diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7061 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.18851918182675087,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00018851918182675087,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5e-06,
+      "loss": 2.7659,
+      "step": 1
+    },
+    {
+      "epoch": 0.00037703836365350174,
+      "grad_norm": 1.6640625,
+      "learning_rate": 1e-05,
+      "loss": 2.5842,
+      "step": 2
+    },
+    {
+      "epoch": 0.0005655575454802526,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5e-05,
+      "loss": 2.8169,
+      "step": 3
+    },
+    {
+      "epoch": 0.0007540767273070035,
+      "grad_norm": 1.640625,
+      "learning_rate": 2e-05,
+      "loss": 2.6938,
+      "step": 4
+    },
+    {
+      "epoch": 0.0009425959091337543,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.5e-05,
+      "loss": 2.7862,
+      "step": 5
+    },
+    {
+      "epoch": 0.0011311150909605052,
+      "grad_norm": 1.671875,
+      "learning_rate": 3e-05,
+      "loss": 2.8844,
+      "step": 6
+    },
+    {
+      "epoch": 0.0013196342727872562,
+      "grad_norm": 1.5703125,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 2.8254,
+      "step": 7
+    },
+    {
+      "epoch": 0.001508153454614007,
+      "grad_norm": 1.5625,
+      "learning_rate": 4e-05,
+      "loss": 2.7735,
+      "step": 8
+    },
+    {
+      "epoch": 0.001696672636440758,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 2.8222,
+      "step": 9
+    },
+    {
+      "epoch": 0.0018851918182675087,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5e-05,
+      "loss": 2.6943,
+      "step": 10
+    },
+    {
+      "epoch": 0.0020737110000942595,
+      "grad_norm": 1.5546875,
+      "learning_rate": 5.5e-05,
+      "loss": 2.6735,
+      "step": 11
+    },
+    {
+      "epoch": 0.0022622301819210104,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6e-05,
+      "loss": 2.6482,
+      "step": 12
+    },
+    {
+      "epoch": 0.0024507493637477614,
+      "grad_norm": 1.4375,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 2.8788,
+      "step": 13
+    },
+    {
+      "epoch": 0.0026392685455745124,
+      "grad_norm": 1.6328125,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 2.7531,
+      "step": 14
+    },
+    {
+      "epoch": 0.002827787727401263,
+      "grad_norm": 1.53125,
+      "learning_rate": 7.5e-05,
+      "loss": 2.7911,
+      "step": 15
+    },
+    {
+      "epoch": 0.003016306909228014,
+      "grad_norm": 1.5703125,
+      "learning_rate": 8e-05,
+      "loss": 2.7358,
+      "step": 16
+    },
+    {
+      "epoch": 0.003204826091054765,
+      "grad_norm": 1.5859375,
+      "learning_rate": 8.5e-05,
+      "loss": 2.7272,
+      "step": 17
+    },
+    {
+      "epoch": 0.003393345272881516,
+      "grad_norm": 1.515625,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 2.7176,
+      "step": 18
+    },
+    {
+      "epoch": 0.0035818644547082664,
+      "grad_norm": 1.5,
+      "learning_rate": 9.5e-05,
+      "loss": 2.8573,
+      "step": 19
+    },
+    {
+      "epoch": 0.0037703836365350174,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.0001,
+      "loss": 2.7512,
+      "step": 20
+    },
+    {
+      "epoch": 0.003958902818361768,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.000105,
+      "loss": 2.7962,
+      "step": 21
+    },
+    {
+      "epoch": 0.004147422000188519,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00011,
+      "loss": 2.7,
+      "step": 22
+    },
+    {
+      "epoch": 0.00433594118201527,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.000115,
+      "loss": 2.7128,
+      "step": 23
+    },
+    {
+      "epoch": 0.004524460363842021,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00012,
+      "loss": 2.729,
+      "step": 24
+    },
+    {
+      "epoch": 0.004712979545668771,
+      "grad_norm": 1.25,
+      "learning_rate": 0.000125,
+      "loss": 2.698,
+      "step": 25
+    },
+    {
+      "epoch": 0.004901498727495523,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 2.7461,
+      "step": 26
+    },
+    {
+      "epoch": 0.005090017909322273,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.000135,
+      "loss": 2.7315,
+      "step": 27
+    },
+    {
+      "epoch": 0.005278537091149025,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00014000000000000001,
+      "loss": 2.7089,
+      "step": 28
+    },
+    {
+      "epoch": 0.005467056272975775,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.000145,
+      "loss": 2.6724,
+      "step": 29
+    },
+    {
+      "epoch": 0.005655575454802526,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00015,
+      "loss": 2.799,
+      "step": 30
+    },
+    {
+      "epoch": 0.005844094636629277,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.000155,
+      "loss": 2.7939,
+      "step": 31
+    },
+    {
+      "epoch": 0.006032613818456028,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00016,
+      "loss": 2.8004,
+      "step": 32
+    },
+    {
+      "epoch": 0.006221133000282778,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.000165,
+      "loss": 2.6322,
+      "step": 33
+    },
+    {
+      "epoch": 0.00640965218210953,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00017,
+      "loss": 2.7095,
+      "step": 34
+    },
+    {
+      "epoch": 0.00659817136393628,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.000175,
+      "loss": 2.7111,
+      "step": 35
+    },
+    {
+      "epoch": 0.006786690545763032,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 2.7666,
+      "step": 36
+    },
+    {
+      "epoch": 0.006975209727589782,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.000185,
+      "loss": 2.7779,
+      "step": 37
+    },
+    {
+      "epoch": 0.007163728909416533,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.00019,
+      "loss": 2.7684,
+      "step": 38
+    },
+    {
+      "epoch": 0.007352248091243284,
+      "grad_norm": 0.875,
+      "learning_rate": 0.00019500000000000002,
+      "loss": 2.8674,
+      "step": 39
+    },
+    {
+      "epoch": 0.007540767273070035,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0002,
+      "loss": 2.7694,
+      "step": 40
+    },
+    {
+      "epoch": 0.007729286454896786,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.000205,
+      "loss": 2.6799,
+      "step": 41
+    },
+    {
+      "epoch": 0.007917805636723537,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00021,
+      "loss": 2.6289,
+      "step": 42
+    },
+    {
+      "epoch": 0.008106324818550288,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.000215,
+      "loss": 2.7937,
+      "step": 43
+    },
+    {
+      "epoch": 0.008294844000377038,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00022,
+      "loss": 2.78,
+      "step": 44
+    },
+    {
+      "epoch": 0.00848336318220379,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00022500000000000002,
+      "loss": 2.6351,
+      "step": 45
+    },
+    {
+      "epoch": 0.00867188236403054,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00023,
+      "loss": 2.8156,
+      "step": 46
+    },
+    {
+      "epoch": 0.00886040154585729,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.000235,
+      "loss": 2.8304,
+      "step": 47
+    },
+    {
+      "epoch": 0.009048920727684042,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00024,
+      "loss": 2.7148,
+      "step": 48
+    },
+    {
+      "epoch": 0.009237439909510793,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.000245,
+      "loss": 2.7169,
+      "step": 49
+    },
+    {
+      "epoch": 0.009425959091337543,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00025,
+      "loss": 2.8345,
+      "step": 50
+    },
+    {
+      "epoch": 0.009614478273164294,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.000255,
+      "loss": 2.8149,
+      "step": 51
+    },
+    {
+      "epoch": 0.009802997454991046,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 2.8182,
+      "step": 52
+    },
+    {
+      "epoch": 0.009991516636817797,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00026500000000000004,
+      "loss": 2.8114,
+      "step": 53
+    },
+    {
+      "epoch": 0.010180035818644547,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00027,
+      "loss": 2.803,
+      "step": 54
+    },
+    {
+      "epoch": 0.010368555000471298,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.000275,
+      "loss": 2.7979,
+      "step": 55
+    },
+    {
+      "epoch": 0.01055707418229805,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 2.8062,
+      "step": 56
+    },
+    {
+      "epoch": 0.0107455933641248,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.000285,
+      "loss": 2.6728,
+      "step": 57
+    },
+    {
+      "epoch": 0.01093411254595155,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00029,
+      "loss": 2.7547,
+      "step": 58
+    },
+    {
+      "epoch": 0.011122631727778302,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.000295,
+      "loss": 2.6773,
+      "step": 59
+    },
+    {
+      "epoch": 0.011311150909605052,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0003,
+      "loss": 2.7238,
+      "step": 60
+    },
+    {
+      "epoch": 0.011499670091431803,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.000305,
+      "loss": 2.6842,
+      "step": 61
+    },
+    {
+      "epoch": 0.011688189273258555,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00031,
+      "loss": 2.8449,
+      "step": 62
+    },
+    {
+      "epoch": 0.011876708455085304,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.000315,
+      "loss": 2.6828,
+      "step": 63
+    },
+    {
+      "epoch": 0.012065227636912056,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00032,
+      "loss": 2.7663,
+      "step": 64
+    },
+    {
+      "epoch": 0.012253746818738807,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00032500000000000004,
+      "loss": 2.6127,
+      "step": 65
+    },
+    {
+      "epoch": 0.012442266000565557,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00033,
+      "loss": 2.6333,
+      "step": 66
+    },
+    {
+      "epoch": 0.012630785182392308,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.000335,
+      "loss": 2.7669,
+      "step": 67
+    },
+    {
+      "epoch": 0.01281930436421906,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00034,
+      "loss": 2.7363,
+      "step": 68
+    },
+    {
+      "epoch": 0.013007823546045811,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.000345,
+      "loss": 2.6626,
+      "step": 69
+    },
+    {
+      "epoch": 0.01319634272787256,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00035,
+      "loss": 2.7896,
+      "step": 70
+    },
+    {
+      "epoch": 0.013384861909699312,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000355,
+      "loss": 2.7407,
+      "step": 71
+    },
+    {
+      "epoch": 0.013573381091526063,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 2.804,
+      "step": 72
+    },
+    {
+      "epoch": 0.013761900273352813,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000365,
+      "loss": 2.781,
+      "step": 73
+    },
+    {
+      "epoch": 0.013950419455179565,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00037,
+      "loss": 2.5436,
+      "step": 74
+    },
+    {
+      "epoch": 0.014138938637006316,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.000375,
+      "loss": 2.7272,
+      "step": 75
+    },
+    {
+      "epoch": 0.014327457818833066,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00038,
+      "loss": 2.6777,
+      "step": 76
+    },
+    {
+      "epoch": 0.014515977000659817,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00038500000000000003,
+      "loss": 2.8211,
+      "step": 77
+    },
+    {
+      "epoch": 0.014704496182486568,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 2.7639,
+      "step": 78
+    },
+    {
+      "epoch": 0.014893015364313318,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.000395,
+      "loss": 2.6884,
+      "step": 79
+    },
+    {
+      "epoch": 0.01508153454614007,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0004,
+      "loss": 2.6492,
+      "step": 80
+    },
+    {
+      "epoch": 0.015270053727966821,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00040500000000000003,
+      "loss": 2.8072,
+      "step": 81
+    },
+    {
+      "epoch": 0.015458572909793572,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00041,
+      "loss": 2.7446,
+      "step": 82
+    },
+    {
+      "epoch": 0.015647092091620324,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000415,
+      "loss": 2.7554,
+      "step": 83
+    },
+    {
+      "epoch": 0.015835611273447073,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00042,
+      "loss": 2.7212,
+      "step": 84
+    },
+    {
+      "epoch": 0.016024130455273823,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.000425,
+      "loss": 2.6933,
+      "step": 85
+    },
+    {
+      "epoch": 0.016212649637100576,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00043,
+      "loss": 2.7461,
+      "step": 86
+    },
+    {
+      "epoch": 0.016401168818927326,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.000435,
+      "loss": 2.7079,
+      "step": 87
+    },
+    {
+      "epoch": 0.016589688000754076,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00044,
+      "loss": 2.8562,
+      "step": 88
+    },
+    {
+      "epoch": 0.01677820718258083,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00044500000000000003,
+      "loss": 2.6606,
+      "step": 89
+    },
+    {
+      "epoch": 0.01696672636440758,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 2.7817,
+      "step": 90
+    },
+    {
+      "epoch": 0.017155245546234328,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.000455,
+      "loss": 2.7714,
+      "step": 91
+    },
+    {
+      "epoch": 0.01734376472806108,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00046,
+      "loss": 2.7217,
+      "step": 92
+    },
+    {
+      "epoch": 0.01753228390988783,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.000465,
+      "loss": 2.6855,
+      "step": 93
+    },
+    {
+      "epoch": 0.01772080309171458,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00047,
+      "loss": 2.7111,
+      "step": 94
+    },
+    {
+      "epoch": 0.017909322273541334,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.000475,
+      "loss": 2.6868,
+      "step": 95
+    },
+    {
+      "epoch": 0.018097841455368083,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00048,
+      "loss": 2.7355,
+      "step": 96
+    },
+    {
+      "epoch": 0.018286360637194833,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00048499999999999997,
+      "loss": 2.7172,
+      "step": 97
+    },
+    {
+      "epoch": 0.018474879819021586,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00049,
+      "loss": 2.8204,
+      "step": 98
+    },
+    {
+      "epoch": 0.018663399000848336,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.000495,
+      "loss": 2.6965,
+      "step": 99
+    },
+    {
+      "epoch": 0.018851918182675086,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0005,
+      "loss": 2.7988,
+      "step": 100
+    },
+    {
+      "epoch": 0.01904043736450184,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.000505,
+      "loss": 2.7069,
+      "step": 101
+    },
+    {
+      "epoch": 0.01922895654632859,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00051,
+      "loss": 2.6942,
+      "step": 102
+    },
+    {
+      "epoch": 0.019417475728155338,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.000515,
+      "loss": 2.7497,
+      "step": 103
+    },
+    {
+      "epoch": 0.01960599490998209,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0005200000000000001,
+      "loss": 2.6381,
+      "step": 104
+    },
+    {
+      "epoch": 0.01979451409180884,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0005250000000000001,
+      "loss": 2.6969,
+      "step": 105
+    },
+    {
+      "epoch": 0.019983033273635594,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0005300000000000001,
+      "loss": 2.7247,
+      "step": 106
+    },
+    {
+      "epoch": 0.020171552455462344,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.000535,
+      "loss": 2.828,
+      "step": 107
+    },
+    {
+      "epoch": 0.020360071637289093,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00054,
+      "loss": 2.7309,
+      "step": 108
+    },
+    {
+      "epoch": 0.020548590819115847,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.000545,
+      "loss": 2.8354,
+      "step": 109
+    },
+    {
+      "epoch": 0.020737110000942596,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00055,
+      "loss": 2.8101,
+      "step": 110
+    },
+    {
+      "epoch": 0.020925629182769346,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.000555,
+      "loss": 2.7837,
+      "step": 111
+    },
+    {
+      "epoch": 0.0211141483645961,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0005600000000000001,
+      "loss": 2.6813,
+      "step": 112
+    },
+    {
+      "epoch": 0.02130266754642285,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.000565,
+      "loss": 2.7035,
+      "step": 113
+    },
+    {
+      "epoch": 0.0214911867282496,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00057,
+      "loss": 2.6901,
+      "step": 114
+    },
+    {
+      "epoch": 0.02167970591007635,
+      "grad_norm": 0.75,
+      "learning_rate": 0.000575,
+      "loss": 2.7001,
+      "step": 115
+    },
+    {
+      "epoch": 0.0218682250919031,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00058,
+      "loss": 2.7508,
+      "step": 116
+    },
+    {
+      "epoch": 0.02205674427372985,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.000585,
+      "loss": 2.7348,
+      "step": 117
+    },
+    {
+      "epoch": 0.022245263455556604,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00059,
+      "loss": 2.7434,
+      "step": 118
+    },
+    {
+      "epoch": 0.022433782637383354,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0005949999999999999,
+      "loss": 2.6735,
+      "step": 119
+    },
+    {
+      "epoch": 0.022622301819210103,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0006,
+      "loss": 2.6258,
+      "step": 120
+    },
+    {
+      "epoch": 0.022810821001036857,
+      "grad_norm": 0.75,
+      "learning_rate": 0.000605,
+      "loss": 2.7676,
+      "step": 121
+    },
+    {
+      "epoch": 0.022999340182863606,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00061,
+      "loss": 2.7045,
+      "step": 122
+    },
+    {
+      "epoch": 0.023187859364690356,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.000615,
+      "loss": 2.6322,
+      "step": 123
+    },
+    {
+      "epoch": 0.02337637854651711,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00062,
+      "loss": 2.6953,
+      "step": 124
+    },
+    {
+      "epoch": 0.02356489772834386,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.000625,
+      "loss": 2.6045,
+      "step": 125
+    },
+    {
+      "epoch": 0.02375341691017061,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00063,
+      "loss": 2.6551,
+      "step": 126
+    },
+    {
+      "epoch": 0.02394193609199736,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.000635,
+      "loss": 2.656,
+      "step": 127
+    },
+    {
+      "epoch": 0.02413045527382411,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00064,
+      "loss": 2.791,
+      "step": 128
+    },
+    {
+      "epoch": 0.02431897445565086,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0006450000000000001,
+      "loss": 2.6599,
+      "step": 129
+    },
+    {
+      "epoch": 0.024507493637477614,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0006500000000000001,
+      "loss": 2.633,
+      "step": 130
+    },
+    {
+      "epoch": 0.024696012819304364,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0006550000000000001,
+      "loss": 2.6002,
+      "step": 131
+    },
+    {
+      "epoch": 0.024884532001131113,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00066,
+      "loss": 2.7593,
+      "step": 132
+    },
+    {
+      "epoch": 0.025073051182957867,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.000665,
+      "loss": 2.706,
+      "step": 133
+    },
+    {
+      "epoch": 0.025261570364784616,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00067,
+      "loss": 2.7094,
+      "step": 134
+    },
+    {
+      "epoch": 0.02545008954661137,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.000675,
+      "loss": 2.6961,
+      "step": 135
+    },
+    {
+      "epoch": 0.02563860872843812,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00068,
+      "loss": 2.7805,
+      "step": 136
+    },
+    {
+      "epoch": 0.02582712791026487,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0006850000000000001,
+      "loss": 2.6559,
+      "step": 137
+    },
+    {
+      "epoch": 0.026015647092091622,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00069,
+      "loss": 2.7455,
+      "step": 138
+    },
+    {
+      "epoch": 0.02620416627391837,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.000695,
+      "loss": 2.7533,
+      "step": 139
+    },
+    {
+      "epoch": 0.02639268545574512,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0007,
+      "loss": 2.7434,
+      "step": 140
+    },
+    {
+      "epoch": 0.026581204637571874,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.000705,
+      "loss": 2.7018,
+      "step": 141
+    },
+    {
+      "epoch": 0.026769723819398624,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00071,
+      "loss": 2.6182,
+      "step": 142
+    },
+    {
+      "epoch": 0.026958243001225374,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.000715,
+      "loss": 2.5742,
+      "step": 143
+    },
+    {
+      "epoch": 0.027146762183052127,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 2.6547,
+      "step": 144
+    },
+    {
+      "epoch": 0.027335281364878877,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.000725,
+      "loss": 2.7054,
+      "step": 145
+    },
+    {
+      "epoch": 0.027523800546705626,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00073,
+      "loss": 2.5809,
+      "step": 146
+    },
+    {
+      "epoch": 0.02771231972853238,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.000735,
+      "loss": 2.6474,
+      "step": 147
+    },
+    {
+      "epoch": 0.02790083891035913,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00074,
+      "loss": 2.7606,
+      "step": 148
+    },
+    {
+      "epoch": 0.02808935809218588,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.000745,
+      "loss": 2.6923,
+      "step": 149
+    },
+    {
+      "epoch": 0.028277877274012632,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00075,
+      "loss": 2.782,
+      "step": 150
+    },
+    {
+      "epoch": 0.02846639645583938,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000755,
+      "loss": 2.7369,
+      "step": 151
+    },
+    {
+      "epoch": 0.02865491563766613,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00076,
+      "loss": 2.6287,
+      "step": 152
+    },
+    {
+      "epoch": 0.028843434819492884,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0007650000000000001,
+      "loss": 2.6649,
+      "step": 153
+    },
+    {
+      "epoch": 0.029031954001319634,
+      "grad_norm": 0.875,
+      "learning_rate": 0.0007700000000000001,
+      "loss": 2.7421,
+      "step": 154
+    },
+    {
+      "epoch": 0.029220473183146384,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0007750000000000001,
+      "loss": 2.5988,
+      "step": 155
+    },
+    {
+      "epoch": 0.029408992364973137,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 2.6876,
+      "step": 156
+    },
+    {
+      "epoch": 0.029597511546799887,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.000785,
+      "loss": 2.6846,
+      "step": 157
+    },
+    {
+      "epoch": 0.029786030728626636,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00079,
+      "loss": 2.7869,
+      "step": 158
+    },
+    {
+      "epoch": 0.02997454991045339,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.000795,
+      "loss": 2.6972,
+      "step": 159
+    },
+    {
+      "epoch": 0.03016306909228014,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0008,
+      "loss": 2.7664,
+      "step": 160
+    },
+    {
+      "epoch": 0.03035158827410689,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000805,
+      "loss": 2.6554,
+      "step": 161
+    },
+    {
+      "epoch": 0.030540107455933642,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 2.662,
+      "step": 162
+    },
+    {
+      "epoch": 0.03072862663776039,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.000815,
+      "loss": 2.622,
+      "step": 163
+    },
+    {
+      "epoch": 0.030917145819587145,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00082,
+      "loss": 2.6071,
+      "step": 164
+    },
+    {
+      "epoch": 0.031105665001413894,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.000825,
+      "loss": 2.6724,
+      "step": 165
+    },
+    {
+      "epoch": 0.03129418418324065,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00083,
+      "loss": 2.5888,
+      "step": 166
+    },
+    {
+      "epoch": 0.031482703365067394,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.000835,
+      "loss": 2.7932,
+      "step": 167
+    },
+    {
+      "epoch": 0.03167122254689415,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00084,
+      "loss": 2.6234,
+      "step": 168
+    },
+    {
+      "epoch": 0.0318597417287209,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0008449999999999999,
+      "loss": 2.6725,
+      "step": 169
+    },
+    {
+      "epoch": 0.032048260910547646,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00085,
+      "loss": 2.6502,
+      "step": 170
+    },
+    {
+      "epoch": 0.0322367800923744,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.000855,
+      "loss": 2.7151,
+      "step": 171
+    },
+    {
+      "epoch": 0.03242529927420115,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00086,
+      "loss": 2.8332,
+      "step": 172
+    },
+    {
+      "epoch": 0.0326138184560279,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.000865,
+      "loss": 2.8183,
+      "step": 173
+    },
+    {
+      "epoch": 0.03280233763785465,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00087,
+      "loss": 2.6777,
+      "step": 174
+    },
+    {
+      "epoch": 0.032990856819681405,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.000875,
+      "loss": 2.6281,
+      "step": 175
+    },
+    {
+      "epoch": 0.03317937600150815,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00088,
+      "loss": 2.7047,
+      "step": 176
+    },
+    {
+      "epoch": 0.033367895183334904,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.000885,
+      "loss": 2.6637,
+      "step": 177
+    },
+    {
+      "epoch": 0.03355641436516166,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0008900000000000001,
+      "loss": 2.7817,
+      "step": 178
+    },
+    {
+      "epoch": 0.033744933546988404,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0008950000000000001,
+      "loss": 2.6216,
+      "step": 179
+    },
+    {
+      "epoch": 0.03393345272881516,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009000000000000001,
+      "loss": 2.6608,
+      "step": 180
+    },
+    {
+      "epoch": 0.03412197191064191,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009050000000000001,
+      "loss": 2.712,
+      "step": 181
+    },
+    {
+      "epoch": 0.034310491092468656,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00091,
+      "loss": 2.6812,
+      "step": 182
+    },
+    {
+      "epoch": 0.03449901027429541,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.000915,
+      "loss": 2.6181,
+      "step": 183
+    },
+    {
+      "epoch": 0.03468752945612216,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00092,
+      "loss": 2.5939,
+      "step": 184
+    },
+    {
+      "epoch": 0.03487604863794891,
+      "grad_norm": 0.75,
+      "learning_rate": 0.000925,
+      "loss": 2.6378,
+      "step": 185
+    },
+    {
+      "epoch": 0.03506456781977566,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00093,
+      "loss": 2.658,
+      "step": 186
+    },
+    {
+      "epoch": 0.035253087001602415,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0009350000000000001,
+      "loss": 2.6324,
+      "step": 187
+    },
+    {
+      "epoch": 0.03544160618342916,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00094,
+      "loss": 2.7615,
+      "step": 188
+    },
+    {
+      "epoch": 0.035630125365255914,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.000945,
+      "loss": 2.8334,
+      "step": 189
+    },
+    {
+      "epoch": 0.03581864454708267,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00095,
+      "loss": 2.8026,
+      "step": 190
+    },
+    {
+      "epoch": 0.036007163728909414,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.000955,
+      "loss": 2.6532,
+      "step": 191
+    },
+    {
+      "epoch": 0.03619568291073617,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00096,
+      "loss": 2.5541,
+      "step": 192
+    },
+    {
+      "epoch": 0.03638420209256292,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.000965,
+      "loss": 2.6375,
+      "step": 193
+    },
+    {
+      "epoch": 0.036572721274389666,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009699999999999999,
+      "loss": 2.5705,
+      "step": 194
+    },
+    {
+      "epoch": 0.03676124045621642,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.000975,
+      "loss": 2.6405,
+      "step": 195
+    },
+    {
+      "epoch": 0.03694975963804317,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00098,
+      "loss": 2.7821,
+      "step": 196
+    },
+    {
+      "epoch": 0.03713827881986992,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000985,
+      "loss": 2.6889,
+      "step": 197
+    },
+    {
+      "epoch": 0.03732679800169667,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00099,
+      "loss": 2.6658,
+      "step": 198
+    },
+    {
+      "epoch": 0.037515317183523425,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.000995,
+      "loss": 2.6969,
+      "step": 199
+    },
+    {
+      "epoch": 0.03770383636535017,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.001,
+      "loss": 2.5479,
+      "step": 200
+    },
+    {
+      "epoch": 0.037892355547176924,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0009998040752351098,
+      "loss": 2.7177,
+      "step": 201
+    },
+    {
+      "epoch": 0.03808087472900368,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0009996081504702195,
+      "loss": 2.7224,
+      "step": 202
+    },
+    {
+      "epoch": 0.038269393910830424,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009994122257053293,
+      "loss": 2.6316,
+      "step": 203
+    },
+    {
+      "epoch": 0.03845791309265718,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0009992163009404388,
+      "loss": 2.8178,
+      "step": 204
+    },
+    {
+      "epoch": 0.03864643227448393,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009990203761755486,
+      "loss": 2.7619,
+      "step": 205
+    },
+    {
+      "epoch": 0.038834951456310676,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009988244514106584,
+      "loss": 2.5739,
+      "step": 206
+    },
+    {
+      "epoch": 0.03902347063813743,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0009986285266457681,
+      "loss": 2.7797,
+      "step": 207
+    },
+    {
+      "epoch": 0.03921198981996418,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0009984326018808779,
+      "loss": 2.695,
+      "step": 208
+    },
+    {
+      "epoch": 0.039400509001790936,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009982366771159876,
+      "loss": 2.7551,
+      "step": 209
+    },
+    {
+      "epoch": 0.03958902818361768,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0009980407523510972,
+      "loss": 2.7898,
+      "step": 210
+    },
+    {
+      "epoch": 0.039777547365444435,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.000997844827586207,
+      "loss": 2.6824,
+      "step": 211
+    },
+    {
+      "epoch": 0.03996606654727119,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009976489028213167,
+      "loss": 2.8341,
+      "step": 212
+    },
+    {
+      "epoch": 0.040154585729097934,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009974529780564262,
+      "loss": 2.6885,
+      "step": 213
+    },
+    {
+      "epoch": 0.04034310491092469,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.000997257053291536,
+      "loss": 2.5722,
+      "step": 214
+    },
+    {
+      "epoch": 0.04053162409275144,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009970611285266457,
+      "loss": 2.7023,
+      "step": 215
+    },
+    {
+      "epoch": 0.04072014327457819,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009968652037617555,
+      "loss": 2.6429,
+      "step": 216
+    },
+    {
+      "epoch": 0.04090866245640494,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0009966692789968653,
+      "loss": 2.7053,
+      "step": 217
+    },
+    {
+      "epoch": 0.04109718163823169,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.000996473354231975,
+      "loss": 2.7841,
+      "step": 218
+    },
+    {
+      "epoch": 0.04128570082005844,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009962774294670846,
+      "loss": 2.6687,
+      "step": 219
+    },
+    {
+      "epoch": 0.04147422000188519,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009960815047021943,
+      "loss": 2.7893,
+      "step": 220
+    },
+    {
+      "epoch": 0.041662739183711946,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.000995885579937304,
+      "loss": 2.5992,
+      "step": 221
+    },
+    {
+      "epoch": 0.04185125836553869,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009956896551724138,
+      "loss": 2.7238,
+      "step": 222
+    },
+    {
+      "epoch": 0.042039777547365445,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009954937304075236,
+      "loss": 2.7477,
+      "step": 223
+    },
+    {
+      "epoch": 0.0422282967291922,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0009952978056426334,
+      "loss": 2.6079,
+      "step": 224
+    },
+    {
+      "epoch": 0.042416815911018944,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.000995101880877743,
+      "loss": 2.6389,
+      "step": 225
+    },
+    {
+      "epoch": 0.0426053350928457,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009949059561128527,
+      "loss": 2.6014,
+      "step": 226
+    },
+    {
+      "epoch": 0.04279385427467245,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009947100313479624,
+      "loss": 2.6708,
+      "step": 227
+    },
+    {
+      "epoch": 0.0429823734564992,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009945141065830722,
+      "loss": 2.7032,
+      "step": 228
+    },
+    {
+      "epoch": 0.04317089263832595,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009943181818181817,
+      "loss": 2.7911,
+      "step": 229
+    },
+    {
+      "epoch": 0.0433594118201527,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009941222570532915,
+      "loss": 2.5071,
+      "step": 230
+    },
+    {
+      "epoch": 0.04354793100197945,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009939263322884012,
+      "loss": 2.695,
+      "step": 231
+    },
+    {
+      "epoch": 0.0437364501838062,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.000993730407523511,
+      "loss": 2.5969,
+      "step": 232
+    },
+    {
+      "epoch": 0.043924969365632956,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009935344827586207,
+      "loss": 2.6602,
+      "step": 233
+    },
+    {
+      "epoch": 0.0441134885474597,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009933385579937305,
+      "loss": 2.6561,
+      "step": 234
+    },
+    {
+      "epoch": 0.044302007729286455,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00099314263322884,
+      "loss": 2.6442,
+      "step": 235
+    },
+    {
+      "epoch": 0.04449052691111321,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0009929467084639498,
+      "loss": 2.7465,
+      "step": 236
+    },
+    {
+      "epoch": 0.044679046092939954,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0009927507836990596,
+      "loss": 2.7102,
+      "step": 237
+    },
+    {
+      "epoch": 0.04486756527476671,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009925548589341693,
+      "loss": 2.7074,
+      "step": 238
+    },
+    {
+      "epoch": 0.04505608445659346,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.000992358934169279,
+      "loss": 2.6626,
+      "step": 239
+    },
+    {
+      "epoch": 0.04524460363842021,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009921630094043888,
+      "loss": 2.5579,
+      "step": 240
+    },
+    {
+      "epoch": 0.04543312282024696,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0009919670846394984,
+      "loss": 2.7225,
+      "step": 241
+    },
+    {
+      "epoch": 0.04562164200207371,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009917711598746081,
+      "loss": 2.6952,
+      "step": 242
+    },
+    {
+      "epoch": 0.04581016118390046,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.000991575235109718,
+      "loss": 2.6886,
+      "step": 243
+    },
+    {
+      "epoch": 0.04599868036572721,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009913793103448277,
+      "loss": 2.6096,
+      "step": 244
+    },
+    {
+      "epoch": 0.046187199547553966,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0009911833855799374,
+      "loss": 2.7612,
+      "step": 245
+    },
+    {
+      "epoch": 0.04637571872938071,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009909874608150472,
+      "loss": 2.6082,
+      "step": 246
+    },
+    {
+      "epoch": 0.046564237911207465,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0009907915360501567,
+      "loss": 2.7621,
+      "step": 247
+    },
+    {
+      "epoch": 0.04675275709303422,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009905956112852665,
+      "loss": 2.6764,
+      "step": 248
+    },
+    {
+      "epoch": 0.046941276274860964,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009903996865203762,
+      "loss": 2.6527,
+      "step": 249
+    },
+    {
+      "epoch": 0.04712979545668772,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009902037617554858,
+      "loss": 2.5762,
+      "step": 250
+    },
+    {
+      "epoch": 0.04731831463851447,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0009900078369905955,
+      "loss": 2.7241,
+      "step": 251
+    },
+    {
+      "epoch": 0.04750683382034122,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009898119122257053,
+      "loss": 2.6935,
+      "step": 252
+    },
+    {
+      "epoch": 0.04769535300216797,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.000989615987460815,
+      "loss": 2.776,
+      "step": 253
+    },
+    {
+      "epoch": 0.04788387218399472,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009894200626959248,
+      "loss": 2.7799,
+      "step": 254
+    },
+    {
+      "epoch": 0.04807239136582147,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009892241379310346,
+      "loss": 2.7589,
+      "step": 255
+    },
+    {
+      "epoch": 0.04826091054764822,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009890282131661443,
+      "loss": 2.646,
+      "step": 256
+    },
+    {
+      "epoch": 0.048449429729474976,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009888322884012539,
+      "loss": 2.7226,
+      "step": 257
+    },
+    {
+      "epoch": 0.04863794891130172,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009886363636363636,
+      "loss": 2.6825,
+      "step": 258
+    },
+    {
+      "epoch": 0.048826468093128475,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009884404388714734,
+      "loss": 2.6494,
+      "step": 259
+    },
+    {
+      "epoch": 0.04901498727495523,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009882445141065831,
+      "loss": 2.7586,
+      "step": 260
+    },
+    {
+      "epoch": 0.049203506456781974,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.000988048589341693,
+      "loss": 2.7986,
+      "step": 261
+    },
+    {
+      "epoch": 0.04939202563860873,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009878526645768027,
+      "loss": 2.624,
+      "step": 262
+    },
+    {
+      "epoch": 0.04958054482043548,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009876567398119122,
+      "loss": 2.4967,
+      "step": 263
+    },
+    {
+      "epoch": 0.04976906400226223,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000987460815047022,
+      "loss": 2.5694,
+      "step": 264
+    },
+    {
+      "epoch": 0.04995758318408898,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009872648902821317,
+      "loss": 2.7369,
+      "step": 265
+    },
+    {
+      "epoch": 0.05014610236591573,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0009870689655172413,
+      "loss": 2.641,
+      "step": 266
+    },
+    {
+      "epoch": 0.05033462154774248,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.000986873040752351,
+      "loss": 2.5988,
+      "step": 267
+    },
+    {
+      "epoch": 0.05052314072956923,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0009866771159874608,
+      "loss": 2.6935,
+      "step": 268
+    },
+    {
+      "epoch": 0.050711659911395986,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009864811912225705,
+      "loss": 2.6573,
+      "step": 269
+    },
+    {
+      "epoch": 0.05090017909322274,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009862852664576803,
+      "loss": 2.5501,
+      "step": 270
+    },
+    {
+      "epoch": 0.051088698275049485,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00098608934169279,
+      "loss": 2.7173,
+      "step": 271
+    },
+    {
+      "epoch": 0.05127721745687624,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0009858934169278996,
+      "loss": 2.7147,
+      "step": 272
+    },
+    {
+      "epoch": 0.05146573663870299,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009856974921630094,
+      "loss": 2.6823,
+      "step": 273
+    },
+    {
+      "epoch": 0.05165425582052974,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009855015673981191,
+      "loss": 2.7399,
+      "step": 274
+    },
+    {
+      "epoch": 0.05184277500235649,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0009853056426332289,
+      "loss": 2.8052,
+      "step": 275
+    },
+    {
+      "epoch": 0.052031294184183244,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0009851097178683386,
+      "loss": 2.6471,
+      "step": 276
+    },
+    {
+      "epoch": 0.05221981336600999,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009849137931034484,
+      "loss": 2.5997,
+      "step": 277
+    },
+    {
+      "epoch": 0.05240833254783674,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.000984717868338558,
+      "loss": 2.6933,
+      "step": 278
+    },
+    {
+      "epoch": 0.052596851729663496,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009845219435736677,
+      "loss": 2.7849,
+      "step": 279
+    },
+    {
+      "epoch": 0.05278537091149024,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0009843260188087774,
+      "loss": 2.7277,
+      "step": 280
+    },
+    {
+      "epoch": 0.052973890093316996,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009841300940438872,
+      "loss": 2.7328,
+      "step": 281
+    },
+    {
+      "epoch": 0.05316240927514375,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.000983934169278997,
+      "loss": 2.8041,
+      "step": 282
+    },
+    {
+      "epoch": 0.053350928456970495,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009837382445141067,
+      "loss": 2.6497,
+      "step": 283
+    },
+    {
+      "epoch": 0.05353944763879725,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009835423197492165,
+      "loss": 2.6852,
+      "step": 284
+    },
+    {
+      "epoch": 0.053727966820624,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.000983346394984326,
+      "loss": 2.6116,
+      "step": 285
+    },
+    {
+      "epoch": 0.05391648600245075,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009831504702194358,
+      "loss": 2.5864,
+      "step": 286
+    },
+    {
+      "epoch": 0.0541050051842775,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009829545454545455,
+      "loss": 2.6291,
+      "step": 287
+    },
+    {
+      "epoch": 0.054293524366104254,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.000982758620689655,
+      "loss": 2.672,
+      "step": 288
+    },
+    {
+      "epoch": 0.054482043547931,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0009825626959247648,
+      "loss": 2.6036,
+      "step": 289
+    },
+    {
+      "epoch": 0.05467056272975775,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0009823667711598746,
+      "loss": 2.4802,
+      "step": 290
+    },
+    {
+      "epoch": 0.054859081911584506,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0009821708463949844,
+      "loss": 2.721,
+      "step": 291
+    },
+    {
+      "epoch": 0.05504760109341125,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009819749216300941,
+      "loss": 2.6039,
+      "step": 292
+    },
+    {
+      "epoch": 0.055236120275238006,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009817789968652039,
+      "loss": 2.7125,
+      "step": 293
+    },
+    {
+      "epoch": 0.05542463945706476,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009815830721003134,
+      "loss": 2.7176,
+      "step": 294
+    },
+    {
+      "epoch": 0.055613158638891505,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0009813871473354232,
+      "loss": 2.7061,
+      "step": 295
+    },
+    {
+      "epoch": 0.05580167782071826,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000981191222570533,
+      "loss": 2.7324,
+      "step": 296
+    },
+    {
+      "epoch": 0.05599019700254501,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009809952978056427,
+      "loss": 2.6318,
+      "step": 297
+    },
+    {
+      "epoch": 0.05617871618437176,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009807993730407524,
+      "loss": 2.637,
+      "step": 298
+    },
+    {
+      "epoch": 0.05636723536619851,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009806034482758622,
+      "loss": 2.5645,
+      "step": 299
+    },
+    {
+      "epoch": 0.056555754548025264,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009804075235109717,
+      "loss": 2.7011,
+      "step": 300
+    },
+    {
+      "epoch": 0.05674427372985201,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009802115987460815,
+      "loss": 2.7293,
+      "step": 301
+    },
+    {
+      "epoch": 0.05693279291167876,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009800156739811913,
+      "loss": 2.5779,
+      "step": 302
+    },
+    {
+      "epoch": 0.057121312093505516,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.000979819749216301,
+      "loss": 2.7574,
+      "step": 303
+    },
+    {
+      "epoch": 0.05730983127533226,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009796238244514106,
+      "loss": 2.7168,
+      "step": 304
+    },
+    {
+      "epoch": 0.057498350457159016,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009794278996865203,
+      "loss": 2.6531,
+      "step": 305
+    },
+    {
+      "epoch": 0.05768686963898577,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00097923197492163,
+      "loss": 2.6852,
+      "step": 306
+    },
+    {
+      "epoch": 0.057875388820812515,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009790360501567398,
+      "loss": 2.8098,
+      "step": 307
+    },
+    {
+      "epoch": 0.05806390800263927,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009788401253918496,
+      "loss": 2.5938,
+      "step": 308
+    },
+    {
+      "epoch": 0.05825242718446602,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009786442006269591,
+      "loss": 2.6858,
+      "step": 309
+    },
+    {
+      "epoch": 0.05844094636629277,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.000978448275862069,
+      "loss": 2.6455,
+      "step": 310
+    },
+    {
+      "epoch": 0.05862946554811952,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009782523510971787,
+      "loss": 2.7194,
+      "step": 311
+    },
+    {
+      "epoch": 0.058817984729946274,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009780564263322884,
+      "loss": 2.5933,
+      "step": 312
+    },
+    {
+      "epoch": 0.05900650391177302,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009778605015673982,
+      "loss": 2.7103,
+      "step": 313
+    },
+    {
+      "epoch": 0.05919502309359977,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.000977664576802508,
+      "loss": 2.7317,
+      "step": 314
+    },
+    {
+      "epoch": 0.059383542275426526,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0009774686520376177,
+      "loss": 2.6629,
+      "step": 315
+    },
+    {
+      "epoch": 0.05957206145725327,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009772727272727272,
+      "loss": 2.811,
+      "step": 316
+    },
+    {
+      "epoch": 0.059760580639080026,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.000977076802507837,
+      "loss": 2.679,
+      "step": 317
+    },
+    {
+      "epoch": 0.05994909982090678,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009768808777429468,
+      "loss": 2.7421,
+      "step": 318
+    },
+    {
+      "epoch": 0.060137619002733525,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0009766849529780565,
+      "loss": 2.7717,
+      "step": 319
+    },
+    {
+      "epoch": 0.06032613818456028,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0009764890282131662,
+      "loss": 2.7456,
+      "step": 320
+    },
+    {
+      "epoch": 0.06051465736638703,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009762931034482759,
+      "loss": 2.6342,
+      "step": 321
+    },
+    {
+      "epoch": 0.06070317654821378,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0009760971786833856,
+      "loss": 2.7088,
+      "step": 322
+    },
+    {
+      "epoch": 0.06089169573004053,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009759012539184952,
+      "loss": 2.651,
+      "step": 323
+    },
+    {
+      "epoch": 0.061080214911867284,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.000975705329153605,
+      "loss": 2.7472,
+      "step": 324
+    },
+    {
+      "epoch": 0.06126873409369403,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0009755094043887147,
+      "loss": 2.8417,
+      "step": 325
+    },
+    {
+      "epoch": 0.06145725327552078,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009753134796238245,
+      "loss": 2.7298,
+      "step": 326
+    },
+    {
+      "epoch": 0.061645772457347536,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009751175548589341,
+      "loss": 2.4934,
+      "step": 327
+    },
+    {
+      "epoch": 0.06183429163917429,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0009749216300940439,
+      "loss": 2.6646,
+      "step": 328
+    },
+    {
+      "epoch": 0.062022810821001036,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0009747257053291537,
+      "loss": 2.6718,
+      "step": 329
+    },
+    {
+      "epoch": 0.06221133000282779,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009745297805642633,
+      "loss": 2.5689,
+      "step": 330
+    },
+    {
+      "epoch": 0.06239984918465454,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009743338557993731,
+      "loss": 2.5188,
+      "step": 331
+    },
+    {
+      "epoch": 0.0625883683664813,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0009741379310344828,
+      "loss": 2.7064,
+      "step": 332
+    },
+    {
+      "epoch": 0.06277688754830804,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009739420062695925,
+      "loss": 2.726,
+      "step": 333
+    },
+    {
+      "epoch": 0.06296540673013479,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009737460815047022,
+      "loss": 2.7389,
+      "step": 334
+    },
+    {
+      "epoch": 0.06315392591196155,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.000973550156739812,
+      "loss": 2.8134,
+      "step": 335
+    },
+    {
+      "epoch": 0.0633424450937883,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009733542319749216,
+      "loss": 2.7394,
+      "step": 336
+    },
+    {
+      "epoch": 0.06353096427561504,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009731583072100314,
+      "loss": 2.6256,
+      "step": 337
+    },
+    {
+      "epoch": 0.0637194834574418,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0009729623824451412,
+      "loss": 2.7413,
+      "step": 338
+    },
+    {
+      "epoch": 0.06390800263926855,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009727664576802508,
+      "loss": 2.7725,
+      "step": 339
+    },
+    {
+      "epoch": 0.06409652182109529,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009725705329153606,
+      "loss": 2.8092,
+      "step": 340
+    },
+    {
+      "epoch": 0.06428504100292205,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009723746081504702,
+      "loss": 2.7276,
+      "step": 341
+    },
+    {
+      "epoch": 0.0644735601847488,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009721786833855799,
+      "loss": 2.5861,
+      "step": 342
+    },
+    {
+      "epoch": 0.06466207936657555,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0009719827586206896,
+      "loss": 2.6467,
+      "step": 343
+    },
+    {
+      "epoch": 0.0648505985484023,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009717868338557994,
+      "loss": 2.7404,
+      "step": 344
+    },
+    {
+      "epoch": 0.06503911773022905,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.000971590909090909,
+      "loss": 2.6333,
+      "step": 345
+    },
+    {
+      "epoch": 0.0652276369120558,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009713949843260188,
+      "loss": 2.6079,
+      "step": 346
+    },
+    {
+      "epoch": 0.06541615609388256,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009711990595611286,
+      "loss": 2.5708,
+      "step": 347
+    },
+    {
+      "epoch": 0.0656046752757093,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009710031347962382,
+      "loss": 2.6675,
+      "step": 348
+    },
+    {
+      "epoch": 0.06579319445753605,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.000970807210031348,
+      "loss": 2.7782,
+      "step": 349
+    },
+    {
+      "epoch": 0.06598171363936281,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009706112852664577,
+      "loss": 2.6853,
+      "step": 350
+    },
+    {
+      "epoch": 0.06617023282118956,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009704153605015674,
+      "loss": 2.7684,
+      "step": 351
+    },
+    {
+      "epoch": 0.0663587520030163,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009702194357366771,
+      "loss": 2.5759,
+      "step": 352
+    },
+    {
+      "epoch": 0.06654727118484306,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0009700235109717869,
+      "loss": 2.7151,
+      "step": 353
+    },
+    {
+      "epoch": 0.06673579036666981,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009698275862068966,
+      "loss": 2.6346,
+      "step": 354
+    },
+    {
+      "epoch": 0.06692430954849656,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009696316614420063,
+      "loss": 2.5878,
+      "step": 355
+    },
+    {
+      "epoch": 0.06711282873032332,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009694357366771161,
+      "loss": 2.6841,
+      "step": 356
+    },
+    {
+      "epoch": 0.06730134791215006,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009692398119122258,
+      "loss": 2.5688,
+      "step": 357
+    },
+    {
+      "epoch": 0.06748986709397681,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009690438871473355,
+      "loss": 2.5057,
+      "step": 358
+    },
+    {
+      "epoch": 0.06767838627580357,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0009688479623824452,
+      "loss": 2.6444,
+      "step": 359
+    },
+    {
+      "epoch": 0.06786690545763031,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009686520376175549,
+      "loss": 2.6894,
+      "step": 360
+    },
+    {
+      "epoch": 0.06805542463945706,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0009684561128526645,
+      "loss": 2.5921,
+      "step": 361
+    },
+    {
+      "epoch": 0.06824394382128382,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0009682601880877743,
+      "loss": 2.7547,
+      "step": 362
+    },
+    {
+      "epoch": 0.06843246300311057,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.000968064263322884,
+      "loss": 2.7235,
+      "step": 363
+    },
+    {
+      "epoch": 0.06862098218493731,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009678683385579937,
+      "loss": 2.6726,
+      "step": 364
+    },
+    {
+      "epoch": 0.06880950136676407,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0009676724137931034,
+      "loss": 2.7688,
+      "step": 365
+    },
+    {
+      "epoch": 0.06899802054859082,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009674764890282132,
+      "loss": 2.6567,
+      "step": 366
+    },
+    {
+      "epoch": 0.06918653973041756,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009672805642633229,
+      "loss": 2.7241,
+      "step": 367
+    },
+    {
+      "epoch": 0.06937505891224433,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009670846394984326,
+      "loss": 2.603,
+      "step": 368
+    },
+    {
+      "epoch": 0.06956357809407107,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009668887147335424,
+      "loss": 2.6863,
+      "step": 369
+    },
+    {
+      "epoch": 0.06975209727589782,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.000966692789968652,
+      "loss": 2.6655,
+      "step": 370
+    },
+    {
+      "epoch": 0.06994061645772458,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009664968652037618,
+      "loss": 2.5301,
+      "step": 371
+    },
+    {
+      "epoch": 0.07012913563955132,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009663009404388715,
+      "loss": 2.7405,
+      "step": 372
+    },
+    {
+      "epoch": 0.07031765482137807,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009661050156739812,
+      "loss": 2.7326,
+      "step": 373
+    },
+    {
+      "epoch": 0.07050617400320483,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.000965909090909091,
+      "loss": 2.6685,
+      "step": 374
+    },
+    {
+      "epoch": 0.07069469318503158,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009657131661442007,
+      "loss": 2.6664,
+      "step": 375
+    },
+    {
+      "epoch": 0.07088321236685832,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009655172413793104,
+      "loss": 2.5892,
+      "step": 376
+    },
+    {
+      "epoch": 0.07107173154868508,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009653213166144201,
+      "loss": 2.6351,
+      "step": 377
+    },
+    {
+      "epoch": 0.07126025073051183,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0009651253918495299,
+      "loss": 2.6587,
+      "step": 378
+    },
+    {
+      "epoch": 0.07144876991233857,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009649294670846394,
+      "loss": 2.7744,
+      "step": 379
+    },
+    {
+      "epoch": 0.07163728909416534,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009647335423197492,
+      "loss": 2.7516,
+      "step": 380
+    },
+    {
+      "epoch": 0.07182580827599208,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0009645376175548589,
+      "loss": 2.6607,
+      "step": 381
+    },
+    {
+      "epoch": 0.07201432745781883,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009643416927899687,
+      "loss": 2.7513,
+      "step": 382
+    },
+    {
+      "epoch": 0.07220284663964559,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0009641457680250783,
+      "loss": 2.607,
+      "step": 383
+    },
+    {
+      "epoch": 0.07239136582147233,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009639498432601881,
+      "loss": 2.5463,
+      "step": 384
+    },
+    {
+      "epoch": 0.07257988500329908,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0009637539184952979,
+      "loss": 2.6368,
+      "step": 385
+    },
+    {
+      "epoch": 0.07276840418512584,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009635579937304075,
+      "loss": 2.5846,
+      "step": 386
+    },
+    {
+      "epoch": 0.07295692336695259,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009633620689655173,
+      "loss": 2.7072,
+      "step": 387
+    },
+    {
+      "epoch": 0.07314544254877933,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.000963166144200627,
+      "loss": 2.6918,
+      "step": 388
+    },
+    {
+      "epoch": 0.07333396173060609,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009629702194357367,
+      "loss": 2.6682,
+      "step": 389
+    },
+    {
+      "epoch": 0.07352248091243284,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009627742946708464,
+      "loss": 2.6512,
+      "step": 390
+    },
+    {
+      "epoch": 0.07371100009425958,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0009625783699059562,
+      "loss": 2.718,
+      "step": 391
+    },
+    {
+      "epoch": 0.07389951927608635,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0009623824451410658,
+      "loss": 2.7208,
+      "step": 392
+    },
+    {
+      "epoch": 0.07408803845791309,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009621865203761756,
+      "loss": 2.7411,
+      "step": 393
+    },
+    {
+      "epoch": 0.07427655763973984,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009619905956112854,
+      "loss": 2.6763,
+      "step": 394
+    },
+    {
+      "epoch": 0.0744650768215666,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.000961794670846395,
+      "loss": 2.6919,
+      "step": 395
+    },
+    {
+      "epoch": 0.07465359600339334,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0009615987460815048,
+      "loss": 2.767,
+      "step": 396
+    },
+    {
+      "epoch": 0.07484211518522009,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009614028213166145,
+      "loss": 2.6868,
+      "step": 397
+    },
+    {
+      "epoch": 0.07503063436704685,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009612068965517241,
+      "loss": 2.6393,
+      "step": 398
+    },
+    {
+      "epoch": 0.0752191535488736,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009610109717868338,
+      "loss": 2.5917,
+      "step": 399
+    },
+    {
+      "epoch": 0.07540767273070034,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009608150470219436,
+      "loss": 2.709,
+      "step": 400
+    },
+    {
+      "epoch": 0.0755961919125271,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009606191222570532,
+      "loss": 2.6591,
+      "step": 401
+    },
+    {
+      "epoch": 0.07578471109435385,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.000960423197492163,
+      "loss": 2.7638,
+      "step": 402
+    },
+    {
+      "epoch": 0.0759732302761806,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009602272727272728,
+      "loss": 2.58,
+      "step": 403
+    },
+    {
+      "epoch": 0.07616174945800736,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009600313479623824,
+      "loss": 2.5257,
+      "step": 404
+    },
+    {
+      "epoch": 0.0763502686398341,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0009598354231974922,
+      "loss": 2.6512,
+      "step": 405
+    },
+    {
+      "epoch": 0.07653878782166085,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009596394984326019,
+      "loss": 2.6432,
+      "step": 406
+    },
+    {
+      "epoch": 0.07672730700348761,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0009594435736677116,
+      "loss": 2.6028,
+      "step": 407
+    },
+    {
+      "epoch": 0.07691582618531435,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.0009592476489028213,
+      "loss": 2.707,
+      "step": 408
+    },
+    {
+      "epoch": 0.0771043453671411,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0009590517241379311,
+      "loss": 2.5831,
+      "step": 409
+    },
+    {
+      "epoch": 0.07729286454896786,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009588557993730408,
+      "loss": 2.7447,
+      "step": 410
+    },
+    {
+      "epoch": 0.0774813837307946,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0009586598746081505,
+      "loss": 2.6452,
+      "step": 411
+    },
+    {
+      "epoch": 0.07766990291262135,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0009584639498432603,
+      "loss": 2.7138,
+      "step": 412
+    },
+    {
+      "epoch": 0.07785842209444811,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00095826802507837,
+      "loss": 2.5726,
+      "step": 413
+    },
+    {
+      "epoch": 0.07804694127627486,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009580721003134797,
+      "loss": 2.7128,
+      "step": 414
+    },
+    {
+      "epoch": 0.0782354604581016,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0009578761755485894,
+      "loss": 2.5482,
+      "step": 415
+    },
+    {
+      "epoch": 0.07842397963992837,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0009576802507836991,
+      "loss": 2.768,
+      "step": 416
+    },
+    {
+      "epoch": 0.07861249882175511,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009574843260188087,
+      "loss": 2.774,
+      "step": 417
+    },
+    {
+      "epoch": 0.07880101800358187,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009572884012539185,
+      "loss": 2.7388,
+      "step": 418
+    },
+    {
+      "epoch": 0.07898953718540862,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0009570924764890282,
+      "loss": 2.5905,
+      "step": 419
+    },
+    {
+      "epoch": 0.07917805636723536,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009568965517241379,
+      "loss": 2.7478,
+      "step": 420
+    },
+    {
+      "epoch": 0.07936657554906212,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009567006269592476,
+      "loss": 2.699,
+      "step": 421
+    },
+    {
+      "epoch": 0.07955509473088887,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009565047021943574,
+      "loss": 2.5614,
+      "step": 422
+    },
+    {
+      "epoch": 0.07974361391271562,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009563087774294671,
+      "loss": 2.6036,
+      "step": 423
+    },
+    {
+      "epoch": 0.07993213309454238,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009561128526645768,
+      "loss": 2.6515,
+      "step": 424
+    },
+    {
+      "epoch": 0.08012065227636912,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009559169278996866,
+      "loss": 2.7026,
+      "step": 425
+    },
+    {
+      "epoch": 0.08030917145819587,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009557210031347962,
+      "loss": 2.7497,
+      "step": 426
+    },
+    {
+      "epoch": 0.08049769064002263,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.000955525078369906,
+      "loss": 2.6129,
+      "step": 427
+    },
+    {
+      "epoch": 0.08068620982184938,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009553291536050157,
+      "loss": 2.6113,
+      "step": 428
+    },
+    {
+      "epoch": 0.08087472900367612,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009551332288401254,
+      "loss": 2.4547,
+      "step": 429
+    },
+    {
+      "epoch": 0.08106324818550288,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009549373040752351,
+      "loss": 2.6197,
+      "step": 430
+    },
+    {
+      "epoch": 0.08125176736732963,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009547413793103449,
+      "loss": 2.684,
+      "step": 431
+    },
+    {
+      "epoch": 0.08144028654915637,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009545454545454546,
+      "loss": 2.6874,
+      "step": 432
+    },
+    {
+      "epoch": 0.08162880573098313,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009543495297805643,
+      "loss": 2.6019,
+      "step": 433
+    },
+    {
+      "epoch": 0.08181732491280988,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009541536050156741,
+      "loss": 2.6309,
+      "step": 434
+    },
+    {
+      "epoch": 0.08200584409463663,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0009539576802507836,
+      "loss": 2.6848,
+      "step": 435
+    },
+    {
+      "epoch": 0.08219436327646339,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009537617554858934,
+      "loss": 2.7124,
+      "step": 436
+    },
+    {
+      "epoch": 0.08238288245829013,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009535658307210031,
+      "loss": 2.5744,
+      "step": 437
+    },
+    {
+      "epoch": 0.08257140164011688,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0009533699059561129,
+      "loss": 2.7689,
+      "step": 438
+    },
+    {
+      "epoch": 0.08275992082194364,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009531739811912225,
+      "loss": 2.7709,
+      "step": 439
+    },
+    {
+      "epoch": 0.08294844000377039,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009529780564263323,
+      "loss": 2.5495,
+      "step": 440
+    },
+    {
+      "epoch": 0.08313695918559713,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009527821316614421,
+      "loss": 2.696,
+      "step": 441
+    },
+    {
+      "epoch": 0.08332547836742389,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009525862068965517,
+      "loss": 2.6657,
+      "step": 442
+    },
+    {
+      "epoch": 0.08351399754925064,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009523902821316615,
+      "loss": 2.6998,
+      "step": 443
+    },
+    {
+      "epoch": 0.08370251673107738,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009521943573667712,
+      "loss": 2.7154,
+      "step": 444
+    },
+    {
+      "epoch": 0.08389103591290414,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009519984326018809,
+      "loss": 2.6478,
+      "step": 445
+    },
+    {
+      "epoch": 0.08407955509473089,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009518025078369906,
+      "loss": 2.6899,
+      "step": 446
+    },
+    {
+      "epoch": 0.08426807427655764,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0009516065830721004,
+      "loss": 2.7137,
+      "step": 447
+    },
+    {
+      "epoch": 0.0844565934583844,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00095141065830721,
+      "loss": 2.6207,
+      "step": 448
+    },
+    {
+      "epoch": 0.08464511264021114,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009512147335423198,
+      "loss": 2.7149,
+      "step": 449
+    },
+    {
+      "epoch": 0.08483363182203789,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009510188087774296,
+      "loss": 2.7011,
+      "step": 450
+    },
+    {
+      "epoch": 0.08502215100386465,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0009508228840125392,
+      "loss": 2.6496,
+      "step": 451
+    },
+    {
+      "epoch": 0.0852106701856914,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.000950626959247649,
+      "loss": 2.6714,
+      "step": 452
+    },
+    {
+      "epoch": 0.08539918936751814,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0009504310344827587,
+      "loss": 2.6271,
+      "step": 453
+    },
+    {
+      "epoch": 0.0855877085493449,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009502351097178683,
+      "loss": 2.6513,
+      "step": 454
+    },
+    {
+      "epoch": 0.08577622773117165,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.000950039184952978,
+      "loss": 2.6638,
+      "step": 455
+    },
+    {
+      "epoch": 0.0859647469129984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009498432601880878,
+      "loss": 2.7398,
+      "step": 456
+    },
+    {
+      "epoch": 0.08615326609482515,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009496473354231974,
+      "loss": 2.7013,
+      "step": 457
+    },
+    {
+      "epoch": 0.0863417852766519,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009494514106583072,
+      "loss": 2.6336,
+      "step": 458
+    },
+    {
+      "epoch": 0.08653030445847865,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.000949255485893417,
+      "loss": 2.5915,
+      "step": 459
+    },
+    {
+      "epoch": 0.0867188236403054,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009490595611285266,
+      "loss": 2.6545,
+      "step": 460
+    },
+    {
+      "epoch": 0.08690734282213215,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009488636363636364,
+      "loss": 2.6792,
+      "step": 461
+    },
+    {
+      "epoch": 0.0870958620039589,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009486677115987461,
+      "loss": 2.6238,
+      "step": 462
+    },
+    {
+      "epoch": 0.08728438118578566,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009484717868338558,
+      "loss": 2.6929,
+      "step": 463
+    },
+    {
+      "epoch": 0.0874729003676124,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009482758620689655,
+      "loss": 2.7269,
+      "step": 464
+    },
+    {
+      "epoch": 0.08766141954943915,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009480799373040753,
+      "loss": 2.6728,
+      "step": 465
+    },
+    {
+      "epoch": 0.08784993873126591,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.000947884012539185,
+      "loss": 2.667,
+      "step": 466
+    },
+    {
+      "epoch": 0.08803845791309266,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0009476880877742947,
+      "loss": 2.7706,
+      "step": 467
+    },
+    {
+      "epoch": 0.0882269770949194,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0009474921630094045,
+      "loss": 2.7464,
+      "step": 468
+    },
+    {
+      "epoch": 0.08841549627674616,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0009472962382445142,
+      "loss": 2.6004,
+      "step": 469
+    },
+    {
+      "epoch": 0.08860401545857291,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009471003134796239,
+      "loss": 2.6237,
+      "step": 470
+    },
+    {
+      "epoch": 0.08879253464039966,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009469043887147336,
+      "loss": 2.6628,
+      "step": 471
+    },
+    {
+      "epoch": 0.08898105382222642,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009467084639498434,
+      "loss": 2.7066,
+      "step": 472
+    },
+    {
+      "epoch": 0.08916957300405316,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009465125391849529,
+      "loss": 2.6655,
+      "step": 473
+    },
+    {
+      "epoch": 0.08935809218587991,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009463166144200627,
+      "loss": 2.6333,
+      "step": 474
+    },
+    {
+      "epoch": 0.08954661136770667,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009461206896551724,
+      "loss": 2.5766,
+      "step": 475
+    },
+    {
+      "epoch": 0.08973513054953342,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0009459247648902821,
+      "loss": 2.7387,
+      "step": 476
+    },
+    {
+      "epoch": 0.08992364973136016,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0009457288401253918,
+      "loss": 2.7342,
+      "step": 477
+    },
+    {
+      "epoch": 0.09011216891318692,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0009455329153605016,
+      "loss": 2.6416,
+      "step": 478
+    },
+    {
+      "epoch": 0.09030068809501367,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009453369905956113,
+      "loss": 2.6143,
+      "step": 479
+    },
+    {
+      "epoch": 0.09048920727684041,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.000945141065830721,
+      "loss": 2.7185,
+      "step": 480
+    },
+    {
+      "epoch": 0.09067772645866717,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0009449451410658308,
+      "loss": 2.6152,
+      "step": 481
+    },
+    {
+      "epoch": 0.09086624564049392,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009447492163009404,
+      "loss": 2.6592,
+      "step": 482
+    },
+    {
+      "epoch": 0.09105476482232067,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009445532915360502,
+      "loss": 2.5181,
+      "step": 483
+    },
+    {
+      "epoch": 0.09124328400414743,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009443573667711599,
+      "loss": 2.6332,
+      "step": 484
+    },
+    {
+      "epoch": 0.09143180318597417,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0009441614420062696,
+      "loss": 2.521,
+      "step": 485
+    },
+    {
+      "epoch": 0.09162032236780092,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009439655172413793,
+      "loss": 2.6339,
+      "step": 486
+    },
+    {
+      "epoch": 0.09180884154962768,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009437695924764891,
+      "loss": 2.6627,
+      "step": 487
+    },
+    {
+      "epoch": 0.09199736073145443,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009435736677115988,
+      "loss": 2.6227,
+      "step": 488
+    },
+    {
+      "epoch": 0.09218587991328117,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009433777429467085,
+      "loss": 2.784,
+      "step": 489
+    },
+    {
+      "epoch": 0.09237439909510793,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.0009431818181818183,
+      "loss": 2.5622,
+      "step": 490
+    },
+    {
+      "epoch": 0.09256291827693468,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009429858934169278,
+      "loss": 2.6712,
+      "step": 491
+    },
+    {
+      "epoch": 0.09275143745876142,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009427899686520376,
+      "loss": 2.5781,
+      "step": 492
+    },
+    {
+      "epoch": 0.09293995664058818,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0009425940438871473,
+      "loss": 2.6193,
+      "step": 493
+    },
+    {
+      "epoch": 0.09312847582241493,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009423981191222571,
+      "loss": 2.716,
+      "step": 494
+    },
+    {
+      "epoch": 0.09331699500424168,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0009422021943573667,
+      "loss": 2.745,
+      "step": 495
+    },
+    {
+      "epoch": 0.09350551418606844,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009420062695924765,
+      "loss": 2.5251,
+      "step": 496
+    },
+    {
+      "epoch": 0.09369403336789518,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009418103448275863,
+      "loss": 2.7023,
+      "step": 497
+    },
+    {
+      "epoch": 0.09388255254972193,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009416144200626959,
+      "loss": 2.7697,
+      "step": 498
+    },
+    {
+      "epoch": 0.09407107173154869,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009414184952978057,
+      "loss": 2.6253,
+      "step": 499
+    },
+    {
+      "epoch": 0.09425959091337544,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0009412225705329154,
+      "loss": 2.6668,
+      "step": 500
+    },
+    {
+      "epoch": 0.09425959091337544,
+      "eval_runtime": 58.5785,
+      "eval_samples_per_second": 17.481,
+      "eval_steps_per_second": 0.546,
+      "step": 500
+    },
+    {
+      "epoch": 0.09425959091337544,
+      "eval/hellaswag_acc": 0.37572196773551086,
+      "eval/hellaswag_acc_norm": 0.4714200358494324,
+      "eval_hellaswag_elapsed_time": 195.95180106163025,
+      "step": 500
+    },
+    {
+      "epoch": 0.09444811009520218,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009410266457680251,
+      "loss": 2.6645,
+      "step": 501
+    },
+    {
+      "epoch": 0.09463662927702894,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0009408307210031348,
+      "loss": 2.7233,
+      "step": 502
+    },
+    {
+      "epoch": 0.09482514845885569,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009406347962382446,
+      "loss": 2.6959,
+      "step": 503
+    },
+    {
+      "epoch": 0.09501366764068243,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009404388714733542,
+      "loss": 2.747,
+      "step": 504
+    },
+    {
+      "epoch": 0.0952021868225092,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.000940242946708464,
+      "loss": 2.521,
+      "step": 505
+    },
+    {
+      "epoch": 0.09539070600433594,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009400470219435738,
+      "loss": 2.7368,
+      "step": 506
+    },
+    {
+      "epoch": 0.09557922518616269,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009398510971786834,
+      "loss": 2.6509,
+      "step": 507
+    },
+    {
+      "epoch": 0.09576774436798945,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009396551724137932,
+      "loss": 2.785,
+      "step": 508
+    },
+    {
+      "epoch": 0.09595626354981619,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009394592476489029,
+      "loss": 2.5647,
+      "step": 509
+    },
+    {
+      "epoch": 0.09614478273164294,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009392633228840125,
+      "loss": 2.6087,
+      "step": 510
+    },
+    {
+      "epoch": 0.0963333019134697,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009390673981191222,
+      "loss": 2.6032,
+      "step": 511
+    },
+    {
+      "epoch": 0.09652182109529645,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.000938871473354232,
+      "loss": 2.6934,
+      "step": 512
+    },
+    {
+      "epoch": 0.09671034027712319,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009386755485893416,
+      "loss": 2.7077,
+      "step": 513
+    },
+    {
+      "epoch": 0.09689885945894995,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009384796238244514,
+      "loss": 2.7372,
+      "step": 514
+    },
+    {
+      "epoch": 0.0970873786407767,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009382836990595611,
+      "loss": 2.5907,
+      "step": 515
+    },
+    {
+      "epoch": 0.09727589782260344,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0009380877742946708,
+      "loss": 2.5623,
+      "step": 516
+    },
+    {
+      "epoch": 0.0974644170044302,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009378918495297806,
+      "loss": 2.6949,
+      "step": 517
+    },
+    {
+      "epoch": 0.09765293618625695,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0009376959247648903,
+      "loss": 2.6505,
+      "step": 518
+    },
+    {
+      "epoch": 0.0978414553680837,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009375,
+      "loss": 2.6902,
+      "step": 519
+    },
+    {
+      "epoch": 0.09802997454991046,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009373040752351097,
+      "loss": 2.6529,
+      "step": 520
+    },
+    {
+      "epoch": 0.0982184937317372,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009371081504702195,
+      "loss": 2.5571,
+      "step": 521
+    },
+    {
+      "epoch": 0.09840701291356395,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009369122257053292,
+      "loss": 2.7718,
+      "step": 522
+    },
+    {
+      "epoch": 0.09859553209539071,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009367163009404389,
+      "loss": 2.7112,
+      "step": 523
+    },
+    {
+      "epoch": 0.09878405127721746,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009365203761755486,
+      "loss": 2.5318,
+      "step": 524
+    },
+    {
+      "epoch": 0.0989725704590442,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009363244514106584,
+      "loss": 2.6242,
+      "step": 525
+    },
+    {
+      "epoch": 0.09916108964087096,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009361285266457681,
+      "loss": 2.6603,
+      "step": 526
+    },
+    {
+      "epoch": 0.09934960882269771,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009359326018808778,
+      "loss": 2.7204,
+      "step": 527
+    },
+    {
+      "epoch": 0.09953812800452445,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009357366771159876,
+      "loss": 2.6355,
+      "step": 528
+    },
+    {
+      "epoch": 0.09972664718635121,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009355407523510971,
+      "loss": 2.6126,
+      "step": 529
+    },
+    {
+      "epoch": 0.09991516636817796,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0009353448275862069,
+      "loss": 2.5042,
+      "step": 530
+    },
+    {
+      "epoch": 0.1001036855500047,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009351489028213166,
+      "loss": 2.639,
+      "step": 531
+    },
+    {
+      "epoch": 0.10029220473183147,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009349529780564263,
+      "loss": 2.6981,
+      "step": 532
+    },
+    {
+      "epoch": 0.10048072391365821,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.000934757053291536,
+      "loss": 2.6578,
+      "step": 533
+    },
+    {
+      "epoch": 0.10066924309548496,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009345611285266458,
+      "loss": 2.7651,
+      "step": 534
+    },
+    {
+      "epoch": 0.10085776227731172,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009343652037617555,
+      "loss": 2.6639,
+      "step": 535
+    },
+    {
+      "epoch": 0.10104628145913847,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009341692789968652,
+      "loss": 2.6911,
+      "step": 536
+    },
+    {
+      "epoch": 0.10123480064096523,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.000933973354231975,
+      "loss": 2.6213,
+      "step": 537
+    },
+    {
+      "epoch": 0.10142331982279197,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009337774294670846,
+      "loss": 2.6084,
+      "step": 538
+    },
+    {
+      "epoch": 0.10161183900461872,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009335815047021944,
+      "loss": 2.6893,
+      "step": 539
+    },
+    {
+      "epoch": 0.10180035818644548,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0009333855799373041,
+      "loss": 2.547,
+      "step": 540
+    },
+    {
+      "epoch": 0.10198887736827222,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009331896551724138,
+      "loss": 2.7084,
+      "step": 541
+    },
+    {
+      "epoch": 0.10217739655009897,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0009329937304075235,
+      "loss": 2.6611,
+      "step": 542
+    },
+    {
+      "epoch": 0.10236591573192573,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009327978056426333,
+      "loss": 2.658,
+      "step": 543
+    },
+    {
+      "epoch": 0.10255443491375248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.000932601880877743,
+      "loss": 2.7423,
+      "step": 544
+    },
+    {
+      "epoch": 0.10274295409557922,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0009324059561128527,
+      "loss": 2.6237,
+      "step": 545
+    },
+    {
+      "epoch": 0.10293147327740598,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009322100313479625,
+      "loss": 2.6846,
+      "step": 546
+    },
+    {
+      "epoch": 0.10311999245923273,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009320141065830722,
+      "loss": 2.5963,
+      "step": 547
+    },
+    {
+      "epoch": 0.10330851164105948,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009318181818181818,
+      "loss": 2.6334,
+      "step": 548
+    },
+    {
+      "epoch": 0.10349703082288624,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009316222570532915,
+      "loss": 2.6657,
+      "step": 549
+    },
+    {
+      "epoch": 0.10368555000471298,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009314263322884013,
+      "loss": 2.7307,
+      "step": 550
+    },
+    {
+      "epoch": 0.10387406918653973,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009312304075235109,
+      "loss": 2.7773,
+      "step": 551
+    },
+    {
+      "epoch": 0.10406258836836649,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009310344827586207,
+      "loss": 2.7054,
+      "step": 552
+    },
+    {
+      "epoch": 0.10425110755019323,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009308385579937305,
+      "loss": 2.6325,
+      "step": 553
+    },
+    {
+      "epoch": 0.10443962673201998,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0009306426332288401,
+      "loss": 2.5011,
+      "step": 554
+    },
+    {
+      "epoch": 0.10462814591384674,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009304467084639499,
+      "loss": 2.6252,
+      "step": 555
+    },
+    {
+      "epoch": 0.10481666509567349,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009302507836990596,
+      "loss": 2.6176,
+      "step": 556
+    },
+    {
+      "epoch": 0.10500518427750023,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009300548589341693,
+      "loss": 2.699,
+      "step": 557
+    },
+    {
+      "epoch": 0.10519370345932699,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.000929858934169279,
+      "loss": 2.5058,
+      "step": 558
+    },
+    {
+      "epoch": 0.10538222264115374,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009296630094043888,
+      "loss": 2.6034,
+      "step": 559
+    },
+    {
+      "epoch": 0.10557074182298049,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0009294670846394984,
+      "loss": 2.5785,
+      "step": 560
+    },
+    {
+      "epoch": 0.10575926100480725,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0009292711598746082,
+      "loss": 2.7846,
+      "step": 561
+    },
+    {
+      "epoch": 0.10594778018663399,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.000929075235109718,
+      "loss": 2.7049,
+      "step": 562
+    },
+    {
+      "epoch": 0.10613629936846074,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009288793103448276,
+      "loss": 2.72,
+      "step": 563
+    },
+    {
+      "epoch": 0.1063248185502875,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009286833855799374,
+      "loss": 2.7468,
+      "step": 564
+    },
+    {
+      "epoch": 0.10651333773211424,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009284874608150471,
+      "loss": 2.706,
+      "step": 565
+    },
+    {
+      "epoch": 0.10670185691394099,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009282915360501567,
+      "loss": 2.6664,
+      "step": 566
+    },
+    {
+      "epoch": 0.10689037609576775,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0009280956112852664,
+      "loss": 2.6685,
+      "step": 567
+    },
+    {
+      "epoch": 0.1070788952775945,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0009278996865203762,
+      "loss": 2.6295,
+      "step": 568
+    },
+    {
+      "epoch": 0.10726741445942124,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0009277037617554858,
+      "loss": 2.7556,
+      "step": 569
+    },
+    {
+      "epoch": 0.107455933641248,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009275078369905956,
+      "loss": 2.6027,
+      "step": 570
+    },
+    {
+      "epoch": 0.10764445282307475,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0009273119122257053,
+      "loss": 2.622,
+      "step": 571
+    },
+    {
+      "epoch": 0.1078329720049015,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.000927115987460815,
+      "loss": 2.6136,
+      "step": 572
+    },
+    {
+      "epoch": 0.10802149118672826,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0009269200626959248,
+      "loss": 2.6196,
+      "step": 573
+    },
+    {
+      "epoch": 0.108210010368555,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0009267241379310345,
+      "loss": 2.6569,
+      "step": 574
+    },
+    {
+      "epoch": 0.10839852955038175,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009265282131661443,
+      "loss": 2.7018,
+      "step": 575
+    },
+    {
+      "epoch": 0.10858704873220851,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009263322884012539,
+      "loss": 2.5521,
+      "step": 576
+    },
+    {
+      "epoch": 0.10877556791403525,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009261363636363637,
+      "loss": 2.6091,
+      "step": 577
+    },
+    {
+      "epoch": 0.108964087095862,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009259404388714734,
+      "loss": 2.5969,
+      "step": 578
+    },
+    {
+      "epoch": 0.10915260627768876,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009257445141065831,
+      "loss": 2.6032,
+      "step": 579
+    },
+    {
+      "epoch": 0.1093411254595155,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009255485893416928,
+      "loss": 2.6755,
+      "step": 580
+    },
+    {
+      "epoch": 0.10952964464134225,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009253526645768026,
+      "loss": 2.6504,
+      "step": 581
+    },
+    {
+      "epoch": 0.10971816382316901,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009251567398119123,
+      "loss": 2.7218,
+      "step": 582
+    },
+    {
+      "epoch": 0.10990668300499576,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.000924960815047022,
+      "loss": 2.6733,
+      "step": 583
+    },
+    {
+      "epoch": 0.1100952021868225,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009247648902821318,
+      "loss": 2.7395,
+      "step": 584
+    },
+    {
+      "epoch": 0.11028372136864927,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009245689655172413,
+      "loss": 2.7391,
+      "step": 585
+    },
+    {
+      "epoch": 0.11047224055047601,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009243730407523511,
+      "loss": 2.5436,
+      "step": 586
+    },
+    {
+      "epoch": 0.11066075973230276,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009241771159874608,
+      "loss": 2.6671,
+      "step": 587
+    },
+    {
+      "epoch": 0.11084927891412952,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0009239811912225705,
+      "loss": 2.7918,
+      "step": 588
+    },
+    {
+      "epoch": 0.11103779809595626,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0009237852664576802,
+      "loss": 2.6591,
+      "step": 589
+    },
+    {
+      "epoch": 0.11122631727778301,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00092358934169279,
+      "loss": 2.5917,
+      "step": 590
+    },
+    {
+      "epoch": 0.11141483645960977,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009233934169278996,
+      "loss": 2.5555,
+      "step": 591
+    },
+    {
+      "epoch": 0.11160335564143652,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0009231974921630094,
+      "loss": 2.7138,
+      "step": 592
+    },
+    {
+      "epoch": 0.11179187482326326,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009230015673981192,
+      "loss": 2.7638,
+      "step": 593
+    },
+    {
+      "epoch": 0.11198039400509002,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0009228056426332288,
+      "loss": 2.6662,
+      "step": 594
+    },
+    {
+      "epoch": 0.11216891318691677,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0009226097178683386,
+      "loss": 2.7772,
+      "step": 595
+    },
+    {
+      "epoch": 0.11235743236874352,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0009224137931034483,
+      "loss": 2.7929,
+      "step": 596
+    },
+    {
+      "epoch": 0.11254595155057028,
+      "grad_norm": 0.75,
+      "learning_rate": 0.000922217868338558,
+      "loss": 2.7971,
+      "step": 597
+    },
+    {
+      "epoch": 0.11273447073239702,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009220219435736677,
+      "loss": 2.7352,
+      "step": 598
+    },
+    {
+      "epoch": 0.11292298991422377,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009218260188087775,
+      "loss": 2.6516,
+      "step": 599
+    },
+    {
+      "epoch": 0.11311150909605053,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0009216300940438871,
+      "loss": 2.7391,
+      "step": 600
+    },
+    {
+      "epoch": 0.11330002827787727,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0009214341692789969,
+      "loss": 2.6186,
+      "step": 601
+    },
+    {
+      "epoch": 0.11348854745970402,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009212382445141067,
+      "loss": 2.7238,
+      "step": 602
+    },
+    {
+      "epoch": 0.11367706664153078,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0009210423197492164,
+      "loss": 2.5381,
+      "step": 603
+    },
+    {
+      "epoch": 0.11386558582335753,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.000920846394984326,
+      "loss": 2.6816,
+      "step": 604
+    },
+    {
+      "epoch": 0.11405410500518427,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0009206504702194357,
+      "loss": 2.5738,
+      "step": 605
+    },
+    {
+      "epoch": 0.11424262418701103,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009204545454545455,
+      "loss": 2.6701,
+      "step": 606
+    },
+    {
+      "epoch": 0.11443114336883778,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0009202586206896551,
+      "loss": 2.668,
+      "step": 607
+    },
+    {
+      "epoch": 0.11461966255066453,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0009200626959247649,
+      "loss": 2.6853,
+      "step": 608
+    },
+    {
+      "epoch": 0.11480818173249129,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009198667711598747,
+      "loss": 2.6926,
+      "step": 609
+    },
+    {
+      "epoch": 0.11499670091431803,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009196708463949843,
+      "loss": 2.7387,
+      "step": 610
+    },
+    {
+      "epoch": 0.11518522009614478,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0009194749216300941,
+      "loss": 2.613,
+      "step": 611
+    },
+    {
+      "epoch": 0.11537373927797154,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009192789968652038,
+      "loss": 2.5144,
+      "step": 612
+    },
+    {
+      "epoch": 0.11556225845979828,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0009190830721003135,
+      "loss": 2.7777,
+      "step": 613
+    },
+    {
+      "epoch": 0.11575077764162503,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009188871473354232,
+      "loss": 2.6103,
+      "step": 614
+    },
+    {
+      "epoch": 0.11593929682345179,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.000918691222570533,
+      "loss": 2.6806,
+      "step": 615
+    },
+    {
+      "epoch": 0.11612781600527854,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009184952978056426,
+      "loss": 2.6122,
+      "step": 616
+    },
+    {
+      "epoch": 0.11631633518710528,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009182993730407524,
+      "loss": 2.6198,
+      "step": 617
+    },
+    {
+      "epoch": 0.11650485436893204,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009181034482758622,
+      "loss": 2.5324,
+      "step": 618
+    },
+    {
+      "epoch": 0.11669337355075879,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0009179075235109718,
+      "loss": 2.6424,
+      "step": 619
+    },
+    {
+      "epoch": 0.11688189273258554,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009177115987460816,
+      "loss": 2.6544,
+      "step": 620
+    },
+    {
+      "epoch": 0.1170704119144123,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009175156739811913,
+      "loss": 2.6725,
+      "step": 621
+    },
+    {
+      "epoch": 0.11725893109623904,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.000917319749216301,
+      "loss": 2.6113,
+      "step": 622
+    },
+    {
+      "epoch": 0.11744745027806579,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009171238244514106,
+      "loss": 2.6232,
+      "step": 623
+    },
+    {
+      "epoch": 0.11763596945989255,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009169278996865204,
+      "loss": 2.5457,
+      "step": 624
+    },
+    {
+      "epoch": 0.1178244886417193,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00091673197492163,
+      "loss": 2.5777,
+      "step": 625
+    },
+    {
+      "epoch": 0.11801300782354604,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0009165360501567398,
+      "loss": 2.6668,
+      "step": 626
+    },
+    {
+      "epoch": 0.1182015270053728,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0009163401253918495,
+      "loss": 2.7571,
+      "step": 627
+    },
+    {
+      "epoch": 0.11839004618719955,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009161442006269592,
+      "loss": 2.6457,
+      "step": 628
+    },
+    {
+      "epoch": 0.11857856536902629,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.000915948275862069,
+      "loss": 2.5717,
+      "step": 629
+    },
+    {
+      "epoch": 0.11876708455085305,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0009157523510971787,
+      "loss": 2.6518,
+      "step": 630
+    },
+    {
+      "epoch": 0.1189556037326798,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009155564263322885,
+      "loss": 2.6215,
+      "step": 631
+    },
+    {
+      "epoch": 0.11914412291450655,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009153605015673981,
+      "loss": 2.6829,
+      "step": 632
+    },
+    {
+      "epoch": 0.1193326420963333,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0009151645768025079,
+      "loss": 2.6905,
+      "step": 633
+    },
+    {
+      "epoch": 0.11952116127816005,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009149686520376176,
+      "loss": 2.5993,
+      "step": 634
+    },
+    {
+      "epoch": 0.1197096804599868,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009147727272727273,
+      "loss": 2.6115,
+      "step": 635
+    },
+    {
+      "epoch": 0.11989819964181356,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.000914576802507837,
+      "loss": 2.7553,
+      "step": 636
+    },
+    {
+      "epoch": 0.1200867188236403,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0009143808777429468,
+      "loss": 2.685,
+      "step": 637
+    },
+    {
+      "epoch": 0.12027523800546705,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009141849529780565,
+      "loss": 2.6026,
+      "step": 638
+    },
+    {
+      "epoch": 0.12046375718729381,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0009139890282131662,
+      "loss": 2.7128,
+      "step": 639
+    },
+    {
+      "epoch": 0.12065227636912056,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.000913793103448276,
+      "loss": 2.5892,
+      "step": 640
+    },
+    {
+      "epoch": 0.1208407955509473,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009135971786833855,
+      "loss": 2.6472,
+      "step": 641
+    },
+    {
+      "epoch": 0.12102931473277406,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009134012539184953,
+      "loss": 2.6738,
+      "step": 642
+    },
+    {
+      "epoch": 0.12121783391460081,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.000913205329153605,
+      "loss": 2.6913,
+      "step": 643
+    },
+    {
+      "epoch": 0.12140635309642756,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009130094043887147,
+      "loss": 2.6355,
+      "step": 644
+    },
+    {
+      "epoch": 0.12159487227825432,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0009128134796238244,
+      "loss": 2.5979,
+      "step": 645
+    },
+    {
+      "epoch": 0.12178339146008106,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009126175548589342,
+      "loss": 2.7308,
+      "step": 646
+    },
+    {
+      "epoch": 0.12197191064190781,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009124216300940438,
+      "loss": 2.6809,
+      "step": 647
+    },
+    {
+      "epoch": 0.12216042982373457,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009122257053291536,
+      "loss": 2.7713,
+      "step": 648
+    },
+    {
+      "epoch": 0.12234894900556131,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0009120297805642634,
+      "loss": 2.4518,
+      "step": 649
+    },
+    {
+      "epoch": 0.12253746818738806,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.000911833855799373,
+      "loss": 2.5884,
+      "step": 650
+    },
+    {
+      "epoch": 0.12272598736921482,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009116379310344828,
+      "loss": 2.6641,
+      "step": 651
+    },
+    {
+      "epoch": 0.12291450655104157,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009114420062695925,
+      "loss": 2.5444,
+      "step": 652
+    },
+    {
+      "epoch": 0.12310302573286831,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009112460815047022,
+      "loss": 2.6773,
+      "step": 653
+    },
+    {
+      "epoch": 0.12329154491469507,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009110501567398119,
+      "loss": 2.6689,
+      "step": 654
+    },
+    {
+      "epoch": 0.12348006409652182,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009108542319749217,
+      "loss": 2.8371,
+      "step": 655
+    },
+    {
+      "epoch": 0.12366858327834858,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009106583072100313,
+      "loss": 2.6491,
+      "step": 656
+    },
+    {
+      "epoch": 0.12385710246017533,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009104623824451411,
+      "loss": 2.7192,
+      "step": 657
+    },
+    {
+      "epoch": 0.12404562164200207,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009102664576802509,
+      "loss": 2.7254,
+      "step": 658
+    },
+    {
+      "epoch": 0.12423414082382883,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0009100705329153606,
+      "loss": 2.6403,
+      "step": 659
+    },
+    {
+      "epoch": 0.12442266000565558,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009098746081504702,
+      "loss": 2.6217,
+      "step": 660
+    },
+    {
+      "epoch": 0.12461117918748232,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0009096786833855799,
+      "loss": 2.7402,
+      "step": 661
+    },
+    {
+      "epoch": 0.12479969836930908,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009094827586206897,
+      "loss": 2.7237,
+      "step": 662
+    },
+    {
+      "epoch": 0.12498821755113583,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009092868338557993,
+      "loss": 2.5321,
+      "step": 663
+    },
+    {
+      "epoch": 0.1251767367329626,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0009090909090909091,
+      "loss": 2.4766,
+      "step": 664
+    },
+    {
+      "epoch": 0.12536525591478934,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0009088949843260188,
+      "loss": 2.5655,
+      "step": 665
+    },
+    {
+      "epoch": 0.12555377509661608,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009086990595611285,
+      "loss": 2.4495,
+      "step": 666
+    },
+    {
+      "epoch": 0.12574229427844283,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009085031347962383,
+      "loss": 2.7035,
+      "step": 667
+    },
+    {
+      "epoch": 0.12593081346026958,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.000908307210031348,
+      "loss": 2.5528,
+      "step": 668
+    },
+    {
+      "epoch": 0.12611933264209632,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009081112852664577,
+      "loss": 2.5787,
+      "step": 669
+    },
+    {
+      "epoch": 0.1263078518239231,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0009079153605015674,
+      "loss": 2.6167,
+      "step": 670
+    },
+    {
+      "epoch": 0.12649637100574984,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009077194357366772,
+      "loss": 2.7147,
+      "step": 671
+    },
+    {
+      "epoch": 0.1266848901875766,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0009075235109717868,
+      "loss": 2.7819,
+      "step": 672
+    },
+    {
+      "epoch": 0.12687340936940333,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009073275862068966,
+      "loss": 2.5718,
+      "step": 673
+    },
+    {
+      "epoch": 0.12706192855123008,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0009071316614420063,
+      "loss": 2.6887,
+      "step": 674
+    },
+    {
+      "epoch": 0.12725044773305683,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.000906935736677116,
+      "loss": 2.6037,
+      "step": 675
+    },
+    {
+      "epoch": 0.1274389669148836,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0009067398119122258,
+      "loss": 2.6148,
+      "step": 676
+    },
+    {
+      "epoch": 0.12762748609671035,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0009065438871473355,
+      "loss": 2.6737,
+      "step": 677
+    },
+    {
+      "epoch": 0.1278160052785371,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0009063479623824452,
+      "loss": 2.7679,
+      "step": 678
+    },
+    {
+      "epoch": 0.12800452446036384,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0009061520376175548,
+      "loss": 2.6948,
+      "step": 679
+    },
+    {
+      "epoch": 0.12819304364219059,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009059561128526646,
+      "loss": 2.6185,
+      "step": 680
+    },
+    {
+      "epoch": 0.12838156282401733,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0009057601880877742,
+      "loss": 2.7351,
+      "step": 681
+    },
+    {
+      "epoch": 0.1285700820058441,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.000905564263322884,
+      "loss": 2.6394,
+      "step": 682
+    },
+    {
+      "epoch": 0.12875860118767085,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009053683385579937,
+      "loss": 2.7473,
+      "step": 683
+    },
+    {
+      "epoch": 0.1289471203694976,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0009051724137931034,
+      "loss": 2.5965,
+      "step": 684
+    },
+    {
+      "epoch": 0.12913563955132434,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0009049764890282132,
+      "loss": 2.8092,
+      "step": 685
+    },
+    {
+      "epoch": 0.1293241587331511,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009047805642633229,
+      "loss": 2.58,
+      "step": 686
+    },
+    {
+      "epoch": 0.12951267791497784,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009045846394984327,
+      "loss": 2.6549,
+      "step": 687
+    },
+    {
+      "epoch": 0.1297011970968046,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009043887147335423,
+      "loss": 2.6374,
+      "step": 688
+    },
+    {
+      "epoch": 0.12988971627863136,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009041927899686521,
+      "loss": 2.5968,
+      "step": 689
+    },
+    {
+      "epoch": 0.1300782354604581,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009039968652037618,
+      "loss": 2.6792,
+      "step": 690
+    },
+    {
+      "epoch": 0.13026675464228485,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0009038009404388715,
+      "loss": 2.6631,
+      "step": 691
+    },
+    {
+      "epoch": 0.1304552738241116,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0009036050156739812,
+      "loss": 2.7008,
+      "step": 692
+    },
+    {
+      "epoch": 0.13064379300593834,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.000903409090909091,
+      "loss": 2.5822,
+      "step": 693
+    },
+    {
+      "epoch": 0.13083231218776512,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0009032131661442007,
+      "loss": 2.5259,
+      "step": 694
+    },
+    {
+      "epoch": 0.13102083136959186,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0009030172413793104,
+      "loss": 2.5717,
+      "step": 695
+    },
+    {
+      "epoch": 0.1312093505514186,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009028213166144202,
+      "loss": 2.7658,
+      "step": 696
+    },
+    {
+      "epoch": 0.13139786973324535,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0009026253918495298,
+      "loss": 2.6304,
+      "step": 697
+    },
+    {
+      "epoch": 0.1315863889150721,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0009024294670846395,
+      "loss": 2.6773,
+      "step": 698
+    },
+    {
+      "epoch": 0.13177490809689885,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0009022335423197492,
+      "loss": 2.6761,
+      "step": 699
+    },
+    {
+      "epoch": 0.13196342727872562,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0009020376175548589,
+      "loss": 2.6741,
+      "step": 700
+    },
+    {
+      "epoch": 0.13215194646055237,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0009018416927899686,
+      "loss": 2.5493,
+      "step": 701
+    },
+    {
+      "epoch": 0.1323404656423791,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0009016457680250784,
+      "loss": 2.6143,
+      "step": 702
+    },
+    {
+      "epoch": 0.13252898482420586,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.000901449843260188,
+      "loss": 2.6671,
+      "step": 703
+    },
+    {
+      "epoch": 0.1327175040060326,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0009012539184952978,
+      "loss": 2.6108,
+      "step": 704
+    },
+    {
+      "epoch": 0.13290602318785935,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009010579937304076,
+      "loss": 2.6649,
+      "step": 705
+    },
+    {
+      "epoch": 0.13309454236968613,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0009008620689655172,
+      "loss": 2.6925,
+      "step": 706
+    },
+    {
+      "epoch": 0.13328306155151287,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.000900666144200627,
+      "loss": 2.6506,
+      "step": 707
+    },
+    {
+      "epoch": 0.13347158073333962,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0009004702194357367,
+      "loss": 2.7378,
+      "step": 708
+    },
+    {
+      "epoch": 0.13366009991516636,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0009002742946708464,
+      "loss": 2.6705,
+      "step": 709
+    },
+    {
+      "epoch": 0.1338486190969931,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0009000783699059561,
+      "loss": 2.7019,
+      "step": 710
+    },
+    {
+      "epoch": 0.13403713827881986,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008998824451410659,
+      "loss": 2.6219,
+      "step": 711
+    },
+    {
+      "epoch": 0.13422565746064663,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0008996865203761755,
+      "loss": 2.5641,
+      "step": 712
+    },
+    {
+      "epoch": 0.13441417664247338,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0008994905956112853,
+      "loss": 2.6977,
+      "step": 713
+    },
+    {
+      "epoch": 0.13460269582430012,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008992946708463951,
+      "loss": 2.7332,
+      "step": 714
+    },
+    {
+      "epoch": 0.13479121500612687,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008990987460815048,
+      "loss": 2.5868,
+      "step": 715
+    },
+    {
+      "epoch": 0.13497973418795361,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008989028213166145,
+      "loss": 2.6311,
+      "step": 716
+    },
+    {
+      "epoch": 0.1351682533697804,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008987068965517241,
+      "loss": 2.6369,
+      "step": 717
+    },
+    {
+      "epoch": 0.13535677255160714,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0008985109717868339,
+      "loss": 2.778,
+      "step": 718
+    },
+    {
+      "epoch": 0.13554529173343388,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0008983150470219435,
+      "loss": 2.6355,
+      "step": 719
+    },
+    {
+      "epoch": 0.13573381091526063,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008981191222570533,
+      "loss": 2.6524,
+      "step": 720
+    },
+    {
+      "epoch": 0.13592233009708737,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.000897923197492163,
+      "loss": 2.7543,
+      "step": 721
+    },
+    {
+      "epoch": 0.13611084927891412,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0008977272727272727,
+      "loss": 2.7062,
+      "step": 722
+    },
+    {
+      "epoch": 0.1362993684607409,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0008975313479623825,
+      "loss": 2.7506,
+      "step": 723
+    },
+    {
+      "epoch": 0.13648788764256764,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0008973354231974922,
+      "loss": 2.6015,
+      "step": 724
+    },
+    {
+      "epoch": 0.1366764068243944,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0008971394984326019,
+      "loss": 2.6391,
+      "step": 725
+    },
+    {
+      "epoch": 0.13686492600622113,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0008969435736677116,
+      "loss": 2.7587,
+      "step": 726
+    },
+    {
+      "epoch": 0.13705344518804788,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008967476489028214,
+      "loss": 2.7175,
+      "step": 727
+    },
+    {
+      "epoch": 0.13724196436987462,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.000896551724137931,
+      "loss": 2.6185,
+      "step": 728
+    },
+    {
+      "epoch": 0.1374304835517014,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0008963557993730408,
+      "loss": 2.5831,
+      "step": 729
+    },
+    {
+      "epoch": 0.13761900273352815,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0008961598746081505,
+      "loss": 2.6364,
+      "step": 730
+    },
+    {
+      "epoch": 0.1378075219153549,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008959639498432602,
+      "loss": 2.5928,
+      "step": 731
+    },
+    {
+      "epoch": 0.13799604109718164,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00089576802507837,
+      "loss": 2.6144,
+      "step": 732
+    },
+    {
+      "epoch": 0.13818456027900838,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008955721003134797,
+      "loss": 2.5887,
+      "step": 733
+    },
+    {
+      "epoch": 0.13837307946083513,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0008953761755485894,
+      "loss": 2.5888,
+      "step": 734
+    },
+    {
+      "epoch": 0.1385615986426619,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.000895180250783699,
+      "loss": 2.5938,
+      "step": 735
+    },
+    {
+      "epoch": 0.13875011782448865,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0008949843260188088,
+      "loss": 2.5876,
+      "step": 736
+    },
+    {
+      "epoch": 0.1389386370063154,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0008947884012539184,
+      "loss": 2.6677,
+      "step": 737
+    },
+    {
+      "epoch": 0.13912715618814214,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0008945924764890282,
+      "loss": 2.5932,
+      "step": 738
+    },
+    {
+      "epoch": 0.1393156753699689,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008943965517241379,
+      "loss": 2.6358,
+      "step": 739
+    },
+    {
+      "epoch": 0.13950419455179563,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008942006269592476,
+      "loss": 2.6529,
+      "step": 740
+    },
+    {
+      "epoch": 0.1396927137336224,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0008940047021943573,
+      "loss": 2.6557,
+      "step": 741
+    },
+    {
+      "epoch": 0.13988123291544916,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0008938087774294671,
+      "loss": 2.6333,
+      "step": 742
+    },
+    {
+      "epoch": 0.1400697520972759,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0008936128526645769,
+      "loss": 2.5974,
+      "step": 743
+    },
+    {
+      "epoch": 0.14025827127910265,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0008934169278996865,
+      "loss": 2.6484,
+      "step": 744
+    },
+    {
+      "epoch": 0.1404467904609294,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0008932210031347963,
+      "loss": 2.617,
+      "step": 745
+    },
+    {
+      "epoch": 0.14063530964275614,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.000893025078369906,
+      "loss": 2.6803,
+      "step": 746
+    },
+    {
+      "epoch": 0.14082382882458291,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0008928291536050157,
+      "loss": 2.6882,
+      "step": 747
+    },
+    {
+      "epoch": 0.14101234800640966,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0008926332288401254,
+      "loss": 2.6814,
+      "step": 748
+    },
+    {
+      "epoch": 0.1412008671882364,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0008924373040752352,
+      "loss": 2.7618,
+      "step": 749
+    },
+    {
+      "epoch": 0.14138938637006315,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008922413793103448,
+      "loss": 2.7598,
+      "step": 750
+    },
+    {
+      "epoch": 0.1415779055518899,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0008920454545454546,
+      "loss": 2.6827,
+      "step": 751
+    },
+    {
+      "epoch": 0.14176642473371664,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0008918495297805644,
+      "loss": 2.7531,
+      "step": 752
+    },
+    {
+      "epoch": 0.14195494391554342,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.000891653605015674,
+      "loss": 2.5641,
+      "step": 753
+    },
+    {
+      "epoch": 0.14214346309737017,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008914576802507837,
+      "loss": 2.6333,
+      "step": 754
+    },
+    {
+      "epoch": 0.1423319822791969,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008912617554858934,
+      "loss": 2.7232,
+      "step": 755
+    },
+    {
+      "epoch": 0.14252050146102366,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008910658307210031,
+      "loss": 2.6999,
+      "step": 756
+    },
+    {
+      "epoch": 0.1427090206428504,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008908699059561128,
+      "loss": 2.6267,
+      "step": 757
+    },
+    {
+      "epoch": 0.14289753982467715,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0008906739811912226,
+      "loss": 2.6947,
+      "step": 758
+    },
+    {
+      "epoch": 0.14308605900650392,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0008904780564263322,
+      "loss": 2.6448,
+      "step": 759
+    },
+    {
+      "epoch": 0.14327457818833067,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.000890282131661442,
+      "loss": 2.7433,
+      "step": 760
+    },
+    {
+      "epoch": 0.14346309737015742,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0008900862068965518,
+      "loss": 2.6372,
+      "step": 761
+    },
+    {
+      "epoch": 0.14365161655198416,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008898902821316614,
+      "loss": 2.7054,
+      "step": 762
+    },
+    {
+      "epoch": 0.1438401357338109,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008896943573667712,
+      "loss": 2.6433,
+      "step": 763
+    },
+    {
+      "epoch": 0.14402865491563765,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008894984326018809,
+      "loss": 2.6648,
+      "step": 764
+    },
+    {
+      "epoch": 0.14421717409746443,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0008893025078369906,
+      "loss": 2.6961,
+      "step": 765
+    },
+    {
+      "epoch": 0.14440569327929118,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0008891065830721003,
+      "loss": 2.5835,
+      "step": 766
+    },
+    {
+      "epoch": 0.14459421246111792,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008889106583072101,
+      "loss": 2.6922,
+      "step": 767
+    },
+    {
+      "epoch": 0.14478273164294467,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008887147335423197,
+      "loss": 2.5899,
+      "step": 768
+    },
+    {
+      "epoch": 0.1449712508247714,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0008885188087774295,
+      "loss": 2.6164,
+      "step": 769
+    },
+    {
+      "epoch": 0.14515977000659816,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008883228840125393,
+      "loss": 2.708,
+      "step": 770
+    },
+    {
+      "epoch": 0.14534828918842493,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000888126959247649,
+      "loss": 2.59,
+      "step": 771
+    },
+    {
+      "epoch": 0.14553680837025168,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008879310344827587,
+      "loss": 2.5905,
+      "step": 772
+    },
+    {
+      "epoch": 0.14572532755207843,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008877351097178683,
+      "loss": 2.6069,
+      "step": 773
+    },
+    {
+      "epoch": 0.14591384673390517,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0008875391849529781,
+      "loss": 2.695,
+      "step": 774
+    },
+    {
+      "epoch": 0.14610236591573192,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008873432601880877,
+      "loss": 2.7418,
+      "step": 775
+    },
+    {
+      "epoch": 0.14629088509755866,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0008871473354231975,
+      "loss": 2.6499,
+      "step": 776
+    },
+    {
+      "epoch": 0.14647940427938544,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008869514106583072,
+      "loss": 2.6959,
+      "step": 777
+    },
+    {
+      "epoch": 0.14666792346121219,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0008867554858934169,
+      "loss": 2.6255,
+      "step": 778
+    },
+    {
+      "epoch": 0.14685644264303893,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0008865595611285267,
+      "loss": 2.6885,
+      "step": 779
+    },
+    {
+      "epoch": 0.14704496182486568,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008863636363636364,
+      "loss": 2.6154,
+      "step": 780
+    },
+    {
+      "epoch": 0.14723348100669242,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0008861677115987461,
+      "loss": 2.6715,
+      "step": 781
+    },
+    {
+      "epoch": 0.14742200018851917,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008859717868338558,
+      "loss": 2.6356,
+      "step": 782
+    },
+    {
+      "epoch": 0.14761051937034594,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008857758620689656,
+      "loss": 2.6181,
+      "step": 783
+    },
+    {
+      "epoch": 0.1477990385521727,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0008855799373040752,
+      "loss": 2.6921,
+      "step": 784
+    },
+    {
+      "epoch": 0.14798755773399944,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.000885384012539185,
+      "loss": 2.5067,
+      "step": 785
+    },
+    {
+      "epoch": 0.14817607691582618,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0008851880877742947,
+      "loss": 2.729,
+      "step": 786
+    },
+    {
+      "epoch": 0.14836459609765293,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0008849921630094044,
+      "loss": 2.6272,
+      "step": 787
+    },
+    {
+      "epoch": 0.14855311527947967,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0008847962382445142,
+      "loss": 2.5603,
+      "step": 788
+    },
+    {
+      "epoch": 0.14874163446130645,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008846003134796239,
+      "loss": 2.7072,
+      "step": 789
+    },
+    {
+      "epoch": 0.1489301536431332,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008844043887147336,
+      "loss": 2.6431,
+      "step": 790
+    },
+    {
+      "epoch": 0.14911867282495994,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0008842084639498433,
+      "loss": 2.6058,
+      "step": 791
+    },
+    {
+      "epoch": 0.1493071920067867,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.000884012539184953,
+      "loss": 2.6369,
+      "step": 792
+    },
+    {
+      "epoch": 0.14949571118861343,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0008838166144200626,
+      "loss": 2.5803,
+      "step": 793
+    },
+    {
+      "epoch": 0.14968423037044018,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008836206896551724,
+      "loss": 2.6228,
+      "step": 794
+    },
+    {
+      "epoch": 0.14987274955226695,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0008834247648902821,
+      "loss": 2.5875,
+      "step": 795
+    },
+    {
+      "epoch": 0.1500612687340937,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0008832288401253918,
+      "loss": 2.6705,
+      "step": 796
+    },
+    {
+      "epoch": 0.15024978791592045,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0008830329153605015,
+      "loss": 2.6356,
+      "step": 797
+    },
+    {
+      "epoch": 0.1504383070977472,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008828369905956113,
+      "loss": 2.6354,
+      "step": 798
+    },
+    {
+      "epoch": 0.15062682627957394,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0008826410658307211,
+      "loss": 2.7054,
+      "step": 799
+    },
+    {
+      "epoch": 0.15081534546140068,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008824451410658307,
+      "loss": 2.5472,
+      "step": 800
+    },
+    {
+      "epoch": 0.15100386464322746,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0008822492163009405,
+      "loss": 2.5917,
+      "step": 801
+    },
+    {
+      "epoch": 0.1511923838250542,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008820532915360502,
+      "loss": 2.6797,
+      "step": 802
+    },
+    {
+      "epoch": 0.15138090300688095,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008818573667711599,
+      "loss": 2.6894,
+      "step": 803
+    },
+    {
+      "epoch": 0.1515694221887077,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0008816614420062696,
+      "loss": 2.6542,
+      "step": 804
+    },
+    {
+      "epoch": 0.15175794137053444,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008814655172413794,
+      "loss": 2.7165,
+      "step": 805
+    },
+    {
+      "epoch": 0.1519464605523612,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.000881269592476489,
+      "loss": 2.6396,
+      "step": 806
+    },
+    {
+      "epoch": 0.15213497973418796,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008810736677115988,
+      "loss": 2.5562,
+      "step": 807
+    },
+    {
+      "epoch": 0.1523234989160147,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008808777429467086,
+      "loss": 2.5097,
+      "step": 808
+    },
+    {
+      "epoch": 0.15251201809784146,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0008806818181818182,
+      "loss": 2.6022,
+      "step": 809
+    },
+    {
+      "epoch": 0.1527005372796682,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008804858934169279,
+      "loss": 2.6144,
+      "step": 810
+    },
+    {
+      "epoch": 0.15288905646149495,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008802899686520376,
+      "loss": 2.7956,
+      "step": 811
+    },
+    {
+      "epoch": 0.1530775756433217,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008800940438871473,
+      "loss": 2.6763,
+      "step": 812
+    },
+    {
+      "epoch": 0.15326609482514847,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.000879898119122257,
+      "loss": 2.7261,
+      "step": 813
+    },
+    {
+      "epoch": 0.15345461400697522,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0008797021943573668,
+      "loss": 2.5969,
+      "step": 814
+    },
+    {
+      "epoch": 0.15364313318880196,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008795062695924764,
+      "loss": 2.5595,
+      "step": 815
+    },
+    {
+      "epoch": 0.1538316523706287,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0008793103448275862,
+      "loss": 2.4956,
+      "step": 816
+    },
+    {
+      "epoch": 0.15402017155245545,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000879114420062696,
+      "loss": 2.6691,
+      "step": 817
+    },
+    {
+      "epoch": 0.1542086907342822,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0008789184952978056,
+      "loss": 2.5205,
+      "step": 818
+    },
+    {
+      "epoch": 0.15439720991610897,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0008787225705329154,
+      "loss": 2.5277,
+      "step": 819
+    },
+    {
+      "epoch": 0.15458572909793572,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0008785266457680251,
+      "loss": 2.6255,
+      "step": 820
+    },
+    {
+      "epoch": 0.15477424827976247,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008783307210031348,
+      "loss": 2.6624,
+      "step": 821
+    },
+    {
+      "epoch": 0.1549627674615892,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0008781347962382445,
+      "loss": 2.6192,
+      "step": 822
+    },
+    {
+      "epoch": 0.15515128664341596,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008779388714733543,
+      "loss": 2.6637,
+      "step": 823
+    },
+    {
+      "epoch": 0.1553398058252427,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.000877742946708464,
+      "loss": 2.6116,
+      "step": 824
+    },
+    {
+      "epoch": 0.15552832500706948,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008775470219435737,
+      "loss": 2.7295,
+      "step": 825
+    },
+    {
+      "epoch": 0.15571684418889623,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008773510971786835,
+      "loss": 2.6604,
+      "step": 826
+    },
+    {
+      "epoch": 0.15590536337072297,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0008771551724137932,
+      "loss": 2.661,
+      "step": 827
+    },
+    {
+      "epoch": 0.15609388255254972,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0008769592476489029,
+      "loss": 2.6784,
+      "step": 828
+    },
+    {
+      "epoch": 0.15628240173437646,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008767633228840125,
+      "loss": 2.6322,
+      "step": 829
+    },
+    {
+      "epoch": 0.1564709209162032,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008765673981191223,
+      "loss": 2.6958,
+      "step": 830
+    },
+    {
+      "epoch": 0.15665944009802998,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0008763714733542319,
+      "loss": 2.614,
+      "step": 831
+    },
+    {
+      "epoch": 0.15684795927985673,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008761755485893417,
+      "loss": 2.61,
+      "step": 832
+    },
+    {
+      "epoch": 0.15703647846168348,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0008759796238244514,
+      "loss": 2.6638,
+      "step": 833
+    },
+    {
+      "epoch": 0.15722499764351022,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0008757836990595611,
+      "loss": 2.5964,
+      "step": 834
+    },
+    {
+      "epoch": 0.15741351682533697,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008755877742946709,
+      "loss": 2.6957,
+      "step": 835
+    },
+    {
+      "epoch": 0.15760203600716374,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0008753918495297806,
+      "loss": 2.5888,
+      "step": 836
+    },
+    {
+      "epoch": 0.1577905551889905,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008751959247648903,
+      "loss": 2.644,
+      "step": 837
+    },
+    {
+      "epoch": 0.15797907437081724,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.000875,
+      "loss": 2.772,
+      "step": 838
+    },
+    {
+      "epoch": 0.15816759355264398,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0008748040752351098,
+      "loss": 2.6483,
+      "step": 839
+    },
+    {
+      "epoch": 0.15835611273447073,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008746081504702194,
+      "loss": 2.5755,
+      "step": 840
+    },
+    {
+      "epoch": 0.15854463191629747,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0008744122257053292,
+      "loss": 2.6584,
+      "step": 841
+    },
+    {
+      "epoch": 0.15873315109812425,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0008742163009404389,
+      "loss": 2.6448,
+      "step": 842
+    },
+    {
+      "epoch": 0.158921670279951,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008740203761755486,
+      "loss": 2.6187,
+      "step": 843
+    },
+    {
+      "epoch": 0.15911018946177774,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008738244514106584,
+      "loss": 2.7043,
+      "step": 844
+    },
+    {
+      "epoch": 0.1592987086436045,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0008736285266457681,
+      "loss": 2.5718,
+      "step": 845
+    },
+    {
+      "epoch": 0.15948722782543123,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008734326018808778,
+      "loss": 2.6218,
+      "step": 846
+    },
+    {
+      "epoch": 0.15967574700725798,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0008732366771159875,
+      "loss": 2.7162,
+      "step": 847
+    },
+    {
+      "epoch": 0.15986426618908475,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008730407523510972,
+      "loss": 2.7002,
+      "step": 848
+    },
+    {
+      "epoch": 0.1600527853709115,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0008728448275862068,
+      "loss": 2.5591,
+      "step": 849
+    },
+    {
+      "epoch": 0.16024130455273825,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0008726489028213166,
+      "loss": 2.68,
+      "step": 850
+    },
+    {
+      "epoch": 0.160429823734565,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0008724529780564263,
+      "loss": 2.6753,
+      "step": 851
+    },
+    {
+      "epoch": 0.16061834291639174,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.000872257053291536,
+      "loss": 2.6676,
+      "step": 852
+    },
+    {
+      "epoch": 0.16080686209821848,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0008720611285266457,
+      "loss": 2.6622,
+      "step": 853
+    },
+    {
+      "epoch": 0.16099538128004526,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0008718652037617555,
+      "loss": 2.8021,
+      "step": 854
+    },
+    {
+      "epoch": 0.161183900461872,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0008716692789968653,
+      "loss": 2.6823,
+      "step": 855
+    },
+    {
+      "epoch": 0.16137241964369875,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0008714733542319749,
+      "loss": 2.6915,
+      "step": 856
+    },
+    {
+      "epoch": 0.1615609388255255,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0008712774294670847,
+      "loss": 2.6373,
+      "step": 857
+    },
+    {
+      "epoch": 0.16174945800735224,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008710815047021944,
+      "loss": 2.6126,
+      "step": 858
+    },
+    {
+      "epoch": 0.161937977189179,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008708855799373041,
+      "loss": 2.544,
+      "step": 859
+    },
+    {
+      "epoch": 0.16212649637100576,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0008706896551724138,
+      "loss": 2.5462,
+      "step": 860
+    },
+    {
+      "epoch": 0.1623150155528325,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008704937304075236,
+      "loss": 2.6703,
+      "step": 861
+    },
+    {
+      "epoch": 0.16250353473465926,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0008702978056426332,
+      "loss": 2.606,
+      "step": 862
+    },
+    {
+      "epoch": 0.162692053916486,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.000870101880877743,
+      "loss": 2.5978,
+      "step": 863
+    },
+    {
+      "epoch": 0.16288057309831275,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0008699059561128528,
+      "loss": 2.6027,
+      "step": 864
+    },
+    {
+      "epoch": 0.1630690922801395,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0008697100313479624,
+      "loss": 2.654,
+      "step": 865
+    },
+    {
+      "epoch": 0.16325761146196627,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0008695141065830722,
+      "loss": 2.4922,
+      "step": 866
+    },
+    {
+      "epoch": 0.16344613064379301,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0008693181818181818,
+      "loss": 2.54,
+      "step": 867
+    },
+    {
+      "epoch": 0.16363464982561976,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008691222570532915,
+      "loss": 2.6802,
+      "step": 868
+    },
+    {
+      "epoch": 0.1638231690074465,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0008689263322884012,
+      "loss": 2.6451,
+      "step": 869
+    },
+    {
+      "epoch": 0.16401168818927325,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.000868730407523511,
+      "loss": 2.6301,
+      "step": 870
+    },
+    {
+      "epoch": 0.1642002073711,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008685344827586206,
+      "loss": 2.6155,
+      "step": 871
+    },
+    {
+      "epoch": 0.16438872655292677,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0008683385579937304,
+      "loss": 2.5815,
+      "step": 872
+    },
+    {
+      "epoch": 0.16457724573475352,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008681426332288402,
+      "loss": 2.6225,
+      "step": 873
+    },
+    {
+      "epoch": 0.16476576491658027,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008679467084639498,
+      "loss": 2.6562,
+      "step": 874
+    },
+    {
+      "epoch": 0.164954284098407,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0008677507836990596,
+      "loss": 2.6155,
+      "step": 875
+    },
+    {
+      "epoch": 0.16514280328023376,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008675548589341693,
+      "loss": 2.6055,
+      "step": 876
+    },
+    {
+      "epoch": 0.1653313224620605,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.000867358934169279,
+      "loss": 2.637,
+      "step": 877
+    },
+    {
+      "epoch": 0.16551984164388728,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008671630094043887,
+      "loss": 2.6913,
+      "step": 878
+    },
+    {
+      "epoch": 0.16570836082571402,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0008669670846394985,
+      "loss": 2.5859,
+      "step": 879
+    },
+    {
+      "epoch": 0.16589688000754077,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008667711598746082,
+      "loss": 2.7039,
+      "step": 880
+    },
+    {
+      "epoch": 0.16608539918936752,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008665752351097179,
+      "loss": 2.662,
+      "step": 881
+    },
+    {
+      "epoch": 0.16627391837119426,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008663793103448277,
+      "loss": 2.6888,
+      "step": 882
+    },
+    {
+      "epoch": 0.166462437553021,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0008661833855799374,
+      "loss": 2.6346,
+      "step": 883
+    },
+    {
+      "epoch": 0.16665095673484778,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008659874608150471,
+      "loss": 2.6536,
+      "step": 884
+    },
+    {
+      "epoch": 0.16683947591667453,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008657915360501567,
+      "loss": 2.5995,
+      "step": 885
+    },
+    {
+      "epoch": 0.16702799509850128,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0008655956112852665,
+      "loss": 2.7111,
+      "step": 886
+    },
+    {
+      "epoch": 0.16721651428032802,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0008653996865203761,
+      "loss": 2.6629,
+      "step": 887
+    },
+    {
+      "epoch": 0.16740503346215477,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008652037617554859,
+      "loss": 2.5553,
+      "step": 888
+    },
+    {
+      "epoch": 0.1675935526439815,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0008650078369905956,
+      "loss": 2.7594,
+      "step": 889
+    },
+    {
+      "epoch": 0.1677820718258083,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008648119122257053,
+      "loss": 2.6729,
+      "step": 890
+    },
+    {
+      "epoch": 0.16797059100763503,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.000864615987460815,
+      "loss": 2.7915,
+      "step": 891
+    },
+    {
+      "epoch": 0.16815911018946178,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008644200626959248,
+      "loss": 2.5758,
+      "step": 892
+    },
+    {
+      "epoch": 0.16834762937128853,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0008642241379310345,
+      "loss": 2.5708,
+      "step": 893
+    },
+    {
+      "epoch": 0.16853614855311527,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008640282131661442,
+      "loss": 2.6254,
+      "step": 894
+    },
+    {
+      "epoch": 0.16872466773494202,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.000863832288401254,
+      "loss": 2.5748,
+      "step": 895
+    },
+    {
+      "epoch": 0.1689131869167688,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0008636363636363636,
+      "loss": 2.7726,
+      "step": 896
+    },
+    {
+      "epoch": 0.16910170609859554,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008634404388714734,
+      "loss": 2.6658,
+      "step": 897
+    },
+    {
+      "epoch": 0.16929022528042229,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0008632445141065831,
+      "loss": 2.6508,
+      "step": 898
+    },
+    {
+      "epoch": 0.16947874446224903,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0008630485893416928,
+      "loss": 2.6266,
+      "step": 899
+    },
+    {
+      "epoch": 0.16966726364407578,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0008628526645768026,
+      "loss": 2.7459,
+      "step": 900
+    },
+    {
+      "epoch": 0.16985578282590252,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008626567398119123,
+      "loss": 2.587,
+      "step": 901
+    },
+    {
+      "epoch": 0.1700443020077293,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.000862460815047022,
+      "loss": 2.6079,
+      "step": 902
+    },
+    {
+      "epoch": 0.17023282118955604,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0008622648902821317,
+      "loss": 2.7354,
+      "step": 903
+    },
+    {
+      "epoch": 0.1704213403713828,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0008620689655172414,
+      "loss": 2.5455,
+      "step": 904
+    },
+    {
+      "epoch": 0.17060985955320954,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.000861873040752351,
+      "loss": 2.6302,
+      "step": 905
+    },
+    {
+      "epoch": 0.17079837873503628,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0008616771159874608,
+      "loss": 2.4757,
+      "step": 906
+    },
+    {
+      "epoch": 0.17098689791686303,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0008614811912225705,
+      "loss": 2.6884,
+      "step": 907
+    },
+    {
+      "epoch": 0.1711754170986898,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008612852664576803,
+      "loss": 2.5877,
+      "step": 908
+    },
+    {
+      "epoch": 0.17136393628051655,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0008610893416927899,
+      "loss": 2.6323,
+      "step": 909
+    },
+    {
+      "epoch": 0.1715524554623433,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0008608934169278997,
+      "loss": 2.657,
+      "step": 910
+    },
+    {
+      "epoch": 0.17174097464417004,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008606974921630095,
+      "loss": 2.6591,
+      "step": 911
+    },
+    {
+      "epoch": 0.1719294938259968,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008605015673981191,
+      "loss": 2.6385,
+      "step": 912
+    },
+    {
+      "epoch": 0.17211801300782353,
+      "grad_norm": 1.859375,
+      "learning_rate": 0.0008603056426332289,
+      "loss": 2.7853,
+      "step": 913
+    },
+    {
+      "epoch": 0.1723065321896503,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0008601097178683386,
+      "loss": 2.7299,
+      "step": 914
+    },
+    {
+      "epoch": 0.17249505137147705,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008599137931034483,
+      "loss": 2.7032,
+      "step": 915
+    },
+    {
+      "epoch": 0.1726835705533038,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.000859717868338558,
+      "loss": 2.6568,
+      "step": 916
+    },
+    {
+      "epoch": 0.17287208973513055,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008595219435736678,
+      "loss": 2.8574,
+      "step": 917
+    },
+    {
+      "epoch": 0.1730606089169573,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0008593260188087774,
+      "loss": 2.7283,
+      "step": 918
+    },
+    {
+      "epoch": 0.17324912809878404,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0008591300940438872,
+      "loss": 2.712,
+      "step": 919
+    },
+    {
+      "epoch": 0.1734376472806108,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.000858934169278997,
+      "loss": 2.5807,
+      "step": 920
+    },
+    {
+      "epoch": 0.17362616646243756,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008587382445141066,
+      "loss": 2.6998,
+      "step": 921
+    },
+    {
+      "epoch": 0.1738146856442643,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0008585423197492164,
+      "loss": 2.5734,
+      "step": 922
+    },
+    {
+      "epoch": 0.17400320482609105,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.000858346394984326,
+      "loss": 2.6374,
+      "step": 923
+    },
+    {
+      "epoch": 0.1741917240079178,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0008581504702194357,
+      "loss": 2.5408,
+      "step": 924
+    },
+    {
+      "epoch": 0.17438024318974454,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0008579545454545454,
+      "loss": 2.661,
+      "step": 925
+    },
+    {
+      "epoch": 0.17456876237157132,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008577586206896552,
+      "loss": 2.701,
+      "step": 926
+    },
+    {
+      "epoch": 0.17475728155339806,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008575626959247648,
+      "loss": 2.71,
+      "step": 927
+    },
+    {
+      "epoch": 0.1749458007352248,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008573667711598746,
+      "loss": 2.6468,
+      "step": 928
+    },
+    {
+      "epoch": 0.17513431991705156,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0008571708463949844,
+      "loss": 2.5039,
+      "step": 929
+    },
+    {
+      "epoch": 0.1753228390988783,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.000856974921630094,
+      "loss": 2.6299,
+      "step": 930
+    },
+    {
+      "epoch": 0.17551135828070505,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0008567789968652038,
+      "loss": 2.7097,
+      "step": 931
+    },
+    {
+      "epoch": 0.17569987746253182,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0008565830721003135,
+      "loss": 2.662,
+      "step": 932
+    },
+    {
+      "epoch": 0.17588839664435857,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0008563871473354232,
+      "loss": 2.6554,
+      "step": 933
+    },
+    {
+      "epoch": 0.17607691582618532,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008561912225705329,
+      "loss": 2.5157,
+      "step": 934
+    },
+    {
+      "epoch": 0.17626543500801206,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0008559952978056427,
+      "loss": 2.658,
+      "step": 935
+    },
+    {
+      "epoch": 0.1764539541898388,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0008557993730407524,
+      "loss": 2.5697,
+      "step": 936
+    },
+    {
+      "epoch": 0.17664247337166555,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008556034482758621,
+      "loss": 2.6467,
+      "step": 937
+    },
+    {
+      "epoch": 0.17683099255349233,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008554075235109719,
+      "loss": 2.7308,
+      "step": 938
+    },
+    {
+      "epoch": 0.17701951173531907,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0008552115987460816,
+      "loss": 2.5552,
+      "step": 939
+    },
+    {
+      "epoch": 0.17720803091714582,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0008550156739811913,
+      "loss": 2.7391,
+      "step": 940
+    },
+    {
+      "epoch": 0.17739655009897257,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.000854819749216301,
+      "loss": 2.5914,
+      "step": 941
+    },
+    {
+      "epoch": 0.1775850692807993,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0008546238244514107,
+      "loss": 2.6652,
+      "step": 942
+    },
+    {
+      "epoch": 0.17777358846262606,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008544278996865203,
+      "loss": 2.7119,
+      "step": 943
+    },
+    {
+      "epoch": 0.17796210764445283,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008542319749216301,
+      "loss": 2.7099,
+      "step": 944
+    },
+    {
+      "epoch": 0.17815062682627958,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0008540360501567398,
+      "loss": 2.6452,
+      "step": 945
+    },
+    {
+      "epoch": 0.17833914600810633,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0008538401253918495,
+      "loss": 2.5148,
+      "step": 946
+    },
+    {
+      "epoch": 0.17852766518993307,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008536442006269592,
+      "loss": 2.6379,
+      "step": 947
+    },
+    {
+      "epoch": 0.17871618437175982,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.000853448275862069,
+      "loss": 2.6155,
+      "step": 948
+    },
+    {
+      "epoch": 0.17890470355358656,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008532523510971787,
+      "loss": 2.5122,
+      "step": 949
+    },
+    {
+      "epoch": 0.17909322273541334,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008530564263322884,
+      "loss": 2.7232,
+      "step": 950
+    },
+    {
+      "epoch": 0.17928174191724008,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0008528605015673982,
+      "loss": 2.7196,
+      "step": 951
+    },
+    {
+      "epoch": 0.17947026109906683,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0008526645768025078,
+      "loss": 2.5332,
+      "step": 952
+    },
+    {
+      "epoch": 0.17965878028089358,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0008524686520376176,
+      "loss": 2.7149,
+      "step": 953
+    },
+    {
+      "epoch": 0.17984729946272032,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008522727272727273,
+      "loss": 2.7601,
+      "step": 954
+    },
+    {
+      "epoch": 0.1800358186445471,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.000852076802507837,
+      "loss": 2.6729,
+      "step": 955
+    },
+    {
+      "epoch": 0.18022433782637384,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008518808777429467,
+      "loss": 2.5701,
+      "step": 956
+    },
+    {
+      "epoch": 0.1804128570082006,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0008516849529780565,
+      "loss": 2.6001,
+      "step": 957
+    },
+    {
+      "epoch": 0.18060137619002734,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0008514890282131662,
+      "loss": 2.7079,
+      "step": 958
+    },
+    {
+      "epoch": 0.18078989537185408,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0008512931034482759,
+      "loss": 2.6712,
+      "step": 959
+    },
+    {
+      "epoch": 0.18097841455368083,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0008510971786833856,
+      "loss": 2.722,
+      "step": 960
+    },
+    {
+      "epoch": 0.1811669337355076,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0008509012539184952,
+      "loss": 2.5951,
+      "step": 961
+    },
+    {
+      "epoch": 0.18135545291733435,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.000850705329153605,
+      "loss": 2.6927,
+      "step": 962
+    },
+    {
+      "epoch": 0.1815439720991611,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008505094043887147,
+      "loss": 2.6872,
+      "step": 963
+    },
+    {
+      "epoch": 0.18173249128098784,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0008503134796238245,
+      "loss": 2.498,
+      "step": 964
+    },
+    {
+      "epoch": 0.1819210104628146,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008501175548589341,
+      "loss": 2.7254,
+      "step": 965
+    },
+    {
+      "epoch": 0.18210952964464133,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008499216300940439,
+      "loss": 2.6044,
+      "step": 966
+    },
+    {
+      "epoch": 0.1822980488264681,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0008497257053291537,
+      "loss": 2.6202,
+      "step": 967
+    },
+    {
+      "epoch": 0.18248656800829485,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0008495297805642633,
+      "loss": 2.6671,
+      "step": 968
+    },
+    {
+      "epoch": 0.1826750871901216,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0008493338557993731,
+      "loss": 2.7925,
+      "step": 969
+    },
+    {
+      "epoch": 0.18286360637194835,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0008491379310344828,
+      "loss": 2.5893,
+      "step": 970
+    },
+    {
+      "epoch": 0.1830521255537751,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0008489420062695925,
+      "loss": 2.6704,
+      "step": 971
+    },
+    {
+      "epoch": 0.18324064473560184,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008487460815047022,
+      "loss": 2.6642,
+      "step": 972
+    },
+    {
+      "epoch": 0.1834291639174286,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.000848550156739812,
+      "loss": 2.6579,
+      "step": 973
+    },
+    {
+      "epoch": 0.18361768309925536,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0008483542319749216,
+      "loss": 2.47,
+      "step": 974
+    },
+    {
+      "epoch": 0.1838062022810821,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0008481583072100314,
+      "loss": 2.5976,
+      "step": 975
+    },
+    {
+      "epoch": 0.18399472146290885,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0008479623824451412,
+      "loss": 2.7708,
+      "step": 976
+    },
+    {
+      "epoch": 0.1841832406447356,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008477664576802508,
+      "loss": 2.6823,
+      "step": 977
+    },
+    {
+      "epoch": 0.18437175982656234,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0008475705329153606,
+      "loss": 2.6927,
+      "step": 978
+    },
+    {
+      "epoch": 0.18456027900838912,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0008473746081504702,
+      "loss": 2.682,
+      "step": 979
+    },
+    {
+      "epoch": 0.18474879819021586,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008471786833855799,
+      "loss": 2.6509,
+      "step": 980
+    },
+    {
+      "epoch": 0.1849373173720426,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008469827586206896,
+      "loss": 2.7625,
+      "step": 981
+    },
+    {
+      "epoch": 0.18512583655386936,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0008467868338557994,
+      "loss": 2.7144,
+      "step": 982
+    },
+    {
+      "epoch": 0.1853143557356961,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000846590909090909,
+      "loss": 2.7065,
+      "step": 983
+    },
+    {
+      "epoch": 0.18550287491752285,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0008463949843260188,
+      "loss": 2.5608,
+      "step": 984
+    },
+    {
+      "epoch": 0.18569139409934962,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0008461990595611286,
+      "loss": 2.587,
+      "step": 985
+    },
+    {
+      "epoch": 0.18587991328117637,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008460031347962382,
+      "loss": 2.5719,
+      "step": 986
+    },
+    {
+      "epoch": 0.18606843246300311,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.000845807210031348,
+      "loss": 2.6672,
+      "step": 987
+    },
+    {
+      "epoch": 0.18625695164482986,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0008456112852664577,
+      "loss": 2.6857,
+      "step": 988
+    },
+    {
+      "epoch": 0.1864454708266566,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0008454153605015674,
+      "loss": 2.6708,
+      "step": 989
+    },
+    {
+      "epoch": 0.18663399000848335,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0008452194357366771,
+      "loss": 2.6183,
+      "step": 990
+    },
+    {
+      "epoch": 0.18682250919031013,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008450235109717869,
+      "loss": 2.6452,
+      "step": 991
+    },
+    {
+      "epoch": 0.18701102837213687,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0008448275862068966,
+      "loss": 2.704,
+      "step": 992
+    },
+    {
+      "epoch": 0.18719954755396362,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008446316614420063,
+      "loss": 2.674,
+      "step": 993
+    },
+    {
+      "epoch": 0.18738806673579037,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.000844435736677116,
+      "loss": 2.7137,
+      "step": 994
+    },
+    {
+      "epoch": 0.1875765859176171,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008442398119122258,
+      "loss": 2.7582,
+      "step": 995
+    },
+    {
+      "epoch": 0.18776510509944386,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0008440438871473355,
+      "loss": 2.6009,
+      "step": 996
+    },
+    {
+      "epoch": 0.18795362428127063,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0008438479623824452,
+      "loss": 2.6595,
+      "step": 997
+    },
+    {
+      "epoch": 0.18814214346309738,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0008436520376175549,
+      "loss": 2.552,
+      "step": 998
+    },
+    {
+      "epoch": 0.18833066264492412,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0008434561128526645,
+      "loss": 2.5806,
+      "step": 999
+    },
+    {
+      "epoch": 0.18851918182675087,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0008432601880877743,
+      "loss": 2.6469,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18851918182675087,
+      "eval_runtime": 16.219,
+      "eval_samples_per_second": 63.136,
+      "eval_steps_per_second": 1.973,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18851918182675087,
+      "eval/hellaswag_acc": 0.3743278231428002,
+      "eval/hellaswag_acc_norm": 0.4706233817964549,
+      "eval_hellaswag_elapsed_time": 116.27660393714905,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 5304,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.325965577388032e+18,
+  "train_batch_size": 12,
+  "trial_name": null,
+  "trial_params": null
+}