402953862,
+      "loss": 0.8185,
+      "step": 842
+    },
+    {
+      "epoch": 0.4690499374043678,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001884053679735371,
+      "loss": 0.948,
+      "step": 843
+    },
+    {
+      "epoch": 0.46960634302406457,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001883780655179932,
+      "loss": 1.0652,
+      "step": 844
+    },
+    {
+      "epoch": 0.4701627486437613,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00018835073293806075,
+      "loss": 0.9291,
+      "step": 845
+    },
+    {
+      "epoch": 0.47071915426345806,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018832337024305638,
+      "loss": 0.9503,
+      "step": 846
+    },
+    {
+      "epoch": 0.47127555988315484,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018829597744230678,
+      "loss": 0.8685,
+      "step": 847
+    },
+    {
+      "epoch": 0.47183196550285156,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018826855454514907,
+      "loss": 0.8796,
+      "step": 848
+    },
+    {
+      "epoch": 0.47238837112254833,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00018824110156093053,
+      "loss": 0.7906,
+      "step": 849
+    },
+    {
+      "epoch": 0.4729447767422451,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018821361849900877,
+      "loss": 0.9085,
+      "step": 850
+    },
+    {
+      "epoch": 0.47350118236194183,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00018818610536875164,
+      "loss": 0.9018,
+      "step": 851
+    },
+    {
+      "epoch": 0.4740575879816386,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00018815856217953714,
+      "loss": 0.8632,
+      "step": 852
+    },
+    {
+      "epoch": 0.4746139936013354,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018813098894075365,
+      "loss": 0.9047,
+      "step": 853
+    },
+    {
+      "epoch": 0.47517039922103216,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018810338566179973,
+      "loss": 0.9046,
+      "step": 854
+    },
+    {
+      "epoch": 0.4757268048407289,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018807575235208415,
+      "loss": 0.8683,
+      "step": 855
+    },
+    {
+      "epoch": 0.47628321046042565,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00018804808902102597,
+      "loss": 0.8541,
+      "step": 856
+    },
+    {
+      "epoch": 0.47683961608012243,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00018802039567805448,
+      "loss": 0.8484,
+      "step": 857
+    },
+    {
+      "epoch": 0.47739602169981915,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018799267233260918,
+      "loss": 0.8559,
+      "step": 858
+    },
+    {
+      "epoch": 0.4779524273195159,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018796491899413976,
+      "loss": 0.8646,
+      "step": 859
+    },
+    {
+      "epoch": 0.4785088329392127,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001879371356721062,
+      "loss": 0.9625,
+      "step": 860
+    },
+    {
+      "epoch": 0.4790652385589094,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001879093223759787,
+      "loss": 0.7927,
+      "step": 861
+    },
+    {
+      "epoch": 0.4796216441786062,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00018788147911523762,
+      "loss": 0.8716,
+      "step": 862
+    },
+    {
+      "epoch": 0.480178049798303,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00018785360589937354,
+      "loss": 0.7913,
+      "step": 863
+    },
+    {
+      "epoch": 0.4807344554179997,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018782570273788736,
+      "loss": 0.8636,
+      "step": 864
+    },
+    {
+      "epoch": 0.48129086103769647,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00018779776964029,
+      "loss": 0.8082,
+      "step": 865
+    },
+    {
+      "epoch": 0.48184726665739325,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018776980661610273,
+      "loss": 0.8549,
+      "step": 866
+    },
+    {
+      "epoch": 0.48240367227709,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.000187741813674857,
+      "loss": 0.8543,
+      "step": 867
+    },
+    {
+      "epoch": 0.48296007789678674,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018771379082609436,
+      "loss": 0.8735,
+      "step": 868
+    },
+    {
+      "epoch": 0.4835164835164835,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001876857380793667,
+      "loss": 0.8338,
+      "step": 869
+    },
+    {
+      "epoch": 0.4840728891361803,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00018765765544423606,
+      "loss": 0.8904,
+      "step": 870
+    },
+    {
+      "epoch": 0.484629294755877,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00018762954293027453,
+      "loss": 0.7981,
+      "step": 871
+    },
+    {
+      "epoch": 0.4851857003755738,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018760140054706453,
+      "loss": 0.9093,
+      "step": 872
+    },
+    {
+      "epoch": 0.48574210599527057,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018757322830419867,
+      "loss": 0.9216,
+      "step": 873
+    },
+    {
+      "epoch": 0.4862985116149673,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00018754502621127966,
+      "loss": 0.7949,
+      "step": 874
+    },
+    {
+      "epoch": 0.48685491723466406,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018751679427792038,
+      "loss": 0.9682,
+      "step": 875
+    },
+    {
+      "epoch": 0.48741132285436084,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018748853251374396,
+      "loss": 0.792,
+      "step": 876
+    },
+    {
+      "epoch": 0.48796772847405756,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018746024092838366,
+      "loss": 0.8901,
+      "step": 877
+    },
+    {
+      "epoch": 0.48852413409375434,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018743191953148286,
+      "loss": 0.9955,
+      "step": 878
+    },
+    {
+      "epoch": 0.4890805397134511,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018740356833269516,
+      "loss": 0.9246,
+      "step": 879
+    },
+    {
+      "epoch": 0.4896369453331479,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018737518734168428,
+      "loss": 0.9122,
+      "step": 880
+    },
+    {
+      "epoch": 0.4901933509528446,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001873467765681242,
+      "loss": 0.9119,
+      "step": 881
+    },
+    {
+      "epoch": 0.4907497565725414,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00018731833602169884,
+      "loss": 0.7738,
+      "step": 882
+    },
+    {
+      "epoch": 0.49130616219223816,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00018728986571210244,
+      "loss": 0.8866,
+      "step": 883
+    },
+    {
+      "epoch": 0.4918625678119349,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00018726136564903938,
+      "loss": 0.9322,
+      "step": 884
+    },
+    {
+      "epoch": 0.49241897343163166,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0001872328358422241,
+      "loss": 0.7841,
+      "step": 885
+    },
+    {
+      "epoch": 0.49297537905132843,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018720427630138122,
+      "loss": 0.9459,
+      "step": 886
+    },
+    {
+      "epoch": 0.49353178467102515,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001871756870362455,
+      "loss": 0.8372,
+      "step": 887
+    },
+    {
+      "epoch": 0.49408819029072193,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018714706805656185,
+      "loss": 0.8801,
+      "step": 888
+    },
+    {
+      "epoch": 0.4946445959104187,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00018711841937208527,
+      "loss": 0.7583,
+      "step": 889
+    },
+    {
+      "epoch": 0.4952010015301155,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00018708974099258086,
+      "loss": 0.8007,
+      "step": 890
+    },
+    {
+      "epoch": 0.4957574071498122,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001870610329278239,
+      "loss": 0.9087,
+      "step": 891
+    },
+    {
+      "epoch": 0.496313812769509,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00018703229518759985,
+      "loss": 0.8809,
+      "step": 892
+    },
+    {
+      "epoch": 0.49687021838920575,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00018700352778170407,
+      "loss": 0.8517,
+      "step": 893
+    },
+    {
+      "epoch": 0.4974266240089025,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018697473071994225,
+      "loss": 0.9004,
+      "step": 894
+    },
+    {
+      "epoch": 0.49798302962859925,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001869459040121301,
+      "loss": 0.8786,
+      "step": 895
+    },
+    {
+      "epoch": 0.498539435248296,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001869170476680934,
+      "loss": 0.911,
+      "step": 896
+    },
+    {
+      "epoch": 0.49909584086799275,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001868881616976681,
+      "loss": 0.8994,
+      "step": 897
+    },
+    {
+      "epoch": 0.4996522464876895,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018685924611070025,
+      "loss": 0.8162,
+      "step": 898
+    },
+    {
+      "epoch": 0.5002086521073863,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018683030091704592,
+      "loss": 0.891,
+      "step": 899
+    },
+    {
+      "epoch": 0.5002086521073863,
+      "eval_loss": 0.8054929971694946,
+      "eval_runtime": 1766.0961,
+      "eval_samples_per_second": 2.656,
+      "eval_steps_per_second": 1.328,
+      "step": 899
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 5391,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 899,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.746246500779295e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}