diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8665 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9968782518210197,
+  "global_step": 1440,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5454545454545457e-07,
+      "loss": 0.862,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 0.8906,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 0.8815,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 0.7889,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 0.8425,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 0.8451,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.181818181818182e-06,
+      "loss": 0.9298,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.8441,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0909090909090915e-06,
+      "loss": 0.7971,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.8787,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.7311,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.8228,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.90909090909091e-06,
+      "loss": 0.7754,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 0.8758,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.818181818181818e-06,
+      "loss": 0.7815,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.8514,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.727272727272727e-06,
+      "loss": 0.855,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 0.8809,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.636363636363637e-06,
+      "loss": 0.8084,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.8265,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.545454545454547e-06,
+      "loss": 0.8748,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.8812,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0454545454545455e-05,
+      "loss": 0.8223,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 0.7911,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1363636363636366e-05,
+      "loss": 0.7482,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.181818181818182e-05,
+      "loss": 0.8515,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2272727272727274e-05,
+      "loss": 0.7573,
+      "step": 27
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 0.8365,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3181818181818183e-05,
+      "loss": 0.831,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.9328,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4090909090909092e-05,
+      "loss": 0.8345,
+      "step": 31
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 0.8601,
+      "step": 32
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.8383,
+      "step": 33
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 0.8194,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 0.8306,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 0.8341,
+      "step": 36
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.681818181818182e-05,
+      "loss": 0.861,
+      "step": 37
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 0.8035,
+      "step": 38
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.772727272727273e-05,
+      "loss": 0.9271,
+      "step": 39
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.8661,
+      "step": 40
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8636363636363638e-05,
+      "loss": 0.916,
+      "step": 41
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 0.9532,
+      "step": 42
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9545454545454546e-05,
+      "loss": 0.8439,
+      "step": 43
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.9363,
+      "step": 44
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999974677957704e-05,
+      "loss": 0.8266,
+      "step": 45
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999989871195906e-05,
+      "loss": 0.8091,
+      "step": 46
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999772102388785e-05,
+      "loss": 0.9211,
+      "step": 47
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999594849888083e-05,
+      "loss": 0.878,
+      "step": 48
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999366955354638e-05,
+      "loss": 0.9082,
+      "step": 49
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999088419942598e-05,
+      "loss": 0.7854,
+      "step": 50
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998759245062583e-05,
+      "loss": 0.8015,
+      "step": 51
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999837943238166e-05,
+      "loss": 0.8802,
+      "step": 52
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997948983823363e-05,
+      "loss": 0.9109,
+      "step": 53
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999746790156766e-05,
+      "loss": 0.9564,
+      "step": 54
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996936188050947e-05,
+      "loss": 0.9064,
+      "step": 55
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996353845966033e-05,
+      "loss": 0.8017,
+      "step": 56
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999572087826214e-05,
+      "loss": 0.9049,
+      "step": 57
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999503728814488e-05,
+      "loss": 0.94,
+      "step": 58
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999430307907623e-05,
+      "loss": 0.7948,
+      "step": 59
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993518254774517e-05,
+      "loss": 0.8214,
+      "step": 60
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999268281921442e-05,
+      "loss": 0.9716,
+      "step": 61
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999179677662692e-05,
+      "loss": 0.8468,
+      "step": 62
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990860131499306e-05,
+      "loss": 0.918,
+      "step": 63
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998987288857513e-05,
+      "loss": 0.8961,
+      "step": 64
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988835052854188e-05,
+      "loss": 0.8048,
+      "step": 65
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987746629592506e-05,
+      "loss": 0.9069,
+      "step": 66
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986607624302305e-05,
+      "loss": 0.9128,
+      "step": 67
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985418042751975e-05,
+      "loss": 0.9182,
+      "step": 68
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998417789096604e-05,
+      "loss": 0.9286,
+      "step": 69
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982887175225136e-05,
+      "loss": 0.8841,
+      "step": 70
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981545902065974e-05,
+      "loss": 0.8429,
+      "step": 71
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998015407828131e-05,
+      "loss": 0.8837,
+      "step": 72
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997871171091991e-05,
+      "loss": 0.8242,
+      "step": 73
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977218807286507e-05,
+      "loss": 0.875,
+      "step": 74
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975675374941778e-05,
+      "loss": 0.8317,
+      "step": 75
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974081421702296e-05,
+      "loss": 0.846,
+      "step": 76
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972436955640488e-05,
+      "loss": 0.8815,
+      "step": 77
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99707419850846e-05,
+      "loss": 0.9116,
+      "step": 78
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968996518618664e-05,
+      "loss": 0.9262,
+      "step": 79
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967200565082426e-05,
+      "loss": 0.7856,
+      "step": 80
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965354133571334e-05,
+      "loss": 0.8215,
+      "step": 81
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963457233436468e-05,
+      "loss": 0.8985,
+      "step": 82
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961509874284508e-05,
+      "loss": 0.8959,
+      "step": 83
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959512065977673e-05,
+      "loss": 0.8295,
+      "step": 84
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957463818633683e-05,
+      "loss": 0.9238,
+      "step": 85
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955365142625694e-05,
+      "loss": 0.9243,
+      "step": 86
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995321604858227e-05,
+      "loss": 0.8294,
+      "step": 87
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951016547387286e-05,
+      "loss": 0.8749,
+      "step": 88
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9948766650179924e-05,
+      "loss": 0.9327,
+      "step": 89
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994646636835458e-05,
+      "loss": 0.8618,
+      "step": 90
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944115713560825e-05,
+      "loss": 0.8898,
+      "step": 91
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941714697703333e-05,
+      "loss": 0.822,
+      "step": 92
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993926333294182e-05,
+      "loss": 0.8817,
+      "step": 93
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9936761631691007e-05,
+      "loss": 0.8315,
+      "step": 94
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934209606620532e-05,
+      "loss": 0.9069,
+      "step": 95
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993160727065489e-05,
+      "loss": 0.8181,
+      "step": 96
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928954636973373e-05,
+      "loss": 0.994,
+      "step": 97
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992625171901e-05,
+      "loss": 0.863,
+      "step": 98
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9923498530453456e-05,
+      "loss": 0.9428,
+      "step": 99
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920695085247012e-05,
+      "loss": 0.9404,
+      "step": 100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917841397588453e-05,
+      "loss": 0.9322,
+      "step": 101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991493748193002e-05,
+      "loss": 0.8693,
+      "step": 102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991198335297834e-05,
+      "loss": 0.9035,
+      "step": 103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908979025694312e-05,
+      "loss": 0.8077,
+      "step": 104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990592451529309e-05,
+      "loss": 0.8423,
+      "step": 105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902819837243954e-05,
+      "loss": 0.9348,
+      "step": 106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899665007270266e-05,
+      "loss": 0.8275,
+      "step": 107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989646004134937e-05,
+      "loss": 0.9544,
+      "step": 108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893204955712525e-05,
+      "loss": 0.9553,
+      "step": 109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889899766844817e-05,
+      "loss": 0.8593,
+      "step": 110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886544491485068e-05,
+      "loss": 0.9636,
+      "step": 111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883139146625763e-05,
+      "loss": 0.9157,
+      "step": 112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9879683749512962e-05,
+      "loss": 0.8436,
+      "step": 113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876178317646203e-05,
+      "loss": 0.917,
+      "step": 114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987262286877843e-05,
+      "loss": 0.8816,
+      "step": 115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869017420915888e-05,
+      "loss": 0.9751,
+      "step": 116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865361992318032e-05,
+      "loss": 0.9259,
+      "step": 117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9861656601497452e-05,
+      "loss": 0.9112,
+      "step": 118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857901267219756e-05,
+      "loss": 0.8553,
+      "step": 119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854096008503495e-05,
+      "loss": 0.8797,
+      "step": 120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850240844620046e-05,
+      "loss": 1.0086,
+      "step": 121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846335795093547e-05,
+      "loss": 0.8952,
+      "step": 122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984238087970075e-05,
+      "loss": 0.812,
+      "step": 123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838376118470965e-05,
+      "loss": 0.8831,
+      "step": 124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834321531685944e-05,
+      "loss": 0.9386,
+      "step": 125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830217139879768e-05,
+      "loss": 0.9607,
+      "step": 126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826062963838756e-05,
+      "loss": 0.883,
+      "step": 127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821859024601345e-05,
+      "loss": 0.9199,
+      "step": 128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817605343458008e-05,
+      "loss": 0.9299,
+      "step": 129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981330194195112e-05,
+      "loss": 0.9255,
+      "step": 130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808948841874863e-05,
+      "loss": 1.0294,
+      "step": 131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804546065275116e-05,
+      "loss": 0.9103,
+      "step": 132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980009363444934e-05,
+      "loss": 0.9002,
+      "step": 133
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795591571946454e-05,
+      "loss": 0.9831,
+      "step": 134
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791039900566755e-05,
+      "loss": 0.9152,
+      "step": 135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978643864336176e-05,
+      "loss": 0.9599,
+      "step": 136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.978178782363411e-05,
+      "loss": 0.9653,
+      "step": 137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777087464937464e-05,
+      "loss": 0.9409,
+      "step": 138
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977233759107635e-05,
+      "loss": 0.9401,
+      "step": 139
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767538226106078e-05,
+      "loss": 0.9049,
+      "step": 140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762689394332587e-05,
+      "loss": 0.8994,
+      "step": 141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9757791120312344e-05,
+      "loss": 0.8553,
+      "step": 142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752843428852208e-05,
+      "loss": 0.9076,
+      "step": 143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747846345009306e-05,
+      "loss": 0.9675,
+      "step": 144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974279989409092e-05,
+      "loss": 0.8934,
+      "step": 145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737704101654335e-05,
+      "loss": 0.945,
+      "step": 146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9732558993506723e-05,
+      "loss": 0.8367,
+      "step": 147
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727364595705012e-05,
+      "loss": 0.9107,
+      "step": 148
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972212093455576e-05,
+      "loss": 0.7714,
+      "step": 149
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716828036615006e-05,
+      "loss": 0.9865,
+      "step": 150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711485928688148e-05,
+      "loss": 0.9583,
+      "step": 151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706094637829797e-05,
+      "loss": 0.8547,
+      "step": 152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700654191343662e-05,
+      "loss": 0.949,
+      "step": 153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695164616782378e-05,
+      "loss": 0.8695,
+      "step": 154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689625941947397e-05,
+      "loss": 0.922,
+      "step": 155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684038194888827e-05,
+      "loss": 0.9375,
+      "step": 156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678401403905307e-05,
+      "loss": 0.9839,
+      "step": 157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672715597543845e-05,
+      "loss": 0.9473,
+      "step": 158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966698080459969e-05,
+      "loss": 0.9679,
+      "step": 159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661197054116165e-05,
+      "loss": 0.8366,
+      "step": 160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965536437538456e-05,
+      "loss": 0.9778,
+      "step": 161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964948279794393e-05,
+      "loss": 0.9235,
+      "step": 162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643552351581003e-05,
+      "loss": 0.9808,
+      "step": 163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963757306632996e-05,
+      "loss": 0.9195,
+      "step": 164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631544972472357e-05,
+      "loss": 0.9478,
+      "step": 165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962546810053692e-05,
+      "loss": 0.9075,
+      "step": 166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961934248129941e-05,
+      "loss": 0.9441,
+      "step": 167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613168145782468e-05,
+      "loss": 0.8808,
+      "step": 168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960694512525545e-05,
+      "loss": 0.9136,
+      "step": 169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960067345123427e-05,
+      "loss": 0.9336,
+      "step": 170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959435315548125e-05,
+      "loss": 0.9673,
+      "step": 171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958798427000495e-05,
+      "loss": 0.9764,
+      "step": 172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958156682706001e-05,
+      "loss": 0.8567,
+      "step": 173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575100859146974e-05,
+      "loss": 0.899,
+      "step": 174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568586399012152e-05,
+      "loss": 0.9528,
+      "step": 175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956202347964743e-05,
+      "loss": 0.9559,
+      "step": 176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555412134290103e-05,
+      "loss": 0.9491,
+      "step": 177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954875239642274e-05,
+      "loss": 0.9203,
+      "step": 178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542044299772963e-05,
+      "loss": 0.9215,
+      "step": 179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535287878313315e-05,
+      "loss": 0.9526,
+      "step": 180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952848316626108e-05,
+      "loss": 0.9186,
+      "step": 181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952163019807809e-05,
+      "loss": 0.891,
+      "step": 182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514729008470584e-05,
+      "loss": 0.9222,
+      "step": 183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507779632388997e-05,
+      "loss": 0.9907,
+      "step": 184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9500782105027812e-05,
+      "loss": 0.8791,
+      "step": 185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493736461825366e-05,
+      "loss": 0.9428,
+      "step": 186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486642738463675e-05,
+      "loss": 0.9729,
+      "step": 187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947950097086825e-05,
+      "loss": 0.9668,
+      "step": 188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472311195207918e-05,
+      "loss": 0.972,
+      "step": 189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946507344789464e-05,
+      "loss": 0.8848,
+      "step": 190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457787765583326e-05,
+      "loss": 0.9581,
+      "step": 191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945045418517165e-05,
+      "loss": 0.9263,
+      "step": 192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443072743799855e-05,
+      "loss": 0.885,
+      "step": 193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435643478850573e-05,
+      "loss": 0.9069,
+      "step": 194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9428166427948643e-05,
+      "loss": 0.9419,
+      "step": 195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420641628960897e-05,
+      "loss": 0.9211,
+      "step": 196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413069119995994e-05,
+      "loss": 0.9004,
+      "step": 197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405448939404215e-05,
+      "loss": 0.8916,
+      "step": 198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397781125777265e-05,
+      "loss": 0.8475,
+      "step": 199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390065717948084e-05,
+      "loss": 0.9358,
+      "step": 200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938230275499065e-05,
+      "loss": 0.9872,
+      "step": 201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374492276219776e-05,
+      "loss": 0.944,
+      "step": 202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366634321190914e-05,
+      "loss": 0.8526,
+      "step": 203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358728929699966e-05,
+      "loss": 0.9562,
+      "step": 204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350776141783055e-05,
+      "loss": 1.0946,
+      "step": 205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342775997716357e-05,
+      "loss": 0.9664,
+      "step": 206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933472853801586e-05,
+      "loss": 1.0158,
+      "step": 207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326633803437197e-05,
+      "loss": 0.9521,
+      "step": 208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318491834975396e-05,
+      "loss": 0.8175,
+      "step": 209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310302673864724e-05,
+      "loss": 0.9353,
+      "step": 210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302066361578433e-05,
+      "loss": 0.9757,
+      "step": 211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929378293982857e-05,
+      "loss": 0.8769,
+      "step": 212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285452450565772e-05,
+      "loss": 0.9397,
+      "step": 213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9277074935979034e-05,
+      "loss": 0.9353,
+      "step": 214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9268650438495517e-05,
+      "loss": 0.9003,
+      "step": 215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926017900078031e-05,
+      "loss": 0.8505,
+      "step": 216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925166066573624e-05,
+      "loss": 0.9931,
+      "step": 217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924309547650363e-05,
+      "loss": 0.8249,
+      "step": 218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9234483476460102e-05,
+      "loss": 0.9438,
+      "step": 219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922582470922034e-05,
+      "loss": 0.9678,
+      "step": 220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921711921863588e-05,
+      "loss": 0.9648,
+      "step": 221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208367048794878e-05,
+      "loss": 0.9831,
+      "step": 222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.91995682440219e-05,
+      "loss": 0.9613,
+      "step": 223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190722848877683e-05,
+      "loss": 0.8756,
+      "step": 224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181830908158926e-05,
+      "loss": 0.9365,
+      "step": 225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172892466898047e-05,
+      "loss": 0.9263,
+      "step": 226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163907570362964e-05,
+      "loss": 1.0134,
+      "step": 227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154876264056863e-05,
+      "loss": 0.8728,
+      "step": 228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145798593717963e-05,
+      "loss": 0.8961,
+      "step": 229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136674605319304e-05,
+      "loss": 0.8716,
+      "step": 230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127504345068484e-05,
+      "loss": 1.0111,
+      "step": 231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911828785940745e-05,
+      "loss": 0.9168,
+      "step": 232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9109025195012244e-05,
+      "loss": 0.947,
+      "step": 233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099716398792788e-05,
+      "loss": 1.0074,
+      "step": 234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090361517892623e-05,
+      "loss": 1.0229,
+      "step": 235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908096059968869e-05,
+      "loss": 0.9255,
+      "step": 236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907151369179108e-05,
+      "loss": 0.9936,
+      "step": 237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.906202084204279e-05,
+      "loss": 0.8688,
+      "step": 238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905248209851949e-05,
+      "loss": 0.9057,
+      "step": 239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904289750952928e-05,
+      "loss": 0.9435,
+      "step": 240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903326712361242e-05,
+      "loss": 0.9164,
+      "step": 241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023590989541126e-05,
+      "loss": 0.9033,
+      "step": 242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9013869156319296e-05,
+      "loss": 0.9085,
+      "step": 243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900410167318226e-05,
+      "loss": 0.9262,
+      "step": 244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899428858959654e-05,
+      "loss": 0.8613,
+      "step": 245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8984429955259607e-05,
+      "loss": 0.9152,
+      "step": 246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974525820099608e-05,
+      "loss": 0.8743,
+      "step": 247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8964576234275123e-05,
+      "loss": 0.9574,
+      "step": 248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895458124817493e-05,
+      "loss": 0.9238,
+      "step": 249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894454091241771e-05,
+      "loss": 0.9502,
+      "step": 250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934455277851835e-05,
+      "loss": 1.0451,
+      "step": 251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924324395555066e-05,
+      "loss": 0.9825,
+      "step": 252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891414831683434e-05,
+      "loss": 0.9385,
+      "step": 253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903927093225474e-05,
+      "loss": 0.8589,
+      "step": 254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8893660776492913e-05,
+      "loss": 1.0166,
+      "step": 255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883349418629487e-05,
+      "loss": 0.9775,
+      "step": 256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8872993071856113e-05,
+      "loss": 0.9641,
+      "step": 257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8862591788621572e-05,
+      "loss": 0.974,
+      "step": 258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852145621602208e-05,
+      "loss": 0.9263,
+      "step": 259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841654623701673e-05,
+      "loss": 0.9005,
+      "step": 260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883111884805067e-05,
+      "loss": 0.9284,
+      "step": 261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8820538348006666e-05,
+      "loss": 0.8282,
+      "step": 262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809913177153644e-05,
+      "loss": 1.0034,
+      "step": 263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799243389301796e-05,
+      "loss": 0.9435,
+      "step": 264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788529038487297e-05,
+      "loss": 0.8977,
+      "step": 265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877777017897199e-05,
+      "loss": 0.8815,
+      "step": 266
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8766966865243138e-05,
+      "loss": 0.9596,
+      "step": 267
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756119152013134e-05,
+      "loss": 0.9779,
+      "step": 268
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874522709421922e-05,
+      "loss": 0.8715,
+      "step": 269
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873429074702324e-05,
+      "loss": 0.836,
+      "step": 270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872331016581131e-05,
+      "loss": 0.9368,
+      "step": 271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8712285406193585e-05,
+      "loss": 0.9233,
+      "step": 272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701216524003956e-05,
+      "loss": 1.0079,
+      "step": 273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8690103575299754e-05,
+      "loss": 0.9296,
+      "step": 274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678946616361492e-05,
+      "loss": 0.9434,
+      "step": 275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866774570369257e-05,
+      "loss": 0.986,
+      "step": 276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656500894018986e-05,
+      "loss": 0.9588,
+      "step": 277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645212244289047e-05,
+      "loss": 0.864,
+      "step": 278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633879811673086e-05,
+      "loss": 0.9181,
+      "step": 279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8622503653563173e-05,
+      "loss": 0.8731,
+      "step": 280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8611083827572818e-05,
+      "loss": 0.8241,
+      "step": 281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599620391536682e-05,
+      "loss": 0.9319,
+      "step": 282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858811340351029e-05,
+      "loss": 0.9154,
+      "step": 283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8576562921769727e-05,
+      "loss": 0.9902,
+      "step": 284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564969004811355e-05,
+      "loss": 0.8483,
+      "step": 285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8553331711351502e-05,
+      "loss": 0.9915,
+      "step": 286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8541651100326175e-05,
+      "loss": 0.9597,
+      "step": 287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529927230890757e-05,
+      "loss": 1.011,
+      "step": 288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8518160162419722e-05,
+      "loss": 0.8831,
+      "step": 289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85063499545063e-05,
+      "loss": 0.9561,
+      "step": 290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849449666696221e-05,
+      "loss": 0.9443,
+      "step": 291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8482600359817344e-05,
+      "loss": 0.9633,
+      "step": 292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470661093319462e-05,
+      "loss": 0.9759,
+      "step": 293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8458678927933884e-05,
+      "loss": 0.9752,
+      "step": 294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446653924343188e-05,
+      "loss": 0.9684,
+      "step": 295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843458614344691e-05,
+      "loss": 0.8356,
+      "step": 296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842247564636121e-05,
+      "loss": 1.0318,
+      "step": 297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8410322494418606e-05,
+      "loss": 0.9808,
+      "step": 298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8398126749167615e-05,
+      "loss": 0.873,
+      "step": 299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8385888472372474e-05,
+      "loss": 0.9989,
+      "step": 300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373607726012816e-05,
+      "loss": 0.961,
+      "step": 301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361284572283356e-05,
+      "loss": 0.978,
+      "step": 302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834891907359358e-05,
+      "loss": 0.9402,
+      "step": 303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833651129256742e-05,
+      "loss": 0.9715,
+      "step": 304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8324061292042953e-05,
+      "loss": 0.9632,
+      "step": 305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831156913507206e-05,
+      "loss": 0.9161,
+      "step": 306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829903488492013e-05,
+      "loss": 0.9706,
+      "step": 307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828645860506573e-05,
+      "loss": 0.9416,
+      "step": 308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8273840359200273e-05,
+      "loss": 0.8993,
+      "step": 309
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.826118021122771e-05,
+      "loss": 0.8861,
+      "step": 310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82484782252642e-05,
+      "loss": 0.8936,
+      "step": 311
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8235734465637794e-05,
+      "loss": 1.0163,
+      "step": 312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822294899688809e-05,
+      "loss": 0.9154,
+      "step": 313
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821012188376593e-05,
+      "loss": 0.9331,
+      "step": 314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8197253191233054e-05,
+      "loss": 1.0286,
+      "step": 315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8184342984461766e-05,
+      "loss": 0.8369,
+      "step": 316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817139132883464e-05,
+      "loss": 1.0157,
+      "step": 317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8158398289944145e-05,
+      "loss": 0.9841,
+      "step": 318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8145363933592338e-05,
+      "loss": 0.9957,
+      "step": 319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132288325790518e-05,
+      "loss": 0.9361,
+      "step": 320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811917153275891e-05,
+      "loss": 0.9471,
+      "step": 321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8106013620926312e-05,
+      "loss": 0.9691,
+      "step": 322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809281465692976e-05,
+      "loss": 1.0162,
+      "step": 323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8079574707614202e-05,
+      "loss": 0.8634,
+      "step": 324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8066293840032146e-05,
+      "loss": 1.0123,
+      "step": 325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8052972121443337e-05,
+      "loss": 0.9297,
+      "step": 326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803960961931439e-05,
+      "loss": 0.8908,
+      "step": 327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802620640131848e-05,
+      "loss": 1.0379,
+      "step": 328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8012762535334978e-05,
+      "loss": 0.8826,
+      "step": 329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799927808944911e-05,
+      "loss": 1.0062,
+      "step": 330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7985753131951617e-05,
+      "loss": 0.9014,
+      "step": 331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797218773133841e-05,
+      "loss": 0.9044,
+      "step": 332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795858195631022e-05,
+      "loss": 0.9804,
+      "step": 333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7944935875772244e-05,
+      "loss": 0.9798,
+      "step": 334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931249558833815e-05,
+      "loss": 0.9973,
+      "step": 335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7917523074808024e-05,
+      "loss": 0.9562,
+      "step": 336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79037564932114e-05,
+      "loss": 0.9356,
+      "step": 337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7889949883763532e-05,
+      "loss": 0.9679,
+      "step": 338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787610331638673e-05,
+      "loss": 0.927,
+      "step": 339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786221686120567e-05,
+      "loss": 0.8715,
+      "step": 340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784829058854703e-05,
+      "loss": 0.9505,
+      "step": 341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7834324568939137e-05,
+      "loss": 0.8715,
+      "step": 342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7820318873111627e-05,
+      "loss": 0.9385,
+      "step": 343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7806273571995066e-05,
+      "loss": 0.9108,
+      "step": 344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792188736720593e-05,
+      "loss": 0.9772,
+      "step": 345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7778064438619562e-05,
+      "loss": 0.9675,
+      "step": 346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7763900749223194e-05,
+      "loss": 1.0191,
+      "step": 347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7749697740262197e-05,
+      "loss": 0.8954,
+      "step": 348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735455483666408e-05,
+      "loss": 0.9897,
+      "step": 349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772117405156443e-05,
+      "loss": 0.9625,
+      "step": 350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770685351628327e-05,
+      "loss": 0.9269,
+      "step": 351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769249395034797e-05,
+      "loss": 0.8875,
+      "step": 352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678095426481237e-05,
+      "loss": 0.9461,
+      "step": 353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7663658017603073e-05,
+      "loss": 1.0047,
+      "step": 354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649181796830418e-05,
+      "loss": 0.9555,
+      "step": 355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7634666837476765e-05,
+      "loss": 0.9941,
+      "step": 356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7620113213051798e-05,
+      "loss": 0.8917,
+      "step": 357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7605520997261014e-05,
+      "loss": 0.9091,
+      "step": 358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7590890264005357e-05,
+      "loss": 0.8904,
+      "step": 359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757622108738083e-05,
+      "loss": 0.9379,
+      "step": 360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7561513541678142e-05,
+      "loss": 0.9434,
+      "step": 361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754676770138231e-05,
+      "loss": 0.9773,
+      "step": 362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7531983641172294e-05,
+      "loss": 0.9735,
+      "step": 363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7517161435920606e-05,
+      "loss": 0.9642,
+      "step": 364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7502301160692954e-05,
+      "loss": 0.8944,
+      "step": 365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7487402890747843e-05,
+      "loss": 0.918,
+      "step": 366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7472466701536193e-05,
+      "loss": 0.9872,
+      "step": 367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457492668700967e-05,
+      "loss": 0.9949,
+      "step": 368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744248086807679e-05,
+      "loss": 0.8549,
+      "step": 369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7427431375689544e-05,
+      "loss": 0.9385,
+      "step": 370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741234426775601e-05,
+      "loss": 1.1029,
+      "step": 371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397219620683465e-05,
+      "loss": 0.9848,
+      "step": 372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73820575110693e-05,
+      "loss": 0.9114,
+      "step": 373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7366858015700626e-05,
+      "loss": 1.0012,
+      "step": 374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73516212115539e-05,
+      "loss": 0.9107,
+      "step": 375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336347175794523e-05,
+      "loss": 0.9813,
+      "step": 376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732103598577645e-05,
+      "loss": 0.9791,
+      "step": 377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.73056877190418e-05,
+      "loss": 0.8552,
+      "step": 378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7290302453320468e-05,
+      "loss": 1.006,
+      "step": 379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7274880266529716e-05,
+      "loss": 0.9354,
+      "step": 380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7259421236773807e-05,
+      "loss": 1.0066,
+      "step": 381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7243925442343578e-05,
+      "loss": 0.9847,
+      "step": 382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722839296171606e-05,
+      "loss": 0.8935,
+      "step": 383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721282387355408e-05,
+      "loss": 0.873,
+      "step": 384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719721825670586e-05,
+      "loss": 0.9384,
+      "step": 385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718157619020462e-05,
+      "loss": 0.8872,
+      "step": 386
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7165897753268168e-05,
+      "loss": 0.9777,
+      "step": 387
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715018302529852e-05,
+      "loss": 0.9095,
+      "step": 388
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713443208588147e-05,
+      "loss": 0.9033,
+      "step": 389
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711864501478622e-05,
+      "loss": 0.9215,
+      "step": 390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7102821891964937e-05,
+      "loss": 0.9063,
+      "step": 391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7086962797552376e-05,
+      "loss": 0.9396,
+      "step": 392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.859,
+      "step": 393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7055137015402935e-05,
+      "loss": 0.9462,
+      "step": 394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703917048884481e-05,
+      "loss": 1.0144,
+      "step": 395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7023168313052118e-05,
+      "loss": 0.8752,
+      "step": 396
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700713056906641e-05,
+      "loss": 0.9419,
+      "step": 397
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6991057338109376e-05,
+      "loss": 0.9295,
+      "step": 398
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697494870158242e-05,
+      "loss": 0.9291,
+      "step": 399
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6958804741066254e-05,
+      "loss": 0.9928,
+      "step": 400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6942625538320495e-05,
+      "loss": 0.9423,
+      "step": 401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6926411175283227e-05,
+      "loss": 1.002,
+      "step": 402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6910161734070614e-05,
+      "loss": 0.9062,
+      "step": 403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689387729697646e-05,
+      "loss": 0.9349,
+      "step": 404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6877557946471808e-05,
+      "loss": 0.8808,
+      "step": 405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686120376520451e-05,
+      "loss": 0.9542,
+      "step": 406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6844814835998825e-05,
+      "loss": 0.9387,
+      "step": 407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6828391241854983e-05,
+      "loss": 0.8898,
+      "step": 408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6811933065948777e-05,
+      "loss": 0.9248,
+      "step": 409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6795440391631122e-05,
+      "loss": 0.9436,
+      "step": 410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6778913302427668e-05,
+      "loss": 0.8674,
+      "step": 411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6762351882038342e-05,
+      "loss": 0.9529,
+      "step": 412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6745756214336938e-05,
+      "loss": 0.8856,
+      "step": 413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6729126383370696e-05,
+      "loss": 0.9027,
+      "step": 414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671246247335988e-05,
+      "loss": 0.9386,
+      "step": 415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669576456869733e-05,
+      "loss": 0.8582,
+      "step": 416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6679032753948056e-05,
+      "loss": 0.9375,
+      "step": 417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666226711384881e-05,
+      "loss": 0.8675,
+      "step": 418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664546773330763e-05,
+      "loss": 1.0055,
+      "step": 419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6628634697403447e-05,
+      "loss": 0.9511,
+      "step": 420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661176809138563e-05,
+      "loss": 1.0775,
+      "step": 421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6594868000673562e-05,
+      "loss": 0.9636,
+      "step": 422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6577934510856205e-05,
+      "loss": 1.0389,
+      "step": 423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560967707691663e-05,
+      "loss": 1.0259,
+      "step": 424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6543967677106767e-05,
+      "loss": 0.9391,
+      "step": 425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6526934505196605e-05,
+      "loss": 0.9565,
+      "step": 426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6509868278224125e-05,
+      "loss": 1.0194,
+      "step": 427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649276908261967e-05,
+      "loss": 0.9288,
+      "step": 428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475637004980547e-05,
+      "loss": 0.9685,
+      "step": 429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64584721320706e-05,
+      "loss": 0.9009,
+      "step": 430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6441274550819756e-05,
+      "loss": 0.9512,
+      "step": 431
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642404434832358e-05,
+      "loss": 0.9748,
+      "step": 432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6406781611842874e-05,
+      "loss": 0.9136,
+      "step": 433
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6389486428803173e-05,
+      "loss": 0.9356,
+      "step": 434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372158886794348e-05,
+      "loss": 0.925,
+      "step": 435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635479907357016e-05,
+      "loss": 0.9704,
+      "step": 436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6337407077047783e-05,
+      "loss": 0.9933,
+      "step": 437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63199829853074e-05,
+      "loss": 0.9215,
+      "step": 438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630252688659172e-05,
+      "loss": 0.9324,
+      "step": 439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285038869305565e-05,
+      "loss": 0.9964,
+      "step": 440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6267519022015394e-05,
+      "loss": 0.9284,
+      "step": 441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624996743344887e-05,
+      "loss": 0.9563,
+      "step": 442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6232384192494404e-05,
+      "loss": 0.8989,
+      "step": 443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621476938820071e-05,
+      "loss": 0.9737,
+      "step": 444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6197123109776362e-05,
+      "loss": 0.9878,
+      "step": 445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6179445446589308e-05,
+      "loss": 1.0366,
+      "step": 446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161736488166463e-05,
+      "loss": 0.9869,
+      "step": 447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6143996324193227e-05,
+      "loss": 0.9227,
+      "step": 448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612622504451303e-05,
+      "loss": 1.0251,
+      "step": 449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6108422739126896e-05,
+      "loss": 0.9149,
+      "step": 450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609058949819297e-05,
+      "loss": 0.9265,
+      "step": 451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6072725412026066e-05,
+      "loss": 0.9905,
+      "step": 452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605483057109722e-05,
+      "loss": 0.9826,
+      "step": 453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6036905066033207e-05,
+      "loss": 0.9663,
+      "step": 454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6018948987616108e-05,
+      "loss": 0.9652,
+      "step": 455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6000962426782844e-05,
+      "loss": 0.9741,
+      "step": 456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5982945474624702e-05,
+      "loss": 0.93,
+      "step": 457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596489822238689e-05,
+      "loss": 0.9246,
+      "step": 458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594682076146806e-05,
+      "loss": 0.9065,
+      "step": 459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592871318341986e-05,
+      "loss": 0.9947,
+      "step": 460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5910575579946462e-05,
+      "loss": 0.9181,
+      "step": 461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5892408042904098e-05,
+      "loss": 0.9076,
+      "step": 462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.58742106643006e-05,
+      "loss": 0.9491,
+      "step": 463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585598353629492e-05,
+      "loss": 0.9746,
+      "step": 464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583772675119668e-05,
+      "loss": 0.9725,
+      "step": 465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58194404014657e-05,
+      "loss": 1.1093,
+      "step": 466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5801124579711527e-05,
+      "loss": 0.912,
+      "step": 467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5782779378692957e-05,
+      "loss": 0.9891,
+      "step": 468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5764404891317583e-05,
+      "loss": 0.9395,
+      "step": 469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5746001210641316e-05,
+      "loss": 0.9278,
+      "step": 470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5727568429867914e-05,
+      "loss": 1.0108,
+      "step": 471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57091066423485e-05,
+      "loss": 0.8841,
+      "step": 472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569061594158112e-05,
+      "loss": 0.9375,
+      "step": 473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5672096421210217e-05,
+      "loss": 0.8543,
+      "step": 474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5653548175026226e-05,
+      "loss": 0.9097,
+      "step": 475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5634971296965027e-05,
+      "loss": 0.9258,
+      "step": 476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561636588110753e-05,
+      "loss": 0.9161,
+      "step": 477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5597732021679153e-05,
+      "loss": 0.8855,
+      "step": 478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557906981304937e-05,
+      "loss": 0.9972,
+      "step": 479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5560379349731234e-05,
+      "loss": 1.0628,
+      "step": 480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5541660726380884e-05,
+      "loss": 0.7049,
+      "step": 481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552291403779707e-05,
+      "loss": 0.4825,
+      "step": 482
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550413937892069e-05,
+      "loss": 0.4481,
+      "step": 483
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5485336844834274e-05,
+      "loss": 0.4417,
+      "step": 484
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546650653076154e-05,
+      "loss": 0.5073,
+      "step": 485
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544764853206689e-05,
+      "loss": 0.4679,
+      "step": 486
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542876294425493e-05,
+      "loss": 0.4235,
+      "step": 487
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5409849862969994e-05,
+      "loss": 0.472,
+      "step": 488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5390909383995648e-05,
+      "loss": 0.4321,
+      "step": 489
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5371941603254215e-05,
+      "loss": 0.4257,
+      "step": 490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5352946616806284e-05,
+      "loss": 0.397,
+      "step": 491
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5333924520850227e-05,
+      "loss": 0.4081,
+      "step": 492
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5314875411721705e-05,
+      "loss": 0.4274,
+      "step": 493
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529579938589319e-05,
+      "loss": 0.4633,
+      "step": 494
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5276696539973466e-05,
+      "loss": 0.4458,
+      "step": 495
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5257566970707147e-05,
+      "loss": 0.4029,
+      "step": 496
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5238410774974188e-05,
+      "loss": 0.3795,
+      "step": 497
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5219228049789388e-05,
+      "loss": 0.3949,
+      "step": 498
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5200018892301903e-05,
+      "loss": 0.4275,
+      "step": 499
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5180783399794749e-05,
+      "loss": 0.386,
+      "step": 500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5161521669684324e-05,
+      "loss": 0.4296,
+      "step": 501
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514223379951989e-05,
+      "loss": 0.3951,
+      "step": 502
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5122919886983103e-05,
+      "loss": 0.3857,
+      "step": 503
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103580029887504e-05,
+      "loss": 0.42,
+      "step": 504
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084214326178033e-05,
+      "loss": 0.4164,
+      "step": 505
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5064822873930516e-05,
+      "loss": 0.3741,
+      "step": 506
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5045405771351194e-05,
+      "loss": 0.4453,
+      "step": 507
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5025963116776203e-05,
+      "loss": 0.4425,
+      "step": 508
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5006495008671089e-05,
+      "loss": 0.418,
+      "step": 509
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.49870015456303e-05,
+      "loss": 0.4239,
+      "step": 510
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4967482826376698e-05,
+      "loss": 0.3721,
+      "step": 511
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4947938949761054e-05,
+      "loss": 0.4399,
+      "step": 512
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4928370014761533e-05,
+      "loss": 0.4596,
+      "step": 513
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490877612048322e-05,
+      "loss": 0.4163,
+      "step": 514
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4889157366157597e-05,
+      "loss": 0.4167,
+      "step": 515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4869513851142051e-05,
+      "loss": 0.4211,
+      "step": 516
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4849845674919365e-05,
+      "loss": 0.3894,
+      "step": 517
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.483015293709722e-05,
+      "loss": 0.3833,
+      "step": 518
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4810435737407679e-05,
+      "loss": 0.4314,
+      "step": 519
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790694175706698e-05,
+      "loss": 0.4456,
+      "step": 520
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4770928351973605e-05,
+      "loss": 0.4001,
+      "step": 521
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4751138366310612e-05,
+      "loss": 0.4039,
+      "step": 522
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4731324318942284e-05,
+      "loss": 0.3601,
+      "step": 523
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4711486310215053e-05,
+      "loss": 0.4098,
+      "step": 524
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4691624440596699e-05,
+      "loss": 0.4034,
+      "step": 525
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4671738810675838e-05,
+      "loss": 0.4215,
+      "step": 526
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4651829521161424e-05,
+      "loss": 0.4936,
+      "step": 527
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4631896672882235e-05,
+      "loss": 0.4182,
+      "step": 528
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4611940366786353e-05,
+      "loss": 0.3938,
+      "step": 529
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4591960703940662e-05,
+      "loss": 0.4405,
+      "step": 530
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4571957785530342e-05,
+      "loss": 0.415,
+      "step": 531
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4551931712858334e-05,
+      "loss": 0.4081,
+      "step": 532
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453188258734486e-05,
+      "loss": 0.4478,
+      "step": 533
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4511810510526869e-05,
+      "loss": 0.3728,
+      "step": 534
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4491715584057562e-05,
+      "loss": 0.3363,
+      "step": 535
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4471597909705858e-05,
+      "loss": 0.4494,
+      "step": 536
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4451457589355873e-05,
+      "loss": 0.4371,
+      "step": 537
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4431294725006415e-05,
+      "loss": 0.4261,
+      "step": 538
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4411109418770468e-05,
+      "loss": 0.4078,
+      "step": 539
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4390901772874668e-05,
+      "loss": 0.3529,
+      "step": 540
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4370671889658783e-05,
+      "loss": 0.4159,
+      "step": 541
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435041987157521e-05,
+      "loss": 0.4073,
+      "step": 542
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4330145821188437e-05,
+      "loss": 0.4827,
+      "step": 543
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4309849841174538e-05,
+      "loss": 0.4802,
+      "step": 544
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4289532034320649e-05,
+      "loss": 0.4123,
+      "step": 545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4269192503524435e-05,
+      "loss": 0.4248,
+      "step": 546
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4248831351793593e-05,
+      "loss": 0.4399,
+      "step": 547
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422844868224531e-05,
+      "loss": 0.4324,
+      "step": 548
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4208044598105755e-05,
+      "loss": 0.3905,
+      "step": 549
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4187619202709538e-05,
+      "loss": 0.422,
+      "step": 550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4167172599499209e-05,
+      "loss": 0.4109,
+      "step": 551
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4146704892024714e-05,
+      "loss": 0.3866,
+      "step": 552
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126216183942889e-05,
+      "loss": 0.3672,
+      "step": 553
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4105706579016916e-05,
+      "loss": 0.3688,
+      "step": 554
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4085176181115812e-05,
+      "loss": 0.4385,
+      "step": 555
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40646250942139e-05,
+      "loss": 0.3453,
+      "step": 556
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.404405342239028e-05,
+      "loss": 0.4604,
+      "step": 557
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4023461269828297e-05,
+      "loss": 0.4166,
+      "step": 558
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4002848740815023e-05,
+      "loss": 0.4709,
+      "step": 559
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3982215939740726e-05,
+      "loss": 0.3525,
+      "step": 560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3961562971098337e-05,
+      "loss": 0.4314,
+      "step": 561
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3940889939482925e-05,
+      "loss": 0.4369,
+      "step": 562
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3920196949591166e-05,
+      "loss": 0.3846,
+      "step": 563
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3899484106220816e-05,
+      "loss": 0.4244,
+      "step": 564
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387875151427017e-05,
+      "loss": 0.3922,
+      "step": 565
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3857999278737546e-05,
+      "loss": 0.4203,
+      "step": 566
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3837227504720739e-05,
+      "loss": 0.4391,
+      "step": 567
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3816436297416496e-05,
+      "loss": 0.3467,
+      "step": 568
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3795625762119987e-05,
+      "loss": 0.3743,
+      "step": 569
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3774796004224258e-05,
+      "loss": 0.3716,
+      "step": 570
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753947129219713e-05,
+      "loss": 0.4121,
+      "step": 571
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3733079242693572e-05,
+      "loss": 0.331,
+      "step": 572
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3712192450329338e-05,
+      "loss": 0.4222,
+      "step": 573
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3691286857906254e-05,
+      "loss": 0.3869,
+      "step": 574
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3670362571298781e-05,
+      "loss": 0.3986,
+      "step": 575
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3649419696476057e-05,
+      "loss": 0.4474,
+      "step": 576
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362845833950135e-05,
+      "loss": 0.4791,
+      "step": 577
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3607478606531533e-05,
+      "loss": 0.4392,
+      "step": 578
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3586480603816545e-05,
+      "loss": 0.3602,
+      "step": 579
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356546443769885e-05,
+      "loss": 0.4234,
+      "step": 580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3544430214612896e-05,
+      "loss": 0.4063,
+      "step": 581
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3523378041084576e-05,
+      "loss": 0.4235,
+      "step": 582
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502308023730703e-05,
+      "loss": 0.3335,
+      "step": 583
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3481220269258449e-05,
+      "loss": 0.3894,
+      "step": 584
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3460114884464814e-05,
+      "loss": 0.4579,
+      "step": 585
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3438991976236087e-05,
+      "loss": 0.4406,
+      "step": 586
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417851651547306e-05,
+      "loss": 0.46,
+      "step": 587
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3396694017461708e-05,
+      "loss": 0.4724,
+      "step": 588
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3375519181130194e-05,
+      "loss": 0.4395,
+      "step": 589
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3354327249790786e-05,
+      "loss": 0.3886,
+      "step": 590
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3333118330768082e-05,
+      "loss": 0.4184,
+      "step": 591
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3311892531472705e-05,
+      "loss": 0.4415,
+      "step": 592
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3290649959400777e-05,
+      "loss": 0.4539,
+      "step": 593
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3269390722133358e-05,
+      "loss": 0.4399,
+      "step": 594
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3248114927335909e-05,
+      "loss": 0.3937,
+      "step": 595
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3226822682757745e-05,
+      "loss": 0.3891,
+      "step": 596
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320551409623149e-05,
+      "loss": 0.3631,
+      "step": 597
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3184189275672532e-05,
+      "loss": 0.4561,
+      "step": 598
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316284832907847e-05,
+      "loss": 0.3919,
+      "step": 599
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3141491364528576e-05,
+      "loss": 0.4112,
+      "step": 600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3120118490183238e-05,
+      "loss": 0.3782,
+      "step": 601
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3098729814283426e-05,
+      "loss": 0.4207,
+      "step": 602
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3077325445150133e-05,
+      "loss": 0.3905,
+      "step": 603
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055905491183822e-05,
+      "loss": 0.4208,
+      "step": 604
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3034470060863891e-05,
+      "loss": 0.3682,
+      "step": 605
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3013019262748112e-05,
+      "loss": 0.4489,
+      "step": 606
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2991553205472091e-05,
+      "loss": 0.4073,
+      "step": 607
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2970071997748712e-05,
+      "loss": 0.4442,
+      "step": 608
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2948575748367585e-05,
+      "loss": 0.4569,
+      "step": 609
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2927064566194493e-05,
+      "loss": 0.3878,
+      "step": 610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2905538560170855e-05,
+      "loss": 0.4247,
+      "step": 611
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288399783931315e-05,
+      "loss": 0.3893,
+      "step": 612
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2862442512712394e-05,
+      "loss": 0.3758,
+      "step": 613
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2840872689533562e-05,
+      "loss": 0.4214,
+      "step": 614
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819288479015048e-05,
+      "loss": 0.3903,
+      "step": 615
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2797689990468113e-05,
+      "loss": 0.3964,
+      "step": 616
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2776077333276326e-05,
+      "loss": 0.4012,
+      "step": 617
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2754450616895006e-05,
+      "loss": 0.3732,
+      "step": 618
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2732809950850685e-05,
+      "loss": 0.4625,
+      "step": 619
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2711155444740529e-05,
+      "loss": 0.4015,
+      "step": 620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2689487208231805e-05,
+      "loss": 0.4655,
+      "step": 621
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2667805351061314e-05,
+      "loss": 0.4486,
+      "step": 622
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2646109983034833e-05,
+      "loss": 0.3759,
+      "step": 623
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2624401214026574e-05,
+      "loss": 0.4325,
+      "step": 624
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2602679153978602e-05,
+      "loss": 0.4113,
+      "step": 625
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2580943912900309e-05,
+      "loss": 0.414,
+      "step": 626
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.255919560086783e-05,
+      "loss": 0.4371,
+      "step": 627
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2537434328023501e-05,
+      "loss": 0.3822,
+      "step": 628
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515660204575298e-05,
+      "loss": 0.4596,
+      "step": 629
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2493873340796271e-05,
+      "loss": 0.4196,
+      "step": 630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2472073847024004e-05,
+      "loss": 0.4373,
+      "step": 631
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2450261833660033e-05,
+      "loss": 0.4279,
+      "step": 632
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2428437411169306e-05,
+      "loss": 0.4424,
+      "step": 633
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2406600690079608e-05,
+      "loss": 0.4165,
+      "step": 634
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238475178098102e-05,
+      "loss": 0.4107,
+      "step": 635
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2362890794525342e-05,
+      "loss": 0.3784,
+      "step": 636
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.234101784142553e-05,
+      "loss": 0.4449,
+      "step": 637
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2319133032455164e-05,
+      "loss": 0.3852,
+      "step": 638
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2297236478447848e-05,
+      "loss": 0.4102,
+      "step": 639
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2275328290296677e-05,
+      "loss": 0.3537,
+      "step": 640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2253408578953667e-05,
+      "loss": 0.3405,
+      "step": 641
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2231477455429185e-05,
+      "loss": 0.4722,
+      "step": 642
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2209535030791404e-05,
+      "loss": 0.4768,
+      "step": 643
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2187581416165721e-05,
+      "loss": 0.3828,
+      "step": 644
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2165616722734211e-05,
+      "loss": 0.4437,
+      "step": 645
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2143641061735048e-05,
+      "loss": 0.4225,
+      "step": 646
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212165454446196e-05,
+      "loss": 0.4594,
+      "step": 647
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2099657282263651e-05,
+      "loss": 0.4256,
+      "step": 648
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2077649386543238e-05,
+      "loss": 0.4465,
+      "step": 649
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2055630968757696e-05,
+      "loss": 0.4011,
+      "step": 650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2033602140417288e-05,
+      "loss": 0.3896,
+      "step": 651
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2011563013084996e-05,
+      "loss": 0.395,
+      "step": 652
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1989513698375968e-05,
+      "loss": 0.4412,
+      "step": 653
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967454307956933e-05,
+      "loss": 0.4305,
+      "step": 654
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194538495354566e-05,
+      "loss": 0.4026,
+      "step": 655
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1923305746910372e-05,
+      "loss": 0.3974,
+      "step": 656
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1901216799869188e-05,
+      "loss": 0.4072,
+      "step": 657
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1879118224289563e-05,
+      "loss": 0.4573,
+      "step": 658
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1857010132087704e-05,
+      "loss": 0.4181,
+      "step": 659
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1834892635228024e-05,
+      "loss": 0.4152,
+      "step": 660
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812765845722561e-05,
+      "loss": 0.3975,
+      "step": 661
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1790629875630412e-05,
+      "loss": 0.3084,
+      "step": 662
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1768484837057176e-05,
+      "loss": 0.4854,
+      "step": 663
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1746330842154371e-05,
+      "loss": 0.4309,
+      "step": 664
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1724168003118876e-05,
+      "loss": 0.4201,
+      "step": 665
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1701996432192363e-05,
+      "loss": 0.376,
+      "step": 666
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1679816241660719e-05,
+      "loss": 0.4495,
+      "step": 667
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1657627543853491e-05,
+      "loss": 0.3998,
+      "step": 668
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1635430451143308e-05,
+      "loss": 0.3632,
+      "step": 669
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1613225075945316e-05,
+      "loss": 0.4364,
+      "step": 670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.15910115307166e-05,
+      "loss": 0.4696,
+      "step": 671
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156878992795563e-05,
+      "loss": 0.4335,
+      "step": 672
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154656038020168e-05,
+      "loss": 0.4057,
+      "step": 673
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1524323000034256e-05,
+      "loss": 0.4275,
+      "step": 674
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1502077900072534e-05,
+      "loss": 0.3872,
+      "step": 675
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1479825192974791e-05,
+      "loss": 0.4417,
+      "step": 676
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1457564991437825e-05,
+      "loss": 0.4162,
+      "step": 677
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1435297408196382e-05,
+      "loss": 0.3493,
+      "step": 678
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1413022556022606e-05,
+      "loss": 0.4046,
+      "step": 679
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1390740547725443e-05,
+      "loss": 0.371,
+      "step": 680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1368451496150088e-05,
+      "loss": 0.424,
+      "step": 681
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13461555141774e-05,
+      "loss": 0.3573,
+      "step": 682
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1323852714723335e-05,
+      "loss": 0.4638,
+      "step": 683
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1301543210738383e-05,
+      "loss": 0.3928,
+      "step": 684
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1279227115206988e-05,
+      "loss": 0.3628,
+      "step": 685
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1256904541146966e-05,
+      "loss": 0.4568,
+      "step": 686
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1234575601608956e-05,
+      "loss": 0.3592,
+      "step": 687
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1212240409675825e-05,
+      "loss": 0.3669,
+      "step": 688
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1189899078462107e-05,
+      "loss": 0.4385,
+      "step": 689
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1167551721113435e-05,
+      "loss": 0.3741,
+      "step": 690
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1145198450805948e-05,
+      "loss": 0.4703,
+      "step": 691
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1122839380745738e-05,
+      "loss": 0.4382,
+      "step": 692
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1100474624168269e-05,
+      "loss": 0.3887,
+      "step": 693
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1078104294337806e-05,
+      "loss": 0.3969,
+      "step": 694
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1055728504546835e-05,
+      "loss": 0.3652,
+      "step": 695
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1033347368115494e-05,
+      "loss": 0.3843,
+      "step": 696
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1010960998391003e-05,
+      "loss": 0.4162,
+      "step": 697
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0988569508747075e-05,
+      "loss": 0.3777,
+      "step": 698
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0966173012583367e-05,
+      "loss": 0.3709,
+      "step": 699
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0943771623324884e-05,
+      "loss": 0.4504,
+      "step": 700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0921365454421404e-05,
+      "loss": 0.374,
+      "step": 701
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0898954619346924e-05,
+      "loss": 0.3568,
+      "step": 702
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0876539231599069e-05,
+      "loss": 0.4537,
+      "step": 703
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.085411940469851e-05,
+      "loss": 0.3287,
+      "step": 704
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0831695252188415e-05,
+      "loss": 0.4042,
+      "step": 705
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809266887633849e-05,
+      "loss": 0.3993,
+      "step": 706
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.078683442462121e-05,
+      "loss": 0.409,
+      "step": 707
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0764397976757658e-05,
+      "loss": 0.3566,
+      "step": 708
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0741957657670523e-05,
+      "loss": 0.4301,
+      "step": 709
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0719513581006751e-05,
+      "loss": 0.3776,
+      "step": 710
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0697065860432314e-05,
+      "loss": 0.3645,
+      "step": 711
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0674614609631634e-05,
+      "loss": 0.3891,
+      "step": 712
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.065215994230702e-05,
+      "loss": 0.4682,
+      "step": 713
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062970197217808e-05,
+      "loss": 0.4094,
+      "step": 714
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0607240812981145e-05,
+      "loss": 0.3954,
+      "step": 715
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0584776578468698e-05,
+      "loss": 0.507,
+      "step": 716
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05623093824088e-05,
+      "loss": 0.4511,
+      "step": 717
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0539839338584509e-05,
+      "loss": 0.3269,
+      "step": 718
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0517366560793305e-05,
+      "loss": 0.3342,
+      "step": 719
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0494891162846515e-05,
+      "loss": 0.4233,
+      "step": 720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0472413258568734e-05,
+      "loss": 0.4296,
+      "step": 721
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0449932961797249e-05,
+      "loss": 0.3722,
+      "step": 722
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0427450386381463e-05,
+      "loss": 0.4086,
+      "step": 723
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040496564618233e-05,
+      "loss": 0.3875,
+      "step": 724
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0382478855071751e-05,
+      "loss": 0.3279,
+      "step": 725
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0359990126932022e-05,
+      "loss": 0.4163,
+      "step": 726
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033749957565525e-05,
+      "loss": 0.4412,
+      "step": 727
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0315007315142772e-05,
+      "loss": 0.415,
+      "step": 728
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029251345930458e-05,
+      "loss": 0.4251,
+      "step": 729
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0270018122058753e-05,
+      "loss": 0.4075,
+      "step": 730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0247521417330864e-05,
+      "loss": 0.3823,
+      "step": 731
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0225023459053416e-05,
+      "loss": 0.4515,
+      "step": 732
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0202524361165255e-05,
+      "loss": 0.4079,
+      "step": 733
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.018002423761101e-05,
+      "loss": 0.3785,
+      "step": 734
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0157523202340491e-05,
+      "loss": 0.3989,
+      "step": 735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0135021369308138e-05,
+      "loss": 0.415,
+      "step": 736
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0112518852472414e-05,
+      "loss": 0.3894,
+      "step": 737
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0090015765795265e-05,
+      "loss": 0.3863,
+      "step": 738
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0067512223241509e-05,
+      "loss": 0.3902,
+      "step": 739
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004500833877828e-05,
+      "loss": 0.4072,
+      "step": 740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002250422637444e-05,
+      "loss": 0.3688,
+      "step": 741
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.412,
+      "step": 742
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.977495773625562e-06,
+      "loss": 0.3986,
+      "step": 743
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.954991661221724e-06,
+      "loss": 0.4599,
+      "step": 744
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93248777675849e-06,
+      "loss": 0.4113,
+      "step": 745
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.909984234204738e-06,
+      "loss": 0.4713,
+      "step": 746
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.887481147527587e-06,
+      "loss": 0.3793,
+      "step": 747
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.864978630691865e-06,
+      "loss": 0.4093,
+      "step": 748
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.842476797659509e-06,
+      "loss": 0.3732,
+      "step": 749
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.819975762388993e-06,
+      "loss": 0.3746,
+      "step": 750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.797475638834745e-06,
+      "loss": 0.3627,
+      "step": 751
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.774976540946589e-06,
+      "loss": 0.4231,
+      "step": 752
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.752478582669136e-06,
+      "loss": 0.4466,
+      "step": 753
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.729981877941249e-06,
+      "loss": 0.4179,
+      "step": 754
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70748654069542e-06,
+      "loss": 0.4146,
+      "step": 755
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.684992684857232e-06,
+      "loss": 0.4035,
+      "step": 756
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.662500424344751e-06,
+      "loss": 0.3877,
+      "step": 757
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640009873067981e-06,
+      "loss": 0.4306,
+      "step": 758
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.617521144928252e-06,
+      "loss": 0.3997,
+      "step": 759
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.595034353817673e-06,
+      "loss": 0.4332,
+      "step": 760
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.572549613618537e-06,
+      "loss": 0.4538,
+      "step": 761
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.550067038202756e-06,
+      "loss": 0.3556,
+      "step": 762
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52758674143127e-06,
+      "loss": 0.4435,
+      "step": 763
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.505108837153489e-06,
+      "loss": 0.368,
+      "step": 764
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.482633439206697e-06,
+      "loss": 0.3727,
+      "step": 765
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.460160661415496e-06,
+      "loss": 0.4113,
+      "step": 766
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.437690617591202e-06,
+      "loss": 0.3771,
+      "step": 767
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.415223421531308e-06,
+      "loss": 0.4213,
+      "step": 768
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.392759187018858e-06,
+      "loss": 0.4006,
+      "step": 769
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.370298027821924e-06,
+      "loss": 0.3888,
+      "step": 770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.347840057692981e-06,
+      "loss": 0.4213,
+      "step": 771
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.325385390368367e-06,
+      "loss": 0.39,
+      "step": 772
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30293413956769e-06,
+      "loss": 0.3883,
+      "step": 773
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280486418993254e-06,
+      "loss": 0.3724,
+      "step": 774
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.25804234232948e-06,
+      "loss": 0.4529,
+      "step": 775
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.23560202324235e-06,
+      "loss": 0.4267,
+      "step": 776
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.213165575378793e-06,
+      "loss": 0.4244,
+      "step": 777
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190733112366158e-06,
+      "loss": 0.4018,
+      "step": 778
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.168304747811588e-06,
+      "loss": 0.4573,
+      "step": 779
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.145880595301495e-06,
+      "loss": 0.4155,
+      "step": 780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.123460768400935e-06,
+      "loss": 0.3583,
+      "step": 781
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.101045380653076e-06,
+      "loss": 0.4034,
+      "step": 782
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.078634545578598e-06,
+      "loss": 0.4101,
+      "step": 783
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.056228376675118e-06,
+      "loss": 0.3948,
+      "step": 784
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.033826987416635e-06,
+      "loss": 0.4181,
+      "step": 785
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.011430491252924e-06,
+      "loss": 0.4648,
+      "step": 786
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.989039001609002e-06,
+      "loss": 0.4065,
+      "step": 787
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.966652631884506e-06,
+      "loss": 0.3337,
+      "step": 788
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.944271495453167e-06,
+      "loss": 0.4112,
+      "step": 789
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.921895705662194e-06,
+      "loss": 0.3682,
+      "step": 790
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.899525375831732e-06,
+      "loss": 0.388,
+      "step": 791
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.877160619254264e-06,
+      "loss": 0.3774,
+      "step": 792
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.854801549194055e-06,
+      "loss": 0.3638,
+      "step": 793
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.832448278886567e-06,
+      "loss": 0.4091,
+      "step": 794
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.810100921537895e-06,
+      "loss": 0.3823,
+      "step": 795
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.787759590324177e-06,
+      "loss": 0.4022,
+      "step": 796
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.765424398391047e-06,
+      "loss": 0.4058,
+      "step": 797
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.743095458853034e-06,
+      "loss": 0.4222,
+      "step": 798
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.720772884793017e-06,
+      "loss": 0.4015,
+      "step": 799
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.698456789261617e-06,
+      "loss": 0.3815,
+      "step": 800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.676147285276668e-06,
+      "loss": 0.4609,
+      "step": 801
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.653844485822603e-06,
+      "loss": 0.3784,
+      "step": 802
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.631548503849915e-06,
+      "loss": 0.4792,
+      "step": 803
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.609259452274559e-06,
+      "loss": 0.4391,
+      "step": 804
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.586977443977396e-06,
+      "loss": 0.3597,
+      "step": 805
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56470259180362e-06,
+      "loss": 0.4319,
+      "step": 806
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54243500856218e-06,
+      "loss": 0.3945,
+      "step": 807
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52017480702521e-06,
+      "loss": 0.3338,
+      "step": 808
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.497922099927469e-06,
+      "loss": 0.3916,
+      "step": 809
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.475676999965747e-06,
+      "loss": 0.4171,
+      "step": 810
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.453439619798325e-06,
+      "loss": 0.3987,
+      "step": 811
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.431210072044371e-06,
+      "loss": 0.3594,
+      "step": 812
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.408988469283403e-06,
+      "loss": 0.4404,
+      "step": 813
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.386774924054686e-06,
+      "loss": 0.3953,
+      "step": 814
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.364569548856695e-06,
+      "loss": 0.422,
+      "step": 815
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.342372456146512e-06,
+      "loss": 0.3908,
+      "step": 816
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.320183758339284e-06,
+      "loss": 0.4302,
+      "step": 817
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.29800356780764e-06,
+      "loss": 0.4296,
+      "step": 818
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.275831996881129e-06,
+      "loss": 0.3956,
+      "step": 819
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.253669157845632e-06,
+      "loss": 0.3831,
+      "step": 820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.231515162942824e-06,
+      "loss": 0.392,
+      "step": 821
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.20937012436959e-06,
+      "loss": 0.3982,
+      "step": 822
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18723415427744e-06,
+      "loss": 0.367,
+      "step": 823
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.165107364771979e-06,
+      "loss": 0.3784,
+      "step": 824
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.142989867912299e-06,
+      "loss": 0.3956,
+      "step": 825
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12088177571044e-06,
+      "loss": 0.3855,
+      "step": 826
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.098783200130814e-06,
+      "loss": 0.3981,
+      "step": 827
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.076694253089632e-06,
+      "loss": 0.3995,
+      "step": 828
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05461504645434e-06,
+      "loss": 0.4358,
+      "step": 829
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.032545692043068e-06,
+      "loss": 0.4232,
+      "step": 830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.010486301624034e-06,
+      "loss": 0.3855,
+      "step": 831
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.988436986915005e-06,
+      "loss": 0.383,
+      "step": 832
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.966397859582713e-06,
+      "loss": 0.4065,
+      "step": 833
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.944369031242307e-06,
+      "loss": 0.3865,
+      "step": 834
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.922350613456763e-06,
+      "loss": 0.4409,
+      "step": 835
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.900342717736354e-06,
+      "loss": 0.3738,
+      "step": 836
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.878345455538043e-06,
+      "loss": 0.3814,
+      "step": 837
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.856358938264953e-06,
+      "loss": 0.4189,
+      "step": 838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.834383277265794e-06,
+      "loss": 0.3726,
+      "step": 839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.812418583834282e-06,
+      "loss": 0.4467,
+      "step": 840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.790464969208598e-06,
+      "loss": 0.4064,
+      "step": 841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.768522544570818e-06,
+      "loss": 0.3909,
+      "step": 842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.746591421046336e-06,
+      "loss": 0.3512,
+      "step": 843
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.724671709703328e-06,
+      "loss": 0.3641,
+      "step": 844
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.702763521552154e-06,
+      "loss": 0.422,
+      "step": 845
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680866967544841e-06,
+      "loss": 0.3568,
+      "step": 846
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.658982158574471e-06,
+      "loss": 0.4292,
+      "step": 847
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.637109205474665e-06,
+      "loss": 0.3753,
+      "step": 848
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.615248219018982e-06,
+      "loss": 0.4646,
+      "step": 849
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5933993099203936e-06,
+      "loss": 0.3828,
+      "step": 850
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.571562588830698e-06,
+      "loss": 0.3927,
+      "step": 851
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5497381663399716e-06,
+      "loss": 0.4506,
+      "step": 852
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.527926152975999e-06,
+      "loss": 0.4226,
+      "step": 853
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506126659203733e-06,
+      "loss": 0.384,
+      "step": 854
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484339795424707e-06,
+      "loss": 0.3545,
+      "step": 855
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.462565671976504e-06,
+      "loss": 0.3547,
+      "step": 856
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.440804399132173e-06,
+      "loss": 0.3905,
+      "step": 857
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.419056087099695e-06,
+      "loss": 0.4131,
+      "step": 858
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3973208460213985e-06,
+      "loss": 0.359,
+      "step": 859
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.375598785973429e-06,
+      "loss": 0.3926,
+      "step": 860
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.35389001696517e-06,
+      "loss": 0.4022,
+      "step": 861
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.332194648938689e-06,
+      "loss": 0.3788,
+      "step": 862
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.310512791768199e-06,
+      "loss": 0.4007,
+      "step": 863
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.288844555259471e-06,
+      "loss": 0.3935,
+      "step": 864
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.267190049149319e-06,
+      "loss": 0.3516,
+      "step": 865
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.245549383104993e-06,
+      "loss": 0.3984,
+      "step": 866
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.223922666723677e-06,
+      "loss": 0.3763,
+      "step": 867
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.202310009531886e-06,
+      "loss": 0.4442,
+      "step": 868
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.180711520984953e-06,
+      "loss": 0.3793,
+      "step": 869
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.159127310466441e-06,
+      "loss": 0.4404,
+      "step": 870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.137557487287609e-06,
+      "loss": 0.3862,
+      "step": 871
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.116002160686851e-06,
+      "loss": 0.3876,
+      "step": 872
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.09446143982915e-06,
+      "loss": 0.4251,
+      "step": 873
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.072935433805508e-06,
+      "loss": 0.4505,
+      "step": 874
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0514242516324184e-06,
+      "loss": 0.4018,
+      "step": 875
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0299280022512875e-06,
+      "loss": 0.4151,
+      "step": 876
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.00844679452791e-06,
+      "loss": 0.3755,
+      "step": 877
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.986980737251889e-06,
+      "loss": 0.4132,
+      "step": 878
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9655299391361156e-06,
+      "loss": 0.4649,
+      "step": 879
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.944094508816182e-06,
+      "loss": 0.4246,
+      "step": 880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.922674554849871e-06,
+      "loss": 0.431,
+      "step": 881
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9012701857165755e-06,
+      "loss": 0.4421,
+      "step": 882
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.879881509816764e-06,
+      "loss": 0.4143,
+      "step": 883
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.858508635471428e-06,
+      "loss": 0.3797,
+      "step": 884
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837151670921534e-06,
+      "loss": 0.3635,
+      "step": 885
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.815810724327469e-06,
+      "loss": 0.3592,
+      "step": 886
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.794485903768513e-06,
+      "loss": 0.4198,
+      "step": 887
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.773177317242257e-06,
+      "loss": 0.4122,
+      "step": 888
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7518850726640964e-06,
+      "loss": 0.3734,
+      "step": 889
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.730609277866644e-06,
+      "loss": 0.4239,
+      "step": 890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.709350040599227e-06,
+      "loss": 0.3756,
+      "step": 891
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.688107468527297e-06,
+      "loss": 0.3468,
+      "step": 892
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.666881669231922e-06,
+      "loss": 0.3824,
+      "step": 893
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.645672750209216e-06,
+      "loss": 0.4066,
+      "step": 894
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6244808188698075e-06,
+      "loss": 0.4047,
+      "step": 895
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.603305982538295e-06,
+      "loss": 0.413,
+      "step": 896
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.582148348452699e-06,
+      "loss": 0.3811,
+      "step": 897
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.561008023763915e-06,
+      "loss": 0.3428,
+      "step": 898
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.539885115535187e-06,
+      "loss": 0.3861,
+      "step": 899
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.518779730741555e-06,
+      "loss": 0.4412,
+      "step": 900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4976919762692965e-06,
+      "loss": 0.3891,
+      "step": 901
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.476621958915426e-06,
+      "loss": 0.4305,
+      "step": 902
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4555697853871055e-06,
+      "loss": 0.3233,
+      "step": 903
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.434535562301153e-06,
+      "loss": 0.4035,
+      "step": 904
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4135193961834565e-06,
+      "loss": 0.3441,
+      "step": 905
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.392521393468471e-06,
+      "loss": 0.3551,
+      "step": 906
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.371541660498652e-06,
+      "loss": 0.3629,
+      "step": 907
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.350580303523947e-06,
+      "loss": 0.3458,
+      "step": 908
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.329637428701219e-06,
+      "loss": 0.3772,
+      "step": 909
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.308713142093749e-06,
+      "loss": 0.3876,
+      "step": 910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.287807549670663e-06,
+      "loss": 0.403,
+      "step": 911
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.266920757306429e-06,
+      "loss": 0.3873,
+      "step": 912
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.246052870780288e-06,
+      "loss": 0.4317,
+      "step": 913
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.225203995775746e-06,
+      "loss": 0.3689,
+      "step": 914
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2043742378800154e-06,
+      "loss": 0.3835,
+      "step": 915
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.183563702583506e-06,
+      "loss": 0.4217,
+      "step": 916
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.162772495279265e-06,
+      "loss": 0.3891,
+      "step": 917
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1420007212624584e-06,
+      "loss": 0.3735,
+      "step": 918
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.121248485729832e-06,
+      "loss": 0.3979,
+      "step": 919
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.100515893779188e-06,
+      "loss": 0.4062,
+      "step": 920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.079803050408837e-06,
+      "loss": 0.4235,
+      "step": 921
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.05911006051708e-06,
+      "loss": 0.3945,
+      "step": 922
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.038437028901667e-06,
+      "loss": 0.4598,
+      "step": 923
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.01778406025928e-06,
+      "loss": 0.39,
+      "step": 924
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.99715125918498e-06,
+      "loss": 0.3637,
+      "step": 925
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.976538730171708e-06,
+      "loss": 0.4071,
+      "step": 926
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.955946577609721e-06,
+      "loss": 0.3802,
+      "step": 927
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.935374905786102e-06,
+      "loss": 0.3767,
+      "step": 928
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.91482381888419e-06,
+      "loss": 0.4016,
+      "step": 929
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.89429342098309e-06,
+      "loss": 0.3565,
+      "step": 930
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.873783816057115e-06,
+      "loss": 0.3799,
+      "step": 931
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8532951079752895e-06,
+      "loss": 0.3815,
+      "step": 932
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.832827400500794e-06,
+      "loss": 0.4156,
+      "step": 933
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.812380797290465e-06,
+      "loss": 0.3372,
+      "step": 934
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.791955401894248e-06,
+      "loss": 0.4424,
+      "step": 935
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.771551317754691e-06,
+      "loss": 0.3707,
+      "step": 936
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.75116864820641e-06,
+      "loss": 0.3603,
+      "step": 937
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.730807496475568e-06,
+      "loss": 0.3933,
+      "step": 938
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.710467965679356e-06,
+      "loss": 0.4183,
+      "step": 939
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.690150158825462e-06,
+      "loss": 0.3446,
+      "step": 940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6698541788115644e-06,
+      "loss": 0.4104,
+      "step": 941
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.649580128424792e-06,
+      "loss": 0.3721,
+      "step": 942
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.629328110341218e-06,
+      "loss": 0.3887,
+      "step": 943
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.609098227125334e-06,
+      "loss": 0.4046,
+      "step": 944
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.588890581229531e-06,
+      "loss": 0.3951,
+      "step": 945
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.568705274993584e-06,
+      "loss": 0.4098,
+      "step": 946
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.548542410644132e-06,
+      "loss": 0.4195,
+      "step": 947
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.528402090294142e-06,
+      "loss": 0.3786,
+      "step": 948
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.508284415942442e-06,
+      "loss": 0.3708,
+      "step": 949
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.488189489473133e-06,
+      "loss": 0.3573,
+      "step": 950
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.468117412655147e-06,
+      "loss": 0.4249,
+      "step": 951
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.448068287141663e-06,
+      "loss": 0.4163,
+      "step": 952
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.428042214469662e-06,
+      "loss": 0.4761,
+      "step": 953
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4080392960593355e-06,
+      "loss": 0.4418,
+      "step": 954
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.388059633213651e-06,
+      "loss": 0.3294,
+      "step": 955
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.368103327117768e-06,
+      "loss": 0.3234,
+      "step": 956
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.348170478838579e-06,
+      "loss": 0.4148,
+      "step": 957
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.328261189324166e-06,
+      "loss": 0.4313,
+      "step": 958
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.308375559403307e-06,
+      "loss": 0.3532,
+      "step": 959
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.288513689784951e-06,
+      "loss": 0.4148,
+      "step": 960
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.26867568105772e-06,
+      "loss": 0.3999,
+      "step": 961
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2488616336893915e-06,
+      "loss": 0.1375,
+      "step": 962
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.229071648026398e-06,
+      "loss": 0.1576,
+      "step": 963
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.209305824293307e-06,
+      "loss": 0.155,
+      "step": 964
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.189564262592326e-06,
+      "loss": 0.1728,
+      "step": 965
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1698470629027845e-06,
+      "loss": 0.156,
+      "step": 966
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.150154325080636e-06,
+      "loss": 0.137,
+      "step": 967
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.130486148857952e-06,
+      "loss": 0.1431,
+      "step": 968
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.110842633842406e-06,
+      "loss": 0.1087,
+      "step": 969
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0912238795167845e-06,
+      "loss": 0.1651,
+      "step": 970
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.071629985238474e-06,
+      "loss": 0.1251,
+      "step": 971
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.05206105023895e-06,
+      "loss": 0.1702,
+      "step": 972
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.032517173623306e-06,
+      "loss": 0.1325,
+      "step": 973
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.012998454369701e-06,
+      "loss": 0.1513,
+      "step": 974
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.993504991328914e-06,
+      "loss": 0.1285,
+      "step": 975
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.974036883223798e-06,
+      "loss": 0.1377,
+      "step": 976
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.954594228648807e-06,
+      "loss": 0.1354,
+      "step": 977
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.935177126069485e-06,
+      "loss": 0.1333,
+      "step": 978
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.91578567382197e-06,
+      "loss": 0.1567,
+      "step": 979
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896419970112499e-06,
+      "loss": 0.1529,
+      "step": 980
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.877080113016898e-06,
+      "loss": 0.1289,
+      "step": 981
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.857766200480115e-06,
+      "loss": 0.1245,
+      "step": 982
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.838478330315677e-06,
+      "loss": 0.1376,
+      "step": 983
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.819216600205254e-06,
+      "loss": 0.1443,
+      "step": 984
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.799981107698097e-06,
+      "loss": 0.1374,
+      "step": 985
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.780771950210616e-06,
+      "loss": 0.1448,
+      "step": 986
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.761589225025811e-06,
+      "loss": 0.1322,
+      "step": 987
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.742433029292856e-06,
+      "loss": 0.1591,
+      "step": 988
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.723303460026538e-06,
+      "loss": 0.1313,
+      "step": 989
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.704200614106813e-06,
+      "loss": 0.1467,
+      "step": 990
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685124588278297e-06,
+      "loss": 0.1279,
+      "step": 991
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6660754791497755e-06,
+      "loss": 0.1491,
+      "step": 992
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.647053383193717e-06,
+      "loss": 0.1476,
+      "step": 993
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.628058396745787e-06,
+      "loss": 0.1399,
+      "step": 994
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.609090616004355e-06,
+      "loss": 0.1502,
+      "step": 995
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.590150137030009e-06,
+      "loss": 0.1557,
+      "step": 996
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.571237055745074e-06,
+      "loss": 0.1463,
+      "step": 997
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.552351467933115e-06,
+      "loss": 0.1479,
+      "step": 998
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.533493469238465e-06,
+      "loss": 0.1242,
+      "step": 999
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.514663155165731e-06,
+      "loss": 0.1287,
+      "step": 1000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.495860621079315e-06,
+      "loss": 0.1331,
+      "step": 1001
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.477085962202931e-06,
+      "loss": 0.1428,
+      "step": 1002
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.458339273619119e-06,
+      "loss": 0.1703,
+      "step": 1003
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.439620650268771e-06,
+      "loss": 0.1218,
+      "step": 1004
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.420930186950631e-06,
+      "loss": 0.1369,
+      "step": 1005
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.402267978320854e-06,
+      "loss": 0.1469,
+      "step": 1006
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.383634118892472e-06,
+      "loss": 0.1457,
+      "step": 1007
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.365028703034976e-06,
+      "loss": 0.1226,
+      "step": 1008
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3464518249737766e-06,
+      "loss": 0.1276,
+      "step": 1009
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.327903578789785e-06,
+      "loss": 0.123,
+      "step": 1010
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.309384058418884e-06,
+      "loss": 0.1359,
+      "step": 1011
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.290893357651502e-06,
+      "loss": 0.1397,
+      "step": 1012
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.272431570132092e-06,
+      "loss": 0.1304,
+      "step": 1013
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.253998789358683e-06,
+      "loss": 0.138,
+      "step": 1014
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.23559510868242e-06,
+      "loss": 0.1149,
+      "step": 1015
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.217220621307043e-06,
+      "loss": 0.144,
+      "step": 1016
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.198875420288477e-06,
+      "loss": 0.1167,
+      "step": 1017
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.180559598534297e-06,
+      "loss": 0.1324,
+      "step": 1018
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.162273248803322e-06,
+      "loss": 0.1249,
+      "step": 1019
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144016463705081e-06,
+      "loss": 0.1625,
+      "step": 1020
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.125789335699404e-06,
+      "loss": 0.1343,
+      "step": 1021
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.107591957095903e-06,
+      "loss": 0.1267,
+      "step": 1022
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.08942442005354e-06,
+      "loss": 0.1405,
+      "step": 1023
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.071286816580142e-06,
+      "loss": 0.1118,
+      "step": 1024
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.053179238531943e-06,
+      "loss": 0.133,
+      "step": 1025
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.035101777613113e-06,
+      "loss": 0.1288,
+      "step": 1026
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017054525375299e-06,
+      "loss": 0.1323,
+      "step": 1027
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.999037573217157e-06,
+      "loss": 0.1009,
+      "step": 1028
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.981051012383893e-06,
+      "loss": 0.1326,
+      "step": 1029
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.963094933966797e-06,
+      "loss": 0.1204,
+      "step": 1030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.945169428902784e-06,
+      "loss": 0.1299,
+      "step": 1031
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927274587973935e-06,
+      "loss": 0.1232,
+      "step": 1032
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.909410501807033e-06,
+      "loss": 0.1109,
+      "step": 1033
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8915772608731066e-06,
+      "loss": 0.1447,
+      "step": 1034
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.873774955486973e-06,
+      "loss": 0.1354,
+      "step": 1035
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.856003675806777e-06,
+      "loss": 0.1197,
+      "step": 1036
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.838263511833542e-06,
+      "loss": 0.1201,
+      "step": 1037
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.820554553410693e-06,
+      "loss": 0.1136,
+      "step": 1038
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8028768902236458e-06,
+      "loss": 0.1357,
+      "step": 1039
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.78523061179929e-06,
+      "loss": 0.1155,
+      "step": 1040
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7676158075056024e-06,
+      "loss": 0.1288,
+      "step": 1041
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7500325665511337e-06,
+      "loss": 0.1269,
+      "step": 1042
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7324809779846116e-06,
+      "loss": 0.1224,
+      "step": 1043
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7149611306944356e-06,
+      "loss": 0.151,
+      "step": 1044
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.697473113408282e-06,
+      "loss": 0.1219,
+      "step": 1045
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.680017014692604e-06,
+      "loss": 0.1214,
+      "step": 1046
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.662592922952218e-06,
+      "loss": 0.1232,
+      "step": 1047
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.645200926429844e-06,
+      "loss": 0.1389,
+      "step": 1048
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.627841113205652e-06,
+      "loss": 0.1211,
+      "step": 1049
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.610513571196832e-06,
+      "loss": 0.1298,
+      "step": 1050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5932183881571305e-06,
+      "loss": 0.1422,
+      "step": 1051
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5759556516764205e-06,
+      "loss": 0.1224,
+      "step": 1052
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.558725449180247e-06,
+      "loss": 0.1414,
+      "step": 1053
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.541527867929403e-06,
+      "loss": 0.1035,
+      "step": 1054
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.524362995019455e-06,
+      "loss": 0.1411,
+      "step": 1055
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.507230917380332e-06,
+      "loss": 0.1205,
+      "step": 1056
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4901317217758767e-06,
+      "loss": 0.1295,
+      "step": 1057
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4730654948033957e-06,
+      "loss": 0.1374,
+      "step": 1058
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.456032322893237e-06,
+      "loss": 0.1247,
+      "step": 1059
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4390322923083385e-06,
+      "loss": 0.18,
+      "step": 1060
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.422065489143799e-06,
+      "loss": 0.1515,
+      "step": 1061
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4051319993264397e-06,
+      "loss": 0.1152,
+      "step": 1062
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.388231908614371e-06,
+      "loss": 0.1433,
+      "step": 1063
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3713653025965544e-06,
+      "loss": 0.122,
+      "step": 1064
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3545322666923717e-06,
+      "loss": 0.1623,
+      "step": 1065
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3377328861511927e-06,
+      "loss": 0.1137,
+      "step": 1066
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3209672460519426e-06,
+      "loss": 0.1218,
+      "step": 1067
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3042354313026702e-06,
+      "loss": 0.1392,
+      "step": 1068
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.287537526640121e-06,
+      "loss": 0.1291,
+      "step": 1069
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2708736166293064e-06,
+      "loss": 0.1232,
+      "step": 1070
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.254243785663065e-06,
+      "loss": 0.142,
+      "step": 1071
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.237648117961665e-06,
+      "loss": 0.1385,
+      "step": 1072
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2210866975723333e-06,
+      "loss": 0.1123,
+      "step": 1073
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2045596083688814e-06,
+      "loss": 0.154,
+      "step": 1074
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.188066934051226e-06,
+      "loss": 0.1344,
+      "step": 1075
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1716087581450193e-06,
+      "loss": 0.1303,
+      "step": 1076
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1551851640011755e-06,
+      "loss": 0.1296,
+      "step": 1077
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1387962347954936e-06,
+      "loss": 0.1253,
+      "step": 1078
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1224420535281974e-06,
+      "loss": 0.1329,
+      "step": 1079
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1061227030235442e-06,
+      "loss": 0.1223,
+      "step": 1080
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0898382659293902e-06,
+      "loss": 0.1265,
+      "step": 1081
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.073588824716777e-06,
+      "loss": 0.1363,
+      "step": 1082
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.05737446167951e-06,
+      "loss": 0.1272,
+      "step": 1083
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.041195258933749e-06,
+      "loss": 0.1149,
+      "step": 1084
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.025051298417585e-06,
+      "loss": 0.1465,
+      "step": 1085
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.008942661890627e-06,
+      "loss": 0.1324,
+      "step": 1086
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9928694309335917e-06,
+      "loss": 0.1371,
+      "step": 1087
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.976831686947884e-06,
+      "loss": 0.1914,
+      "step": 1088
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.960829511155191e-06,
+      "loss": 0.1327,
+      "step": 1089
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.944862984597068e-06,
+      "loss": 0.1497,
+      "step": 1090
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.1337,
+      "step": 1091
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.913037202447625e-06,
+      "loss": 0.1646,
+      "step": 1092
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.897178108035067e-06,
+      "loss": 0.1406,
+      "step": 1093
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8813549852137824e-06,
+      "loss": 0.1359,
+      "step": 1094
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8655679141185287e-06,
+      "loss": 0.1446,
+      "step": 1095
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8498169747014824e-06,
+      "loss": 0.1546,
+      "step": 1096
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8341022467318335e-06,
+      "loss": 0.1327,
+      "step": 1097
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.818423809795384e-06,
+      "loss": 0.1183,
+      "step": 1098
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.802781743294143e-06,
+      "loss": 0.125,
+      "step": 1099
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.787176126445923e-06,
+      "loss": 0.1312,
+      "step": 1100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.771607038283942e-06,
+      "loss": 0.1296,
+      "step": 1101
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.756074557656424e-06,
+      "loss": 0.1401,
+      "step": 1102
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7405787632261936e-06,
+      "loss": 0.1421,
+      "step": 1103
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.725119733470284e-06,
+      "loss": 0.1454,
+      "step": 1104
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.709697546679537e-06,
+      "loss": 0.1125,
+      "step": 1105
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6943122809582e-06,
+      "loss": 0.1072,
+      "step": 1106
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.678964014223553e-06,
+      "loss": 0.1211,
+      "step": 1107
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.663652824205476e-06,
+      "loss": 0.1639,
+      "step": 1108
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6483787884461022e-06,
+      "loss": 0.1464,
+      "step": 1109
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6331419842993746e-06,
+      "loss": 0.1379,
+      "step": 1110
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.617942488930705e-06,
+      "loss": 0.171,
+      "step": 1111
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6027803793165353e-06,
+      "loss": 0.1443,
+      "step": 1112
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5876557322439922e-06,
+      "loss": 0.1848,
+      "step": 1113
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5725686243104586e-06,
+      "loss": 0.0993,
+      "step": 1114
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5575191319232127e-06,
+      "loss": 0.1149,
+      "step": 1115
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5425073312990334e-06,
+      "loss": 0.1031,
+      "step": 1116
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.52753329846381e-06,
+      "loss": 0.1051,
+      "step": 1117
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5125971092521607e-06,
+      "loss": 0.1666,
+      "step": 1118
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4976988393070477e-06,
+      "loss": 0.121,
+      "step": 1119
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4828385640793974e-06,
+      "loss": 0.1195,
+      "step": 1120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.468016358827712e-06,
+      "loss": 0.1772,
+      "step": 1121
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4532322986176926e-06,
+      "loss": 0.1517,
+      "step": 1122
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4384864583218592e-06,
+      "loss": 0.1352,
+      "step": 1123
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4237789126191715e-06,
+      "loss": 0.1263,
+      "step": 1124
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.409109735994647e-06,
+      "loss": 0.1167,
+      "step": 1125
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3944790027389888e-06,
+      "loss": 0.1322,
+      "step": 1126
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3798867869482047e-06,
+      "loss": 0.1407,
+      "step": 1127
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3653331625232367e-06,
+      "loss": 0.1179,
+      "step": 1128
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.350818203169586e-06,
+      "loss": 0.138,
+      "step": 1129
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.336341982396928e-06,
+      "loss": 0.1265,
+      "step": 1130
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.321904573518765e-06,
+      "loss": 0.152,
+      "step": 1131
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.307506049652031e-06,
+      "loss": 0.1226,
+      "step": 1132
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2931464837167303e-06,
+      "loss": 0.1395,
+      "step": 1133
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2788259484355712e-06,
+      "loss": 0.1726,
+      "step": 1134
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2645445163335944e-06,
+      "loss": 0.1148,
+      "step": 1135
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.250302259737803e-06,
+      "loss": 0.141,
+      "step": 1136
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2360992507768055e-06,
+      "loss": 0.1247,
+      "step": 1137
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2219355613804406e-06,
+      "loss": 0.1256,
+      "step": 1138
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2078112632794092e-06,
+      "loss": 0.1472,
+      "step": 1139
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1937264280049365e-06,
+      "loss": 0.1645,
+      "step": 1140
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.179681126888371e-06,
+      "loss": 0.125,
+      "step": 1141
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1656754310608663e-06,
+      "loss": 0.1197,
+      "step": 1142
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1517094114529746e-06,
+      "loss": 0.1462,
+      "step": 1143
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.137783138794335e-06,
+      "loss": 0.1217,
+      "step": 1144
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1238966836132693e-06,
+      "loss": 0.1334,
+      "step": 1145
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1100501162364707e-06,
+      "loss": 0.1265,
+      "step": 1146
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.096243506788602e-06,
+      "loss": 0.1675,
+      "step": 1147
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0824769251919775e-06,
+      "loss": 0.1263,
+      "step": 1148
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0687504411661897e-06,
+      "loss": 0.1424,
+      "step": 1149
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.055064124227758e-06,
+      "loss": 0.1503,
+      "step": 1150
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0414180436897846e-06,
+      "loss": 0.1276,
+      "step": 1151
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.027812268661592e-06,
+      "loss": 0.1288,
+      "step": 1152
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.014246868048385e-06,
+      "loss": 0.1315,
+      "step": 1153
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0007219105508935e-06,
+      "loss": 0.0959,
+      "step": 1154
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.987237464665024e-06,
+      "loss": 0.1163,
+      "step": 1155
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9737935986815205e-06,
+      "loss": 0.1613,
+      "step": 1156
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9603903806856106e-06,
+      "loss": 0.1671,
+      "step": 1157
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9470278785566653e-06,
+      "loss": 0.1055,
+      "step": 1158
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.933706159967854e-06,
+      "loss": 0.1701,
+      "step": 1159
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9204252923858003e-06,
+      "loss": 0.1371,
+      "step": 1160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9071853430702415e-06,
+      "loss": 0.1302,
+      "step": 1161
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8939863790736923e-06,
+      "loss": 0.1125,
+      "step": 1162
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8808284672410904e-06,
+      "loss": 0.1174,
+      "step": 1163
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8677116742094858e-06,
+      "loss": 0.1314,
+      "step": 1164
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8546360664076657e-06,
+      "loss": 0.1172,
+      "step": 1165
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8416017100558592e-06,
+      "loss": 0.155,
+      "step": 1166
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8286086711653605e-06,
+      "loss": 0.1217,
+      "step": 1167
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8156570155382357e-06,
+      "loss": 0.136,
+      "step": 1168
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8027468087669486e-06,
+      "loss": 0.1284,
+      "step": 1169
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7898781162340683e-06,
+      "loss": 0.1278,
+      "step": 1170
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7770510031119103e-06,
+      "loss": 0.1354,
+      "step": 1171
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.764265534362205e-06,
+      "loss": 0.1142,
+      "step": 1172
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7515217747358015e-06,
+      "loss": 0.1251,
+      "step": 1173
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7388197887722914e-06,
+      "loss": 0.1586,
+      "step": 1174
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7261596407997306e-06,
+      "loss": 0.1449,
+      "step": 1175
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7135413949342706e-06,
+      "loss": 0.1258,
+      "step": 1176
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7009651150798713e-06,
+      "loss": 0.1236,
+      "step": 1177
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6884308649279413e-06,
+      "loss": 0.1471,
+      "step": 1178
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6759387079570532e-06,
+      "loss": 0.1411,
+      "step": 1179
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6634887074325844e-06,
+      "loss": 0.1245,
+      "step": 1180
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6510809264064253e-06,
+      "loss": 0.1505,
+      "step": 1181
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6387154277166484e-06,
+      "loss": 0.1303,
+      "step": 1182
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6263922739871886e-06,
+      "loss": 0.109,
+      "step": 1183
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6141115276275298e-06,
+      "loss": 0.1146,
+      "step": 1184
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6018732508323886e-06,
+      "loss": 0.1434,
+      "step": 1185
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5896775055813973e-06,
+      "loss": 0.1185,
+      "step": 1186
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5775243536387908e-06,
+      "loss": 0.1584,
+      "step": 1187
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.565413856553095e-06,
+      "loss": 0.1305,
+      "step": 1188
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.553346075656813e-06,
+      "loss": 0.1384,
+      "step": 1189
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5413210720661187e-06,
+      "loss": 0.1438,
+      "step": 1190
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5293389066805397e-06,
+      "loss": 0.1819,
+      "step": 1191
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5173996401826563e-06,
+      "loss": 0.1646,
+      "step": 1192
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5055033330377912e-06,
+      "loss": 0.1529,
+      "step": 1193
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.493650045493703e-06,
+      "loss": 0.1514,
+      "step": 1194
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4818398375802834e-06,
+      "loss": 0.1263,
+      "step": 1195
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.470072769109242e-06,
+      "loss": 0.1204,
+      "step": 1196
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4583488996738292e-06,
+      "loss": 0.1314,
+      "step": 1197
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4466682886485007e-06,
+      "loss": 0.1135,
+      "step": 1198
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4350309951886488e-06,
+      "loss": 0.1354,
+      "step": 1199
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4234370782302742e-06,
+      "loss": 0.1266,
+      "step": 1200
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4118865964897143e-06,
+      "loss": 0.1401,
+      "step": 1201
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4003796084633204e-06,
+      "loss": 0.1403,
+      "step": 1202
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.388916172427187e-06,
+      "loss": 0.1738,
+      "step": 1203
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3774963464368295e-06,
+      "loss": 0.1419,
+      "step": 1204
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3661201883269159e-06,
+      "loss": 0.1468,
+      "step": 1205
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3547877557109546e-06,
+      "loss": 0.1182,
+      "step": 1206
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3434991059810153e-06,
+      "loss": 0.1221,
+      "step": 1207
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3322542963074314e-06,
+      "loss": 0.1207,
+      "step": 1208
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3210533836385086e-06,
+      "loss": 0.1427,
+      "step": 1209
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3098964247002499e-06,
+      "loss": 0.1529,
+      "step": 1210
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.298783475996046e-06,
+      "loss": 0.1199,
+      "step": 1211
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.287714593806415e-06,
+      "loss": 0.1342,
+      "step": 1212
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2766898341886914e-06,
+      "loss": 0.1159,
+      "step": 1213
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2657092529767644e-06,
+      "loss": 0.1635,
+      "step": 1214
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2547729057807811e-06,
+      "loss": 0.1378,
+      "step": 1215
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2438808479868715e-06,
+      "loss": 0.1098,
+      "step": 1216
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2330331347568635e-06,
+      "loss": 0.1323,
+      "step": 1217
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2222298210280103e-06,
+      "loss": 0.1139,
+      "step": 1218
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2114709615127052e-06,
+      "loss": 0.1151,
+      "step": 1219
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.200756610698205e-06,
+      "loss": 0.1329,
+      "step": 1220
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1900868228463603e-06,
+      "loss": 0.1559,
+      "step": 1221
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1794616519933344e-06,
+      "loss": 0.1805,
+      "step": 1222
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1688811519493326e-06,
+      "loss": 0.1445,
+      "step": 1223
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1583453762983289e-06,
+      "loss": 0.1545,
+      "step": 1224
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1478543783977948e-06,
+      "loss": 0.1395,
+      "step": 1225
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.137408211378429e-06,
+      "loss": 0.1438,
+      "step": 1226
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1270069281438867e-06,
+      "loss": 0.1279,
+      "step": 1227
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1166505813705187e-06,
+      "loss": 0.1738,
+      "step": 1228
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.106339223507088e-06,
+      "loss": 0.139,
+      "step": 1229
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.09607290677453e-06,
+      "loss": 0.1378,
+      "step": 1230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0858516831656595e-06,
+      "loss": 0.1033,
+      "step": 1231
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0756756044449358e-06,
+      "loss": 0.1385,
+      "step": 1232
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0655447221481685e-06,
+      "loss": 0.1129,
+      "step": 1233
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0554590875822924e-06,
+      "loss": 0.1304,
+      "step": 1234
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0454187518250735e-06,
+      "loss": 0.1216,
+      "step": 1235
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.035423765724879e-06,
+      "loss": 0.1086,
+      "step": 1236
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0254741799003975e-06,
+      "loss": 0.133,
+      "step": 1237
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.015570044740396e-06,
+      "loss": 0.1455,
+      "step": 1238
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0057114104034605e-06,
+      "loss": 0.1397,
+      "step": 1239
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.958983268177425e-07,
+      "loss": 0.1206,
+      "step": 1240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.86130843680706e-07,
+      "loss": 0.2065,
+      "step": 1241
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.764090104588741e-07,
+      "loss": 0.1011,
+      "step": 1242
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.667328763875816e-07,
+      "loss": 0.1228,
+      "step": 1243
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.571024904707238e-07,
+      "loss": 0.1419,
+      "step": 1244
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.475179014805092e-07,
+      "loss": 0.1416,
+      "step": 1245
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.379791579572118e-07,
+      "loss": 0.1389,
+      "step": 1246
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.284863082089223e-07,
+      "loss": 0.1305,
+      "step": 1247
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.190394003113123e-07,
+      "loss": 0.1265,
+      "step": 1248
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.0963848210738e-07,
+      "loss": 0.1218,
+      "step": 1249
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.002836012072169e-07,
+      "loss": 0.1164,
+      "step": 1250
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.909748049877587e-07,
+      "loss": 0.1538,
+      "step": 1251
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.817121405925543e-07,
+      "loss": 0.1311,
+      "step": 1252
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.724956549315177e-07,
+      "loss": 0.1437,
+      "step": 1253
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.633253946806974e-07,
+      "loss": 0.1079,
+      "step": 1254
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.542014062820369e-07,
+      "loss": 0.1087,
+      "step": 1255
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.451237359431397e-07,
+      "loss": 0.145,
+      "step": 1256
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.360924296370376e-07,
+      "loss": 0.1226,
+      "step": 1257
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.271075331019541e-07,
+      "loss": 0.1472,
+      "step": 1258
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.181690918410756e-07,
+      "loss": 0.1189,
+      "step": 1259
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.092771511223185e-07,
+      "loss": 0.1354,
+      "step": 1260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.004317559781049e-07,
+      "loss": 0.1332,
+      "step": 1261
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.916329512051235e-07,
+      "loss": 0.101,
+      "step": 1262
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.828807813641226e-07,
+      "loss": 0.1158,
+      "step": 1263
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.741752907796584e-07,
+      "loss": 0.1123,
+      "step": 1264
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.655165235398987e-07,
+      "loss": 0.1483,
+      "step": 1265
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.56904523496369e-07,
+      "loss": 0.1442,
+      "step": 1266
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.483393342637635e-07,
+      "loss": 0.1097,
+      "step": 1267
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.398209992196914e-07,
+      "loss": 0.1645,
+      "step": 1268
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.313495615044874e-07,
+      "loss": 0.1222,
+      "step": 1269
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.229250640209661e-07,
+      "loss": 0.1379,
+      "step": 1270
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.145475494342302e-07,
+      "loss": 0.1517,
+      "step": 1271
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.062170601714302e-07,
+      "loss": 0.1115,
+      "step": 1272
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.979336384215695e-07,
+      "loss": 0.1118,
+      "step": 1273
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.896973261352779e-07,
+      "loss": 0.1123,
+      "step": 1274
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.815081650246048e-07,
+      "loss": 0.1162,
+      "step": 1275
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.73366196562808e-07,
+      "loss": 0.1254,
+      "step": 1276
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.652714619841405e-07,
+      "loss": 0.1539,
+      "step": 1277
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.572240022836451e-07,
+      "loss": 0.1437,
+      "step": 1278
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.492238582169453e-07,
+      "loss": 0.1425,
+      "step": 1279
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.412710703000368e-07,
+      "loss": 0.1366,
+      "step": 1280
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.33365678809088e-07,
+      "loss": 0.0997,
+      "step": 1281
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.255077237802288e-07,
+      "loss": 0.1183,
+      "step": 1282
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.176972450093543e-07,
+      "loss": 0.1212,
+      "step": 1283
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.099342820519183e-07,
+      "loss": 0.1736,
+      "step": 1284
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.022188742227375e-07,
+      "loss": 0.126,
+      "step": 1285
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.945510605957861e-07,
+      "loss": 0.1263,
+      "step": 1286
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.869308800040075e-07,
+      "loss": 0.134,
+      "step": 1287
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.79358371039106e-07,
+      "loss": 0.1443,
+      "step": 1288
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.718335720513602e-07,
+      "loss": 0.1356,
+      "step": 1289
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.643565211494285e-07,
+      "loss": 0.1171,
+      "step": 1290
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.569272562001471e-07,
+      "loss": 0.1389,
+      "step": 1291
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.495458148283505e-07,
+      "loss": 0.1538,
+      "step": 1292
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.422122344166736e-07,
+      "loss": 0.0964,
+      "step": 1293
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.349265521053603e-07,
+      "loss": 0.142,
+      "step": 1294
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.276888047920836e-07,
+      "loss": 0.1438,
+      "step": 1295
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.204990291317535e-07,
+      "loss": 0.1527,
+      "step": 1296
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.13357261536327e-07,
+      "loss": 0.1116,
+      "step": 1297
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.062635381746362e-07,
+      "loss": 0.1337,
+      "step": 1298
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.992178949721893e-07,
+      "loss": 0.1161,
+      "step": 1299
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.92220367611006e-07,
+      "loss": 0.132,
+      "step": 1300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.852709915294196e-07,
+      "loss": 0.1109,
+      "step": 1301
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.783698019219118e-07,
+      "loss": 0.1475,
+      "step": 1302
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.715168337389231e-07,
+      "loss": 0.131,
+      "step": 1303
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.647121216866857e-07,
+      "loss": 0.139,
+      "step": 1304
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5795570022703964e-07,
+      "loss": 0.144,
+      "step": 1305
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.512476035772628e-07,
+      "loss": 0.1128,
+      "step": 1306
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4458786570989786e-07,
+      "loss": 0.1319,
+      "step": 1307
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3797652035257544e-07,
+      "loss": 0.1509,
+      "step": 1308
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.314136009878511e-07,
+      "loss": 0.1287,
+      "step": 1309
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.248991408530279e-07,
+      "loss": 0.126,
+      "step": 1310
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1843317293999374e-07,
+      "loss": 0.1323,
+      "step": 1311
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1201572999505e-07,
+      "loss": 0.1479,
+      "step": 1312
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0564684451875005e-07,
+      "loss": 0.1047,
+      "step": 1313
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.9932654876573164e-07,
+      "loss": 0.1266,
+      "step": 1314
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9305487474455285e-07,
+      "loss": 0.1307,
+      "step": 1315
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8683185421753313e-07,
+      "loss": 0.114,
+      "step": 1316
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.806575187005901e-07,
+      "loss": 0.1167,
+      "step": 1317
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.74531899463082e-07,
+      "loss": 0.1252,
+      "step": 1318
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.684550275276455e-07,
+      "loss": 0.0919,
+      "step": 1319
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6242693367004365e-07,
+      "loss": 0.1265,
+      "step": 1320
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.564476484190016e-07,
+      "loss": 0.1309,
+      "step": 1321
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5051720205606877e-07,
+      "loss": 0.1299,
+      "step": 1322
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4463562461544255e-07,
+      "loss": 0.1489,
+      "step": 1323
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.38802945883836e-07,
+      "loss": 0.1881,
+      "step": 1324
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.330191954003159e-07,
+      "loss": 0.141,
+      "step": 1325
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2728440245615724e-07,
+      "loss": 0.1507,
+      "step": 1326
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.215985960946943e-07,
+      "loss": 0.1539,
+      "step": 1327
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1596180511117235e-07,
+      "loss": 0.1264,
+      "step": 1328
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.10374058052606e-07,
+      "loss": 0.1448,
+      "step": 1329
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.048353832176221e-07,
+      "loss": 0.1447,
+      "step": 1330
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9934580865634056e-07,
+      "loss": 0.1394,
+      "step": 1331
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.939053621702015e-07,
+      "loss": 0.145,
+      "step": 1332
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.8851407131185506e-07,
+      "loss": 0.1217,
+      "step": 1333
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.83171963384995e-07,
+      "loss": 0.1365,
+      "step": 1334
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7787906544424093e-07,
+      "loss": 0.1375,
+      "step": 1335
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7263540429498747e-07,
+      "loss": 0.1296,
+      "step": 1336
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6744100649327976e-07,
+      "loss": 0.1189,
+      "step": 1337
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6229589834566807e-07,
+      "loss": 0.1422,
+      "step": 1338
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.572001059090812e-07,
+      "loss": 0.1135,
+      "step": 1339
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5215365499069446e-07,
+      "loss": 0.1584,
+      "step": 1340
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4715657114779523e-07,
+      "loss": 0.1633,
+      "step": 1341
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4220887968765873e-07,
+      "loss": 0.1273,
+      "step": 1342
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.373106056674146e-07,
+      "loss": 0.1841,
+      "step": 1343
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3246177389392388e-07,
+      "loss": 0.1159,
+      "step": 1344
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.276624089236501e-07,
+      "loss": 0.149,
+      "step": 1345
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.229125350625394e-07,
+      "loss": 0.1444,
+      "step": 1346
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1821217636589176e-07,
+      "loss": 0.1503,
+      "step": 1347
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1356135663824328e-07,
+      "loss": 0.126,
+      "step": 1348
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0896009943324634e-07,
+      "loss": 0.1204,
+      "step": 1349
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0440842805354522e-07,
+      "loss": 0.1353,
+      "step": 1350
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.99906365550665e-07,
+      "loss": 0.1187,
+      "step": 1351
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9545393472488738e-07,
+      "loss": 0.1273,
+      "step": 1352
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9105115812514065e-07,
+      "loss": 0.1231,
+      "step": 1353
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.866980580488842e-07,
+      "loss": 0.1239,
+      "step": 1354
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8239465654199651e-07,
+      "loss": 0.1326,
+      "step": 1355
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7814097539865626e-07,
+      "loss": 0.1117,
+      "step": 1356
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7393703616124802e-07,
+      "loss": 0.1311,
+      "step": 1357
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6978286012023225e-07,
+      "loss": 0.1222,
+      "step": 1358
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6567846831405664e-07,
+      "loss": 0.1516,
+      "step": 1359
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6162388152903498e-07,
+      "loss": 0.1456,
+      "step": 1360
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5761912029925387e-07,
+      "loss": 0.1675,
+      "step": 1361
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5366420490645738e-07,
+      "loss": 0.122,
+      "step": 1362
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4975915537995266e-07,
+      "loss": 0.1102,
+      "step": 1363
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4590399149650769e-07,
+      "loss": 0.0966,
+      "step": 1364
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4209873278024478e-07,
+      "loss": 0.0929,
+      "step": 1365
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3834339850254952e-07,
+      "loss": 0.1417,
+      "step": 1366
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3463800768196866e-07,
+      "loss": 0.0977,
+      "step": 1367
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.309825790841146e-07,
+      "loss": 0.1279,
+      "step": 1368
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.273771312215699e-07,
+      "loss": 0.1538,
+      "step": 1369
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2382168235379742e-07,
+      "loss": 0.1314,
+      "step": 1370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2031625048704144e-07,
+      "loss": 0.1357,
+      "step": 1371
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1686085337423991e-07,
+      "loss": 0.1181,
+      "step": 1372
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.134555085149347e-07,
+      "loss": 0.1369,
+      "step": 1373
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1010023315518592e-07,
+      "loss": 0.1503,
+      "step": 1374
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0679504428747545e-07,
+      "loss": 0.1277,
+      "step": 1375
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0353995865063138e-07,
+      "loss": 0.0987,
+      "step": 1376
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0033499272973702e-07,
+      "loss": 0.1382,
+      "step": 1377
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.718016275604759e-08,
+      "loss": 0.1172,
+      "step": 1378
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.407548470691252e-08,
+      "loss": 0.1409,
+      "step": 1379
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.10209743056889e-08,
+      "loss": 0.1493,
+      "step": 1380
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.801664702166368e-08,
+      "loss": 0.1508,
+      "step": 1381
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.506251806997934e-08,
+      "loss": 0.1439,
+      "step": 1382
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.21586024115506e-08,
+      "loss": 0.121,
+      "step": 1383
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.930491475299229e-08,
+      "loss": 0.1743,
+      "step": 1384
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.650146954654491e-08,
+      "loss": 0.1235,
+      "step": 1385
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.37482809900003e-08,
+      "loss": 0.1337,
+      "step": 1386
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.104536302662834e-08,
+      "loss": 0.1305,
+      "step": 1387
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.839272934511143e-08,
+      "loss": 0.1374,
+      "step": 1388
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.579039337946791e-08,
+      "loss": 0.1184,
+      "step": 1389
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.323836830899321e-08,
+      "loss": 0.1536,
+      "step": 1390
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.073666705818104e-08,
+      "loss": 0.1178,
+      "step": 1391
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.828530229667228e-08,
+      "loss": 0.1527,
+      "step": 1392
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.588428643917509e-08,
+      "loss": 0.1192,
+      "step": 1393
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.353363164541825e-08,
+      "loss": 0.1218,
+      "step": 1394
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.123334982007566e-08,
+      "loss": 0.1219,
+      "step": 1395
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.898345261271531e-08,
+      "loss": 0.1588,
+      "step": 1396
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.678395141773373e-08,
+      "loss": 0.1137,
+      "step": 1397
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.463485737430606e-08,
+      "loss": 0.1579,
+      "step": 1398
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.253618136631943e-08,
+      "loss": 0.136,
+      "step": 1399
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0487934022328533e-08,
+      "loss": 0.1502,
+      "step": 1400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.849012571549349e-08,
+      "loss": 0.1402,
+      "step": 1401
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.654276656353206e-08,
+      "loss": 0.0991,
+      "step": 1402
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.464586642866752e-08,
+      "loss": 0.1216,
+      "step": 1403
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.27994349175742e-08,
+      "loss": 0.1323,
+      "step": 1404
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1003481381337576e-08,
+      "loss": 0.1336,
+      "step": 1405
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9258014915399813e-08,
+      "loss": 0.1631,
+      "step": 1406
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.756304435951429e-08,
+      "loss": 0.1292,
+      "step": 1407
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.591857829770672e-08,
+      "loss": 0.1323,
+      "step": 1408
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4324625058222974e-08,
+      "loss": 0.1153,
+      "step": 1409
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.278119271349466e-08,
+      "loss": 0.1358,
+      "step": 1410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1288289080092505e-08,
+      "loss": 0.1219,
+      "step": 1411
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.984592171869082e-08,
+      "loss": 0.1123,
+      "step": 1412
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8454097934027524e-08,
+      "loss": 0.1422,
+      "step": 1413
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.711282477486642e-08,
+      "loss": 0.1373,
+      "step": 1414
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.582210903396275e-08,
+      "loss": 0.1324,
+      "step": 1415
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4581957248026579e-08,
+      "loss": 0.1436,
+      "step": 1416
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3392375697696136e-08,
+      "loss": 0.1248,
+      "step": 1417
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2253370407495636e-08,
+      "loss": 0.1508,
+      "step": 1418
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1164947145815287e-08,
+      "loss": 0.1538,
+      "step": 1419
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0127111424872437e-08,
+      "loss": 0.1478,
+      "step": 1420
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.139868500693816e-09,
+      "loss": 0.1182,
+      "step": 1421
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.203223373078883e-09,
+      "loss": 0.158,
+      "step": 1422
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.317180785582079e-09,
+      "loss": 0.157,
+      "step": 1423
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.481745225485059e-09,
+      "loss": 0.1119,
+      "step": 1424
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.696920923774496e-09,
+      "loss": 0.1396,
+      "step": 1425
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.962711855120983e-09,
+      "loss": 0.1264,
+      "step": 1426
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.279121737859049e-09,
+      "loss": 0.1356,
+      "step": 1427
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6461540339682855e-09,
+      "loss": 0.1512,
+      "step": 1428
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.063811949056694e-09,
+      "loss": 0.1287,
+      "step": 1429
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.532098432341812e-09,
+      "loss": 0.1599,
+      "step": 1430
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.051016176637388e-09,
+      "loss": 0.1351,
+      "step": 1431
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6205676183411733e-09,
+      "loss": 0.1352,
+      "step": 1432
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2407549374204852e-09,
+      "loss": 0.1368,
+      "step": 1433
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.115800574022171e-10,
+      "loss": 0.1116,
+      "step": 1434
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.330446453617356e-10,
+      "loss": 0.1175,
+      "step": 1435
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0515011191621933e-10,
+      "loss": 0.1471,
+      "step": 1436
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2789761121688736e-10,
+      "loss": 0.1406,
+      "step": 1437
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0128804094233779e-10,
+      "loss": 0.0964,
+      "step": 1438
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5322042296327398e-11,
+      "loss": 0.103,
+      "step": 1439
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.1147,
+      "step": 1440
+    },
+    {
+      "epoch": 3.0,
+      "step": 1440,
+      "total_flos": 2.2020327372069274e+17,
+      "train_loss": 0.48756657937645087,
+      "train_runtime": 36917.0164,
+      "train_samples_per_second": 2.499,
+      "train_steps_per_second": 0.039
+    }
+  ],
+  "max_steps": 1440,
+  "num_train_epochs": 3,
+  "total_flos": 2.2020327372069274e+17,
+  "trial_name": null,
+  "trial_params": null
+}