diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7410 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998983843105376,
+  "eval_steps": 500,
+  "global_step": 1230,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 1.368,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 1.2883,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 1.3309,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 1.3241,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 1.2976,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 1.2878,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 1.2736,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 1.3015,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 1.3109,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 1.2547,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 1.2891,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 1.224,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 1.2727,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 1.3202,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 1.2584,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 1.2318,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 1.271,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 1.2171,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 1.2778,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 1.2577,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 1.2742,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 1.1896,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 1.2352,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 1.2495,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 1.1847,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 1.2438,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 1.2831,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 1.2669,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 1.2038,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 1.2608,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 1.2226,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 1.2011,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 1.2631,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 1.1941,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 1.259,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 1.2768,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 1.2244,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999653272242e-05,
+      "loss": 1.2453,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999861309137232e-05,
+      "loss": 1.2447,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999968794646042e-05,
+      "loss": 1.2684,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999445240395953e-05,
+      "loss": 1.2455,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999133192626893e-05,
+      "loss": 1.2277,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998751805317152e-05,
+      "loss": 1.233,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999830108111148e-05,
+      "loss": 1.2259,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999778102313545e-05,
+      "loss": 1.2735,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999719163499543e-05,
+      "loss": 1.2376,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999653292077857e-05,
+      "loss": 1.2808,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999580488505276e-05,
+      "loss": 1.2816,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995007532866594e-05,
+      "loss": 1.2007,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994140869749366e-05,
+      "loss": 1.2145,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993204901710995e-05,
+      "loss": 1.2874,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992199635241997e-05,
+      "loss": 1.2482,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999112507731346e-05,
+      "loss": 1.2084,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989981235376956e-05,
+      "loss": 1.2667,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988768117364526e-05,
+      "loss": 1.2539,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987485731688595e-05,
+      "loss": 1.218,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998613408724195e-05,
+      "loss": 1.2402,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998471319339763e-05,
+      "loss": 1.2641,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983223060008908e-05,
+      "loss": 1.2634,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981663697409203e-05,
+      "loss": 1.2185,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998003511641199e-05,
+      "loss": 1.2257,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997833732831076e-05,
+      "loss": 1.2937,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976570344878916e-05,
+      "loss": 1.2116,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974734178369702e-05,
+      "loss": 1.233,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997282884151612e-05,
+      "loss": 1.2553,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970854347530828e-05,
+      "loss": 1.2508,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968810710106065e-05,
+      "loss": 1.2102,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966697943413548e-05,
+      "loss": 1.2336,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964516062104377e-05,
+      "loss": 1.2492,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996226508130892e-05,
+      "loss": 1.2363,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995994501663674e-05,
+      "loss": 1.2638,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995755588417644e-05,
+      "loss": 1.1536,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99550977004956e-05,
+      "loss": 1.2669,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952570482640628e-05,
+      "loss": 1.22,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949974248136655e-05,
+      "loss": 1.1928,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947309014987414e-05,
+      "loss": 1.2036,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944574801675106e-05,
+      "loss": 1.2803,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941771627160287e-05,
+      "loss": 1.2578,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938899510881732e-05,
+      "loss": 1.2423,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935958472756283e-05,
+      "loss": 1.241,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993294853317873e-05,
+      "loss": 1.2408,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929869713021668e-05,
+      "loss": 1.2506,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926722033635343e-05,
+      "loss": 1.2556,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923505516847514e-05,
+      "loss": 1.2058,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920220184963296e-05,
+      "loss": 1.1737,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916866060764994e-05,
+      "loss": 1.2145,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991344316751198e-05,
+      "loss": 1.2288,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909951528940485e-05,
+      "loss": 1.2863,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990639116926348e-05,
+      "loss": 1.1998,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902762113170467e-05,
+      "loss": 1.2705,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989906438582734e-05,
+      "loss": 1.2608,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895298012876192e-05,
+      "loss": 1.2428,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891463020435144e-05,
+      "loss": 1.2147,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887559435098162e-05,
+      "loss": 1.1704,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883587283934875e-05,
+      "loss": 1.2352,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879546594490383e-05,
+      "loss": 1.2747,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987543739478507e-05,
+      "loss": 1.2214,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987125971331441e-05,
+      "loss": 1.1768,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867013579048765e-05,
+      "loss": 1.2477,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862699021433186e-05,
+      "loss": 1.2497,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858316070387208e-05,
+      "loss": 1.2369,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853864756304654e-05,
+      "loss": 1.213,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849345110053405e-05,
+      "loss": 1.2646,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984475716297519e-05,
+      "loss": 1.2429,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984010094688539e-05,
+      "loss": 1.2373,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835376494072788e-05,
+      "loss": 1.2087,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830583837299363e-05,
+      "loss": 1.2192,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825723009800058e-05,
+      "loss": 1.2755,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820794045282553e-05,
+      "loss": 1.213,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815796977927015e-05,
+      "loss": 1.2135,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810731842385892e-05,
+      "loss": 1.1937,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805598673783644e-05,
+      "loss": 1.221,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980039750771651e-05,
+      "loss": 1.2214,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795128380252263e-05,
+      "loss": 1.2292,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978979132792996e-05,
+      "loss": 1.2203,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784386387759684e-05,
+      "loss": 1.271,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977891359722229e-05,
+      "loss": 1.2343,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773372994269147e-05,
+      "loss": 1.2169,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976776461732187e-05,
+      "loss": 1.2817,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976208850527206e-05,
+      "loss": 1.2446,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756344697481027e-05,
+      "loss": 1.1881,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975053323377952e-05,
+      "loss": 1.1776,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744654154467468e-05,
+      "loss": 1.1861,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738707500313655e-05,
+      "loss": 1.2239,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732693312555492e-05,
+      "loss": 1.2387,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726611632898693e-05,
+      "loss": 1.272,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720462503517e-05,
+      "loss": 1.243,
+      "step": 127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971424596705189e-05,
+      "loss": 1.2171,
+      "step": 128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707962066612278e-05,
+      "loss": 1.2162,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970161084577422e-05,
+      "loss": 1.2394,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695192348580606e-05,
+      "loss": 1.2199,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688706619540863e-05,
+      "loss": 1.1925,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968215370363063e-05,
+      "loss": 1.2006,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675533646291463e-05,
+      "loss": 1.2347,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668846493430522e-05,
+      "loss": 1.2254,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662092291420233e-05,
+      "loss": 1.2083,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965527108709798e-05,
+      "loss": 1.1978,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964838292776579e-05,
+      "loss": 1.1688,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641427861189973e-05,
+      "loss": 1.2215,
+      "step": 139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963440593560083e-05,
+      "loss": 1.2278,
+      "step": 140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627317199692287e-05,
+      "loss": 1.2405,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962016170262157e-05,
+      "loss": 1.2181,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961293949400888e-05,
+      "loss": 1.2253,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960565062393701e-05,
+      "loss": 1.242,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598295142951035e-05,
+      "loss": 1.1836,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590873102057948e-05,
+      "loss": 1.2616,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583384552726294e-05,
+      "loss": 1.1986,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957582954688584e-05,
+      "loss": 1.2223,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568208136927177e-05,
+      "loss": 1.188,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560520375701408e-05,
+      "loss": 1.2564,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552766316519726e-05,
+      "loss": 1.2322,
+      "step": 151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544946013153093e-05,
+      "loss": 1.2289,
+      "step": 152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537059519831822e-05,
+      "loss": 1.2271,
+      "step": 153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529106891245244e-05,
+      "loss": 1.2414,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521088182541298e-05,
+      "loss": 1.2172,
+      "step": 155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951300344932616e-05,
+      "loss": 1.2444,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504852747663862e-05,
+      "loss": 1.2653,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496636134075894e-05,
+      "loss": 1.2406,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488353665540813e-05,
+      "loss": 1.1813,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480005399493857e-05,
+      "loss": 1.3069,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471591393826536e-05,
+      "loss": 1.2525,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463111706886234e-05,
+      "loss": 1.2535,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454566397475813e-05,
+      "loss": 1.1938,
+      "step": 163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944595552485319e-05,
+      "loss": 1.1764,
+      "step": 164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943727914873094e-05,
+      "loss": 1.197,
+      "step": 165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428537329275862e-05,
+      "loss": 1.2366,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941973012710859e-05,
+      "loss": 1.2221,
+      "step": 167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941085760330316e-05,
+      "loss": 1.1582,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940191981938657e-05,
+      "loss": 1.1946,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392916837338376e-05,
+      "loss": 1.2467,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383848719590257e-05,
+      "loss": 1.2094,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374715529025575e-05,
+      "loss": 1.2033,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365517328978943e-05,
+      "loss": 1.2388,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356254183235785e-05,
+      "loss": 1.2654,
+      "step": 174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93469261560319e-05,
+      "loss": 1.1707,
+      "step": 175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337533312053002e-05,
+      "loss": 1.2144,
+      "step": 176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328075716434287e-05,
+      "loss": 1.173,
+      "step": 177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931855343475998e-05,
+      "loss": 1.23,
+      "step": 178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930896653306286e-05,
+      "loss": 1.2284,
+      "step": 179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929931507782383e-05,
+      "loss": 1.2648,
+      "step": 180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289599135971437e-05,
+      "loss": 1.2374,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279818774881418e-05,
+      "loss": 1.2341,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269974062376224e-05,
+      "loss": 1.2253,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926006506672456e-05,
+      "loss": 1.1798,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250091856640895e-05,
+      "loss": 1.2063,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240054501285015e-05,
+      "loss": 1.2206,
+      "step": 186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922995307026151e-05,
+      "loss": 1.2651,
+      "step": 187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921978763361931e-05,
+      "loss": 1.2442,
+      "step": 188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209558261851194e-05,
+      "loss": 1.181,
+      "step": 189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919926502589331e-05,
+      "loss": 1.2252,
+      "step": 190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188907997124666e-05,
+      "loss": 1.1908,
+      "step": 191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178487247366652e-05,
+      "loss": 1.1913,
+      "step": 192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916800284888253e-05,
+      "loss": 1.2298,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915745487437694e-05,
+      "loss": 1.2255,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146843396995396e-05,
+      "loss": 1.2144,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136168490323772e-05,
+      "loss": 1.2682,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125430228387794e-05,
+      "loss": 1.2443,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114628685652535e-05,
+      "loss": 1.2539,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103763937021887e-05,
+      "loss": 1.2556,
+      "step": 199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909283605783805e-05,
+      "loss": 1.2261,
+      "step": 200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081845123881002e-05,
+      "loss": 1.2595,
+      "step": 201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070791211367984e-05,
+      "loss": 1.195,
+      "step": 202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059674396952963e-05,
+      "loss": 1.2302,
+      "step": 203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90484947577261e-05,
+      "loss": 1.228,
+      "step": 204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903725237121322e-05,
+      "loss": 1.2262,
+      "step": 205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902594731537527e-05,
+      "loss": 1.2366,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901457966860779e-05,
+      "loss": 1.2131,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003149509740347e-05,
+      "loss": 1.2044,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899165691803601e-05,
+      "loss": 1.2142,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980101973190787e-05,
+      "loss": 1.2214,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896848475533309e-05,
+      "loss": 1.2449,
+      "step": 211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956805345023145e-05,
+      "loss": 1.1548,
+      "step": 212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894506382325248e-05,
+      "loss": 1.1714,
+      "step": 213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933260271443313e-05,
+      "loss": 1.1764,
+      "step": 214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921394771448032e-05,
+      "loss": 1.2932,
+      "step": 215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89094674055486e-05,
+      "loss": 1.2256,
+      "step": 216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889747825645599e-05,
+      "loss": 1.2403,
+      "step": 217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885427407309627e-05,
+      "loss": 1.2661,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887331494167678e-05,
+      "loss": 1.1906,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861140943552014e-05,
+      "loss": 1.2643,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884890549735659e-05,
+      "loss": 1.209,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836608687937883e-05,
+      "loss": 1.1859,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824250600568798e-05,
+      "loss": 1.1723,
+      "step": 223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811831320947174e-05,
+      "loss": 1.2263,
+      "step": 224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879935093519519e-05,
+      "loss": 1.2086,
+      "step": 225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786809529858766e-05,
+      "loss": 1.2279,
+      "step": 226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774207191906976e-05,
+      "loss": 1.1943,
+      "step": 227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761544008731426e-05,
+      "loss": 1.1616,
+      "step": 228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874882006814565e-05,
+      "loss": 1.1972,
+      "step": 229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736035458384528e-05,
+      "loss": 1.2034,
+      "step": 230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723190268103634e-05,
+      "loss": 1.2238,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710284586378645e-05,
+      "loss": 1.1813,
+      "step": 232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697318502704734e-05,
+      "loss": 1.2154,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684292106995916e-05,
+      "loss": 1.1565,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671205489584453e-05,
+      "loss": 1.1861,
+      "step": 235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865805874122021e-05,
+      "loss": 1.245,
+      "step": 236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644851953070045e-05,
+      "loss": 1.1938,
+      "step": 237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863158521671716e-05,
+      "loss": 1.1563,
+      "step": 238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618258624160465e-05,
+      "loss": 1.2807,
+      "step": 239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604872267813954e-05,
+      "loss": 1.2266,
+      "step": 240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859142624050605e-05,
+      "loss": 1.2749,
+      "step": 241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577920635478976e-05,
+      "loss": 1.2349,
+      "step": 242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564355546388094e-05,
+      "loss": 1.2175,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855073106730126e-05,
+      "loss": 1.2175,
+      "step": 244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537047292698175e-05,
+      "loss": 1.1908,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852330431746973e-05,
+      "loss": 1.2191,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509502236917353e-05,
+      "loss": 1.2083,
+      "step": 247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495641146752322e-05,
+      "loss": 1.2827,
+      "step": 248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848172114309513e-05,
+      "loss": 1.2277,
+      "step": 249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467742322474822e-05,
+      "loss": 1.2175,
+      "step": 250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845370478182829e-05,
+      "loss": 1.2498,
+      "step": 251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8439608618499637e-05,
+      "loss": 1.2211,
+      "step": 252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842545393023949e-05,
+      "loss": 1.2003,
+      "step": 253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841124081520431e-05,
+      "loss": 1.215,
+      "step": 254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396969371955724e-05,
+      "loss": 1.1724,
+      "step": 255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838263969945985e-05,
+      "loss": 1.2061,
+      "step": 256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836825189708659e-05,
+      "loss": 1.2767,
+      "step": 257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353806064608953e-05,
+      "loss": 1.2336,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833930230220236e-05,
+      "loss": 1.1731,
+      "step": 259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324740710443955e-05,
+      "loss": 1.2047,
+      "step": 260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831012139031189e-05,
+      "loss": 1.2381,
+      "step": 261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829544444318466e-05,
+      "loss": 1.1746,
+      "step": 262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280709970840352e-05,
+      "loss": 1.2142,
+      "step": 263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265918075455985e-05,
+      "loss": 1.2034,
+      "step": 264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251068859606777e-05,
+      "loss": 1.1801,
+      "step": 265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236162426265424e-05,
+      "loss": 1.2023,
+      "step": 266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221198878801415e-05,
+      "loss": 1.2535,
+      "step": 267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206178320980295e-05,
+      "loss": 1.2258,
+      "step": 268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819110085696295e-05,
+      "loss": 1.197,
+      "step": 269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817596659130489e-05,
+      "loss": 1.214,
+      "step": 270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816077562895551e-05,
+      "loss": 1.2129,
+      "step": 271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814552807525738e-05,
+      "loss": 1.1589,
+      "step": 272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81302240359455e-05,
+      "loss": 1.2022,
+      "step": 273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114863617146576e-05,
+      "loss": 1.2463,
+      "step": 274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099446925378278e-05,
+      "loss": 1.2163,
+      "step": 275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083974067548506e-05,
+      "loss": 1.2308,
+      "step": 276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806844515095465e-05,
+      "loss": 1.127,
+      "step": 277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052860283282832e-05,
+      "loss": 1.2639,
+      "step": 278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037219572607177e-05,
+      "loss": 1.2254,
+      "step": 279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021523127389066e-05,
+      "loss": 1.2645,
+      "step": 280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800577105647635e-05,
+      "loss": 1.1495,
+      "step": 281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989963469102643e-05,
+      "loss": 1.1708,
+      "step": 282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797410047488653e-05,
+      "loss": 1.2124,
+      "step": 283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958182183830816e-05,
+      "loss": 1.2583,
+      "step": 284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794220870632177e-05,
+      "loss": 1.2181,
+      "step": 285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926180153128358e-05,
+      "loss": 1.2052,
+      "step": 286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791009663540146e-05,
+      "loss": 1.2095,
+      "step": 287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893958264673117e-05,
+      "loss": 1.2197,
+      "step": 288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877765152855757e-05,
+      "loss": 1.2569,
+      "step": 289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78615174122414e-05,
+      "loss": 1.2133,
+      "step": 290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78452151555009e-05,
+      "loss": 1.1737,
+      "step": 291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828858495683162e-05,
+      "loss": 1.1945,
+      "step": 292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781244754621434e-05,
+      "loss": 1.1835,
+      "step": 293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779598242089707e-05,
+      "loss": 1.2762,
+      "step": 294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779463233909677e-05,
+      "loss": 1.182,
+      "step": 295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762890099805362e-05,
+      "loss": 1.1733,
+      "step": 296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774626313351145e-05,
+      "loss": 1.2231,
+      "step": 297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729582450328547e-05,
+      "loss": 1.1535,
+      "step": 298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712848165929776e-05,
+      "loss": 1.2462,
+      "step": 299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696060396359956e-05,
+      "loss": 1.2039,
+      "step": 300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679219258034798e-05,
+      "loss": 1.2023,
+      "step": 301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662324867740102e-05,
+      "loss": 1.2235,
+      "step": 302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645377342630956e-05,
+      "loss": 1.1846,
+      "step": 303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76283768002309e-05,
+      "loss": 1.2434,
+      "step": 304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611323358431145e-05,
+      "loss": 1.2344,
+      "step": 305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759421713548971e-05,
+      "loss": 1.2312,
+      "step": 306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577058250030655e-05,
+      "loss": 1.1927,
+      "step": 307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559846821043205e-05,
+      "loss": 1.2007,
+      "step": 308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754258296788097e-05,
+      "loss": 1.2099,
+      "step": 309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525266810261096e-05,
+      "loss": 1.2793,
+      "step": 310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507898468263422e-05,
+      "loss": 1.2342,
+      "step": 311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490478062329686e-05,
+      "loss": 1.2643,
+      "step": 312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473005713262644e-05,
+      "loss": 1.2237,
+      "step": 313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745548154222527e-05,
+      "loss": 1.2549,
+      "step": 314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437905670739893e-05,
+      "loss": 1.2176,
+      "step": 315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420278220687366e-05,
+      "loss": 1.171,
+      "step": 316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402599314306207e-05,
+      "loss": 1.2079,
+      "step": 317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384869074191777e-05,
+      "loss": 1.2201,
+      "step": 318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367087623295394e-05,
+      "loss": 1.2257,
+      "step": 319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7349255084923517e-05,
+      "loss": 1.1221,
+      "step": 320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7331371582736864e-05,
+      "loss": 1.1572,
+      "step": 321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731343724074957e-05,
+      "loss": 1.2267,
+      "step": 322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7295452183328317e-05,
+      "loss": 1.2196,
+      "step": 323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277416535191478e-05,
+      "loss": 1.2231,
+      "step": 324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259330421408247e-05,
+      "loss": 1.1771,
+      "step": 325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7241193967397784e-05,
+      "loss": 1.1737,
+      "step": 326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223007298928322e-05,
+      "loss": 1.2136,
+      "step": 327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7204770542116326e-05,
+      "loss": 1.2632,
+      "step": 328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186483823425582e-05,
+      "loss": 1.2171,
+      "step": 329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168147269666357e-05,
+      "loss": 1.2089,
+      "step": 330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714976100799449e-05,
+      "loss": 1.1739,
+      "step": 331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713132516591053e-05,
+      "loss": 1.228,
+      "step": 332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112839871258838e-05,
+      "loss": 1.1951,
+      "step": 333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094305252226713e-05,
+      "loss": 1.2141,
+      "step": 334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075721437343488e-05,
+      "loss": 1.2067,
+      "step": 335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705708855547966e-05,
+      "loss": 1.1405,
+      "step": 336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7038406735845967e-05,
+      "loss": 1.2045,
+      "step": 337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019676107992523e-05,
+      "loss": 1.1631,
+      "step": 338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.70008968018079e-05,
+      "loss": 1.213,
+      "step": 339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982068947518235e-05,
+      "loss": 1.1659,
+      "step": 340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963192675686312e-05,
+      "loss": 1.187,
+      "step": 341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694426811721069e-05,
+      "loss": 1.2258,
+      "step": 342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692529540332476e-05,
+      "loss": 1.2047,
+      "step": 343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690627466559585e-05,
+      "loss": 1.2238,
+      "step": 344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688720603592432e-05,
+      "loss": 1.196,
+      "step": 345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868089646542632e-05,
+      "loss": 1.2162,
+      "step": 346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6848925630014445e-05,
+      "loss": 1.2097,
+      "step": 347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6829714119233688e-05,
+      "loss": 1.1976,
+      "step": 348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810455247423634e-05,
+      "loss": 1.2536,
+      "step": 349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6791149148136003e-05,
+      "loss": 1.1825,
+      "step": 350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677179595525e-05,
+      "loss": 1.2136,
+      "step": 351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752395802971407e-05,
+      "loss": 1.2162,
+      "step": 352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732948825831657e-05,
+      "loss": 1.2377,
+      "step": 353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713455158686878e-05,
+      "loss": 1.1747,
+      "step": 354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693914936716983e-05,
+      "loss": 1.186,
+      "step": 355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674328295424723e-05,
+      "loss": 1.1659,
+      "step": 356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6654695370634738e-05,
+      "loss": 1.2005,
+      "step": 357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635016298492628e-05,
+      "loss": 1.1649,
+      "step": 358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6615291215464005e-05,
+      "loss": 1.157,
+      "step": 359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6595520258333545e-05,
+      "loss": 1.211,
+      "step": 360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657570356420404e-05,
+      "loss": 1.186,
+      "step": 361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555841270495456e-05,
+      "loss": 1.1863,
+      "step": 362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6535933514943955e-05,
+      "loss": 1.1728,
+      "step": 363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515980435600965e-05,
+      "loss": 1.1642,
+      "step": 364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495982170832224e-05,
+      "loss": 1.1872,
+      "step": 365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475938859316795e-05,
+      "loss": 1.1606,
+      "step": 366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6455850640046134e-05,
+      "loss": 1.211,
+      "step": 367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435717652323097e-05,
+      "loss": 1.1654,
+      "step": 368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415540035761008e-05,
+      "loss": 1.2022,
+      "step": 369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639531793028265e-05,
+      "loss": 1.1947,
+      "step": 370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637505147611934e-05,
+      "loss": 1.1684,
+      "step": 371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6354740813809917e-05,
+      "loss": 1.1719,
+      "step": 372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6334386084199787e-05,
+      "loss": 1.1496,
+      "step": 373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631398742843995e-05,
+      "loss": 1.2575,
+      "step": 374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629354498798601e-05,
+      "loss": 1.1629,
+      "step": 375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627305890459719e-05,
+      "loss": 1.1316,
+      "step": 376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625252932033538e-05,
+      "loss": 1.1437,
+      "step": 377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6231956377564095e-05,
+      "loss": 1.2018,
+      "step": 378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621134021894756e-05,
+      "loss": 1.2288,
+      "step": 379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619068098744965e-05,
+      "loss": 1.1403,
+      "step": 380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169978826332955e-05,
+      "loss": 1.1945,
+      "step": 381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6149233879157747e-05,
+      "loss": 1.1712,
+      "step": 382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128446289781012e-05,
+      "loss": 1.1651,
+      "step": 383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610761620235543e-05,
+      "loss": 1.2365,
+      "step": 384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60867437613284e-05,
+      "loss": 1.2245,
+      "step": 385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6065829111441e-05,
+      "loss": 1.1262,
+      "step": 386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6044872397727037e-05,
+      "loss": 1.1627,
+      "step": 387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6023873765511993e-05,
+      "loss": 1.2391,
+      "step": 388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002833360412044e-05,
+      "loss": 1.2052,
+      "step": 389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5981751328333036e-05,
+      "loss": 1.2199,
+      "step": 390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960627815469482e-05,
+      "loss": 1.1462,
+      "step": 391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939462968303554e-05,
+      "loss": 1.1428,
+      "step": 392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5918256933604047e-05,
+      "loss": 1.1804,
+      "step": 393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897009858425383e-05,
+      "loss": 1.1685,
+      "step": 394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5875721890106574e-05,
+      "loss": 1.1846,
+      "step": 395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854393176270205e-05,
+      "loss": 1.2156,
+      "step": 396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833023864821427e-05,
+      "loss": 1.216,
+      "step": 397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811614103946905e-05,
+      "loss": 1.2288,
+      "step": 398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5790164042113805e-05,
+      "loss": 1.17,
+      "step": 399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576867382806877e-05,
+      "loss": 1.2223,
+      "step": 400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747143610836873e-05,
+      "loss": 1.1373,
+      "step": 401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572557353972059e-05,
+      "loss": 1.2559,
+      "step": 402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570396376429877e-05,
+      "loss": 1.2033,
+      "step": 403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682314434425593e-05,
+      "loss": 1.2259,
+      "step": 404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5660625700229526e-05,
+      "loss": 1.1371,
+      "step": 405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5638897712112303e-05,
+      "loss": 1.1417,
+      "step": 406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561713062074785e-05,
+      "loss": 1.2067,
+      "step": 407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5595324577081262e-05,
+      "loss": 1.2265,
+      "step": 408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5573479732327758e-05,
+      "loss": 1.1781,
+      "step": 409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555159623797161e-05,
+      "loss": 1.1462,
+      "step": 410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552967424576512e-05,
+      "loss": 1.1876,
+      "step": 411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507713907727557e-05,
+      "loss": 1.1872,
+      "step": 412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5485715376144087e-05,
+      "loss": 1.181,
+      "step": 413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463678803564753e-05,
+      "loss": 1.175,
+      "step": 414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544160434280337e-05,
+      "loss": 1.1791,
+      "step": 415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419492146936518e-05,
+      "loss": 1.19,
+      "step": 416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5397342369302425e-05,
+      "loss": 1.245,
+      "step": 417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375155163499953e-05,
+      "loss": 1.1792,
+      "step": 418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5352930683387502e-05,
+      "loss": 1.171,
+      "step": 419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533066908308196e-05,
+      "loss": 1.1415,
+      "step": 420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308370516957617e-05,
+      "loss": 1.2402,
+      "step": 421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528603513964511e-05,
+      "loss": 1.1745,
+      "step": 422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5263663106030347e-05,
+      "loss": 1.1877,
+      "step": 423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241254571253433e-05,
+      "loss": 1.1788,
+      "step": 424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5218809690707583e-05,
+      "loss": 1.2046,
+      "step": 425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196328620038059e-05,
+      "loss": 1.2149,
+      "step": 426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173811515141088e-05,
+      "loss": 1.1763,
+      "step": 427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5151258532162771e-05,
+      "loss": 1.2053,
+      "step": 428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128669827498024e-05,
+      "loss": 1.2089,
+      "step": 429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5106045557789453e-05,
+      "loss": 1.212,
+      "step": 430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5083385879926314e-05,
+      "loss": 1.1271,
+      "step": 431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060690951043385e-05,
+      "loss": 1.1759,
+      "step": 432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037960928519902e-05,
+      "loss": 1.1752,
+      "step": 433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501519596997847e-05,
+      "loss": 1.1923,
+      "step": 434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499239623328394e-05,
+      "loss": 1.1593,
+      "step": 435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4969561876542348e-05,
+      "loss": 1.1768,
+      "step": 436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4946693058099802e-05,
+      "loss": 1.2081,
+      "step": 437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492378993654138e-05,
+      "loss": 1.1572,
+      "step": 438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900852670690044e-05,
+      "loss": 1.162,
+      "step": 439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877881419605531e-05,
+      "loss": 1.1952,
+      "step": 440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854876342583246e-05,
+      "loss": 1.1586,
+      "step": 441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831837599153167e-05,
+      "loss": 1.2054,
+      "step": 442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4808765349078729e-05,
+      "loss": 1.1601,
+      "step": 443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785659752355724e-05,
+      "loss": 1.179,
+      "step": 444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4762520969211186e-05,
+      "loss": 1.1822,
+      "step": 445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4739349160102285e-05,
+      "loss": 1.173,
+      "step": 446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471614448571521e-05,
+      "loss": 1.2011,
+      "step": 447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692907106964051e-05,
+      "loss": 1.1721,
+      "step": 448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46696371849897e-05,
+      "loss": 1.1786,
+      "step": 449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646334881158706e-05,
+      "loss": 1.2071,
+      "step": 450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623000357062184e-05,
+      "loss": 1.2234,
+      "step": 451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459963377451468e-05,
+      "loss": 1.16,
+      "step": 452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457623529555305e-05,
+      "loss": 1.1922,
+      "step": 453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552805082435333e-05,
+      "loss": 1.1721,
+      "step": 454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529343297639638e-05,
+      "loss": 1.2375,
+      "step": 455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505850103863003e-05,
+      "loss": 1.1577,
+      "step": 456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448232566402028e-05,
+      "loss": 1.2174,
+      "step": 457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4458770141242992e-05,
+      "loss": 1.1911,
+      "step": 458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443518369887821e-05,
+      "loss": 1.1826,
+      "step": 459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411566500487425e-05,
+      "loss": 1.1723,
+      "step": 460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387918709845395e-05,
+      "loss": 1.1527,
+      "step": 461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364240490939032e-05,
+      "loss": 1.2365,
+      "step": 462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434053200796625e-05,
+      "loss": 1.2061,
+      "step": 463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316793425334834e-05,
+      "loss": 1.2218,
+      "step": 464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4293024907661295e-05,
+      "loss": 1.1734,
+      "step": 465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269226619769725e-05,
+      "loss": 1.212,
+      "step": 466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424539872669067e-05,
+      "loss": 1.1607,
+      "step": 467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4221541393659966e-05,
+      "loss": 1.1388,
+      "step": 468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4197654786117604e-05,
+      "loss": 1.203,
+      "step": 469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173739069706584e-05,
+      "loss": 1.1801,
+      "step": 470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414979441027176e-05,
+      "loss": 1.1927,
+      "step": 471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4125820973858693e-05,
+      "loss": 1.1707,
+      "step": 472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41018189267125e-05,
+      "loss": 1.2063,
+      "step": 473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077788435276701e-05,
+      "loss": 1.1656,
+      "step": 474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053729666192064e-05,
+      "loss": 1.2206,
+      "step": 475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029642786295452e-05,
+      "loss": 1.1147,
+      "step": 476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400552796261866e-05,
+      "loss": 1.1892,
+      "step": 477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981385362387268e-05,
+      "loss": 1.2071,
+      "step": 478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957215153019462e-05,
+      "loss": 1.2227,
+      "step": 479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933017502124897e-05,
+      "loss": 1.1625,
+      "step": 480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908792577503514e-05,
+      "loss": 1.1982,
+      "step": 481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884540547144393e-05,
+      "loss": 1.2355,
+      "step": 482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860261579224574e-05,
+      "loss": 1.1543,
+      "step": 483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835955842107903e-05,
+      "loss": 1.1948,
+      "step": 484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811623504343845e-05,
+      "loss": 1.2269,
+      "step": 485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787264734666348e-05,
+      "loss": 1.2357,
+      "step": 486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3762879701992642e-05,
+      "loss": 1.1814,
+      "step": 487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373846857542208e-05,
+      "loss": 1.1582,
+      "step": 488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714031524234965e-05,
+      "loss": 1.1656,
+      "step": 489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368956871789138e-05,
+      "loss": 1.174,
+      "step": 490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665080326030001e-05,
+      "loss": 1.136,
+      "step": 491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364056651846693e-05,
+      "loss": 1.2108,
+      "step": 492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616027465194525e-05,
+      "loss": 1.2016,
+      "step": 493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35914633363802e-05,
+      "loss": 1.2034,
+      "step": 494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566874302365262e-05,
+      "loss": 1.1206,
+      "step": 495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542260533663723e-05,
+      "loss": 1.1982,
+      "step": 496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351762220096112e-05,
+      "loss": 1.1445,
+      "step": 497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492959475113332e-05,
+      "loss": 1.1877,
+      "step": 498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468272527145388e-05,
+      "loss": 1.1683,
+      "step": 499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3443561528250295e-05,
+      "loss": 1.1368,
+      "step": 500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418826649787834e-05,
+      "loss": 1.154,
+      "step": 501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3394068063283387e-05,
+      "loss": 1.177,
+      "step": 502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369285940426737e-05,
+      "loss": 1.1707,
+      "step": 503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3344480453070882e-05,
+      "loss": 1.1674,
+      "step": 504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331965177323084e-05,
+      "loss": 1.2047,
+      "step": 505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294800073082465e-05,
+      "loss": 1.1635,
+      "step": 506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3269925524961237e-05,
+      "loss": 1.1921,
+      "step": 507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3245028301361086e-05,
+      "loss": 1.1486,
+      "step": 508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220108574933185e-05,
+      "loss": 1.1625,
+      "step": 509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3195166518484748e-05,
+      "loss": 1.2083,
+      "step": 510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317020230497784e-05,
+      "loss": 1.1531,
+      "step": 511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3145216107528178e-05,
+      "loss": 1.1074,
+      "step": 512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120208099403928e-05,
+      "loss": 1.2205,
+      "step": 513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095178454024499e-05,
+      "loss": 1.181,
+      "step": 514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070127344959348e-05,
+      "loss": 1.1736,
+      "step": 515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045054945926775e-05,
+      "loss": 1.2039,
+      "step": 516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3019961430792711e-05,
+      "loss": 1.2059,
+      "step": 517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2994846973569526e-05,
+      "loss": 1.138,
+      "step": 518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969711748414804e-05,
+      "loss": 1.1806,
+      "step": 519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944555929630149e-05,
+      "loss": 1.1989,
+      "step": 520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291937969165998e-05,
+      "loss": 1.1866,
+      "step": 521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2894183209090304e-05,
+      "loss": 1.195,
+      "step": 522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868966656647519e-05,
+      "loss": 1.1406,
+      "step": 523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2843730209197203e-05,
+      "loss": 1.2052,
+      "step": 524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818474041742885e-05,
+      "loss": 1.196,
+      "step": 525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793198329424858e-05,
+      "loss": 1.1633,
+      "step": 526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767903247518943e-05,
+      "loss": 1.1886,
+      "step": 527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2742588971435276e-05,
+      "loss": 1.1421,
+      "step": 528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717255676717106e-05,
+      "loss": 1.1137,
+      "step": 529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2691903539039561e-05,
+      "loss": 1.2006,
+      "step": 530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2666532734208437e-05,
+      "loss": 1.1825,
+      "step": 531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264114343815898e-05,
+      "loss": 1.2314,
+      "step": 532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2615735826954664e-05,
+      "loss": 1.189,
+      "step": 533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590310076785972e-05,
+      "loss": 1.1816,
+      "step": 534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.256486636396917e-05,
+      "loss": 1.1777,
+      "step": 535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539404864945087e-05,
+      "loss": 1.141,
+      "step": 536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513925756277894e-05,
+      "loss": 1.1549,
+      "step": 537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488429214653871e-05,
+      "loss": 1.2417,
+      "step": 538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462915416880199e-05,
+      "loss": 1.1379,
+      "step": 539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437384539883715e-05,
+      "loss": 1.1966,
+      "step": 540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2411836760709686e-05,
+      "loss": 1.1681,
+      "step": 541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386272256520606e-05,
+      "loss": 1.1873,
+      "step": 542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360691204594934e-05,
+      "loss": 1.1749,
+      "step": 543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335093782325889e-05,
+      "loss": 1.2204,
+      "step": 544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2309480167220203e-05,
+      "loss": 1.1876,
+      "step": 545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283850536896907e-05,
+      "loss": 1.1788,
+      "step": 546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258205069086082e-05,
+      "loss": 1.2119,
+      "step": 547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2232543941627643e-05,
+      "loss": 1.1518,
+      "step": 548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2206867332470091e-05,
+      "loss": 1.1658,
+      "step": 549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181175419669292e-05,
+      "loss": 1.1896,
+      "step": 550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215546838138723e-05,
+      "loss": 1.2005,
+      "step": 551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212974639589078e-05,
+      "loss": 1.1362,
+      "step": 552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104009641550472e-05,
+      "loss": 1.1699,
+      "step": 553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078258296839245e-05,
+      "loss": 1.1544,
+      "step": 554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205249254033122e-05,
+      "loss": 1.1455,
+      "step": 555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026712550700457e-05,
+      "loss": 1.1757,
+      "step": 556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200091850671972e-05,
+      "loss": 1.1273,
+      "step": 557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1975110587259222e-05,
+      "loss": 1.1708,
+      "step": 558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1949288971285413e-05,
+      "loss": 1.1899,
+      "step": 559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923453837859706e-05,
+      "loss": 1.1923,
+      "step": 560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897605366137262e-05,
+      "loss": 1.1824,
+      "step": 561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1871743735365735e-05,
+      "loss": 1.1466,
+      "step": 562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845869124884027e-05,
+      "loss": 1.1514,
+      "step": 563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1819981714121054e-05,
+      "loss": 1.182,
+      "step": 564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794081682594491e-05,
+      "loss": 1.1404,
+      "step": 565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1768169209909544e-05,
+      "loss": 1.1638,
+      "step": 566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174224447575767e-05,
+      "loss": 1.1555,
+      "step": 567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716307659915382e-05,
+      "loss": 1.1643,
+      "step": 568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169035894224295e-05,
+      "loss": 1.1342,
+      "step": 569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664398502683194e-05,
+      "loss": 1.1887,
+      "step": 570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1638426521260211e-05,
+      "loss": 1.1293,
+      "step": 571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1612443178078138e-05,
+      "loss": 1.1726,
+      "step": 572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586448653319908e-05,
+      "loss": 1.1474,
+      "step": 573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156044312724598e-05,
+      "loss": 1.1406,
+      "step": 574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534426780193114e-05,
+      "loss": 1.1759,
+      "step": 575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508399792573095e-05,
+      "loss": 1.1365,
+      "step": 576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1482362344871516e-05,
+      "loss": 1.1808,
+      "step": 577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456314617646485e-05,
+      "loss": 1.206,
+      "step": 578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.143025679152741e-05,
+      "loss": 1.1531,
+      "step": 579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404189047213716e-05,
+      "loss": 1.1951,
+      "step": 580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137811156547362e-05,
+      "loss": 1.1524,
+      "step": 581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1352024527142857e-05,
+      "loss": 1.1732,
+      "step": 582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325928113123431e-05,
+      "loss": 1.1932,
+      "step": 583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1299822504382365e-05,
+      "loss": 1.1737,
+      "step": 584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1273707881950447e-05,
+      "loss": 1.1712,
+      "step": 585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247584426920962e-05,
+      "loss": 1.2088,
+      "step": 586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221452320448447e-05,
+      "loss": 1.127,
+      "step": 587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195311743747445e-05,
+      "loss": 1.2277,
+      "step": 588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116916287809122e-05,
+      "loss": 1.2246,
+      "step": 589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143005904810527e-05,
+      "loss": 1.1174,
+      "step": 590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116841005292335e-05,
+      "loss": 1.1491,
+      "step": 591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090668360978587e-05,
+      "loss": 1.1252,
+      "step": 592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106448815336493e-05,
+      "loss": 1.2098,
+      "step": 593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038300563999453e-05,
+      "loss": 1.1624,
+      "step": 594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1012105774481446e-05,
+      "loss": 1.1954,
+      "step": 595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0985903966460115e-05,
+      "loss": 1.1636,
+      "step": 596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959695321633346e-05,
+      "loss": 1.1576,
+      "step": 597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093348002174643e-05,
+      "loss": 1.1737,
+      "step": 598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0907258248590816e-05,
+      "loss": 1.1534,
+      "step": 599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881030184002827e-05,
+      "loss": 1.1986,
+      "step": 600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854796009862434e-05,
+      "loss": 1.1867,
+      "step": 601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828555908091958e-05,
+      "loss": 1.2107,
+      "step": 602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080231006065483e-05,
+      "loss": 1.2043,
+      "step": 603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776058649554336e-05,
+      "loss": 1.1512,
+      "step": 604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0749801856832325e-05,
+      "loss": 1.1073,
+      "step": 605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723539864567983e-05,
+      "loss": 1.1603,
+      "step": 606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697272854876535e-05,
+      "loss": 1.1944,
+      "step": 607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0671001009908015e-05,
+      "loss": 1.1842,
+      "step": 608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0644724511845976e-05,
+      "loss": 1.1428,
+      "step": 609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618443542906251e-05,
+      "loss": 1.1471,
+      "step": 610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059215828533566e-05,
+      "loss": 1.202,
+      "step": 611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0565868921410777e-05,
+      "loss": 1.1954,
+      "step": 612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0539575633436645e-05,
+      "loss": 1.1587,
+      "step": 613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051327860374552e-05,
+      "loss": 1.164,
+      "step": 614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486978014695606e-05,
+      "loss": 1.2268,
+      "step": 615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0460674048669783e-05,
+      "loss": 1.1936,
+      "step": 616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0434366888074363e-05,
+      "loss": 1.0972,
+      "step": 617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408056715337795e-05,
+      "loss": 1.1468,
+      "step": 618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381743712909427e-05,
+      "loss": 1.2106,
+      "step": 619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355428063258224e-05,
+      "loss": 1.2397,
+      "step": 620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0329109948871512e-05,
+      "loss": 1.1623,
+      "step": 621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302789552253702e-05,
+      "loss": 1.1782,
+      "step": 622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0276467055925046e-05,
+      "loss": 1.1776,
+      "step": 623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0250142642420335e-05,
+      "loss": 1.142,
+      "step": 624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0223816494287673e-05,
+      "loss": 1.1545,
+      "step": 625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197488794087188e-05,
+      "loss": 1.2001,
+      "step": 626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0171159724389766e-05,
+      "loss": 1.2183,
+      "step": 627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144829467775794e-05,
+      "loss": 1.1695,
+      "step": 628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0118498206833886e-05,
+      "loss": 1.1617,
+      "step": 629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0092166124159631e-05,
+      "loss": 1.1075,
+      "step": 630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065833402354302e-05,
+      "loss": 1.1466,
+      "step": 631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039500224023612e-05,
+      "loss": 1.1446,
+      "step": 632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013166771776441e-05,
+      "loss": 1.1887,
+      "step": 633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986833228223566e-06,
+      "loss": 1.2252,
+      "step": 634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960499775976393e-06,
+      "loss": 1.161,
+      "step": 635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934166597645703e-06,
+      "loss": 1.1616,
+      "step": 636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.907833875840374e-06,
+      "loss": 1.1537,
+      "step": 637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881501793166117e-06,
+      "loss": 1.1833,
+      "step": 638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85517053222421e-06,
+      "loss": 1.2053,
+      "step": 639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828840275610238e-06,
+      "loss": 1.1529,
+      "step": 640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802511205912817e-06,
+      "loss": 1.1838,
+      "step": 641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77618350571233e-06,
+      "loss": 1.1906,
+      "step": 642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74985735757967e-06,
+      "loss": 1.2112,
+      "step": 643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.723532944074961e-06,
+      "loss": 1.2169,
+      "step": 644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.6972104477463e-06,
+      "loss": 1.1831,
+      "step": 645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670890051128493e-06,
+      "loss": 1.1826,
+      "step": 646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64457193674178e-06,
+      "loss": 1.183,
+      "step": 647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618256287090576e-06,
+      "loss": 1.1277,
+      "step": 648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591943284662208e-06,
+      "loss": 1.2159,
+      "step": 649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56563311192564e-06,
+      "loss": 1.1904,
+      "step": 650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539325951330217e-06,
+      "loss": 1.1363,
+      "step": 651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513021985304399e-06,
+      "loss": 1.1812,
+      "step": 652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.486721396254482e-06,
+      "loss": 1.1698,
+      "step": 653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460424366563356e-06,
+      "loss": 1.1751,
+      "step": 654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434131078589224e-06,
+      "loss": 1.16,
+      "step": 655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407841714664341e-06,
+      "loss": 1.1353,
+      "step": 656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.381556457093752e-06,
+      "loss": 1.1619,
+      "step": 657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.355275488154024e-06,
+      "loss": 1.2014,
+      "step": 658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.328998990091989e-06,
+      "loss": 1.1892,
+      "step": 659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302727145123466e-06,
+      "loss": 1.2126,
+      "step": 660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27646013543202e-06,
+      "loss": 1.1503,
+      "step": 661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.250198143167675e-06,
+      "loss": 1.149,
+      "step": 662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223941350445666e-06,
+      "loss": 1.1251,
+      "step": 663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19768993934517e-06,
+      "loss": 1.1395,
+      "step": 664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171444091908046e-06,
+      "loss": 1.116,
+      "step": 665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14520399013757e-06,
+      "loss": 1.1682,
+      "step": 666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118969815997171e-06,
+      "loss": 1.1704,
+      "step": 667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.092741751409188e-06,
+      "loss": 1.1565,
+      "step": 668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06651997825357e-06,
+      "loss": 1.1694,
+      "step": 669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040304678366658e-06,
+      "loss": 1.1667,
+      "step": 670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014096033539887e-06,
+      "loss": 1.1558,
+      "step": 671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.987894225518556e-06,
+      "loss": 1.152,
+      "step": 672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961699436000547e-06,
+      "loss": 1.1772,
+      "step": 673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935511846635072e-06,
+      "loss": 1.2031,
+      "step": 674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.909331639021414e-06,
+      "loss": 1.1352,
+      "step": 675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883158994707668e-06,
+      "loss": 1.185,
+      "step": 676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856994095189477e-06,
+      "loss": 1.1511,
+      "step": 677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830837121908781e-06,
+      "loss": 1.1147,
+      "step": 678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804688256252557e-06,
+      "loss": 1.1575,
+      "step": 679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778547679551553e-06,
+      "loss": 1.1913,
+      "step": 680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752415573079043e-06,
+      "loss": 1.1794,
+      "step": 681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.726292118049555e-06,
+      "loss": 1.1114,
+      "step": 682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.700177495617636e-06,
+      "loss": 1.1667,
+      "step": 683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674071886876572e-06,
+      "loss": 1.1204,
+      "step": 684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.647975472857148e-06,
+      "loss": 1.1403,
+      "step": 685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.621888434526382e-06,
+      "loss": 1.2086,
+      "step": 686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595810952786289e-06,
+      "loss": 1.1376,
+      "step": 687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.569743208472596e-06,
+      "loss": 1.1492,
+      "step": 688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54368538235352e-06,
+      "loss": 1.1376,
+      "step": 689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517637655128488e-06,
+      "loss": 1.144,
+      "step": 690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491600207426907e-06,
+      "loss": 1.1479,
+      "step": 691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465573219806893e-06,
+      "loss": 1.1704,
+      "step": 692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439556872754025e-06,
+      "loss": 1.1555,
+      "step": 693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413551346680093e-06,
+      "loss": 1.1537,
+      "step": 694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.387556821921863e-06,
+      "loss": 1.1519,
+      "step": 695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361573478739792e-06,
+      "loss": 1.1028,
+      "step": 696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335601497316812e-06,
+      "loss": 1.1715,
+      "step": 697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309641057757052e-06,
+      "loss": 1.173,
+      "step": 698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283692340084623e-06,
+      "loss": 1.2277,
+      "step": 699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.257755524242333e-06,
+      "loss": 1.1248,
+      "step": 700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231830790090461e-06,
+      "loss": 1.1027,
+      "step": 701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20591831740551e-06,
+      "loss": 1.1983,
+      "step": 702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180018285878951e-06,
+      "loss": 1.1773,
+      "step": 703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15413087511598e-06,
+      "loss": 1.165,
+      "step": 704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12825626463427e-06,
+      "loss": 1.1435,
+      "step": 705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102394633862743e-06,
+      "loss": 1.1666,
+      "step": 706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.0765461621403e-06,
+      "loss": 1.1618,
+      "step": 707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050711028714592e-06,
+      "loss": 1.172,
+      "step": 708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02488941274078e-06,
+      "loss": 1.1194,
+      "step": 709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999081493280285e-06,
+      "loss": 1.1563,
+      "step": 710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973287449299545e-06,
+      "loss": 1.1356,
+      "step": 711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.947507459668782e-06,
+      "loss": 1.1282,
+      "step": 712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.921741703160758e-06,
+      "loss": 1.1395,
+      "step": 713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895990358449532e-06,
+      "loss": 1.1488,
+      "step": 714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870253604109222e-06,
+      "loss": 1.1662,
+      "step": 715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844531618612772e-06,
+      "loss": 1.1319,
+      "step": 716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81882458033071e-06,
+      "loss": 1.1693,
+      "step": 717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79313266752991e-06,
+      "loss": 1.1639,
+      "step": 718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767456058372362e-06,
+      "loss": 1.2031,
+      "step": 719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74179493091392e-06,
+      "loss": 1.1188,
+      "step": 720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716149463103097e-06,
+      "loss": 1.1507,
+      "step": 721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6905198327798e-06,
+      "loss": 1.1559,
+      "step": 722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664906217674111e-06,
+      "loss": 1.1723,
+      "step": 723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.639308795405066e-06,
+      "loss": 1.152,
+      "step": 724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613727743479395e-06,
+      "loss": 1.125,
+      "step": 725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588163239290316e-06,
+      "loss": 1.1368,
+      "step": 726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5626154601162874e-06,
+      "loss": 1.166,
+      "step": 727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537084583119802e-06,
+      "loss": 1.1855,
+      "step": 728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51157078534613e-06,
+      "loss": 1.115,
+      "step": 729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486074243722111e-06,
+      "loss": 1.2038,
+      "step": 730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460595135054914e-06,
+      "loss": 1.2054,
+      "step": 731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435133636030831e-06,
+      "loss": 1.1364,
+      "step": 732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4096899232140295e-06,
+      "loss": 1.1184,
+      "step": 733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.384264173045335e-06,
+      "loss": 1.1973,
+      "step": 734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358856561841021e-06,
+      "loss": 1.1593,
+      "step": 735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333467265791565e-06,
+      "loss": 1.1763,
+      "step": 736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308096460960443e-06,
+      "loss": 1.1532,
+      "step": 737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2827443232828935e-06,
+      "loss": 1.1718,
+      "step": 738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2574110285647244e-06,
+      "loss": 1.1647,
+      "step": 739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232096752481061e-06,
+      "loss": 1.1626,
+      "step": 740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206801670575145e-06,
+      "loss": 1.1236,
+      "step": 741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181525958257116e-06,
+      "loss": 1.156,
+      "step": 742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156269790802801e-06,
+      "loss": 1.1582,
+      "step": 743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131033343352485e-06,
+      "loss": 1.1344,
+      "step": 744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105816790909696e-06,
+      "loss": 1.2174,
+      "step": 745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080620308340022e-06,
+      "loss": 1.143,
+      "step": 746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055444070369852e-06,
+      "loss": 1.1557,
+      "step": 747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0302882515852025e-06,
+      "loss": 1.1574,
+      "step": 748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.005153026430476e-06,
+      "loss": 1.1706,
+      "step": 749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980038569207291e-06,
+      "loss": 1.1525,
+      "step": 750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.954945054073229e-06,
+      "loss": 1.1916,
+      "step": 751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9298726550406524e-06,
+      "loss": 1.1099,
+      "step": 752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.904821545975504e-06,
+      "loss": 1.1596,
+      "step": 753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879791900596077e-06,
+      "loss": 1.1477,
+      "step": 754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.854783892471827e-06,
+      "loss": 1.2362,
+      "step": 755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829797695022163e-06,
+      "loss": 1.1253,
+      "step": 756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.804833481515256e-06,
+      "loss": 1.1082,
+      "step": 757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779891425066818e-06,
+      "loss": 1.1894,
+      "step": 758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.754971698638917e-06,
+      "loss": 1.1347,
+      "step": 759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.730074475038764e-06,
+      "loss": 1.1739,
+      "step": 760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7051999269175405e-06,
+      "loss": 1.1456,
+      "step": 761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680348226769164e-06,
+      "loss": 1.1876,
+      "step": 762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.655519546929121e-06,
+      "loss": 1.1199,
+      "step": 763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630714059573267e-06,
+      "loss": 1.1836,
+      "step": 764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6059319367166165e-06,
+      "loss": 1.2137,
+      "step": 765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5811733502121715e-06,
+      "loss": 1.1481,
+      "step": 766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.556438471749708e-06,
+      "loss": 1.1523,
+      "step": 767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.531727472854614e-06,
+      "loss": 1.1573,
+      "step": 768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507040524886674e-06,
+      "loss": 1.1651,
+      "step": 769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4823777990388835e-06,
+      "loss": 1.1269,
+      "step": 770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45773946633628e-06,
+      "loss": 1.0994,
+      "step": 771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4331256976347434e-06,
+      "loss": 1.1059,
+      "step": 772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408536663619803e-06,
+      "loss": 1.1778,
+      "step": 773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383972534805477e-06,
+      "loss": 1.1603,
+      "step": 774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.359433481533072e-06,
+      "loss": 1.1653,
+      "step": 775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3349196739700024e-06,
+      "loss": 1.1883,
+      "step": 776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310431282108625e-06,
+      "loss": 1.1621,
+      "step": 777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2859684757650365e-06,
+      "loss": 1.1714,
+      "step": 778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26153142457792e-06,
+      "loss": 1.132,
+      "step": 779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2371202980073596e-06,
+      "loss": 1.1728,
+      "step": 780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212735265333655e-06,
+      "loss": 1.1265,
+      "step": 781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188376495656156e-06,
+      "loss": 1.1612,
+      "step": 782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164044157892102e-06,
+      "loss": 1.1061,
+      "step": 783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13973842077543e-06,
+      "loss": 1.0818,
+      "step": 784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1154594528556075e-06,
+      "loss": 1.1483,
+      "step": 785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091207422496487e-06,
+      "loss": 1.1565,
+      "step": 786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066982497875107e-06,
+      "loss": 1.1751,
+      "step": 787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042784846980542e-06,
+      "loss": 1.1733,
+      "step": 788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018614637612733e-06,
+      "loss": 1.1758,
+      "step": 789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.99447203738134e-06,
+      "loss": 1.1392,
+      "step": 790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9703572137045495e-06,
+      "loss": 1.1374,
+      "step": 791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94627033380794e-06,
+      "loss": 1.1059,
+      "step": 792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922211564723299e-06,
+      "loss": 1.1694,
+      "step": 793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8981810732875024e-06,
+      "loss": 1.1634,
+      "step": 794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87417902614131e-06,
+      "loss": 1.1399,
+      "step": 795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.850205589728239e-06,
+      "loss": 1.17,
+      "step": 796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826260930293417e-06,
+      "loss": 1.1192,
+      "step": 797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802345213882399e-06,
+      "loss": 1.1306,
+      "step": 798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.778458606340037e-06,
+      "loss": 1.1511,
+      "step": 799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.754601273309333e-06,
+      "loss": 1.1951,
+      "step": 800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730773380230276e-06,
+      "loss": 1.1594,
+      "step": 801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.70697509233871e-06,
+      "loss": 1.1566,
+      "step": 802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683206574665169e-06,
+      "loss": 1.1454,
+      "step": 803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6594679920337514e-06,
+      "loss": 1.1345,
+      "step": 804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635759509060969e-06,
+      "loss": 1.1423,
+      "step": 805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612081290154607e-06,
+      "loss": 1.096,
+      "step": 806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588433499512576e-06,
+      "loss": 1.1761,
+      "step": 807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564816301121792e-06,
+      "loss": 1.1247,
+      "step": 808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.541229858757011e-06,
+      "loss": 1.168,
+      "step": 809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.517674335979721e-06,
+      "loss": 1.1658,
+      "step": 810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494149896136998e-06,
+      "loss": 1.126,
+      "step": 811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.470656702360363e-06,
+      "loss": 1.1915,
+      "step": 812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447194917564671e-06,
+      "loss": 1.1914,
+      "step": 813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423764704446954e-06,
+      "loss": 1.1638,
+      "step": 814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.400366225485322e-06,
+      "loss": 1.1167,
+      "step": 815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376999642937817e-06,
+      "loss": 1.153,
+      "step": 816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3536651188413e-06,
+      "loss": 1.151,
+      "step": 817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.330362815010302e-06,
+      "loss": 1.1679,
+      "step": 818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307092893035951e-06,
+      "loss": 1.1837,
+      "step": 819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2838555142847925e-06,
+      "loss": 1.1779,
+      "step": 820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.260650839897719e-06,
+      "loss": 1.1897,
+      "step": 821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237479030788813e-06,
+      "loss": 1.1088,
+      "step": 822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214340247644278e-06,
+      "loss": 1.1988,
+      "step": 823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191234650921273e-06,
+      "loss": 1.1311,
+      "step": 824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168162400846839e-06,
+      "loss": 1.1758,
+      "step": 825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145123657416755e-06,
+      "loss": 1.1591,
+      "step": 826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122118580394473e-06,
+      "loss": 1.1308,
+      "step": 827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099147329309959e-06,
+      "loss": 1.1886,
+      "step": 828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076210063458622e-06,
+      "loss": 1.152,
+      "step": 829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0533069419002e-06,
+      "loss": 1.1644,
+      "step": 830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030438123457655e-06,
+      "loss": 1.1246,
+      "step": 831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007603766716063e-06,
+      "loss": 1.1455,
+      "step": 832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.984804030021533e-06,
+      "loss": 1.1194,
+      "step": 833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962039071480098e-06,
+      "loss": 1.1803,
+      "step": 834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939309048956622e-06,
+      "loss": 1.177,
+      "step": 835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.916614120073693e-06,
+      "loss": 1.125,
+      "step": 836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.89395444221055e-06,
+      "loss": 1.1602,
+      "step": 837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871330172501979e-06,
+      "loss": 1.1623,
+      "step": 838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8487414678372315e-06,
+      "loss": 1.1648,
+      "step": 839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826188484858913e-06,
+      "loss": 1.1159,
+      "step": 840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803671379961945e-06,
+      "loss": 1.1565,
+      "step": 841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781190309292421e-06,
+      "loss": 1.1528,
+      "step": 842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758745428746573e-06,
+      "loss": 1.1858,
+      "step": 843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.736336893969652e-06,
+      "loss": 1.133,
+      "step": 844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7139648603548925e-06,
+      "loss": 1.1275,
+      "step": 845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691629483042387e-06,
+      "loss": 1.1115,
+      "step": 846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669330916918047e-06,
+      "loss": 1.1297,
+      "step": 847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647069316612499e-06,
+      "loss": 1.1774,
+      "step": 848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624844836500052e-06,
+      "loss": 1.1648,
+      "step": 849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60265763069758e-06,
+      "loss": 1.1578,
+      "step": 850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580507853063487e-06,
+      "loss": 1.165,
+      "step": 851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5583956571966295e-06,
+      "loss": 1.1498,
+      "step": 852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5363211964352524e-06,
+      "loss": 1.1535,
+      "step": 853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514284623855915e-06,
+      "loss": 1.1439,
+      "step": 854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4922860922724466e-06,
+      "loss": 1.1434,
+      "step": 855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470325754234881e-06,
+      "loss": 1.2336,
+      "step": 856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.448403762028396e-06,
+      "loss": 1.1662,
+      "step": 857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4265202676722475e-06,
+      "loss": 1.126,
+      "step": 858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40467542291874e-06,
+      "loss": 1.1529,
+      "step": 859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382869379252152e-06,
+      "loss": 1.1493,
+      "step": 860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3611022878877015e-06,
+      "loss": 1.1482,
+      "step": 861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339374299770473e-06,
+      "loss": 1.1408,
+      "step": 862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317685565574413e-06,
+      "loss": 1.1513,
+      "step": 863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296036235701235e-06,
+      "loss": 1.1312,
+      "step": 864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.274426460279412e-06,
+      "loss": 1.1597,
+      "step": 865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252856389163128e-06,
+      "loss": 1.1811,
+      "step": 866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231326171931231e-06,
+      "loss": 1.1459,
+      "step": 867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209835957886196e-06,
+      "loss": 1.1522,
+      "step": 868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188385896053098e-06,
+      "loss": 1.1482,
+      "step": 869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166976135178575e-06,
+      "loss": 1.1396,
+      "step": 870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1456068237297964e-06,
+      "loss": 1.1558,
+      "step": 871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124278109893432e-06,
+      "loss": 1.1432,
+      "step": 872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10299014157462e-06,
+      "loss": 1.1315,
+      "step": 873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0817430663959536e-06,
+      "loss": 1.1128,
+      "step": 874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060537031696446e-06,
+      "loss": 1.1229,
+      "step": 875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039372184530521e-06,
+      "loss": 1.1211,
+      "step": 876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018248671666969e-06,
+      "loss": 1.169,
+      "step": 877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9971666395879605e-06,
+      "loss": 1.1072,
+      "step": 878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9761262344880096e-06,
+      "loss": 1.1476,
+      "step": 879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9551276022729644e-06,
+      "loss": 1.0853,
+      "step": 880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934170888559e-06,
+      "loss": 1.1687,
+      "step": 881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913256238671607e-06,
+      "loss": 1.1539,
+      "step": 882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89238379764457e-06,
+      "loss": 1.0958,
+      "step": 883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.871553710218988e-06,
+      "loss": 1.1998,
+      "step": 884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850766120842252e-06,
+      "loss": 1.1589,
+      "step": 885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830021173667048e-06,
+      "loss": 1.1514,
+      "step": 886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809319012550352e-06,
+      "loss": 1.1087,
+      "step": 887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.788659781052444e-06,
+      "loss": 1.1364,
+      "step": 888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.768043622435905e-06,
+      "loss": 1.1737,
+      "step": 889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7474706796646275e-06,
+      "loss": 1.1584,
+      "step": 890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7269410954028073e-06,
+      "loss": 1.1401,
+      "step": 891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.706455012013994e-06,
+      "loss": 1.1456,
+      "step": 892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6860125715600513e-06,
+      "loss": 1.1236,
+      "step": 893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.665613915800217e-06,
+      "loss": 1.1432,
+      "step": 894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6452591861900857e-06,
+      "loss": 1.1848,
+      "step": 895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6249485238806637e-06,
+      "loss": 1.1296,
+      "step": 896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6046820697173514e-06,
+      "loss": 1.1462,
+      "step": 897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.584459964239e-06,
+      "loss": 1.147,
+      "step": 898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.564282347676903e-06,
+      "loss": 1.1399,
+      "step": 899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.54414935995387e-06,
+      "loss": 1.1472,
+      "step": 900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.524061140683206e-06,
+      "loss": 1.1688,
+      "step": 901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5040178291677786e-06,
+      "loss": 1.1588,
+      "step": 902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484019564399035e-06,
+      "loss": 1.129,
+      "step": 903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4640664850560514e-06,
+      "loss": 1.1715,
+      "step": 904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444158729504549e-06,
+      "loss": 1.1075,
+      "step": 905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4242964357959597e-06,
+      "loss": 1.2042,
+      "step": 906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4044797416664564e-06,
+      "loss": 1.1493,
+      "step": 907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3847087845359996e-06,
+      "loss": 1.1522,
+      "step": 908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.364983701507376e-06,
+      "loss": 1.1815,
+      "step": 909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3453046293652657e-06,
+      "loss": 1.1356,
+      "step": 910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3256717045752794e-06,
+      "loss": 1.1131,
+      "step": 911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30608506328302e-06,
+      "loss": 1.1695,
+      "step": 912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.286544841313123e-06,
+      "loss": 1.1028,
+      "step": 913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2670511741683475e-06,
+      "loss": 1.1548,
+      "step": 914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2476041970285945e-06,
+      "loss": 1.1038,
+      "step": 915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2282040447500063e-06,
+      "loss": 1.1464,
+      "step": 916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.208850851863998e-06,
+      "loss": 1.1932,
+      "step": 917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.189544752576369e-06,
+      "loss": 1.1167,
+      "step": 918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1702858807663175e-06,
+      "loss": 1.2171,
+      "step": 919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1510743699855596e-06,
+      "loss": 1.0843,
+      "step": 920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131910353457366e-06,
+      "loss": 1.1415,
+      "step": 921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.112793964075681e-06,
+      "loss": 1.1127,
+      "step": 922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0937253344041507e-06,
+      "loss": 1.1563,
+      "step": 923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.074704596675242e-06,
+      "loss": 1.1567,
+      "step": 924
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055731882789311e-06,
+      "loss": 1.1468,
+      "step": 925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.036807324313691e-06,
+      "loss": 1.1656,
+      "step": 926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0179310524817707e-06,
+      "loss": 1.1455,
+      "step": 927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9991031981921026e-06,
+      "loss": 1.1439,
+      "step": 928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9803238920074784e-06,
+      "loss": 1.1148,
+      "step": 929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.961593264154038e-06,
+      "loss": 1.1496,
+      "step": 930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9429114445203453e-06,
+      "loss": 1.184,
+      "step": 931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.924278562656514e-06,
+      "loss": 1.153,
+      "step": 932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.90569474777329e-06,
+      "loss": 1.1785,
+      "step": 933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8871601287411634e-06,
+      "loss": 1.1467,
+      "step": 934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.868674834089471e-06,
+      "loss": 1.1201,
+      "step": 935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850238992005514e-06,
+      "loss": 1.0852,
+      "step": 936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8318527303336465e-06,
+      "loss": 1.1701,
+      "step": 937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.81351617657442e-06,
+      "loss": 1.1229,
+      "step": 938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.795229457883678e-06,
+      "loss": 1.0841,
+      "step": 939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.776992701071678e-06,
+      "loss": 1.1177,
+      "step": 940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7588060326022205e-06,
+      "loss": 1.1815,
+      "step": 941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.740669578591755e-06,
+      "loss": 1.0721,
+      "step": 942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.722583464808525e-06,
+      "loss": 1.1371,
+      "step": 943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7045478166716843e-06,
+      "loss": 1.1468,
+      "step": 944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.686562759250433e-06,
+      "loss": 1.1387,
+      "step": 945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.668628417263135e-06,
+      "loss": 1.1026,
+      "step": 946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6507449150764852e-06,
+      "loss": 1.1647,
+      "step": 947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.632912376704607e-06,
+      "loss": 1.1644,
+      "step": 948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.615130925808228e-06,
+      "loss": 1.1347,
+      "step": 949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5974006856937917e-06,
+      "loss": 1.1401,
+      "step": 950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5797217793126373e-06,
+      "loss": 1.1395,
+      "step": 951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5620943292601074e-06,
+      "loss": 1.1204,
+      "step": 952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.544518457774734e-06,
+      "loss": 1.173,
+      "step": 953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5269942867373565e-06,
+      "loss": 1.1249,
+      "step": 954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5095219376703183e-06,
+      "loss": 1.1115,
+      "step": 955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4921015317365794e-06,
+      "loss": 1.1363,
+      "step": 956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.474733189738908e-06,
+      "loss": 1.1512,
+      "step": 957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4574170321190305e-06,
+      "loss": 1.1167,
+      "step": 958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440153178956798e-06,
+      "loss": 1.1136,
+      "step": 959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42294174996935e-06,
+      "loss": 1.1459,
+      "step": 960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.40578286451029e-06,
+      "loss": 1.1848,
+      "step": 961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3886766415688567e-06,
+      "loss": 1.1791,
+      "step": 962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3716231997691007e-06,
+      "loss": 1.1055,
+      "step": 963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.354622657369048e-06,
+      "loss": 1.1378,
+      "step": 964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3376751322599e-06,
+      "loss": 1.1409,
+      "step": 965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320780741965206e-06,
+      "loss": 1.0695,
+      "step": 966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3039396036400484e-06,
+      "loss": 1.1121,
+      "step": 967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2871518340702236e-06,
+      "loss": 1.1278,
+      "step": 968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2704175496714552e-06,
+      "loss": 1.1539,
+      "step": 969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2537368664885527e-06,
+      "loss": 1.2019,
+      "step": 970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.237109900194642e-06,
+      "loss": 1.1149,
+      "step": 971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2205367660903267e-06,
+      "loss": 1.1227,
+      "step": 972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2040175791029305e-06,
+      "loss": 1.1557,
+      "step": 973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.187552453785662e-06,
+      "loss": 1.1264,
+      "step": 974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1711415043168425e-06,
+      "loss": 1.1483,
+      "step": 975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1547848444991004e-06,
+      "loss": 1.1494,
+      "step": 976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.138482587758605e-06,
+      "loss": 1.1849,
+      "step": 977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1222348471442477e-06,
+      "loss": 1.1274,
+      "step": 978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1060417353268845e-06,
+      "loss": 1.1215,
+      "step": 979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0899033645985423e-06,
+      "loss": 1.1129,
+      "step": 980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.073819846871646e-06,
+      "loss": 1.1245,
+      "step": 981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0577912936782317e-06,
+      "loss": 1.1335,
+      "step": 982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041817816169187e-06,
+      "loss": 1.118,
+      "step": 983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.025899525113474e-06,
+      "loss": 1.1778,
+      "step": 984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.010036530897361e-06,
+      "loss": 1.0896,
+      "step": 985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994228943523654e-06,
+      "loss": 1.1314,
+      "step": 986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978476872610939e-06,
+      "loss": 1.08,
+      "step": 987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962780427392823e-06,
+      "loss": 1.1247,
+      "step": 988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9471397167171735e-06,
+      "loss": 1.1553,
+      "step": 989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.931554849045353e-06,
+      "loss": 1.146,
+      "step": 990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916025932451496e-06,
+      "loss": 1.1474,
+      "step": 991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9005530746217238e-06,
+      "loss": 1.1209,
+      "step": 992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8851363828534253e-06,
+      "loss": 1.0977,
+      "step": 993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.869775964054501e-06,
+      "loss": 1.1195,
+      "step": 994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8544719247426224e-06,
+      "loss": 1.1426,
+      "step": 995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8392243710444911e-06,
+      "loss": 1.1331,
+      "step": 996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8240334086951117e-06,
+      "loss": 1.1091,
+      "step": 997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8088991430370506e-06,
+      "loss": 1.1769,
+      "step": 998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7938216790197095e-06,
+      "loss": 1.153,
+      "step": 999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.77880112119859e-06,
+      "loss": 1.1231,
+      "step": 1000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7638375737345804e-06,
+      "loss": 1.1393,
+      "step": 1001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7489311403932274e-06,
+      "loss": 1.1331,
+      "step": 1002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7340819245440144e-06,
+      "loss": 1.1521,
+      "step": 1003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7192900291596493e-06,
+      "loss": 1.1218,
+      "step": 1004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7045555568153438e-06,
+      "loss": 1.1196,
+      "step": 1005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6898786096881104e-06,
+      "loss": 1.1207,
+      "step": 1006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6752592895560493e-06,
+      "loss": 1.1627,
+      "step": 1007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6606976977976408e-06,
+      "loss": 1.1614,
+      "step": 1008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6461939353910473e-06,
+      "loss": 1.1378,
+      "step": 1009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.631748102913412e-06,
+      "loss": 1.1669,
+      "step": 1010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6173603005401505e-06,
+      "loss": 1.1814,
+      "step": 1011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6030306280442764e-06,
+      "loss": 1.1652,
+      "step": 1012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.588759184795694e-06,
+      "loss": 1.1286,
+      "step": 1013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.574546069760514e-06,
+      "loss": 1.1506,
+      "step": 1014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5603913815003634e-06,
+      "loss": 1.1357,
+      "step": 1015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5462952181717117e-06,
+      "loss": 1.2027,
+      "step": 1016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5322576775251808e-06,
+      "loss": 1.1373,
+      "step": 1017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5182788569048712e-06,
+      "loss": 1.1319,
+      "step": 1018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5043588532476806e-06,
+      "loss": 1.1618,
+      "step": 1019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49049776308265e-06,
+      "loss": 1.1578,
+      "step": 1020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.476695682530268e-06,
+      "loss": 1.1545,
+      "step": 1021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4629527073018267e-06,
+      "loss": 1.116,
+      "step": 1022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4492689326987408e-06,
+      "loss": 1.1662,
+      "step": 1023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4356444536119085e-06,
+      "loss": 1.1887,
+      "step": 1024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.422079364521024e-06,
+      "loss": 1.111,
+      "step": 1025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4085737594939519e-06,
+      "loss": 1.1384,
+      "step": 1026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3951277321860468e-06,
+      "loss": 1.0839,
+      "step": 1027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.381741375839537e-06,
+      "loss": 1.167,
+      "step": 1028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3684147832828409e-06,
+      "loss": 1.1495,
+      "step": 1029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3551480469299538e-06,
+      "loss": 1.176,
+      "step": 1030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3419412587797887e-06,
+      "loss": 1.0655,
+      "step": 1031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3287945104155508e-06,
+      "loss": 1.1679,
+      "step": 1032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3157078930040856e-06,
+      "loss": 1.1448,
+      "step": 1033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3026814972952674e-06,
+      "loss": 1.1494,
+      "step": 1034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2897154136213542e-06,
+      "loss": 1.121,
+      "step": 1035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2768097318963701e-06,
+      "loss": 1.1679,
+      "step": 1036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2639645416154744e-06,
+      "loss": 1.1609,
+      "step": 1037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2511799318543493e-06,
+      "loss": 1.1445,
+      "step": 1038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2384559912685768e-06,
+      "loss": 1.1041,
+      "step": 1039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.225792808093025e-06,
+      "loss": 1.1081,
+      "step": 1040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2131904701412322e-06,
+      "loss": 1.1605,
+      "step": 1041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2006490648048118e-06,
+      "loss": 1.1238,
+      "step": 1042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1881686790528279e-06,
+      "loss": 1.1187,
+      "step": 1043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1757493994312052e-06,
+      "loss": 1.0732,
+      "step": 1044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1633913120621188e-06,
+      "loss": 1.1456,
+      "step": 1045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.151094502643414e-06,
+      "loss": 1.1755,
+      "step": 1046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1388590564479895e-06,
+      "loss": 1.1334,
+      "step": 1047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1266850583232248e-06,
+      "loss": 1.1008,
+      "step": 1048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.114572592690375e-06,
+      "loss": 1.1489,
+      "step": 1049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1025217435440116e-06,
+      "loss": 1.1248,
+      "step": 1050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0905325944514034e-06,
+      "loss": 1.125,
+      "step": 1051
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.078605228551971e-06,
+      "loss": 1.1097,
+      "step": 1052
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0667397285566893e-06,
+      "loss": 1.1247,
+      "step": 1053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0549361767475252e-06,
+      "loss": 1.1651,
+      "step": 1054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0431946549768567e-06,
+      "loss": 1.1221,
+      "step": 1055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0315152446669142e-06,
+      "loss": 1.1349,
+      "step": 1056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.019898026809214e-06,
+      "loss": 1.1144,
+      "step": 1057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0083430819639962e-06,
+      "loss": 1.1489,
+      "step": 1058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.968504902596576e-07,
+      "loss": 1.1133,
+      "step": 1059
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85420331392214e-07,
+      "loss": 1.1033,
+      "step": 1060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.74052684624731e-07,
+      "loss": 1.109,
+      "step": 1061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.62747628786782e-07,
+      "loss": 1.1,
+      "step": 1062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.515052422739013e-07,
+      "loss": 1.2168,
+      "step": 1063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.403256030470386e-07,
+      "loss": 1.1188,
+      "step": 1064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.292087886320166e-07,
+      "loss": 1.1206,
+      "step": 1065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.181548761189996e-07,
+      "loss": 1.132,
+      "step": 1066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.071639421619527e-07,
+      "loss": 1.1774,
+      "step": 1067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.962360629781153e-07,
+      "loss": 1.1345,
+      "step": 1068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.853713143474685e-07,
+      "loss": 1.1571,
+      "step": 1069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.745697716122081e-07,
+      "loss": 1.1471,
+      "step": 1070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.638315096762306e-07,
+      "loss": 1.0699,
+      "step": 1071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.531566030046035e-07,
+      "loss": 1.1462,
+      "step": 1072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.42545125623061e-07,
+      "loss": 1.1512,
+      "step": 1073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.319971511174696e-07,
+      "loss": 1.1104,
+      "step": 1074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.215127526333499e-07,
+      "loss": 1.1239,
+      "step": 1075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.110920028753355e-07,
+      "loss": 1.1659,
+      "step": 1076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.007349741066939e-07,
+      "loss": 1.0606,
+      "step": 1077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.904417381488072e-07,
+      "loss": 1.1624,
+      "step": 1078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.802123663806938e-07,
+      "loss": 1.0982,
+      "step": 1079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.700469297384927e-07,
+      "loss": 1.1544,
+      "step": 1080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.599454987149879e-07,
+      "loss": 1.1848,
+      "step": 1081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.499081433591049e-07,
+      "loss": 1.1559,
+      "step": 1082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.399349332754458e-07,
+      "loss": 1.0858,
+      "step": 1083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.300259376237795e-07,
+      "loss": 1.1171,
+      "step": 1084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.201812251185847e-07,
+      "loss": 1.1894,
+      "step": 1085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.104008640285642e-07,
+      "loss": 1.1294,
+      "step": 1086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.006849221761736e-07,
+      "loss": 1.1218,
+      "step": 1087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.910334669371433e-07,
+      "loss": 1.1276,
+      "step": 1088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.814465652400237e-07,
+      "loss": 1.1393,
+      "step": 1089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.719242835657125e-07,
+      "loss": 1.1422,
+      "step": 1090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.62466687947001e-07,
+      "loss": 1.1709,
+      "step": 1091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.53073843968104e-07,
+      "loss": 1.1388,
+      "step": 1092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.437458167642164e-07,
+      "loss": 1.0765,
+      "step": 1093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.344826710210584e-07,
+      "loss": 1.0951,
+      "step": 1094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.252844709744266e-07,
+      "loss": 1.101,
+      "step": 1095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.161512804097414e-07,
+      "loss": 1.0931,
+      "step": 1096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.070831626616236e-07,
+      "loss": 1.1475,
+      "step": 1097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.980801806134318e-07,
+      "loss": 1.1936,
+      "step": 1098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.891423966968424e-07,
+      "loss": 1.1667,
+      "step": 1099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.80269872891408e-07,
+      "loss": 1.1223,
+      "step": 1100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.714626707241411e-07,
+      "loss": 1.123,
+      "step": 1101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.627208512690641e-07,
+      "loss": 1.1255,
+      "step": 1102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.540444751468122e-07,
+      "loss": 1.1056,
+      "step": 1103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.454336025241869e-07,
+      "loss": 1.123,
+      "step": 1104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.368882931137675e-07,
+      "loss": 1.0953,
+      "step": 1105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.284086061734672e-07,
+      "loss": 1.1141,
+      "step": 1106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.199946005061462e-07,
+      "loss": 1.0916,
+      "step": 1107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116463344591893e-07,
+      "loss": 1.1653,
+      "step": 1108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.033638659241102e-07,
+      "loss": 1.1423,
+      "step": 1109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.951472523361401e-07,
+      "loss": 1.1569,
+      "step": 1110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.869965506738416e-07,
+      "loss": 1.079,
+      "step": 1111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.789118174587048e-07,
+      "loss": 1.1044,
+      "step": 1112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7089310875475967e-07,
+      "loss": 1.1143,
+      "step": 1113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.629404801681803e-07,
+      "loss": 1.1438,
+      "step": 1114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550539868469106e-07,
+      "loss": 1.0748,
+      "step": 1115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4723368348027375e-07,
+      "loss": 1.1078,
+      "step": 1116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.394796242985955e-07,
+      "loss": 1.1889,
+      "step": 1117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.317918630728224e-07,
+      "loss": 1.1545,
+      "step": 1118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.241704531141644e-07,
+      "loss": 1.1793,
+      "step": 1119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.166154472737061e-07,
+      "loss": 1.14,
+      "step": 1120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0912689794205483e-07,
+      "loss": 1.1623,
+      "step": 1121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0170485704896453e-07,
+      "loss": 1.1723,
+      "step": 1122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.943493760629924e-07,
+      "loss": 1.0986,
+      "step": 1123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8706050599112363e-07,
+      "loss": 1.0803,
+      "step": 1124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.798382973784298e-07,
+      "loss": 1.1279,
+      "step": 1125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7268280030771655e-07,
+      "loss": 1.1194,
+      "step": 1126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.655940643991729e-07,
+      "loss": 1.1201,
+      "step": 1127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.585721388100283e-07,
+      "loss": 1.1357,
+      "step": 1128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.516170722342127e-07,
+      "loss": 1.1804,
+      "step": 1129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4472891290201927e-07,
+      "loss": 1.0892,
+      "step": 1130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3790770857976884e-07,
+      "loss": 1.1181,
+      "step": 1131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3115350656948043e-07,
+      "loss": 1.1336,
+      "step": 1132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.24466353708538e-07,
+      "loss": 1.1421,
+      "step": 1133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1784629636937404e-07,
+      "loss": 1.0703,
+      "step": 1134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1129338045914004e-07,
+      "loss": 1.1196,
+      "step": 1135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0480765141939316e-07,
+      "loss": 1.1564,
+      "step": 1136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9838915422577887e-07,
+      "loss": 1.1543,
+      "step": 1137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.920379333877221e-07,
+      "loss": 1.1087,
+      "step": 1138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8575403294811123e-07,
+      "loss": 1.1275,
+      "step": 1139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.795374964830022e-07,
+      "loss": 1.1603,
+      "step": 1140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.733883671013082e-07,
+      "loss": 1.1318,
+      "step": 1141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.673066874445096e-07,
+      "loss": 1.1621,
+      "step": 1142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.612924996863453e-07,
+      "loss": 1.1318,
+      "step": 1143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5534584553253526e-07,
+      "loss": 1.1347,
+      "step": 1144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.494667662204786e-07,
+      "loss": 1.1108,
+      "step": 1145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4365530251897693e-07,
+      "loss": 1.1349,
+      "step": 1146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3791149472794261e-07,
+      "loss": 1.119,
+      "step": 1147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3223538267813317e-07,
+      "loss": 1.1273,
+      "step": 1148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2662700573085505e-07,
+      "loss": 1.1704,
+      "step": 1149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2108640277771153e-07,
+      "loss": 1.1416,
+      "step": 1150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.156136122403163e-07,
+      "loss": 1.1566,
+      "step": 1151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1020867207004026e-07,
+      "loss": 1.0761,
+      "step": 1152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.048716197477374e-07,
+      "loss": 1.0678,
+      "step": 1153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.996024922834916e-07,
+      "loss": 1.1022,
+      "step": 1154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9440132621635687e-07,
+      "loss": 1.107,
+      "step": 1155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8926815761410867e-07,
+      "loss": 1.125,
+      "step": 1156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8420302207298623e-07,
+      "loss": 1.1162,
+      "step": 1157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.792059547174507e-07,
+      "loss": 1.1517,
+      "step": 1158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7427699019994304e-07,
+      "loss": 1.1779,
+      "step": 1159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6941616270063965e-07,
+      "loss": 1.1355,
+      "step": 1160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6462350592721498e-07,
+      "loss": 1.141,
+      "step": 1161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5989905311461274e-07,
+      "loss": 1.0878,
+      "step": 1162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5524283702481158e-07,
+      "loss": 1.1254,
+      "step": 1163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5065488994659983e-07,
+      "loss": 1.16,
+      "step": 1164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.461352436953478e-07,
+      "loss": 1.1375,
+      "step": 1165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4168392961279254e-07,
+      "loss": 1.1231,
+      "step": 1166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3730097856681668e-07,
+      "loss": 1.1206,
+      "step": 1167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3298642095123882e-07,
+      "loss": 1.1146,
+      "step": 1168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2874028668559136e-07,
+      "loss": 1.1339,
+      "step": 1169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.245626052149318e-07,
+      "loss": 1.2054,
+      "step": 1170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2045340550961958e-07,
+      "loss": 1.1727,
+      "step": 1171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.164127160651285e-07,
+      "loss": 1.1806,
+      "step": 1172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1244056490184008e-07,
+      "loss": 1.1502,
+      "step": 1173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0853697956485942e-07,
+      "loss": 1.1204,
+      "step": 1174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0470198712381086e-07,
+      "loss": 1.1163,
+      "step": 1175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.009356141726625e-07,
+      "loss": 1.142,
+      "step": 1176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.723788682953428e-08,
+      "loss": 1.1344,
+      "step": 1177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.360883073652238e-08,
+      "loss": 1.1607,
+      "step": 1178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.004847105951509e-08,
+      "loss": 1.15,
+      "step": 1179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.655683248802282e-08,
+      "loss": 1.1318,
+      "step": 1180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.313393923500613e-08,
+      "loss": 1.1051,
+      "step": 1181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.977981503670795e-08,
+      "loss": 1.1328,
+      "step": 1182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.64944831524872e-08,
+      "loss": 1.1228,
+      "step": 1183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.327796636465767e-08,
+      "loss": 1.1368,
+      "step": 1184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.01302869783338e-08,
+      "loss": 1.1083,
+      "step": 1185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.705146682127184e-08,
+      "loss": 1.1316,
+      "step": 1186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.404152724371892e-08,
+      "loss": 1.1411,
+      "step": 1187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.110048911826871e-08,
+      "loss": 1.121,
+      "step": 1188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.82283728397115e-08,
+      "loss": 1.0985,
+      "step": 1189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.542519832489546e-08,
+      "loss": 1.1362,
+      "step": 1190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.269098501259007e-08,
+      "loss": 1.1123,
+      "step": 1191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.002575186334735e-08,
+      "loss": 1.1148,
+      "step": 1192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.742951735937418e-08,
+      "loss": 1.153,
+      "step": 1193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.490229950440239e-08,
+      "loss": 1.138,
+      "step": 1194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2444115823562226e-08,
+      "loss": 1.1994,
+      "step": 1195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.005498336326463e-08,
+      "loss": 1.0905,
+      "step": 1196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.773491869108137e-08,
+      "loss": 1.0826,
+      "step": 1197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.548393789562732e-08,
+      "loss": 1.0926,
+      "step": 1198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3302056586453916e-08,
+      "loss": 1.2021,
+      "step": 1199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.118928989393699e-08,
+      "loss": 1.134,
+      "step": 1200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9145652469174666e-08,
+      "loss": 1.1044,
+      "step": 1201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7171158483881855e-08,
+      "loss": 1.1576,
+      "step": 1202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5265821630298116e-08,
+      "loss": 1.1418,
+      "step": 1203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3429655121085525e-08,
+      "loss": 1.1132,
+      "step": 1204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1662671689242076e-08,
+      "loss": 1.0855,
+      "step": 1205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9964883588010632e-08,
+      "loss": 1.1696,
+      "step": 1206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8336302590798992e-08,
+      "loss": 1.1172,
+      "step": 1207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.677693999109109e-08,
+      "loss": 1.0926,
+      "step": 1208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5286806602372583e-08,
+      "loss": 1.1428,
+      "step": 1209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3865912758054267e-08,
+      "loss": 1.1577,
+      "step": 1210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2514268311405452e-08,
+      "loss": 1.1354,
+      "step": 1211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1231882635477364e-08,
+      "loss": 1.131,
+      "step": 1212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0018764623044297e-08,
+      "loss": 1.161,
+      "step": 1213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.874922686541442e-09,
+      "loss": 1.1174,
+      "step": 1214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.800364758002721e-09,
+      "loss": 1.0942,
+      "step": 1215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.795098289008595e-09,
+      "loss": 1.1148,
+      "step": 1216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.859130250636113e-09,
+      "loss": 1.1866,
+      "step": 1217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.992467133406731e-09,
+      "loss": 1.1487,
+      "step": 1218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.195114947244117e-09,
+      "loss": 1.0937,
+      "step": 1219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4670792214297476e-09,
+      "loss": 1.1327,
+      "step": 1220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.808365004569602e-09,
+      "loss": 1.1506,
+      "step": 1221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2189768645519693e-09,
+      "loss": 1.1333,
+      "step": 1222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6989188885219165e-09,
+      "loss": 1.1423,
+      "step": 1223
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2481946828502011e-09,
+      "loss": 1.1475,
+      "step": 1224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.668073731088467e-10,
+      "loss": 1.1123,
+      "step": 1225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.547596040489378e-10,
+      "loss": 1.1709,
+      "step": 1226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1205353958285724e-10,
+      "loss": 1.0993,
+      "step": 1227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3869086276985243e-10,
+      "loss": 1.0955,
+      "step": 1228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.467277580271322e-11,
+      "loss": 1.1693,
+      "step": 1229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.0486,
+      "step": 1230
+    },
+    {
+      "epoch": 1.0,
+      "step": 1230,
+      "total_flos": 4.0987686361796444e+19,
+      "train_loss": 1.175328787167867,
+      "train_runtime": 87109.1288,
+      "train_samples_per_second": 1.807,
+      "train_steps_per_second": 0.014
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1230,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 4.0987686361796444e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}