diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,21859 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9990728340372925,
+  "global_step": 3639,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 1.9719,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.949261039804782e-06,
+      "loss": 1.6092,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.674468152928465e-06,
+      "loss": 1.709,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.898522079609564e-06,
+      "loss": 1.3722,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.847972067479439e-06,
+      "loss": 1.1749,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.623729192733248e-06,
+      "loss": 1.0124,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.279622496528684e-06,
+      "loss": 1.0029,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.847783119414348e-06,
+      "loss": 0.9601,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.34893630585693e-06,
+      "loss": 0.9194,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.797233107284223e-06,
+      "loss": 1.0078,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0202766892715779e-05,
+      "loss": 0.933,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.057299023253803e-05,
+      "loss": 0.9426,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0913562690859714e-05,
+      "loss": 0.9574,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1228883536333466e-05,
+      "loss": 0.9195,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1522440220407905e-05,
+      "loss": 0.9359,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1797044159219129e-05,
+      "loss": 0.9204,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2054994909349387e-05,
+      "loss": 0.8601,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2298197345661713e-05,
+      "loss": 0.9547,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2528247115313971e-05,
+      "loss": 0.882,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2746494147089004e-05,
+      "loss": 0.9337,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.295409064945715e-05,
+      "loss": 0.8774,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3152027932520562e-05,
+      "loss": 0.9026,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.334116503814206e-05,
+      "loss": 0.8901,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3522251272342814e-05,
+      "loss": 0.8583,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3695944134958878e-05,
+      "loss": 0.8826,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3862823730664496e-05,
+      "loss": 0.8762,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4023404458785397e-05,
+      "loss": 0.9377,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4178144576138249e-05,
+      "loss": 0.8693,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4327454081041816e-05,
+      "loss": 0.8836,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4471701260212687e-05,
+      "loss": 0.8541,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4611218161768611e-05,
+      "loss": 0.8616,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4746305199023911e-05,
+      "loss": 0.8333,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4877235045644244e-05,
+      "loss": 0.8331,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.500425594915417e-05,
+      "loss": 0.8442,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5127594564008124e-05,
+      "loss": 0.8231,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5247458385466496e-05,
+      "loss": 0.9136,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.536403784992936e-05,
+      "loss": 0.8697,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5477508155118754e-05,
+      "loss": 0.8289,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5588030843788177e-05,
+      "loss": 0.855,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5695755186893788e-05,
+      "loss": 0.864,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5800819395948124e-05,
+      "loss": 0.8082,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.590335168926193e-05,
+      "loss": 0.7626,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.600347123270875e-05,
+      "loss": 0.8118,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.610128897232534e-05,
+      "loss": 0.8165,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.619690837333637e-05,
+      "loss": 0.8346,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6290426077946845e-05,
+      "loss": 0.8612,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.638193249238684e-05,
+      "loss": 0.8752,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.64715123121476e-05,
+      "loss": 0.8263,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6559244993057368e-05,
+      "loss": 0.8442,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.664520517476366e-05,
+      "loss": 0.8176,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6729463062277852e-05,
+      "loss": 0.8302,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.681208477046928e-05,
+      "loss": 0.8514,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6893132635744143e-05,
+      "loss": 0.8139,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6972665498590178e-05,
+      "loss": 0.8686,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7050738960195217e-05,
+      "loss": 0.8637,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7127405615943035e-05,
+      "loss": 0.7968,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7202715268242437e-05,
+      "loss": 0.869,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.72767151208466e-05,
+      "loss": 0.8382,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.734944995656128e-05,
+      "loss": 0.8222,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.742096230001747e-05,
+      "loss": 0.8081,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7491292566989765e-05,
+      "loss": 0.8425,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7560479201573396e-05,
+      "loss": 0.8434,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7628558802385615e-05,
+      "loss": 0.8068,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7695566238828696e-05,
+      "loss": 0.8422,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.776153475833915e-05,
+      "loss": 0.8064,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7826496085449024e-05,
+      "loss": 0.8109,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.789048051339804e-05,
+      "loss": 0.87,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7953516988958953e-05,
+      "loss": 0.8233,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.801563319107053e-05,
+      "loss": 0.7784,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8076855603812908e-05,
+      "loss": 0.8075,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8137209584206844e-05,
+      "loss": 0.8782,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.819671942527128e-05,
+      "loss": 0.8384,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8255408414731658e-05,
+      "loss": 0.8307,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8313298889734142e-05,
+      "loss": 0.7907,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8370412287887345e-05,
+      "loss": 0.8429,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8426769194923538e-05,
+      "loss": 0.857,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.848238938924446e-05,
+      "loss": 0.8278,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.853729188359296e-05,
+      "loss": 0.8537,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8591494964070173e-05,
+      "loss": 0.8334,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.864501622669857e-05,
+      "loss": 0.8328,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.869787261171386e-05,
+      "loss": 0.7944,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8750080435752912e-05,
+      "loss": 0.8219,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.880165542209072e-05,
+      "loss": 0.8243,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8852612729066712e-05,
+      "loss": 0.8119,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8902966976828826e-05,
+      "loss": 0.7879,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.895273227251353e-05,
+      "loss": 0.8387,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.900192223397028e-05,
+      "loss": 0.8177,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9050550012130125e-05,
+      "loss": 0.7856,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.909862831211055e-05,
+      "loss": 0.8418,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9146169413141155e-05,
+      "loss": 0.8146,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9193185187388398e-05,
+      "loss": 0.8014,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9239687117751626e-05,
+      "loss": 0.8305,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.928568631469708e-05,
+      "loss": 0.8048,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9331193532191624e-05,
+      "loss": 0.7904,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.937621918279341e-05,
+      "loss": 0.783,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.942077335195238e-05,
+      "loss": 0.8395,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9464865811569746e-05,
+      "loss": 0.8221,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9508506032862152e-05,
+      "loss": 0.8109,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9551703198572708e-05,
+      "loss": 0.7325,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9594466214568445e-05,
+      "loss": 0.7972,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.96368037208607e-05,
+      "loss": 0.8008,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9678724102082633e-05,
+      "loss": 0.812,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720235497455603e-05,
+      "loss": 0.81,
+      "step": 103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.976134581027406e-05,
+      "loss": 0.8101,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980206271693659e-05,
+      "loss": 0.8272,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842393675548924e-05,
+      "loss": 0.8249,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9882345934123056e-05,
+      "loss": 0.8585,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9921926538394962e-05,
+      "loss": 0.8452,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9961142339282028e-05,
+      "loss": 0.8195,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.7686,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.8302,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9994332672145086e-05,
+      "loss": 0.8507,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988665344290167e-05,
+      "loss": 0.7795,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9982998016435254e-05,
+      "loss": 0.7935,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9977330688580335e-05,
+      "loss": 0.7661,
+      "step": 115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997166336072542e-05,
+      "loss": 0.7838,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965996032870503e-05,
+      "loss": 0.8302,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9960328705015587e-05,
+      "loss": 0.7688,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.995466137716067e-05,
+      "loss": 0.7932,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9948994049305756e-05,
+      "loss": 0.8479,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9943326721450837e-05,
+      "loss": 0.7638,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.993765939359592e-05,
+      "loss": 0.7822,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9931992065741005e-05,
+      "loss": 0.7676,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992632473788609e-05,
+      "loss": 0.8223,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9920657410031173e-05,
+      "loss": 0.7892,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9914990082176254e-05,
+      "loss": 0.824,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.990932275432134e-05,
+      "loss": 0.744,
+      "step": 127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9903655426466423e-05,
+      "loss": 0.7993,
+      "step": 128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9897988098611507e-05,
+      "loss": 0.7445,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.989232077075659e-05,
+      "loss": 0.7999,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.988665344290167e-05,
+      "loss": 0.7884,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9880986115046756e-05,
+      "loss": 0.7277,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987531878719184e-05,
+      "loss": 0.7851,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9869651459336924e-05,
+      "loss": 0.7269,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.986398413148201e-05,
+      "loss": 0.7999,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9858316803627093e-05,
+      "loss": 0.7717,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9852649475772173e-05,
+      "loss": 0.7529,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.984698214791726e-05,
+      "loss": 0.7585,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.984131482006234e-05,
+      "loss": 0.7773,
+      "step": 139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9835647492207426e-05,
+      "loss": 0.8281,
+      "step": 140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.982998016435251e-05,
+      "loss": 0.7566,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.982431283649759e-05,
+      "loss": 0.7538,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.981864550864268e-05,
+      "loss": 0.7365,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.981297818078776e-05,
+      "loss": 0.793,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9807310852932843e-05,
+      "loss": 0.8285,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9801643525077928e-05,
+      "loss": 0.8424,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9795976197223012e-05,
+      "loss": 0.796,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9790308869368096e-05,
+      "loss": 0.8073,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.978464154151318e-05,
+      "loss": 0.7997,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.977897421365826e-05,
+      "loss": 0.7939,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9773306885803345e-05,
+      "loss": 0.7395,
+      "step": 151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.976763955794843e-05,
+      "loss": 0.7688,
+      "step": 152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9761972230093513e-05,
+      "loss": 0.8181,
+      "step": 153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9756304902238598e-05,
+      "loss": 0.8313,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.975063757438368e-05,
+      "loss": 0.7604,
+      "step": 155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9744970246528763e-05,
+      "loss": 0.7857,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9739302918673847e-05,
+      "loss": 0.7559,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.973363559081893e-05,
+      "loss": 0.7644,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9727968262964015e-05,
+      "loss": 0.776,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.97223009351091e-05,
+      "loss": 0.806,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.971663360725418e-05,
+      "loss": 0.7337,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9710966279399264e-05,
+      "loss": 0.7954,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.970529895154435e-05,
+      "loss": 0.769,
+      "step": 163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9699631623689433e-05,
+      "loss": 0.8161,
+      "step": 164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9693964295834517e-05,
+      "loss": 0.7932,
+      "step": 165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9688296967979598e-05,
+      "loss": 0.7911,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9682629640124682e-05,
+      "loss": 0.7419,
+      "step": 167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9676962312269766e-05,
+      "loss": 0.7574,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.967129498441485e-05,
+      "loss": 0.7748,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9665627656559934e-05,
+      "loss": 0.7115,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9659960328705015e-05,
+      "loss": 0.7662,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.96542930008501e-05,
+      "loss": 0.786,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9648625672995187e-05,
+      "loss": 0.7651,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9642958345140268e-05,
+      "loss": 0.8218,
+      "step": 174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9637291017285352e-05,
+      "loss": 0.7535,
+      "step": 175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9631623689430436e-05,
+      "loss": 0.7802,
+      "step": 176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9625956361575517e-05,
+      "loss": 0.788,
+      "step": 177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9620289033720604e-05,
+      "loss": 0.7947,
+      "step": 178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9614621705865685e-05,
+      "loss": 0.7739,
+      "step": 179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.960895437801077e-05,
+      "loss": 0.7553,
+      "step": 180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9603287050155853e-05,
+      "loss": 0.7523,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9597619722300934e-05,
+      "loss": 0.7907,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9591952394446022e-05,
+      "loss": 0.7646,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9586285066591103e-05,
+      "loss": 0.7823,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9580617738736187e-05,
+      "loss": 0.7704,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.957495041088127e-05,
+      "loss": 0.7673,
+      "step": 186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9569283083026355e-05,
+      "loss": 0.7127,
+      "step": 187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.956361575517144e-05,
+      "loss": 0.7778,
+      "step": 188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9557948427316523e-05,
+      "loss": 0.7445,
+      "step": 189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9552281099461604e-05,
+      "loss": 0.7761,
+      "step": 190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.954661377160669e-05,
+      "loss": 0.7921,
+      "step": 191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9540946443751773e-05,
+      "loss": 0.7865,
+      "step": 192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9535279115896857e-05,
+      "loss": 0.7994,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.952961178804194e-05,
+      "loss": 0.8106,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9523944460187022e-05,
+      "loss": 0.7468,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9518277132332106e-05,
+      "loss": 0.7911,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.951260980447719e-05,
+      "loss": 0.8131,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9506942476622274e-05,
+      "loss": 0.7529,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.950127514876736e-05,
+      "loss": 0.7749,
+      "step": 199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9495607820912443e-05,
+      "loss": 0.736,
+      "step": 200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9489940493057523e-05,
+      "loss": 0.7817,
+      "step": 201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.948427316520261e-05,
+      "loss": 0.6713,
+      "step": 202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9478605837347692e-05,
+      "loss": 0.7621,
+      "step": 203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9472938509492776e-05,
+      "loss": 0.7208,
+      "step": 204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.946727118163786e-05,
+      "loss": 0.7241,
+      "step": 205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.946160385378294e-05,
+      "loss": 0.7727,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.945593652592803e-05,
+      "loss": 0.804,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.945026919807311e-05,
+      "loss": 0.7673,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9444601870218194e-05,
+      "loss": 0.745,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9438934542363278e-05,
+      "loss": 0.7777,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.943326721450836e-05,
+      "loss": 0.771,
+      "step": 211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9427599886653446e-05,
+      "loss": 0.7474,
+      "step": 212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.942193255879853e-05,
+      "loss": 0.6783,
+      "step": 213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.941626523094361e-05,
+      "loss": 0.8169,
+      "step": 214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9410597903088695e-05,
+      "loss": 0.767,
+      "step": 215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.940493057523378e-05,
+      "loss": 0.7824,
+      "step": 216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9399263247378864e-05,
+      "loss": 0.798,
+      "step": 217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9393595919523948e-05,
+      "loss": 0.7943,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.938792859166903e-05,
+      "loss": 0.7446,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9382261263814113e-05,
+      "loss": 0.7085,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9376593935959197e-05,
+      "loss": 0.8119,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.937092660810428e-05,
+      "loss": 0.7651,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9365259280249365e-05,
+      "loss": 0.7585,
+      "step": 223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9359591952394446e-05,
+      "loss": 0.8155,
+      "step": 224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.935392462453953e-05,
+      "loss": 0.7755,
+      "step": 225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9348257296684614e-05,
+      "loss": 0.8351,
+      "step": 226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.93425899688297e-05,
+      "loss": 0.7438,
+      "step": 227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9336922640974783e-05,
+      "loss": 0.7594,
+      "step": 228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9331255313119867e-05,
+      "loss": 0.7617,
+      "step": 229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9325587985264948e-05,
+      "loss": 0.7901,
+      "step": 230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9319920657410032e-05,
+      "loss": 0.7443,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9314253329555116e-05,
+      "loss": 0.7492,
+      "step": 232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.93085860017002e-05,
+      "loss": 0.7063,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9302918673845284e-05,
+      "loss": 0.8104,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9297251345990365e-05,
+      "loss": 0.7244,
+      "step": 235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.929158401813545e-05,
+      "loss": 0.7426,
+      "step": 236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9285916690280534e-05,
+      "loss": 0.7431,
+      "step": 237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9280249362425618e-05,
+      "loss": 0.7034,
+      "step": 238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9274582034570702e-05,
+      "loss": 0.7329,
+      "step": 239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9268914706715786e-05,
+      "loss": 0.7397,
+      "step": 240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9263247378860867e-05,
+      "loss": 0.7531,
+      "step": 241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9257580051005954e-05,
+      "loss": 0.7795,
+      "step": 242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9251912723151035e-05,
+      "loss": 0.7783,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.924624539529612e-05,
+      "loss": 0.7272,
+      "step": 244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9240578067441204e-05,
+      "loss": 0.7508,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9234910739586288e-05,
+      "loss": 0.7737,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9229243411731372e-05,
+      "loss": 0.7039,
+      "step": 247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9223576083876453e-05,
+      "loss": 0.7639,
+      "step": 248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9217908756021537e-05,
+      "loss": 0.7658,
+      "step": 249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.921224142816662e-05,
+      "loss": 0.7511,
+      "step": 250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9206574100311705e-05,
+      "loss": 0.7105,
+      "step": 251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.920090677245679e-05,
+      "loss": 0.7537,
+      "step": 252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.919523944460187e-05,
+      "loss": 0.7093,
+      "step": 253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9189572116746954e-05,
+      "loss": 0.7438,
+      "step": 254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.918390478889204e-05,
+      "loss": 0.6905,
+      "step": 255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9178237461037123e-05,
+      "loss": 0.7615,
+      "step": 256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9172570133182207e-05,
+      "loss": 0.7442,
+      "step": 257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.916690280532729e-05,
+      "loss": 0.7817,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9161235477472372e-05,
+      "loss": 0.7463,
+      "step": 259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9155568149617456e-05,
+      "loss": 0.783,
+      "step": 260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.914990082176254e-05,
+      "loss": 0.7689,
+      "step": 261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9144233493907624e-05,
+      "loss": 0.7547,
+      "step": 262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.913856616605271e-05,
+      "loss": 0.7277,
+      "step": 263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.913289883819779e-05,
+      "loss": 0.7328,
+      "step": 264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9127231510342874e-05,
+      "loss": 0.7368,
+      "step": 265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9121564182487958e-05,
+      "loss": 0.775,
+      "step": 266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9115896854633042e-05,
+      "loss": 0.7221,
+      "step": 267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9110229526778126e-05,
+      "loss": 0.7729,
+      "step": 268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.910456219892321e-05,
+      "loss": 0.7342,
+      "step": 269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.909889487106829e-05,
+      "loss": 0.7247,
+      "step": 270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.909322754321338e-05,
+      "loss": 0.7288,
+      "step": 271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.908756021535846e-05,
+      "loss": 0.7132,
+      "step": 272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9081892887503544e-05,
+      "loss": 0.7461,
+      "step": 273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9076225559648628e-05,
+      "loss": 0.75,
+      "step": 274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.907055823179371e-05,
+      "loss": 0.7639,
+      "step": 275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9064890903938796e-05,
+      "loss": 0.7639,
+      "step": 276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9059223576083877e-05,
+      "loss": 0.7,
+      "step": 277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.905355624822896e-05,
+      "loss": 0.7058,
+      "step": 278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9047888920374045e-05,
+      "loss": 0.7292,
+      "step": 279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.904222159251913e-05,
+      "loss": 0.7612,
+      "step": 280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9036554264664214e-05,
+      "loss": 0.7662,
+      "step": 281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9030886936809298e-05,
+      "loss": 0.7901,
+      "step": 282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.902521960895438e-05,
+      "loss": 0.8134,
+      "step": 283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9019552281099463e-05,
+      "loss": 0.753,
+      "step": 284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9013884953244547e-05,
+      "loss": 0.6812,
+      "step": 285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.900821762538963e-05,
+      "loss": 0.7947,
+      "step": 286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9002550297534715e-05,
+      "loss": 0.7559,
+      "step": 287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8996882969679796e-05,
+      "loss": 0.7069,
+      "step": 288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.899121564182488e-05,
+      "loss": 0.7716,
+      "step": 289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8985548313969965e-05,
+      "loss": 0.6966,
+      "step": 290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.897988098611505e-05,
+      "loss": 0.7294,
+      "step": 291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8974213658260133e-05,
+      "loss": 0.7164,
+      "step": 292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8968546330405214e-05,
+      "loss": 0.7076,
+      "step": 293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8962879002550298e-05,
+      "loss": 0.7305,
+      "step": 294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8957211674695382e-05,
+      "loss": 0.7318,
+      "step": 295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8951544346840466e-05,
+      "loss": 0.7158,
+      "step": 296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.894587701898555e-05,
+      "loss": 0.7716,
+      "step": 297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8940209691130635e-05,
+      "loss": 0.769,
+      "step": 298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8934542363275715e-05,
+      "loss": 0.7418,
+      "step": 299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8928875035420803e-05,
+      "loss": 0.7365,
+      "step": 300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8923207707565884e-05,
+      "loss": 0.7601,
+      "step": 301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8917540379710968e-05,
+      "loss": 0.7331,
+      "step": 302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8911873051856052e-05,
+      "loss": 0.685,
+      "step": 303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8906205724001133e-05,
+      "loss": 0.7869,
+      "step": 304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.890053839614622e-05,
+      "loss": 0.7118,
+      "step": 305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.88948710682913e-05,
+      "loss": 0.7627,
+      "step": 306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8889203740436385e-05,
+      "loss": 0.7407,
+      "step": 307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.888353641258147e-05,
+      "loss": 0.7161,
+      "step": 308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8877869084726554e-05,
+      "loss": 0.7548,
+      "step": 309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8872201756871638e-05,
+      "loss": 0.7818,
+      "step": 310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8866534429016722e-05,
+      "loss": 0.772,
+      "step": 311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8860867101161803e-05,
+      "loss": 0.7839,
+      "step": 312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8855199773306887e-05,
+      "loss": 0.7453,
+      "step": 313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.884953244545197e-05,
+      "loss": 0.7649,
+      "step": 314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8843865117597055e-05,
+      "loss": 0.7784,
+      "step": 315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.883819778974214e-05,
+      "loss": 0.7566,
+      "step": 316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.883253046188722e-05,
+      "loss": 0.8249,
+      "step": 317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8826863134032305e-05,
+      "loss": 0.723,
+      "step": 318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.882119580617739e-05,
+      "loss": 0.6916,
+      "step": 319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8815528478322473e-05,
+      "loss": 0.7069,
+      "step": 320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8809861150467557e-05,
+      "loss": 0.7224,
+      "step": 321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.880419382261264e-05,
+      "loss": 0.7349,
+      "step": 322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8798526494757722e-05,
+      "loss": 0.6954,
+      "step": 323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8792859166902806e-05,
+      "loss": 0.796,
+      "step": 324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.878719183904789e-05,
+      "loss": 0.7254,
+      "step": 325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8781524511192975e-05,
+      "loss": 0.7098,
+      "step": 326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.877585718333806e-05,
+      "loss": 0.7331,
+      "step": 327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.877018985548314e-05,
+      "loss": 0.6958,
+      "step": 328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8764522527628224e-05,
+      "loss": 0.7354,
+      "step": 329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8758855199773308e-05,
+      "loss": 0.701,
+      "step": 330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8753187871918392e-05,
+      "loss": 0.7168,
+      "step": 331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8747520544063476e-05,
+      "loss": 0.7352,
+      "step": 332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8741853216208557e-05,
+      "loss": 0.7042,
+      "step": 333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.873618588835364e-05,
+      "loss": 0.7239,
+      "step": 334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.873051856049873e-05,
+      "loss": 0.7023,
+      "step": 335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.872485123264381e-05,
+      "loss": 0.7255,
+      "step": 336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8719183904788894e-05,
+      "loss": 0.6881,
+      "step": 337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8713516576933978e-05,
+      "loss": 0.7347,
+      "step": 338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.870784924907906e-05,
+      "loss": 0.7321,
+      "step": 339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8702181921224146e-05,
+      "loss": 0.7243,
+      "step": 340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8696514593369227e-05,
+      "loss": 0.7142,
+      "step": 341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.869084726551431e-05,
+      "loss": 0.6813,
+      "step": 342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8685179937659395e-05,
+      "loss": 0.7418,
+      "step": 343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8679512609804476e-05,
+      "loss": 0.7154,
+      "step": 344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8673845281949564e-05,
+      "loss": 0.6925,
+      "step": 345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8668177954094645e-05,
+      "loss": 0.7339,
+      "step": 346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866251062623973e-05,
+      "loss": 0.7184,
+      "step": 347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8656843298384813e-05,
+      "loss": 0.647,
+      "step": 348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8651175970529897e-05,
+      "loss": 0.7431,
+      "step": 349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864550864267498e-05,
+      "loss": 0.701,
+      "step": 350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8639841314820066e-05,
+      "loss": 0.7256,
+      "step": 351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8634173986965146e-05,
+      "loss": 0.7371,
+      "step": 352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862850665911023e-05,
+      "loss": 0.7548,
+      "step": 353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8622839331255315e-05,
+      "loss": 0.7591,
+      "step": 354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.86171720034004e-05,
+      "loss": 0.7309,
+      "step": 355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8611504675545483e-05,
+      "loss": 0.683,
+      "step": 356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8605837347690564e-05,
+      "loss": 0.686,
+      "step": 357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8600170019835648e-05,
+      "loss": 0.7678,
+      "step": 358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8594502691980732e-05,
+      "loss": 0.7882,
+      "step": 359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8588835364125816e-05,
+      "loss": 0.7503,
+      "step": 360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.85831680362709e-05,
+      "loss": 0.7472,
+      "step": 361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8577500708415985e-05,
+      "loss": 0.7547,
+      "step": 362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8571833380561065e-05,
+      "loss": 0.7263,
+      "step": 363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8566166052706153e-05,
+      "loss": 0.7489,
+      "step": 364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8560498724851234e-05,
+      "loss": 0.7261,
+      "step": 365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8554831396996318e-05,
+      "loss": 0.7322,
+      "step": 366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8549164069141402e-05,
+      "loss": 0.7077,
+      "step": 367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8543496741286483e-05,
+      "loss": 0.722,
+      "step": 368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853782941343157e-05,
+      "loss": 0.7306,
+      "step": 369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853216208557665e-05,
+      "loss": 0.673,
+      "step": 370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8526494757721736e-05,
+      "loss": 0.7381,
+      "step": 371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.852082742986682e-05,
+      "loss": 0.6959,
+      "step": 372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.85151601020119e-05,
+      "loss": 0.7527,
+      "step": 373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8509492774156988e-05,
+      "loss": 0.7285,
+      "step": 374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8503825446302072e-05,
+      "loss": 0.7283,
+      "step": 375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8498158118447153e-05,
+      "loss": 0.7418,
+      "step": 376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8492490790592237e-05,
+      "loss": 0.7751,
+      "step": 377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.848682346273732e-05,
+      "loss": 0.7468,
+      "step": 378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8481156134882406e-05,
+      "loss": 0.7521,
+      "step": 379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.847548880702749e-05,
+      "loss": 0.6884,
+      "step": 380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.846982147917257e-05,
+      "loss": 0.7475,
+      "step": 381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8464154151317655e-05,
+      "loss": 0.7062,
+      "step": 382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.845848682346274e-05,
+      "loss": 0.7447,
+      "step": 383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8452819495607823e-05,
+      "loss": 0.741,
+      "step": 384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8447152167752907e-05,
+      "loss": 0.7174,
+      "step": 385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8441484839897988e-05,
+      "loss": 0.769,
+      "step": 386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8435817512043072e-05,
+      "loss": 0.7247,
+      "step": 387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8430150184188156e-05,
+      "loss": 0.7509,
+      "step": 388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.842448285633324e-05,
+      "loss": 0.74,
+      "step": 389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8418815528478325e-05,
+      "loss": 0.695,
+      "step": 390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.841314820062341e-05,
+      "loss": 0.7388,
+      "step": 391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.840748087276849e-05,
+      "loss": 0.6904,
+      "step": 392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8401813544913574e-05,
+      "loss": 0.7235,
+      "step": 393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8396146217058658e-05,
+      "loss": 0.7503,
+      "step": 394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8390478889203742e-05,
+      "loss": 0.6892,
+      "step": 395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8384811561348826e-05,
+      "loss": 0.7069,
+      "step": 396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8379144233493907e-05,
+      "loss": 0.7096,
+      "step": 397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.837347690563899e-05,
+      "loss": 0.7857,
+      "step": 398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8367809577784076e-05,
+      "loss": 0.6896,
+      "step": 399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.836214224992916e-05,
+      "loss": 0.7097,
+      "step": 400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8356474922074244e-05,
+      "loss": 0.7333,
+      "step": 401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8350807594219328e-05,
+      "loss": 0.685,
+      "step": 402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.834514026636441e-05,
+      "loss": 0.7125,
+      "step": 403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8339472938509496e-05,
+      "loss": 0.7269,
+      "step": 404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8333805610654577e-05,
+      "loss": 0.7073,
+      "step": 405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.832813828279966e-05,
+      "loss": 0.7435,
+      "step": 406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8322470954944746e-05,
+      "loss": 0.7232,
+      "step": 407
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.831680362708983e-05,
+      "loss": 0.7612,
+      "step": 408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8311136299234914e-05,
+      "loss": 0.7661,
+      "step": 409
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8305468971379995e-05,
+      "loss": 0.7596,
+      "step": 410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.829980164352508e-05,
+      "loss": 0.7449,
+      "step": 411
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8294134315670163e-05,
+      "loss": 0.7007,
+      "step": 412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8288466987815247e-05,
+      "loss": 0.6853,
+      "step": 413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.828279965996033e-05,
+      "loss": 0.7234,
+      "step": 414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8277132332105412e-05,
+      "loss": 0.7274,
+      "step": 415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8271465004250496e-05,
+      "loss": 0.6875,
+      "step": 416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.826579767639558e-05,
+      "loss": 0.7148,
+      "step": 417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8260130348540665e-05,
+      "loss": 0.7624,
+      "step": 418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.825446302068575e-05,
+      "loss": 0.7006,
+      "step": 419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8248795692830833e-05,
+      "loss": 0.6812,
+      "step": 420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8243128364975914e-05,
+      "loss": 0.7626,
+      "step": 421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8237461037120998e-05,
+      "loss": 0.6867,
+      "step": 422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8231793709266082e-05,
+      "loss": 0.6939,
+      "step": 423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8226126381411166e-05,
+      "loss": 0.7752,
+      "step": 424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.822045905355625e-05,
+      "loss": 0.7097,
+      "step": 425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.821479172570133e-05,
+      "loss": 0.6833,
+      "step": 426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8209124397846416e-05,
+      "loss": 0.6892,
+      "step": 427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.82034570699915e-05,
+      "loss": 0.7055,
+      "step": 428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8197789742136584e-05,
+      "loss": 0.7164,
+      "step": 429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8192122414281668e-05,
+      "loss": 0.729,
+      "step": 430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8186455086426752e-05,
+      "loss": 0.779,
+      "step": 431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8180787758571833e-05,
+      "loss": 0.7523,
+      "step": 432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.817512043071692e-05,
+      "loss": 0.7152,
+      "step": 433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8169453102862e-05,
+      "loss": 0.6811,
+      "step": 434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8163785775007086e-05,
+      "loss": 0.7226,
+      "step": 435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.815811844715217e-05,
+      "loss": 0.7194,
+      "step": 436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.815245111929725e-05,
+      "loss": 0.6955,
+      "step": 437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8146783791442338e-05,
+      "loss": 0.7676,
+      "step": 438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.814111646358742e-05,
+      "loss": 0.7413,
+      "step": 439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8135449135732503e-05,
+      "loss": 0.6911,
+      "step": 440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8129781807877587e-05,
+      "loss": 0.6707,
+      "step": 441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8124114480022668e-05,
+      "loss": 0.6974,
+      "step": 442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8118447152167756e-05,
+      "loss": 0.6775,
+      "step": 443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.811277982431284e-05,
+      "loss": 0.7446,
+      "step": 444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.810711249645792e-05,
+      "loss": 0.6617,
+      "step": 445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8101445168603005e-05,
+      "loss": 0.675,
+      "step": 446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.809577784074809e-05,
+      "loss": 0.6851,
+      "step": 447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8090110512893173e-05,
+      "loss": 0.7094,
+      "step": 448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8084443185038257e-05,
+      "loss": 0.7013,
+      "step": 449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8078775857183338e-05,
+      "loss": 0.708,
+      "step": 450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8073108529328422e-05,
+      "loss": 0.6964,
+      "step": 451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8067441201473507e-05,
+      "loss": 0.6777,
+      "step": 452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.806177387361859e-05,
+      "loss": 0.693,
+      "step": 453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8056106545763675e-05,
+      "loss": 0.74,
+      "step": 454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8050439217908756e-05,
+      "loss": 0.6574,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.804477189005384e-05,
+      "loss": 0.7084,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8039104562198924e-05,
+      "loss": 0.6921,
+      "step": 457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8033437234344008e-05,
+      "loss": 0.7031,
+      "step": 458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8027769906489092e-05,
+      "loss": 0.6624,
+      "step": 459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8022102578634177e-05,
+      "loss": 0.7246,
+      "step": 460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8016435250779257e-05,
+      "loss": 0.7305,
+      "step": 461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8010767922924345e-05,
+      "loss": 0.7943,
+      "step": 462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8005100595069426e-05,
+      "loss": 0.7298,
+      "step": 463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.799943326721451e-05,
+      "loss": 0.7064,
+      "step": 464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7993765939359594e-05,
+      "loss": 0.7212,
+      "step": 465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7988098611504675e-05,
+      "loss": 0.6883,
+      "step": 466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7982431283649762e-05,
+      "loss": 0.6723,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7976763955794843e-05,
+      "loss": 0.6806,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7971096627939927e-05,
+      "loss": 0.7287,
+      "step": 469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.796542930008501e-05,
+      "loss": 0.6785,
+      "step": 470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7959761972230096e-05,
+      "loss": 0.7621,
+      "step": 471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.795409464437518e-05,
+      "loss": 0.6488,
+      "step": 472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7948427316520264e-05,
+      "loss": 0.7509,
+      "step": 473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7942759988665345e-05,
+      "loss": 0.7274,
+      "step": 474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.793709266081043e-05,
+      "loss": 0.6829,
+      "step": 475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7931425332955513e-05,
+      "loss": 0.7675,
+      "step": 476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7925758005100597e-05,
+      "loss": 0.7397,
+      "step": 477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.792009067724568e-05,
+      "loss": 0.7225,
+      "step": 478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7914423349390762e-05,
+      "loss": 0.7404,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7908756021535847e-05,
+      "loss": 0.7049,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.790308869368093e-05,
+      "loss": 0.7422,
+      "step": 481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7897421365826015e-05,
+      "loss": 0.6525,
+      "step": 482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.78917540379711e-05,
+      "loss": 0.7223,
+      "step": 483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7886086710116183e-05,
+      "loss": 0.7108,
+      "step": 484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7880419382261264e-05,
+      "loss": 0.7037,
+      "step": 485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7874752054406348e-05,
+      "loss": 0.6848,
+      "step": 486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7869084726551432e-05,
+      "loss": 0.6648,
+      "step": 487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7863417398696517e-05,
+      "loss": 0.7051,
+      "step": 488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.78577500708416e-05,
+      "loss": 0.7214,
+      "step": 489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.785208274298668e-05,
+      "loss": 0.694,
+      "step": 490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7846415415131766e-05,
+      "loss": 0.7461,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.784074808727685e-05,
+      "loss": 0.7113,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7835080759421934e-05,
+      "loss": 0.717,
+      "step": 493
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7829413431567018e-05,
+      "loss": 0.6968,
+      "step": 494
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.78237461037121e-05,
+      "loss": 0.719,
+      "step": 495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7818078775857183e-05,
+      "loss": 0.6946,
+      "step": 496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.781241144800227e-05,
+      "loss": 0.7232,
+      "step": 497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.780674412014735e-05,
+      "loss": 0.6873,
+      "step": 498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7801076792292436e-05,
+      "loss": 0.6354,
+      "step": 499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.779540946443752e-05,
+      "loss": 0.687,
+      "step": 500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.77897421365826e-05,
+      "loss": 0.6802,
+      "step": 501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.778407480872769e-05,
+      "loss": 0.6908,
+      "step": 502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.777840748087277e-05,
+      "loss": 0.6904,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7772740153017853e-05,
+      "loss": 0.6887,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7767072825162937e-05,
+      "loss": 0.7569,
+      "step": 505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7761405497308018e-05,
+      "loss": 0.7313,
+      "step": 506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7755738169453106e-05,
+      "loss": 0.7095,
+      "step": 507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7750070841598187e-05,
+      "loss": 0.7461,
+      "step": 508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.774440351374327e-05,
+      "loss": 0.6844,
+      "step": 509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7738736185888355e-05,
+      "loss": 0.7011,
+      "step": 510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.773306885803344e-05,
+      "loss": 0.7004,
+      "step": 511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7727401530178523e-05,
+      "loss": 0.7271,
+      "step": 512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7721734202323608e-05,
+      "loss": 0.7029,
+      "step": 513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.771606687446869e-05,
+      "loss": 0.6734,
+      "step": 514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7710399546613772e-05,
+      "loss": 0.6829,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7704732218758857e-05,
+      "loss": 0.7148,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.769906489090394e-05,
+      "loss": 0.6813,
+      "step": 517
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7693397563049025e-05,
+      "loss": 0.707,
+      "step": 518
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7687730235194106e-05,
+      "loss": 0.6793,
+      "step": 519
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.768206290733919e-05,
+      "loss": 0.6756,
+      "step": 520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7676395579484274e-05,
+      "loss": 0.7053,
+      "step": 521
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.767072825162936e-05,
+      "loss": 0.643,
+      "step": 522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7665060923774443e-05,
+      "loss": 0.744,
+      "step": 523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7659393595919527e-05,
+      "loss": 0.6647,
+      "step": 524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7653726268064607e-05,
+      "loss": 0.6684,
+      "step": 525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7648058940209695e-05,
+      "loss": 0.6619,
+      "step": 526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7642391612354776e-05,
+      "loss": 0.7079,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.763672428449986e-05,
+      "loss": 0.7221,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7631056956644944e-05,
+      "loss": 0.7104,
+      "step": 529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7625389628790025e-05,
+      "loss": 0.7065,
+      "step": 530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7619722300935113e-05,
+      "loss": 0.6903,
+      "step": 531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7614054973080193e-05,
+      "loss": 0.7045,
+      "step": 532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7608387645225278e-05,
+      "loss": 0.7001,
+      "step": 533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7602720317370362e-05,
+      "loss": 0.6679,
+      "step": 534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7597052989515443e-05,
+      "loss": 0.7485,
+      "step": 535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.759138566166053e-05,
+      "loss": 0.6639,
+      "step": 536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.758571833380561e-05,
+      "loss": 0.7212,
+      "step": 537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7580051005950695e-05,
+      "loss": 0.6884,
+      "step": 538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.757438367809578e-05,
+      "loss": 0.7019,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7568716350240863e-05,
+      "loss": 0.7168,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7563049022385948e-05,
+      "loss": 0.6954,
+      "step": 541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7557381694531032e-05,
+      "loss": 0.6929,
+      "step": 542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7551714366676113e-05,
+      "loss": 0.6425,
+      "step": 543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7546047038821197e-05,
+      "loss": 0.7062,
+      "step": 544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.754037971096628e-05,
+      "loss": 0.7099,
+      "step": 545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7534712383111365e-05,
+      "loss": 0.7014,
+      "step": 546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.752904505525645e-05,
+      "loss": 0.6498,
+      "step": 547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.752337772740153e-05,
+      "loss": 0.6359,
+      "step": 548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7517710399546614e-05,
+      "loss": 0.6879,
+      "step": 549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.75120430716917e-05,
+      "loss": 0.7067,
+      "step": 550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7506375743836783e-05,
+      "loss": 0.6763,
+      "step": 551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7500708415981867e-05,
+      "loss": 0.7221,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.749504108812695e-05,
+      "loss": 0.6847,
+      "step": 553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7489373760272032e-05,
+      "loss": 0.6781,
+      "step": 554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7483706432417116e-05,
+      "loss": 0.6668,
+      "step": 555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.74780391045622e-05,
+      "loss": 0.7068,
+      "step": 556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7472371776707284e-05,
+      "loss": 0.6901,
+      "step": 557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.746670444885237e-05,
+      "loss": 0.6327,
+      "step": 558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.746103712099745e-05,
+      "loss": 0.6891,
+      "step": 559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7455369793142533e-05,
+      "loss": 0.6922,
+      "step": 560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7449702465287618e-05,
+      "loss": 0.7134,
+      "step": 561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7444035137432702e-05,
+      "loss": 0.6929,
+      "step": 562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7438367809577786e-05,
+      "loss": 0.7003,
+      "step": 563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.743270048172287e-05,
+      "loss": 0.6962,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.742703315386795e-05,
+      "loss": 0.7219,
+      "step": 565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.742136582601304e-05,
+      "loss": 0.7359,
+      "step": 566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.741569849815812e-05,
+      "loss": 0.6739,
+      "step": 567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7410031170303203e-05,
+      "loss": 0.726,
+      "step": 568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7404363842448288e-05,
+      "loss": 0.6881,
+      "step": 569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7398696514593372e-05,
+      "loss": 0.636,
+      "step": 570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7393029186738456e-05,
+      "loss": 0.7186,
+      "step": 571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7387361858883537e-05,
+      "loss": 0.6857,
+      "step": 572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.738169453102862e-05,
+      "loss": 0.6938,
+      "step": 573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7376027203173705e-05,
+      "loss": 0.7115,
+      "step": 574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.737035987531879e-05,
+      "loss": 0.6594,
+      "step": 575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7364692547463873e-05,
+      "loss": 0.6611,
+      "step": 576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7359025219608954e-05,
+      "loss": 0.7315,
+      "step": 577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.735335789175404e-05,
+      "loss": 0.693,
+      "step": 578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7347690563899123e-05,
+      "loss": 0.6941,
+      "step": 579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7342023236044207e-05,
+      "loss": 0.6682,
+      "step": 580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.733635590818929e-05,
+      "loss": 0.7196,
+      "step": 581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7330688580334375e-05,
+      "loss": 0.6354,
+      "step": 582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7325021252479456e-05,
+      "loss": 0.7366,
+      "step": 583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.731935392462454e-05,
+      "loss": 0.6912,
+      "step": 584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7313686596769624e-05,
+      "loss": 0.7209,
+      "step": 585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.730801926891471e-05,
+      "loss": 0.7098,
+      "step": 586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7302351941059793e-05,
+      "loss": 0.7138,
+      "step": 587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7296684613204873e-05,
+      "loss": 0.6981,
+      "step": 588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7291017285349958e-05,
+      "loss": 0.6851,
+      "step": 589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7285349957495042e-05,
+      "loss": 0.7214,
+      "step": 590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7279682629640126e-05,
+      "loss": 0.6436,
+      "step": 591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.727401530178521e-05,
+      "loss": 0.7085,
+      "step": 592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7268347973930294e-05,
+      "loss": 0.6516,
+      "step": 593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7262680646075375e-05,
+      "loss": 0.704,
+      "step": 594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7257013318220463e-05,
+      "loss": 0.7133,
+      "step": 595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7251345990365543e-05,
+      "loss": 0.672,
+      "step": 596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7245678662510628e-05,
+      "loss": 0.6742,
+      "step": 597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7240011334655712e-05,
+      "loss": 0.6849,
+      "step": 598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7234344006800793e-05,
+      "loss": 0.6734,
+      "step": 599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.722867667894588e-05,
+      "loss": 0.6727,
+      "step": 600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.722300935109096e-05,
+      "loss": 0.7105,
+      "step": 601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7217342023236045e-05,
+      "loss": 0.6789,
+      "step": 602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.721167469538113e-05,
+      "loss": 0.6677,
+      "step": 603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.720600736752621e-05,
+      "loss": 0.7034,
+      "step": 604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7200340039671298e-05,
+      "loss": 0.6423,
+      "step": 605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7194672711816382e-05,
+      "loss": 0.6866,
+      "step": 606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7189005383961463e-05,
+      "loss": 0.704,
+      "step": 607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7183338056106547e-05,
+      "loss": 0.6722,
+      "step": 608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.717767072825163e-05,
+      "loss": 0.6959,
+      "step": 609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7172003400396715e-05,
+      "loss": 0.6829,
+      "step": 610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.71663360725418e-05,
+      "loss": 0.6332,
+      "step": 611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.716066874468688e-05,
+      "loss": 0.7192,
+      "step": 612
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7155001416831964e-05,
+      "loss": 0.7122,
+      "step": 613
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.714933408897705e-05,
+      "loss": 0.726,
+      "step": 614
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7143666761122133e-05,
+      "loss": 0.7063,
+      "step": 615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7137999433267217e-05,
+      "loss": 0.6636,
+      "step": 616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7132332105412298e-05,
+      "loss": 0.6879,
+      "step": 617
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7126664777557382e-05,
+      "loss": 0.7195,
+      "step": 618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.712099744970247e-05,
+      "loss": 0.704,
+      "step": 619
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.711533012184755e-05,
+      "loss": 0.6734,
+      "step": 620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7109662793992634e-05,
+      "loss": 0.6831,
+      "step": 621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.710399546613772e-05,
+      "loss": 0.687,
+      "step": 622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.70983281382828e-05,
+      "loss": 0.6804,
+      "step": 623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7092660810427887e-05,
+      "loss": 0.6772,
+      "step": 624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7086993482572968e-05,
+      "loss": 0.6949,
+      "step": 625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7081326154718052e-05,
+      "loss": 0.7178,
+      "step": 626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7075658826863136e-05,
+      "loss": 0.6818,
+      "step": 627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7069991499008217e-05,
+      "loss": 0.6855,
+      "step": 628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7064324171153304e-05,
+      "loss": 0.6812,
+      "step": 629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7058656843298385e-05,
+      "loss": 0.6603,
+      "step": 630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.705298951544347e-05,
+      "loss": 0.6748,
+      "step": 631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7047322187588554e-05,
+      "loss": 0.7188,
+      "step": 632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7041654859733638e-05,
+      "loss": 0.7129,
+      "step": 633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7035987531878722e-05,
+      "loss": 0.6605,
+      "step": 634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7030320204023806e-05,
+      "loss": 0.7157,
+      "step": 635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7024652876168887e-05,
+      "loss": 0.6955,
+      "step": 636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.701898554831397e-05,
+      "loss": 0.6841,
+      "step": 637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7013318220459055e-05,
+      "loss": 0.6979,
+      "step": 638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.700765089260414e-05,
+      "loss": 0.6342,
+      "step": 639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7001983564749224e-05,
+      "loss": 0.6664,
+      "step": 640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6996316236894304e-05,
+      "loss": 0.6954,
+      "step": 641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.699064890903939e-05,
+      "loss": 0.7299,
+      "step": 642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6984981581184473e-05,
+      "loss": 0.6568,
+      "step": 643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6979314253329557e-05,
+      "loss": 0.7126,
+      "step": 644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.697364692547464e-05,
+      "loss": 0.6638,
+      "step": 645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6967979597619725e-05,
+      "loss": 0.6451,
+      "step": 646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6962312269764806e-05,
+      "loss": 0.6811,
+      "step": 647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.695664494190989e-05,
+      "loss": 0.6785,
+      "step": 648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6950977614054974e-05,
+      "loss": 0.728,
+      "step": 649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.694531028620006e-05,
+      "loss": 0.6262,
+      "step": 650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6939642958345143e-05,
+      "loss": 0.6769,
+      "step": 651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6933975630490224e-05,
+      "loss": 0.6822,
+      "step": 652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6928308302635308e-05,
+      "loss": 0.6666,
+      "step": 653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6922640974780392e-05,
+      "loss": 0.7052,
+      "step": 654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6916973646925476e-05,
+      "loss": 0.6825,
+      "step": 655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.691130631907056e-05,
+      "loss": 0.632,
+      "step": 656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.690563899121564e-05,
+      "loss": 0.674,
+      "step": 657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6899971663360725e-05,
+      "loss": 0.7479,
+      "step": 658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6894304335505813e-05,
+      "loss": 0.6912,
+      "step": 659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6888637007650894e-05,
+      "loss": 0.6613,
+      "step": 660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6882969679795978e-05,
+      "loss": 0.7089,
+      "step": 661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6877302351941062e-05,
+      "loss": 0.6982,
+      "step": 662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6871635024086143e-05,
+      "loss": 0.6853,
+      "step": 663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.686596769623123e-05,
+      "loss": 0.6842,
+      "step": 664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.686030036837631e-05,
+      "loss": 0.7298,
+      "step": 665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6854633040521395e-05,
+      "loss": 0.7276,
+      "step": 666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.684896571266648e-05,
+      "loss": 0.6768,
+      "step": 667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.684329838481156e-05,
+      "loss": 0.6632,
+      "step": 668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6837631056956648e-05,
+      "loss": 0.705,
+      "step": 669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.683196372910173e-05,
+      "loss": 0.6841,
+      "step": 670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6826296401246813e-05,
+      "loss": 0.645,
+      "step": 671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6820629073391897e-05,
+      "loss": 0.7083,
+      "step": 672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.681496174553698e-05,
+      "loss": 0.6179,
+      "step": 673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6809294417682065e-05,
+      "loss": 0.631,
+      "step": 674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.680362708982715e-05,
+      "loss": 0.6952,
+      "step": 675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.679795976197223e-05,
+      "loss": 0.7001,
+      "step": 676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6792292434117315e-05,
+      "loss": 0.6958,
+      "step": 677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.67866251062624e-05,
+      "loss": 0.6621,
+      "step": 678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6780957778407483e-05,
+      "loss": 0.6397,
+      "step": 679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6775290450552567e-05,
+      "loss": 0.6382,
+      "step": 680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6769623122697648e-05,
+      "loss": 0.6337,
+      "step": 681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6763955794842732e-05,
+      "loss": 0.6808,
+      "step": 682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6758288466987816e-05,
+      "loss": 0.7024,
+      "step": 683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.67526211391329e-05,
+      "loss": 0.6493,
+      "step": 684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6746953811277985e-05,
+      "loss": 0.704,
+      "step": 685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.674128648342307e-05,
+      "loss": 0.7374,
+      "step": 686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673561915556815e-05,
+      "loss": 0.6799,
+      "step": 687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6729951827713237e-05,
+      "loss": 0.7005,
+      "step": 688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6724284499858318e-05,
+      "loss": 0.6062,
+      "step": 689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6718617172003402e-05,
+      "loss": 0.6319,
+      "step": 690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6712949844148486e-05,
+      "loss": 0.6657,
+      "step": 691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6707282516293567e-05,
+      "loss": 0.696,
+      "step": 692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6701615188438655e-05,
+      "loss": 0.677,
+      "step": 693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6695947860583735e-05,
+      "loss": 0.6441,
+      "step": 694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.669028053272882e-05,
+      "loss": 0.6639,
+      "step": 695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6684613204873904e-05,
+      "loss": 0.6921,
+      "step": 696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6678945877018985e-05,
+      "loss": 0.6648,
+      "step": 697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6673278549164072e-05,
+      "loss": 0.663,
+      "step": 698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6667611221309153e-05,
+      "loss": 0.6989,
+      "step": 699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6661943893454237e-05,
+      "loss": 0.7532,
+      "step": 700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665627656559932e-05,
+      "loss": 0.6709,
+      "step": 701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6650609237744405e-05,
+      "loss": 0.6661,
+      "step": 702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.664494190988949e-05,
+      "loss": 0.6429,
+      "step": 703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6639274582034574e-05,
+      "loss": 0.6802,
+      "step": 704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6633607254179655e-05,
+      "loss": 0.6782,
+      "step": 705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662793992632474e-05,
+      "loss": 0.631,
+      "step": 706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6622272598469823e-05,
+      "loss": 0.6666,
+      "step": 707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6616605270614907e-05,
+      "loss": 0.6847,
+      "step": 708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661093794275999e-05,
+      "loss": 0.6939,
+      "step": 709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6605270614905072e-05,
+      "loss": 0.6483,
+      "step": 710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6599603287050156e-05,
+      "loss": 0.6937,
+      "step": 711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.659393595919524e-05,
+      "loss": 0.6598,
+      "step": 712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6588268631340325e-05,
+      "loss": 0.6557,
+      "step": 713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.658260130348541e-05,
+      "loss": 0.6507,
+      "step": 714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6576933975630493e-05,
+      "loss": 0.7154,
+      "step": 715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6571266647775574e-05,
+      "loss": 0.7608,
+      "step": 716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6565599319920658e-05,
+      "loss": 0.7166,
+      "step": 717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6559931992065742e-05,
+      "loss": 0.709,
+      "step": 718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6554264664210826e-05,
+      "loss": 0.6551,
+      "step": 719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654859733635591e-05,
+      "loss": 0.6906,
+      "step": 720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654293000850099e-05,
+      "loss": 0.7165,
+      "step": 721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6537262680646075e-05,
+      "loss": 0.6544,
+      "step": 722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.653159535279116e-05,
+      "loss": 0.6946,
+      "step": 723
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6525928024936244e-05,
+      "loss": 0.7665,
+      "step": 724
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6520260697081328e-05,
+      "loss": 0.7082,
+      "step": 725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.651459336922641e-05,
+      "loss": 0.6838,
+      "step": 726
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6508926041371496e-05,
+      "loss": 0.6703,
+      "step": 727
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.650325871351658e-05,
+      "loss": 0.6189,
+      "step": 728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.649759138566166e-05,
+      "loss": 0.655,
+      "step": 729
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6491924057806745e-05,
+      "loss": 0.6729,
+      "step": 730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.648625672995183e-05,
+      "loss": 0.6615,
+      "step": 731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6480589402096914e-05,
+      "loss": 0.6798,
+      "step": 732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6474922074241998e-05,
+      "loss": 0.7378,
+      "step": 733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.646925474638708e-05,
+      "loss": 0.7006,
+      "step": 734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6463587418532163e-05,
+      "loss": 0.6932,
+      "step": 735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6457920090677247e-05,
+      "loss": 0.6943,
+      "step": 736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.645225276282233e-05,
+      "loss": 0.6771,
+      "step": 737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6446585434967415e-05,
+      "loss": 0.6838,
+      "step": 738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6440918107112496e-05,
+      "loss": 0.7008,
+      "step": 739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.643525077925758e-05,
+      "loss": 0.5913,
+      "step": 740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6429583451402665e-05,
+      "loss": 0.6234,
+      "step": 741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.642391612354775e-05,
+      "loss": 0.6388,
+      "step": 742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6418248795692833e-05,
+      "loss": 0.6711,
+      "step": 743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6412581467837917e-05,
+      "loss": 0.6868,
+      "step": 744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6406914139982998e-05,
+      "loss": 0.6525,
+      "step": 745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6401246812128082e-05,
+      "loss": 0.6614,
+      "step": 746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6395579484273166e-05,
+      "loss": 0.6612,
+      "step": 747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.638991215641825e-05,
+      "loss": 0.6162,
+      "step": 748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6384244828563335e-05,
+      "loss": 0.6417,
+      "step": 749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6378577500708415e-05,
+      "loss": 0.6877,
+      "step": 750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.63729101728535e-05,
+      "loss": 0.7331,
+      "step": 751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6367242844998584e-05,
+      "loss": 0.665,
+      "step": 752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6361575517143668e-05,
+      "loss": 0.7035,
+      "step": 753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6355908189288752e-05,
+      "loss": 0.658,
+      "step": 754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6350240861433836e-05,
+      "loss": 0.6195,
+      "step": 755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6344573533578917e-05,
+      "loss": 0.6624,
+      "step": 756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6338906205724005e-05,
+      "loss": 0.6306,
+      "step": 757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6333238877869086e-05,
+      "loss": 0.6863,
+      "step": 758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.632757155001417e-05,
+      "loss": 0.7054,
+      "step": 759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6321904222159254e-05,
+      "loss": 0.6632,
+      "step": 760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6316236894304335e-05,
+      "loss": 0.6444,
+      "step": 761
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6310569566449422e-05,
+      "loss": 0.6676,
+      "step": 762
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6304902238594503e-05,
+      "loss": 0.6445,
+      "step": 763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6299234910739587e-05,
+      "loss": 0.6898,
+      "step": 764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.629356758288467e-05,
+      "loss": 0.6832,
+      "step": 765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6287900255029752e-05,
+      "loss": 0.6476,
+      "step": 766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.628223292717484e-05,
+      "loss": 0.7498,
+      "step": 767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6276565599319924e-05,
+      "loss": 0.6687,
+      "step": 768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6270898271465005e-05,
+      "loss": 0.6475,
+      "step": 769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.626523094361009e-05,
+      "loss": 0.6461,
+      "step": 770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6259563615755173e-05,
+      "loss": 0.6707,
+      "step": 771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6253896287900257e-05,
+      "loss": 0.718,
+      "step": 772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.624822896004534e-05,
+      "loss": 0.6783,
+      "step": 773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6242561632190422e-05,
+      "loss": 0.6885,
+      "step": 774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6236894304335506e-05,
+      "loss": 0.6959,
+      "step": 775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.623122697648059e-05,
+      "loss": 0.6857,
+      "step": 776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6225559648625675e-05,
+      "loss": 0.6792,
+      "step": 777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.621989232077076e-05,
+      "loss": 0.6439,
+      "step": 778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.621422499291584e-05,
+      "loss": 0.6366,
+      "step": 779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6208557665060924e-05,
+      "loss": 0.6947,
+      "step": 780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.620289033720601e-05,
+      "loss": 0.6552,
+      "step": 781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6197223009351092e-05,
+      "loss": 0.7057,
+      "step": 782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6191555681496176e-05,
+      "loss": 0.7002,
+      "step": 783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.618588835364126e-05,
+      "loss": 0.6753,
+      "step": 784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.618022102578634e-05,
+      "loss": 0.6166,
+      "step": 785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.617455369793143e-05,
+      "loss": 0.6376,
+      "step": 786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.616888637007651e-05,
+      "loss": 0.6462,
+      "step": 787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6163219042221594e-05,
+      "loss": 0.6704,
+      "step": 788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6157551714366678e-05,
+      "loss": 0.6616,
+      "step": 789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.615188438651176e-05,
+      "loss": 0.7375,
+      "step": 790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6146217058656846e-05,
+      "loss": 0.6966,
+      "step": 791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6140549730801927e-05,
+      "loss": 0.6616,
+      "step": 792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.613488240294701e-05,
+      "loss": 0.6759,
+      "step": 793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6129215075092096e-05,
+      "loss": 0.6233,
+      "step": 794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.612354774723718e-05,
+      "loss": 0.6901,
+      "step": 795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6117880419382264e-05,
+      "loss": 0.6112,
+      "step": 796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6112213091527348e-05,
+      "loss": 0.6364,
+      "step": 797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.610654576367243e-05,
+      "loss": 0.7073,
+      "step": 798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6100878435817513e-05,
+      "loss": 0.7074,
+      "step": 799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6095211107962597e-05,
+      "loss": 0.6651,
+      "step": 800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.608954378010768e-05,
+      "loss": 0.6593,
+      "step": 801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6083876452252766e-05,
+      "loss": 0.6731,
+      "step": 802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6078209124397846e-05,
+      "loss": 0.654,
+      "step": 803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.607254179654293e-05,
+      "loss": 0.7243,
+      "step": 804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6066874468688015e-05,
+      "loss": 0.6711,
+      "step": 805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.60612071408331e-05,
+      "loss": 0.6209,
+      "step": 806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6055539812978183e-05,
+      "loss": 0.6779,
+      "step": 807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6049872485123267e-05,
+      "loss": 0.7375,
+      "step": 808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6044205157268348e-05,
+      "loss": 0.7026,
+      "step": 809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6038537829413432e-05,
+      "loss": 0.6583,
+      "step": 810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6032870501558516e-05,
+      "loss": 0.6394,
+      "step": 811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.60272031737036e-05,
+      "loss": 0.6318,
+      "step": 812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6021535845848685e-05,
+      "loss": 0.6449,
+      "step": 813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6015868517993766e-05,
+      "loss": 0.6615,
+      "step": 814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.601020119013885e-05,
+      "loss": 0.6569,
+      "step": 815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6004533862283934e-05,
+      "loss": 0.6651,
+      "step": 816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5998866534429018e-05,
+      "loss": 0.6798,
+      "step": 817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5993199206574102e-05,
+      "loss": 0.6794,
+      "step": 818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5987531878719183e-05,
+      "loss": 0.6466,
+      "step": 819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5981864550864267e-05,
+      "loss": 0.6737,
+      "step": 820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.597619722300935e-05,
+      "loss": 0.6717,
+      "step": 821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5970529895154436e-05,
+      "loss": 0.7015,
+      "step": 822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.596486256729952e-05,
+      "loss": 0.6521,
+      "step": 823
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5959195239444604e-05,
+      "loss": 0.6821,
+      "step": 824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5953527911589685e-05,
+      "loss": 0.6415,
+      "step": 825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5947860583734772e-05,
+      "loss": 0.616,
+      "step": 826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5942193255879853e-05,
+      "loss": 0.6929,
+      "step": 827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5936525928024937e-05,
+      "loss": 0.6871,
+      "step": 828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.593085860017002e-05,
+      "loss": 0.6085,
+      "step": 829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5925191272315102e-05,
+      "loss": 0.6385,
+      "step": 830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.591952394446019e-05,
+      "loss": 0.6882,
+      "step": 831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.591385661660527e-05,
+      "loss": 0.7004,
+      "step": 832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5908189288750355e-05,
+      "loss": 0.6769,
+      "step": 833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.590252196089544e-05,
+      "loss": 0.6972,
+      "step": 834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5896854633040523e-05,
+      "loss": 0.6904,
+      "step": 835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5891187305185607e-05,
+      "loss": 0.6418,
+      "step": 836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.588551997733069e-05,
+      "loss": 0.6553,
+      "step": 837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5879852649475772e-05,
+      "loss": 0.7047,
+      "step": 838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5874185321620857e-05,
+      "loss": 0.6441,
+      "step": 839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.586851799376594e-05,
+      "loss": 0.678,
+      "step": 840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5862850665911025e-05,
+      "loss": 0.606,
+      "step": 841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.585718333805611e-05,
+      "loss": 0.6222,
+      "step": 842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.585151601020119e-05,
+      "loss": 0.7005,
+      "step": 843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5845848682346274e-05,
+      "loss": 0.7027,
+      "step": 844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5840181354491358e-05,
+      "loss": 0.684,
+      "step": 845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5834514026636442e-05,
+      "loss": 0.7007,
+      "step": 846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5828846698781527e-05,
+      "loss": 0.6845,
+      "step": 847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.582317937092661e-05,
+      "loss": 0.6615,
+      "step": 848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.581751204307169e-05,
+      "loss": 0.6422,
+      "step": 849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.581184471521678e-05,
+      "loss": 0.6032,
+      "step": 850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.580617738736186e-05,
+      "loss": 0.6422,
+      "step": 851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5800510059506944e-05,
+      "loss": 0.6445,
+      "step": 852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5794842731652028e-05,
+      "loss": 0.6431,
+      "step": 853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.578917540379711e-05,
+      "loss": 0.64,
+      "step": 854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5783508075942197e-05,
+      "loss": 0.6623,
+      "step": 855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5777840748087277e-05,
+      "loss": 0.6473,
+      "step": 856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.577217342023236e-05,
+      "loss": 0.6823,
+      "step": 857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5766506092377446e-05,
+      "loss": 0.6818,
+      "step": 858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5760838764522527e-05,
+      "loss": 0.6434,
+      "step": 859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5755171436667614e-05,
+      "loss": 0.6767,
+      "step": 860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5749504108812695e-05,
+      "loss": 0.6449,
+      "step": 861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.574383678095778e-05,
+      "loss": 0.6255,
+      "step": 862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5738169453102863e-05,
+      "loss": 0.7053,
+      "step": 863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5732502125247947e-05,
+      "loss": 0.6632,
+      "step": 864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.572683479739303e-05,
+      "loss": 0.6458,
+      "step": 865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5721167469538116e-05,
+      "loss": 0.6325,
+      "step": 866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5715500141683197e-05,
+      "loss": 0.647,
+      "step": 867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.570983281382828e-05,
+      "loss": 0.6366,
+      "step": 868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5704165485973365e-05,
+      "loss": 0.6197,
+      "step": 869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.569849815811845e-05,
+      "loss": 0.6283,
+      "step": 870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5692830830263533e-05,
+      "loss": 0.67,
+      "step": 871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5687163502408614e-05,
+      "loss": 0.6799,
+      "step": 872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5681496174553698e-05,
+      "loss": 0.6535,
+      "step": 873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5675828846698782e-05,
+      "loss": 0.6426,
+      "step": 874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5670161518843867e-05,
+      "loss": 0.6722,
+      "step": 875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.566449419098895e-05,
+      "loss": 0.6804,
+      "step": 876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5658826863134035e-05,
+      "loss": 0.6512,
+      "step": 877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5653159535279116e-05,
+      "loss": 0.6579,
+      "step": 878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.56474922074242e-05,
+      "loss": 0.6616,
+      "step": 879
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5641824879569284e-05,
+      "loss": 0.6705,
+      "step": 880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5636157551714368e-05,
+      "loss": 0.6902,
+      "step": 881
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5630490223859452e-05,
+      "loss": 0.6678,
+      "step": 882
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5624822896004533e-05,
+      "loss": 0.634,
+      "step": 883
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5619155568149617e-05,
+      "loss": 0.6554,
+      "step": 884
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.56134882402947e-05,
+      "loss": 0.6473,
+      "step": 885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5607820912439786e-05,
+      "loss": 0.6896,
+      "step": 886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.560215358458487e-05,
+      "loss": 0.6307,
+      "step": 887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.559648625672995e-05,
+      "loss": 0.6382,
+      "step": 888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.559081892887504e-05,
+      "loss": 0.6448,
+      "step": 889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5585151601020123e-05,
+      "loss": 0.7105,
+      "step": 890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5579484273165203e-05,
+      "loss": 0.636,
+      "step": 891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5573816945310287e-05,
+      "loss": 0.6882,
+      "step": 892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.556814961745537e-05,
+      "loss": 0.6629,
+      "step": 893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5562482289600456e-05,
+      "loss": 0.6068,
+      "step": 894
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.555681496174554e-05,
+      "loss": 0.6745,
+      "step": 895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.555114763389062e-05,
+      "loss": 0.6659,
+      "step": 896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5545480306035705e-05,
+      "loss": 0.6135,
+      "step": 897
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.553981297818079e-05,
+      "loss": 0.6318,
+      "step": 898
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5534145650325873e-05,
+      "loss": 0.675,
+      "step": 899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5528478322470958e-05,
+      "loss": 0.6624,
+      "step": 900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5522810994616038e-05,
+      "loss": 0.6892,
+      "step": 901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5517143666761122e-05,
+      "loss": 0.6672,
+      "step": 902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5511476338906207e-05,
+      "loss": 0.6673,
+      "step": 903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.550580901105129e-05,
+      "loss": 0.709,
+      "step": 904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5500141683196375e-05,
+      "loss": 0.6423,
+      "step": 905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.549447435534146e-05,
+      "loss": 0.6528,
+      "step": 906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.548880702748654e-05,
+      "loss": 0.6406,
+      "step": 907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5483139699631624e-05,
+      "loss": 0.7049,
+      "step": 908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.547747237177671e-05,
+      "loss": 0.6668,
+      "step": 909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5471805043921793e-05,
+      "loss": 0.7361,
+      "step": 910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5466137716066877e-05,
+      "loss": 0.7029,
+      "step": 911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5460470388211957e-05,
+      "loss": 0.6339,
+      "step": 912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545480306035704e-05,
+      "loss": 0.6667,
+      "step": 913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5449135732502126e-05,
+      "loss": 0.6801,
+      "step": 914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.544346840464721e-05,
+      "loss": 0.6337,
+      "step": 915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5437801076792294e-05,
+      "loss": 0.5723,
+      "step": 916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.543213374893738e-05,
+      "loss": 0.6314,
+      "step": 917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.542646642108246e-05,
+      "loss": 0.6069,
+      "step": 918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5420799093227547e-05,
+      "loss": 0.6345,
+      "step": 919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5415131765372628e-05,
+      "loss": 0.591,
+      "step": 920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5409464437517712e-05,
+      "loss": 0.6199,
+      "step": 921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5403797109662796e-05,
+      "loss": 0.6661,
+      "step": 922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5398129781807877e-05,
+      "loss": 0.7095,
+      "step": 923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5392462453952964e-05,
+      "loss": 0.6791,
+      "step": 924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5386795126098045e-05,
+      "loss": 0.6565,
+      "step": 925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.538112779824313e-05,
+      "loss": 0.6566,
+      "step": 926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5375460470388213e-05,
+      "loss": 0.6779,
+      "step": 927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5369793142533294e-05,
+      "loss": 0.6241,
+      "step": 928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5364125814678382e-05,
+      "loss": 0.6349,
+      "step": 929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5358458486823466e-05,
+      "loss": 0.6449,
+      "step": 930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5352791158968547e-05,
+      "loss": 0.6605,
+      "step": 931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.534712383111363e-05,
+      "loss": 0.6989,
+      "step": 932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5341456503258715e-05,
+      "loss": 0.5694,
+      "step": 933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.53357891754038e-05,
+      "loss": 0.6527,
+      "step": 934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5330121847548883e-05,
+      "loss": 0.7321,
+      "step": 935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5324454519693964e-05,
+      "loss": 0.6731,
+      "step": 936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.531878719183905e-05,
+      "loss": 0.6854,
+      "step": 937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5313119863984133e-05,
+      "loss": 0.7258,
+      "step": 938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5307452536129217e-05,
+      "loss": 0.6418,
+      "step": 939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.53017852082743e-05,
+      "loss": 0.655,
+      "step": 940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5296117880419382e-05,
+      "loss": 0.6549,
+      "step": 941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5290450552564466e-05,
+      "loss": 0.6351,
+      "step": 942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5284783224709553e-05,
+      "loss": 0.6894,
+      "step": 943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5279115896854634e-05,
+      "loss": 0.68,
+      "step": 944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.527344856899972e-05,
+      "loss": 0.6468,
+      "step": 945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5267781241144803e-05,
+      "loss": 0.6885,
+      "step": 946
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5262113913289883e-05,
+      "loss": 0.6714,
+      "step": 947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.525644658543497e-05,
+      "loss": 0.6318,
+      "step": 948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5250779257580053e-05,
+      "loss": 0.6474,
+      "step": 949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5245111929725136e-05,
+      "loss": 0.6871,
+      "step": 950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5239444601870218e-05,
+      "loss": 0.6742,
+      "step": 951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5233777274015303e-05,
+      "loss": 0.6299,
+      "step": 952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5228109946160388e-05,
+      "loss": 0.6468,
+      "step": 953
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5222442618305471e-05,
+      "loss": 0.6077,
+      "step": 954
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5216775290450553e-05,
+      "loss": 0.6801,
+      "step": 955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5211107962595638e-05,
+      "loss": 0.6537,
+      "step": 956
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.520544063474072e-05,
+      "loss": 0.6313,
+      "step": 957
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5199773306885806e-05,
+      "loss": 0.6608,
+      "step": 958
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5194105979030888e-05,
+      "loss": 0.6634,
+      "step": 959
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5188438651175973e-05,
+      "loss": 0.6453,
+      "step": 960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5182771323321055e-05,
+      "loss": 0.6301,
+      "step": 961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5177103995466138e-05,
+      "loss": 0.6383,
+      "step": 962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5171436667611223e-05,
+      "loss": 0.6773,
+      "step": 963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5165769339756306e-05,
+      "loss": 0.6205,
+      "step": 964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.516010201190139e-05,
+      "loss": 0.6566,
+      "step": 965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5154434684046473e-05,
+      "loss": 0.642,
+      "step": 966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5148767356191557e-05,
+      "loss": 0.6801,
+      "step": 967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5143100028336641e-05,
+      "loss": 0.6811,
+      "step": 968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5137432700481725e-05,
+      "loss": 0.6409,
+      "step": 969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5131765372626808e-05,
+      "loss": 0.6322,
+      "step": 970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.512609804477189e-05,
+      "loss": 0.616,
+      "step": 971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5120430716916974e-05,
+      "loss": 0.6406,
+      "step": 972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5114763389062058e-05,
+      "loss": 0.6426,
+      "step": 973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5109096061207143e-05,
+      "loss": 0.5831,
+      "step": 974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5103428733352225e-05,
+      "loss": 0.6559,
+      "step": 975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.509776140549731e-05,
+      "loss": 0.6952,
+      "step": 976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5092094077642392e-05,
+      "loss": 0.6515,
+      "step": 977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5086426749787478e-05,
+      "loss": 0.6666,
+      "step": 978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.508075942193256e-05,
+      "loss": 0.6411,
+      "step": 979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5075092094077644e-05,
+      "loss": 0.6458,
+      "step": 980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5069424766222727e-05,
+      "loss": 0.5928,
+      "step": 981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.506375743836781e-05,
+      "loss": 0.7139,
+      "step": 982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5058090110512895e-05,
+      "loss": 0.6436,
+      "step": 983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5052422782657978e-05,
+      "loss": 0.6651,
+      "step": 984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5046755454803062e-05,
+      "loss": 0.6928,
+      "step": 985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5041088126948144e-05,
+      "loss": 0.6493,
+      "step": 986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5035420799093229e-05,
+      "loss": 0.6287,
+      "step": 987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5029753471238313e-05,
+      "loss": 0.6308,
+      "step": 988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5024086143383397e-05,
+      "loss": 0.6609,
+      "step": 989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.501841881552848e-05,
+      "loss": 0.6462,
+      "step": 990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5012751487673562e-05,
+      "loss": 0.6498,
+      "step": 991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5007084159818646e-05,
+      "loss": 0.6496,
+      "step": 992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.500141683196373e-05,
+      "loss": 0.6243,
+      "step": 993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4995749504108814e-05,
+      "loss": 0.635,
+      "step": 994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4990082176253897e-05,
+      "loss": 0.6874,
+      "step": 995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4984414848398981e-05,
+      "loss": 0.6443,
+      "step": 996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4978747520544065e-05,
+      "loss": 0.6955,
+      "step": 997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.497308019268915e-05,
+      "loss": 0.6319,
+      "step": 998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4967412864834232e-05,
+      "loss": 0.6614,
+      "step": 999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4961745536979316e-05,
+      "loss": 0.6698,
+      "step": 1000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4956078209124399e-05,
+      "loss": 0.7165,
+      "step": 1001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4950410881269484e-05,
+      "loss": 0.6693,
+      "step": 1002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4944743553414567e-05,
+      "loss": 0.6924,
+      "step": 1003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.493907622555965e-05,
+      "loss": 0.6624,
+      "step": 1004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4933408897704734e-05,
+      "loss": 0.696,
+      "step": 1005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4927741569849816e-05,
+      "loss": 0.6148,
+      "step": 1006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4922074241994902e-05,
+      "loss": 0.6553,
+      "step": 1007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4916406914139984e-05,
+      "loss": 0.6322,
+      "step": 1008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4910739586285069e-05,
+      "loss": 0.6234,
+      "step": 1009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4905072258430151e-05,
+      "loss": 0.6098,
+      "step": 1010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4899404930575234e-05,
+      "loss": 0.6654,
+      "step": 1011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.489373760272032e-05,
+      "loss": 0.6008,
+      "step": 1012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4888070274865402e-05,
+      "loss": 0.6311,
+      "step": 1013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4882402947010486e-05,
+      "loss": 0.7052,
+      "step": 1014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4876735619155569e-05,
+      "loss": 0.6401,
+      "step": 1015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4871068291300653e-05,
+      "loss": 0.6988,
+      "step": 1016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4865400963445737e-05,
+      "loss": 0.6148,
+      "step": 1017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4859733635590821e-05,
+      "loss": 0.6904,
+      "step": 1018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4854066307735904e-05,
+      "loss": 0.5993,
+      "step": 1019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4848398979880986e-05,
+      "loss": 0.632,
+      "step": 1020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.484273165202607e-05,
+      "loss": 0.6966,
+      "step": 1021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4837064324171156e-05,
+      "loss": 0.633,
+      "step": 1022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4831396996316239e-05,
+      "loss": 0.6383,
+      "step": 1023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4825729668461321e-05,
+      "loss": 0.6219,
+      "step": 1024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4820062340606405e-05,
+      "loss": 0.6753,
+      "step": 1025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4814395012751488e-05,
+      "loss": 0.6191,
+      "step": 1026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4808727684896574e-05,
+      "loss": 0.6199,
+      "step": 1027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4803060357041656e-05,
+      "loss": 0.6504,
+      "step": 1028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.479739302918674e-05,
+      "loss": 0.6647,
+      "step": 1029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4791725701331823e-05,
+      "loss": 0.635,
+      "step": 1030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4786058373476905e-05,
+      "loss": 0.6895,
+      "step": 1031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4780391045621991e-05,
+      "loss": 0.6681,
+      "step": 1032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4774723717767074e-05,
+      "loss": 0.6598,
+      "step": 1033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4769056389912158e-05,
+      "loss": 0.6282,
+      "step": 1034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.476338906205724e-05,
+      "loss": 0.6309,
+      "step": 1035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4757721734202324e-05,
+      "loss": 0.5984,
+      "step": 1036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4752054406347409e-05,
+      "loss": 0.6444,
+      "step": 1037
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4746387078492493e-05,
+      "loss": 0.6832,
+      "step": 1038
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4740719750637575e-05,
+      "loss": 0.6226,
+      "step": 1039
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4735052422782658e-05,
+      "loss": 0.6599,
+      "step": 1040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4729385094927742e-05,
+      "loss": 0.662,
+      "step": 1041
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4723717767072828e-05,
+      "loss": 0.6371,
+      "step": 1042
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.471805043921791e-05,
+      "loss": 0.667,
+      "step": 1043
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4712383111362993e-05,
+      "loss": 0.6505,
+      "step": 1044
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4706715783508077e-05,
+      "loss": 0.6859,
+      "step": 1045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.470104845565316e-05,
+      "loss": 0.6211,
+      "step": 1046
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4695381127798245e-05,
+      "loss": 0.6616,
+      "step": 1047
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4689713799943328e-05,
+      "loss": 0.6594,
+      "step": 1048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4684046472088412e-05,
+      "loss": 0.6753,
+      "step": 1049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4678379144233494e-05,
+      "loss": 0.6684,
+      "step": 1050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.467271181637858e-05,
+      "loss": 0.6695,
+      "step": 1051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4667044488523663e-05,
+      "loss": 0.659,
+      "step": 1052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4661377160668745e-05,
+      "loss": 0.6693,
+      "step": 1053
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.465570983281383e-05,
+      "loss": 0.6296,
+      "step": 1054
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4650042504958912e-05,
+      "loss": 0.6354,
+      "step": 1055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4644375177103998e-05,
+      "loss": 0.6636,
+      "step": 1056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.463870784924908e-05,
+      "loss": 0.6366,
+      "step": 1057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4633040521394165e-05,
+      "loss": 0.6357,
+      "step": 1058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4627373193539247e-05,
+      "loss": 0.6532,
+      "step": 1059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.462170586568433e-05,
+      "loss": 0.6685,
+      "step": 1060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4616038537829415e-05,
+      "loss": 0.6601,
+      "step": 1061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.46103712099745e-05,
+      "loss": 0.6338,
+      "step": 1062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4604703882119582e-05,
+      "loss": 0.6564,
+      "step": 1063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4599036554264664e-05,
+      "loss": 0.5851,
+      "step": 1064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4593369226409749e-05,
+      "loss": 0.6444,
+      "step": 1065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4587701898554833e-05,
+      "loss": 0.6437,
+      "step": 1066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4582034570699917e-05,
+      "loss": 0.6818,
+      "step": 1067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4576367242845e-05,
+      "loss": 0.6606,
+      "step": 1068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4570699914990084e-05,
+      "loss": 0.6418,
+      "step": 1069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4565032587135166e-05,
+      "loss": 0.6846,
+      "step": 1070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4559365259280252e-05,
+      "loss": 0.6298,
+      "step": 1071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4553697931425335e-05,
+      "loss": 0.7,
+      "step": 1072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4548030603570417e-05,
+      "loss": 0.6662,
+      "step": 1073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4542363275715501e-05,
+      "loss": 0.6562,
+      "step": 1074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4536695947860584e-05,
+      "loss": 0.6585,
+      "step": 1075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.453102862000567e-05,
+      "loss": 0.6877,
+      "step": 1076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4525361292150752e-05,
+      "loss": 0.677,
+      "step": 1077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4519693964295836e-05,
+      "loss": 0.6552,
+      "step": 1078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4514026636440919e-05,
+      "loss": 0.6197,
+      "step": 1079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4508359308586001e-05,
+      "loss": 0.6832,
+      "step": 1080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4502691980731087e-05,
+      "loss": 0.6795,
+      "step": 1081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4497024652876171e-05,
+      "loss": 0.6512,
+      "step": 1082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4491357325021254e-05,
+      "loss": 0.5673,
+      "step": 1083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4485689997166336e-05,
+      "loss": 0.6017,
+      "step": 1084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.448002266931142e-05,
+      "loss": 0.6205,
+      "step": 1085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4474355341456505e-05,
+      "loss": 0.6234,
+      "step": 1086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4468688013601589e-05,
+      "loss": 0.6742,
+      "step": 1087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4463020685746671e-05,
+      "loss": 0.6621,
+      "step": 1088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4457353357891755e-05,
+      "loss": 0.6484,
+      "step": 1089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4451686030036838e-05,
+      "loss": 0.6174,
+      "step": 1090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4446018702181924e-05,
+      "loss": 0.619,
+      "step": 1091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4440351374327006e-05,
+      "loss": 0.6775,
+      "step": 1092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4434684046472089e-05,
+      "loss": 0.6617,
+      "step": 1093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4429016718617173e-05,
+      "loss": 0.6397,
+      "step": 1094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4423349390762255e-05,
+      "loss": 0.6265,
+      "step": 1095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4417682062907341e-05,
+      "loss": 0.6136,
+      "step": 1096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4412014735052424e-05,
+      "loss": 0.5904,
+      "step": 1097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4406347407197508e-05,
+      "loss": 0.6184,
+      "step": 1098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.440068007934259e-05,
+      "loss": 0.6689,
+      "step": 1099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4395012751487673e-05,
+      "loss": 0.6745,
+      "step": 1100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4389345423632759e-05,
+      "loss": 0.6141,
+      "step": 1101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4383678095777843e-05,
+      "loss": 0.6021,
+      "step": 1102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4378010767922925e-05,
+      "loss": 0.6551,
+      "step": 1103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4372343440068008e-05,
+      "loss": 0.6406,
+      "step": 1104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4366676112213094e-05,
+      "loss": 0.6583,
+      "step": 1105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4361008784358176e-05,
+      "loss": 0.627,
+      "step": 1106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.435534145650326e-05,
+      "loss": 0.6712,
+      "step": 1107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4349674128648343e-05,
+      "loss": 0.6773,
+      "step": 1108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4344006800793427e-05,
+      "loss": 0.644,
+      "step": 1109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4338339472938511e-05,
+      "loss": 0.6534,
+      "step": 1110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4332672145083595e-05,
+      "loss": 0.6284,
+      "step": 1111
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4327004817228678e-05,
+      "loss": 0.6473,
+      "step": 1112
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.432133748937376e-05,
+      "loss": 0.6382,
+      "step": 1113
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4315670161518845e-05,
+      "loss": 0.65,
+      "step": 1114
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4310002833663929e-05,
+      "loss": 0.6637,
+      "step": 1115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4304335505809013e-05,
+      "loss": 0.6423,
+      "step": 1116
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4298668177954095e-05,
+      "loss": 0.6687,
+      "step": 1117
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.429300085009918e-05,
+      "loss": 0.6641,
+      "step": 1118
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4287333522244262e-05,
+      "loss": 0.6151,
+      "step": 1119
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4281666194389348e-05,
+      "loss": 0.6341,
+      "step": 1120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.427599886653443e-05,
+      "loss": 0.6757,
+      "step": 1121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4270331538679515e-05,
+      "loss": 0.7004,
+      "step": 1122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4264664210824597e-05,
+      "loss": 0.6307,
+      "step": 1123
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.425899688296968e-05,
+      "loss": 0.6114,
+      "step": 1124
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4253329555114765e-05,
+      "loss": 0.631,
+      "step": 1125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4247662227259848e-05,
+      "loss": 0.6762,
+      "step": 1126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4241994899404932e-05,
+      "loss": 0.6461,
+      "step": 1127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4236327571550015e-05,
+      "loss": 0.6487,
+      "step": 1128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4230660243695099e-05,
+      "loss": 0.6513,
+      "step": 1129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4224992915840183e-05,
+      "loss": 0.6249,
+      "step": 1130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4219325587985267e-05,
+      "loss": 0.6215,
+      "step": 1131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.421365826013035e-05,
+      "loss": 0.6541,
+      "step": 1132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4207990932275432e-05,
+      "loss": 0.6143,
+      "step": 1133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4202323604420516e-05,
+      "loss": 0.6431,
+      "step": 1134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.41966562765656e-05,
+      "loss": 0.6365,
+      "step": 1135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4190988948710685e-05,
+      "loss": 0.6968,
+      "step": 1136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4185321620855767e-05,
+      "loss": 0.6171,
+      "step": 1137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4179654293000851e-05,
+      "loss": 0.6133,
+      "step": 1138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4173986965145934e-05,
+      "loss": 0.6428,
+      "step": 1139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.416831963729102e-05,
+      "loss": 0.6344,
+      "step": 1140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4162652309436102e-05,
+      "loss": 0.6458,
+      "step": 1141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4156984981581186e-05,
+      "loss": 0.6363,
+      "step": 1142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4151317653726269e-05,
+      "loss": 0.6499,
+      "step": 1143
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4145650325871351e-05,
+      "loss": 0.6098,
+      "step": 1144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4139982998016437e-05,
+      "loss": 0.6748,
+      "step": 1145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.413431567016152e-05,
+      "loss": 0.6782,
+      "step": 1146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4128648342306604e-05,
+      "loss": 0.643,
+      "step": 1147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4122981014451686e-05,
+      "loss": 0.6814,
+      "step": 1148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.411731368659677e-05,
+      "loss": 0.6714,
+      "step": 1149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4111646358741855e-05,
+      "loss": 0.654,
+      "step": 1150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4105979030886939e-05,
+      "loss": 0.6131,
+      "step": 1151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4100311703032021e-05,
+      "loss": 0.6502,
+      "step": 1152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4094644375177104e-05,
+      "loss": 0.6335,
+      "step": 1153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.408897704732219e-05,
+      "loss": 0.6012,
+      "step": 1154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4083309719467272e-05,
+      "loss": 0.6048,
+      "step": 1155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4077642391612356e-05,
+      "loss": 0.6905,
+      "step": 1156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4071975063757439e-05,
+      "loss": 0.6588,
+      "step": 1157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4066307735902523e-05,
+      "loss": 0.6746,
+      "step": 1158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4060640408047607e-05,
+      "loss": 0.6346,
+      "step": 1159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4054973080192691e-05,
+      "loss": 0.6517,
+      "step": 1160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4049305752337774e-05,
+      "loss": 0.6227,
+      "step": 1161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4043638424482856e-05,
+      "loss": 0.6759,
+      "step": 1162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.403797109662794e-05,
+      "loss": 0.6185,
+      "step": 1163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4032303768773026e-05,
+      "loss": 0.6382,
+      "step": 1164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4026636440918109e-05,
+      "loss": 0.6281,
+      "step": 1165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4020969113063191e-05,
+      "loss": 0.6503,
+      "step": 1166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4015301785208276e-05,
+      "loss": 0.6368,
+      "step": 1167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4009634457353358e-05,
+      "loss": 0.6497,
+      "step": 1168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4003967129498444e-05,
+      "loss": 0.6238,
+      "step": 1169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3998299801643526e-05,
+      "loss": 0.674,
+      "step": 1170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.399263247378861e-05,
+      "loss": 0.5788,
+      "step": 1171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3986965145933693e-05,
+      "loss": 0.5861,
+      "step": 1172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3981297818078776e-05,
+      "loss": 0.643,
+      "step": 1173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3975630490223861e-05,
+      "loss": 0.6619,
+      "step": 1174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3969963162368944e-05,
+      "loss": 0.6486,
+      "step": 1175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3964295834514028e-05,
+      "loss": 0.6278,
+      "step": 1176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.395862850665911e-05,
+      "loss": 0.6845,
+      "step": 1177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3952961178804195e-05,
+      "loss": 0.661,
+      "step": 1178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3947293850949279e-05,
+      "loss": 0.6697,
+      "step": 1179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3941626523094363e-05,
+      "loss": 0.6145,
+      "step": 1180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3935959195239446e-05,
+      "loss": 0.6224,
+      "step": 1181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3930291867384528e-05,
+      "loss": 0.6739,
+      "step": 1182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3924624539529612e-05,
+      "loss": 0.5997,
+      "step": 1183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3918957211674698e-05,
+      "loss": 0.5833,
+      "step": 1184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.391328988381978e-05,
+      "loss": 0.6778,
+      "step": 1185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3907622555964863e-05,
+      "loss": 0.6786,
+      "step": 1186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3901955228109947e-05,
+      "loss": 0.6426,
+      "step": 1187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.389628790025503e-05,
+      "loss": 0.6271,
+      "step": 1188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3890620572400116e-05,
+      "loss": 0.678,
+      "step": 1189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3884953244545198e-05,
+      "loss": 0.6782,
+      "step": 1190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3879285916690282e-05,
+      "loss": 0.6671,
+      "step": 1191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3873618588835365e-05,
+      "loss": 0.63,
+      "step": 1192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3867951260980447e-05,
+      "loss": 0.6103,
+      "step": 1193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3862283933125533e-05,
+      "loss": 0.596,
+      "step": 1194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3856616605270616e-05,
+      "loss": 0.6375,
+      "step": 1195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.38509492774157e-05,
+      "loss": 0.5896,
+      "step": 1196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3845281949560782e-05,
+      "loss": 0.6363,
+      "step": 1197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3839614621705866e-05,
+      "loss": 0.6261,
+      "step": 1198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.383394729385095e-05,
+      "loss": 0.616,
+      "step": 1199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3828279965996035e-05,
+      "loss": 0.6252,
+      "step": 1200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3822612638141117e-05,
+      "loss": 0.6715,
+      "step": 1201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.38169453102862e-05,
+      "loss": 0.6977,
+      "step": 1202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3811277982431284e-05,
+      "loss": 0.6235,
+      "step": 1203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.380561065457637e-05,
+      "loss": 0.6569,
+      "step": 1204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3799943326721452e-05,
+      "loss": 0.6944,
+      "step": 1205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3794275998866535e-05,
+      "loss": 0.6516,
+      "step": 1206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3788608671011619e-05,
+      "loss": 0.6213,
+      "step": 1207
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3782941343156703e-05,
+      "loss": 0.6445,
+      "step": 1208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3777274015301787e-05,
+      "loss": 0.6571,
+      "step": 1209
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.377160668744687e-05,
+      "loss": 0.6464,
+      "step": 1210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3765939359591954e-05,
+      "loss": 0.636,
+      "step": 1211
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3760272031737036e-05,
+      "loss": 0.6129,
+      "step": 1212
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3754604703882122e-05,
+      "loss": 0.6732,
+      "step": 1213
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3748937376027205e-05,
+      "loss": 0.4677,
+      "step": 1214
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3743270048172287e-05,
+      "loss": 0.3575,
+      "step": 1215
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3737602720317371e-05,
+      "loss": 0.3952,
+      "step": 1216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3731935392462454e-05,
+      "loss": 0.3648,
+      "step": 1217
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.372626806460754e-05,
+      "loss": 0.3626,
+      "step": 1218
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3720600736752622e-05,
+      "loss": 0.3413,
+      "step": 1219
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3714933408897707e-05,
+      "loss": 0.3263,
+      "step": 1220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3709266081042789e-05,
+      "loss": 0.3322,
+      "step": 1221
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3703598753187871e-05,
+      "loss": 0.3494,
+      "step": 1222
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3697931425332957e-05,
+      "loss": 0.3529,
+      "step": 1223
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3692264097478042e-05,
+      "loss": 0.3734,
+      "step": 1224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3686596769623124e-05,
+      "loss": 0.3331,
+      "step": 1225
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3680929441768207e-05,
+      "loss": 0.3516,
+      "step": 1226
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.367526211391329e-05,
+      "loss": 0.322,
+      "step": 1227
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3669594786058375e-05,
+      "loss": 0.343,
+      "step": 1228
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3663927458203459e-05,
+      "loss": 0.3347,
+      "step": 1229
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3658260130348542e-05,
+      "loss": 0.3535,
+      "step": 1230
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3652592802493626e-05,
+      "loss": 0.343,
+      "step": 1231
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3646925474638708e-05,
+      "loss": 0.3409,
+      "step": 1232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3641258146783794e-05,
+      "loss": 0.3384,
+      "step": 1233
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3635590818928877e-05,
+      "loss": 0.3016,
+      "step": 1234
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3629923491073959e-05,
+      "loss": 0.3633,
+      "step": 1235
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3624256163219043e-05,
+      "loss": 0.3701,
+      "step": 1236
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3618588835364126e-05,
+      "loss": 0.3267,
+      "step": 1237
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3612921507509212e-05,
+      "loss": 0.3129,
+      "step": 1238
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3607254179654294e-05,
+      "loss": 0.3124,
+      "step": 1239
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3601586851799378e-05,
+      "loss": 0.3593,
+      "step": 1240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.359591952394446e-05,
+      "loss": 0.3465,
+      "step": 1241
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3590252196089543e-05,
+      "loss": 0.3272,
+      "step": 1242
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3584584868234629e-05,
+      "loss": 0.3561,
+      "step": 1243
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3578917540379713e-05,
+      "loss": 0.3597,
+      "step": 1244
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3573250212524796e-05,
+      "loss": 0.3411,
+      "step": 1245
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3567582884669878e-05,
+      "loss": 0.3661,
+      "step": 1246
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3561915556814962e-05,
+      "loss": 0.3263,
+      "step": 1247
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3556248228960047e-05,
+      "loss": 0.3612,
+      "step": 1248
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.355058090110513e-05,
+      "loss": 0.314,
+      "step": 1249
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3544913573250213e-05,
+      "loss": 0.3268,
+      "step": 1250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3539246245395297e-05,
+      "loss": 0.3338,
+      "step": 1251
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.353357891754038e-05,
+      "loss": 0.285,
+      "step": 1252
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3527911589685466e-05,
+      "loss": 0.3454,
+      "step": 1253
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3522244261830548e-05,
+      "loss": 0.3323,
+      "step": 1254
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.351657693397563e-05,
+      "loss": 0.3446,
+      "step": 1255
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3510909606120715e-05,
+      "loss": 0.3407,
+      "step": 1256
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3505242278265797e-05,
+      "loss": 0.3318,
+      "step": 1257
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3499574950410883e-05,
+      "loss": 0.3647,
+      "step": 1258
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3493907622555966e-05,
+      "loss": 0.3409,
+      "step": 1259
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.348824029470105e-05,
+      "loss": 0.3357,
+      "step": 1260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3482572966846132e-05,
+      "loss": 0.3374,
+      "step": 1261
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3476905638991218e-05,
+      "loss": 0.3317,
+      "step": 1262
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.34712383111363e-05,
+      "loss": 0.352,
+      "step": 1263
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3465570983281385e-05,
+      "loss": 0.3755,
+      "step": 1264
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3459903655426467e-05,
+      "loss": 0.3363,
+      "step": 1265
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.345423632757155e-05,
+      "loss": 0.3583,
+      "step": 1266
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3448568999716636e-05,
+      "loss": 0.3123,
+      "step": 1267
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3442901671861718e-05,
+      "loss": 0.3192,
+      "step": 1268
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3437234344006802e-05,
+      "loss": 0.2851,
+      "step": 1269
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3431567016151885e-05,
+      "loss": 0.3239,
+      "step": 1270
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3425899688296969e-05,
+      "loss": 0.3317,
+      "step": 1271
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3420232360442053e-05,
+      "loss": 0.3664,
+      "step": 1272
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3414565032587137e-05,
+      "loss": 0.3154,
+      "step": 1273
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.340889770473222e-05,
+      "loss": 0.3552,
+      "step": 1274
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3403230376877302e-05,
+      "loss": 0.3399,
+      "step": 1275
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3397563049022387e-05,
+      "loss": 0.3507,
+      "step": 1276
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.339189572116747e-05,
+      "loss": 0.3225,
+      "step": 1277
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3386228393312555e-05,
+      "loss": 0.3257,
+      "step": 1278
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3380561065457637e-05,
+      "loss": 0.3374,
+      "step": 1279
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3374893737602722e-05,
+      "loss": 0.3269,
+      "step": 1280
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3369226409747804e-05,
+      "loss": 0.3177,
+      "step": 1281
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.336355908189289e-05,
+      "loss": 0.3356,
+      "step": 1282
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3357891754037972e-05,
+      "loss": 0.3237,
+      "step": 1283
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3352224426183057e-05,
+      "loss": 0.3663,
+      "step": 1284
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3346557098328139e-05,
+      "loss": 0.3475,
+      "step": 1285
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3340889770473222e-05,
+      "loss": 0.3124,
+      "step": 1286
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3335222442618307e-05,
+      "loss": 0.36,
+      "step": 1287
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.332955511476339e-05,
+      "loss": 0.3567,
+      "step": 1288
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3323887786908474e-05,
+      "loss": 0.3247,
+      "step": 1289
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3318220459053557e-05,
+      "loss": 0.3194,
+      "step": 1290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.331255313119864e-05,
+      "loss": 0.3307,
+      "step": 1291
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3306885803343725e-05,
+      "loss": 0.3496,
+      "step": 1292
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.330121847548881e-05,
+      "loss": 0.3273,
+      "step": 1293
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3295551147633892e-05,
+      "loss": 0.35,
+      "step": 1294
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3289883819778974e-05,
+      "loss": 0.3181,
+      "step": 1295
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3284216491924058e-05,
+      "loss": 0.3443,
+      "step": 1296
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3278549164069143e-05,
+      "loss": 0.3384,
+      "step": 1297
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3272881836214227e-05,
+      "loss": 0.3355,
+      "step": 1298
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.326721450835931e-05,
+      "loss": 0.3395,
+      "step": 1299
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3261547180504393e-05,
+      "loss": 0.3511,
+      "step": 1300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3255879852649476e-05,
+      "loss": 0.3397,
+      "step": 1301
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3250212524794562e-05,
+      "loss": 0.3209,
+      "step": 1302
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3244545196939644e-05,
+      "loss": 0.3329,
+      "step": 1303
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3238877869084727e-05,
+      "loss": 0.346,
+      "step": 1304
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.323321054122981e-05,
+      "loss": 0.3287,
+      "step": 1305
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3227543213374893e-05,
+      "loss": 0.3129,
+      "step": 1306
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.322187588551998e-05,
+      "loss": 0.3357,
+      "step": 1307
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3216208557665062e-05,
+      "loss": 0.3377,
+      "step": 1308
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3210541229810146e-05,
+      "loss": 0.3559,
+      "step": 1309
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3204873901955228e-05,
+      "loss": 0.3293,
+      "step": 1310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3199206574100313e-05,
+      "loss": 0.379,
+      "step": 1311
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3193539246245397e-05,
+      "loss": 0.3223,
+      "step": 1312
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3187871918390481e-05,
+      "loss": 0.3338,
+      "step": 1313
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3182204590535563e-05,
+      "loss": 0.3543,
+      "step": 1314
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3176537262680646e-05,
+      "loss": 0.3348,
+      "step": 1315
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3170869934825732e-05,
+      "loss": 0.3447,
+      "step": 1316
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3165202606970814e-05,
+      "loss": 0.3534,
+      "step": 1317
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3159535279115898e-05,
+      "loss": 0.333,
+      "step": 1318
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3153867951260981e-05,
+      "loss": 0.3573,
+      "step": 1319
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3148200623406065e-05,
+      "loss": 0.3052,
+      "step": 1320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.314253329555115e-05,
+      "loss": 0.3318,
+      "step": 1321
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3136865967696233e-05,
+      "loss": 0.3418,
+      "step": 1322
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3131198639841316e-05,
+      "loss": 0.3958,
+      "step": 1323
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3125531311986398e-05,
+      "loss": 0.3366,
+      "step": 1324
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3119863984131483e-05,
+      "loss": 0.3556,
+      "step": 1325
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3114196656276568e-05,
+      "loss": 0.3252,
+      "step": 1326
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3108529328421651e-05,
+      "loss": 0.3418,
+      "step": 1327
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3102862000566733e-05,
+      "loss": 0.3486,
+      "step": 1328
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3097194672711818e-05,
+      "loss": 0.3183,
+      "step": 1329
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.30915273448569e-05,
+      "loss": 0.326,
+      "step": 1330
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3085860017001986e-05,
+      "loss": 0.3351,
+      "step": 1331
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3080192689147068e-05,
+      "loss": 0.3394,
+      "step": 1332
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3074525361292153e-05,
+      "loss": 0.3507,
+      "step": 1333
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3068858033437235e-05,
+      "loss": 0.3709,
+      "step": 1334
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3063190705582318e-05,
+      "loss": 0.348,
+      "step": 1335
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3057523377727403e-05,
+      "loss": 0.3683,
+      "step": 1336
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3051856049872486e-05,
+      "loss": 0.3255,
+      "step": 1337
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.304618872201757e-05,
+      "loss": 0.3224,
+      "step": 1338
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3040521394162653e-05,
+      "loss": 0.3419,
+      "step": 1339
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3034854066307737e-05,
+      "loss": 0.2878,
+      "step": 1340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3029186738452821e-05,
+      "loss": 0.3451,
+      "step": 1341
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3023519410597905e-05,
+      "loss": 0.3139,
+      "step": 1342
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3017852082742988e-05,
+      "loss": 0.3425,
+      "step": 1343
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.301218475488807e-05,
+      "loss": 0.3597,
+      "step": 1344
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3006517427033154e-05,
+      "loss": 0.3425,
+      "step": 1345
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.300085009917824e-05,
+      "loss": 0.3153,
+      "step": 1346
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2995182771323323e-05,
+      "loss": 0.3437,
+      "step": 1347
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2989515443468405e-05,
+      "loss": 0.35,
+      "step": 1348
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.298384811561349e-05,
+      "loss": 0.3544,
+      "step": 1349
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2978180787758572e-05,
+      "loss": 0.3639,
+      "step": 1350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2972513459903658e-05,
+      "loss": 0.3624,
+      "step": 1351
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.296684613204874e-05,
+      "loss": 0.3247,
+      "step": 1352
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2961178804193824e-05,
+      "loss": 0.3428,
+      "step": 1353
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2955511476338907e-05,
+      "loss": 0.3363,
+      "step": 1354
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.294984414848399e-05,
+      "loss": 0.2946,
+      "step": 1355
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2944176820629075e-05,
+      "loss": 0.326,
+      "step": 1356
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2938509492774158e-05,
+      "loss": 0.371,
+      "step": 1357
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2932842164919242e-05,
+      "loss": 0.3468,
+      "step": 1358
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2927174837064324e-05,
+      "loss": 0.3181,
+      "step": 1359
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2921507509209408e-05,
+      "loss": 0.306,
+      "step": 1360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2915840181354493e-05,
+      "loss": 0.3182,
+      "step": 1361
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2910172853499577e-05,
+      "loss": 0.345,
+      "step": 1362
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.290450552564466e-05,
+      "loss": 0.341,
+      "step": 1363
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2898838197789742e-05,
+      "loss": 0.3813,
+      "step": 1364
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2893170869934826e-05,
+      "loss": 0.3274,
+      "step": 1365
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2887503542079912e-05,
+      "loss": 0.3512,
+      "step": 1366
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2881836214224994e-05,
+      "loss": 0.3297,
+      "step": 1367
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2876168886370077e-05,
+      "loss": 0.3291,
+      "step": 1368
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2870501558515161e-05,
+      "loss": 0.3478,
+      "step": 1369
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2864834230660245e-05,
+      "loss": 0.3377,
+      "step": 1370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.285916690280533e-05,
+      "loss": 0.3507,
+      "step": 1371
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2853499574950412e-05,
+      "loss": 0.3292,
+      "step": 1372
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2847832247095496e-05,
+      "loss": 0.3497,
+      "step": 1373
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2842164919240578e-05,
+      "loss": 0.346,
+      "step": 1374
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2836497591385664e-05,
+      "loss": 0.3275,
+      "step": 1375
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2830830263530747e-05,
+      "loss": 0.3323,
+      "step": 1376
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.282516293567583e-05,
+      "loss": 0.3493,
+      "step": 1377
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2819495607820914e-05,
+      "loss": 0.3675,
+      "step": 1378
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2813828279965996e-05,
+      "loss": 0.342,
+      "step": 1379
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2808160952111082e-05,
+      "loss": 0.3235,
+      "step": 1380
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2802493624256164e-05,
+      "loss": 0.333,
+      "step": 1381
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2796826296401249e-05,
+      "loss": 0.33,
+      "step": 1382
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2791158968546331e-05,
+      "loss": 0.3202,
+      "step": 1383
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2785491640691413e-05,
+      "loss": 0.3499,
+      "step": 1384
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.27798243128365e-05,
+      "loss": 0.3233,
+      "step": 1385
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2774156984981584e-05,
+      "loss": 0.3226,
+      "step": 1386
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2768489657126666e-05,
+      "loss": 0.3201,
+      "step": 1387
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2762822329271749e-05,
+      "loss": 0.3414,
+      "step": 1388
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2757155001416833e-05,
+      "loss": 0.3309,
+      "step": 1389
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2751487673561917e-05,
+      "loss": 0.3606,
+      "step": 1390
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2745820345707001e-05,
+      "loss": 0.372,
+      "step": 1391
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2740153017852084e-05,
+      "loss": 0.3588,
+      "step": 1392
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2734485689997168e-05,
+      "loss": 0.3659,
+      "step": 1393
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.272881836214225e-05,
+      "loss": 0.3347,
+      "step": 1394
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2723151034287336e-05,
+      "loss": 0.3514,
+      "step": 1395
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2717483706432419e-05,
+      "loss": 0.3499,
+      "step": 1396
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2711816378577501e-05,
+      "loss": 0.3463,
+      "step": 1397
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2706149050722585e-05,
+      "loss": 0.3303,
+      "step": 1398
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2700481722867668e-05,
+      "loss": 0.3463,
+      "step": 1399
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2694814395012754e-05,
+      "loss": 0.3302,
+      "step": 1400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2689147067157836e-05,
+      "loss": 0.3291,
+      "step": 1401
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.268347973930292e-05,
+      "loss": 0.3282,
+      "step": 1402
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2677812411448003e-05,
+      "loss": 0.3569,
+      "step": 1403
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2672145083593085e-05,
+      "loss": 0.3328,
+      "step": 1404
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2666477755738171e-05,
+      "loss": 0.2909,
+      "step": 1405
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2660810427883255e-05,
+      "loss": 0.3703,
+      "step": 1406
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2655143100028338e-05,
+      "loss": 0.3179,
+      "step": 1407
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.264947577217342e-05,
+      "loss": 0.3408,
+      "step": 1408
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2643808444318504e-05,
+      "loss": 0.3495,
+      "step": 1409
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2638141116463589e-05,
+      "loss": 0.3352,
+      "step": 1410
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2632473788608673e-05,
+      "loss": 0.331,
+      "step": 1411
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2626806460753755e-05,
+      "loss": 0.3455,
+      "step": 1412
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.262113913289884e-05,
+      "loss": 0.3125,
+      "step": 1413
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2615471805043922e-05,
+      "loss": 0.3346,
+      "step": 1414
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2609804477189008e-05,
+      "loss": 0.3149,
+      "step": 1415
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.260413714933409e-05,
+      "loss": 0.3274,
+      "step": 1416
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2598469821479173e-05,
+      "loss": 0.3077,
+      "step": 1417
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2592802493624257e-05,
+      "loss": 0.3028,
+      "step": 1418
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.258713516576934e-05,
+      "loss": 0.325,
+      "step": 1419
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2581467837914425e-05,
+      "loss": 0.3395,
+      "step": 1420
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2575800510059508e-05,
+      "loss": 0.3409,
+      "step": 1421
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2570133182204592e-05,
+      "loss": 0.333,
+      "step": 1422
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2564465854349674e-05,
+      "loss": 0.2913,
+      "step": 1423
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.255879852649476e-05,
+      "loss": 0.3331,
+      "step": 1424
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2553131198639843e-05,
+      "loss": 0.3387,
+      "step": 1425
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2547463870784927e-05,
+      "loss": 0.3474,
+      "step": 1426
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.254179654293001e-05,
+      "loss": 0.3217,
+      "step": 1427
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2536129215075092e-05,
+      "loss": 0.3457,
+      "step": 1428
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2530461887220178e-05,
+      "loss": 0.321,
+      "step": 1429
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.252479455936526e-05,
+      "loss": 0.3438,
+      "step": 1430
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2519127231510344e-05,
+      "loss": 0.3421,
+      "step": 1431
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2513459903655427e-05,
+      "loss": 0.3184,
+      "step": 1432
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2507792575800511e-05,
+      "loss": 0.3379,
+      "step": 1433
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2502125247945595e-05,
+      "loss": 0.3377,
+      "step": 1434
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.249645792009068e-05,
+      "loss": 0.3086,
+      "step": 1435
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2490790592235762e-05,
+      "loss": 0.3458,
+      "step": 1436
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2485123264380844e-05,
+      "loss": 0.3316,
+      "step": 1437
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2479455936525929e-05,
+      "loss": 0.3512,
+      "step": 1438
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2473788608671013e-05,
+      "loss": 0.3206,
+      "step": 1439
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2468121280816097e-05,
+      "loss": 0.3463,
+      "step": 1440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.246245395296118e-05,
+      "loss": 0.3293,
+      "step": 1441
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2456786625106264e-05,
+      "loss": 0.3329,
+      "step": 1442
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2451119297251346e-05,
+      "loss": 0.3517,
+      "step": 1443
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2445451969396432e-05,
+      "loss": 0.3843,
+      "step": 1444
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2439784641541514e-05,
+      "loss": 0.3349,
+      "step": 1445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2434117313686597e-05,
+      "loss": 0.3531,
+      "step": 1446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2428449985831681e-05,
+      "loss": 0.3453,
+      "step": 1447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2422782657976764e-05,
+      "loss": 0.368,
+      "step": 1448
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.241711533012185e-05,
+      "loss": 0.3503,
+      "step": 1449
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2411448002266932e-05,
+      "loss": 0.3178,
+      "step": 1450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2405780674412016e-05,
+      "loss": 0.3362,
+      "step": 1451
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2400113346557099e-05,
+      "loss": 0.3513,
+      "step": 1452
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2394446018702183e-05,
+      "loss": 0.3694,
+      "step": 1453
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2388778690847267e-05,
+      "loss": 0.3267,
+      "step": 1454
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2383111362992351e-05,
+      "loss": 0.308,
+      "step": 1455
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2377444035137434e-05,
+      "loss": 0.3738,
+      "step": 1456
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2371776707282516e-05,
+      "loss": 0.3166,
+      "step": 1457
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.23661093794276e-05,
+      "loss": 0.333,
+      "step": 1458
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2360442051572685e-05,
+      "loss": 0.3406,
+      "step": 1459
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2354774723717769e-05,
+      "loss": 0.3369,
+      "step": 1460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2349107395862851e-05,
+      "loss": 0.3326,
+      "step": 1461
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2343440068007935e-05,
+      "loss": 0.3424,
+      "step": 1462
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2337772740153018e-05,
+      "loss": 0.3178,
+      "step": 1463
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2332105412298104e-05,
+      "loss": 0.3435,
+      "step": 1464
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2326438084443186e-05,
+      "loss": 0.3334,
+      "step": 1465
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2320770756588269e-05,
+      "loss": 0.3318,
+      "step": 1466
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2315103428733353e-05,
+      "loss": 0.3433,
+      "step": 1467
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2309436100878435e-05,
+      "loss": 0.3777,
+      "step": 1468
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2303768773023521e-05,
+      "loss": 0.3496,
+      "step": 1469
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2298101445168604e-05,
+      "loss": 0.337,
+      "step": 1470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2292434117313688e-05,
+      "loss": 0.3297,
+      "step": 1471
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.228676678945877e-05,
+      "loss": 0.3166,
+      "step": 1472
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2281099461603853e-05,
+      "loss": 0.3472,
+      "step": 1473
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2275432133748939e-05,
+      "loss": 0.3195,
+      "step": 1474
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2269764805894023e-05,
+      "loss": 0.3184,
+      "step": 1475
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2264097478039105e-05,
+      "loss": 0.3411,
+      "step": 1476
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2258430150184188e-05,
+      "loss": 0.3394,
+      "step": 1477
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2252762822329274e-05,
+      "loss": 0.3291,
+      "step": 1478
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2247095494474356e-05,
+      "loss": 0.3212,
+      "step": 1479
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.224142816661944e-05,
+      "loss": 0.3249,
+      "step": 1480
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2235760838764523e-05,
+      "loss": 0.3457,
+      "step": 1481
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2230093510909607e-05,
+      "loss": 0.3063,
+      "step": 1482
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2224426183054691e-05,
+      "loss": 0.3324,
+      "step": 1483
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2218758855199775e-05,
+      "loss": 0.3527,
+      "step": 1484
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2213091527344858e-05,
+      "loss": 0.3339,
+      "step": 1485
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.220742419948994e-05,
+      "loss": 0.3481,
+      "step": 1486
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2201756871635025e-05,
+      "loss": 0.333,
+      "step": 1487
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.219608954378011e-05,
+      "loss": 0.3378,
+      "step": 1488
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2190422215925193e-05,
+      "loss": 0.3438,
+      "step": 1489
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2184754888070275e-05,
+      "loss": 0.3862,
+      "step": 1490
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.217908756021536e-05,
+      "loss": 0.332,
+      "step": 1491
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2173420232360442e-05,
+      "loss": 0.3468,
+      "step": 1492
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2167752904505528e-05,
+      "loss": 0.3069,
+      "step": 1493
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.216208557665061e-05,
+      "loss": 0.3436,
+      "step": 1494
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2156418248795695e-05,
+      "loss": 0.3371,
+      "step": 1495
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2150750920940777e-05,
+      "loss": 0.3287,
+      "step": 1496
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.214508359308586e-05,
+      "loss": 0.317,
+      "step": 1497
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2139416265230945e-05,
+      "loss": 0.3576,
+      "step": 1498
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2133748937376028e-05,
+      "loss": 0.3262,
+      "step": 1499
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2128081609521112e-05,
+      "loss": 0.3217,
+      "step": 1500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2122414281666195e-05,
+      "loss": 0.3378,
+      "step": 1501
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2116746953811279e-05,
+      "loss": 0.3194,
+      "step": 1502
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2111079625956363e-05,
+      "loss": 0.3464,
+      "step": 1503
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2105412298101447e-05,
+      "loss": 0.3414,
+      "step": 1504
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.209974497024653e-05,
+      "loss": 0.3082,
+      "step": 1505
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2094077642391612e-05,
+      "loss": 0.3526,
+      "step": 1506
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2088410314536696e-05,
+      "loss": 0.3323,
+      "step": 1507
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2082742986681782e-05,
+      "loss": 0.3514,
+      "step": 1508
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2077075658826865e-05,
+      "loss": 0.3347,
+      "step": 1509
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2071408330971947e-05,
+      "loss": 0.3689,
+      "step": 1510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2065741003117031e-05,
+      "loss": 0.3063,
+      "step": 1511
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2060073675262114e-05,
+      "loss": 0.3525,
+      "step": 1512
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.20544063474072e-05,
+      "loss": 0.3262,
+      "step": 1513
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2048739019552282e-05,
+      "loss": 0.3306,
+      "step": 1514
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2043071691697366e-05,
+      "loss": 0.3204,
+      "step": 1515
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2037404363842449e-05,
+      "loss": 0.3409,
+      "step": 1516
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2031737035987531e-05,
+      "loss": 0.3347,
+      "step": 1517
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2026069708132617e-05,
+      "loss": 0.3114,
+      "step": 1518
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.20204023802777e-05,
+      "loss": 0.3261,
+      "step": 1519
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2014735052422784e-05,
+      "loss": 0.3446,
+      "step": 1520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2009067724567866e-05,
+      "loss": 0.3484,
+      "step": 1521
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.200340039671295e-05,
+      "loss": 0.3434,
+      "step": 1522
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1997733068858035e-05,
+      "loss": 0.3716,
+      "step": 1523
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1992065741003119e-05,
+      "loss": 0.3267,
+      "step": 1524
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1986398413148201e-05,
+      "loss": 0.3414,
+      "step": 1525
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1980731085293284e-05,
+      "loss": 0.3418,
+      "step": 1526
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1975063757438368e-05,
+      "loss": 0.3416,
+      "step": 1527
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1969396429583454e-05,
+      "loss": 0.3321,
+      "step": 1528
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1963729101728536e-05,
+      "loss": 0.3625,
+      "step": 1529
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1958061773873619e-05,
+      "loss": 0.3532,
+      "step": 1530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1952394446018703e-05,
+      "loss": 0.3337,
+      "step": 1531
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1946727118163787e-05,
+      "loss": 0.3089,
+      "step": 1532
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1941059790308871e-05,
+      "loss": 0.3277,
+      "step": 1533
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1935392462453954e-05,
+      "loss": 0.3491,
+      "step": 1534
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1929725134599038e-05,
+      "loss": 0.3161,
+      "step": 1535
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.192405780674412e-05,
+      "loss": 0.3464,
+      "step": 1536
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1918390478889206e-05,
+      "loss": 0.337,
+      "step": 1537
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1912723151034289e-05,
+      "loss": 0.3148,
+      "step": 1538
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1907055823179371e-05,
+      "loss": 0.3277,
+      "step": 1539
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1901388495324456e-05,
+      "loss": 0.316,
+      "step": 1540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1895721167469538e-05,
+      "loss": 0.3335,
+      "step": 1541
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1890053839614624e-05,
+      "loss": 0.3418,
+      "step": 1542
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1884386511759706e-05,
+      "loss": 0.3369,
+      "step": 1543
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.187871918390479e-05,
+      "loss": 0.3325,
+      "step": 1544
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1873051856049873e-05,
+      "loss": 0.3052,
+      "step": 1545
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1867384528194956e-05,
+      "loss": 0.3354,
+      "step": 1546
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1861717200340041e-05,
+      "loss": 0.3298,
+      "step": 1547
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1856049872485126e-05,
+      "loss": 0.3297,
+      "step": 1548
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1850382544630208e-05,
+      "loss": 0.3534,
+      "step": 1549
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.184471521677529e-05,
+      "loss": 0.3051,
+      "step": 1550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1839047888920375e-05,
+      "loss": 0.3321,
+      "step": 1551
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1833380561065459e-05,
+      "loss": 0.3141,
+      "step": 1552
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1827713233210543e-05,
+      "loss": 0.3305,
+      "step": 1553
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1822045905355626e-05,
+      "loss": 0.3157,
+      "step": 1554
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.181637857750071e-05,
+      "loss": 0.3404,
+      "step": 1555
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1810711249645792e-05,
+      "loss": 0.3603,
+      "step": 1556
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1805043921790878e-05,
+      "loss": 0.357,
+      "step": 1557
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.179937659393596e-05,
+      "loss": 0.3532,
+      "step": 1558
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1793709266081043e-05,
+      "loss": 0.3698,
+      "step": 1559
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1788041938226127e-05,
+      "loss": 0.3002,
+      "step": 1560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.178237461037121e-05,
+      "loss": 0.3349,
+      "step": 1561
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1776707282516296e-05,
+      "loss": 0.3378,
+      "step": 1562
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1771039954661378e-05,
+      "loss": 0.3583,
+      "step": 1563
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1765372626806462e-05,
+      "loss": 0.3484,
+      "step": 1564
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1759705298951545e-05,
+      "loss": 0.3342,
+      "step": 1565
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1754037971096627e-05,
+      "loss": 0.341,
+      "step": 1566
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1748370643241713e-05,
+      "loss": 0.3538,
+      "step": 1567
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1742703315386797e-05,
+      "loss": 0.3209,
+      "step": 1568
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.173703598753188e-05,
+      "loss": 0.3244,
+      "step": 1569
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1731368659676962e-05,
+      "loss": 0.3372,
+      "step": 1570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1725701331822046e-05,
+      "loss": 0.3567,
+      "step": 1571
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.172003400396713e-05,
+      "loss": 0.3531,
+      "step": 1572
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1714366676112215e-05,
+      "loss": 0.3127,
+      "step": 1573
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1708699348257297e-05,
+      "loss": 0.3515,
+      "step": 1574
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1703032020402381e-05,
+      "loss": 0.3235,
+      "step": 1575
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1697364692547464e-05,
+      "loss": 0.3064,
+      "step": 1576
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.169169736469255e-05,
+      "loss": 0.3563,
+      "step": 1577
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1686030036837632e-05,
+      "loss": 0.3438,
+      "step": 1578
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1680362708982715e-05,
+      "loss": 0.3106,
+      "step": 1579
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1674695381127799e-05,
+      "loss": 0.3231,
+      "step": 1580
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1669028053272881e-05,
+      "loss": 0.3235,
+      "step": 1581
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1663360725417967e-05,
+      "loss": 0.3809,
+      "step": 1582
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.165769339756305e-05,
+      "loss": 0.3453,
+      "step": 1583
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1652026069708134e-05,
+      "loss": 0.3172,
+      "step": 1584
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1646358741853216e-05,
+      "loss": 0.35,
+      "step": 1585
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1640691413998302e-05,
+      "loss": 0.3625,
+      "step": 1586
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1635024086143385e-05,
+      "loss": 0.3168,
+      "step": 1587
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1629356758288467e-05,
+      "loss": 0.3665,
+      "step": 1588
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1623689430433551e-05,
+      "loss": 0.3232,
+      "step": 1589
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1618022102578634e-05,
+      "loss": 0.3184,
+      "step": 1590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.161235477472372e-05,
+      "loss": 0.3537,
+      "step": 1591
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1606687446868802e-05,
+      "loss": 0.3309,
+      "step": 1592
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1601020119013886e-05,
+      "loss": 0.3549,
+      "step": 1593
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1595352791158969e-05,
+      "loss": 0.3442,
+      "step": 1594
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1589685463304053e-05,
+      "loss": 0.346,
+      "step": 1595
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1584018135449137e-05,
+      "loss": 0.3355,
+      "step": 1596
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1578350807594221e-05,
+      "loss": 0.3311,
+      "step": 1597
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1572683479739304e-05,
+      "loss": 0.3408,
+      "step": 1598
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1567016151884386e-05,
+      "loss": 0.3603,
+      "step": 1599
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.156134882402947e-05,
+      "loss": 0.3468,
+      "step": 1600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1555681496174555e-05,
+      "loss": 0.3438,
+      "step": 1601
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1550014168319639e-05,
+      "loss": 0.3378,
+      "step": 1602
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1544346840464721e-05,
+      "loss": 0.3383,
+      "step": 1603
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1538679512609806e-05,
+      "loss": 0.3302,
+      "step": 1604
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1533012184754888e-05,
+      "loss": 0.3454,
+      "step": 1605
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1527344856899974e-05,
+      "loss": 0.3361,
+      "step": 1606
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1521677529045056e-05,
+      "loss": 0.368,
+      "step": 1607
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1516010201190139e-05,
+      "loss": 0.33,
+      "step": 1608
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1510342873335223e-05,
+      "loss": 0.3315,
+      "step": 1609
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1504675545480306e-05,
+      "loss": 0.3382,
+      "step": 1610
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1499008217625392e-05,
+      "loss": 0.3418,
+      "step": 1611
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1493340889770474e-05,
+      "loss": 0.3328,
+      "step": 1612
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1487673561915558e-05,
+      "loss": 0.3327,
+      "step": 1613
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.148200623406064e-05,
+      "loss": 0.3389,
+      "step": 1614
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1476338906205723e-05,
+      "loss": 0.33,
+      "step": 1615
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1470671578350809e-05,
+      "loss": 0.34,
+      "step": 1616
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1465004250495893e-05,
+      "loss": 0.3584,
+      "step": 1617
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1459336922640976e-05,
+      "loss": 0.3491,
+      "step": 1618
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1453669594786058e-05,
+      "loss": 0.3141,
+      "step": 1619
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1448002266931142e-05,
+      "loss": 0.323,
+      "step": 1620
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1442334939076227e-05,
+      "loss": 0.3547,
+      "step": 1621
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.143666761122131e-05,
+      "loss": 0.3744,
+      "step": 1622
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1431000283366393e-05,
+      "loss": 0.3134,
+      "step": 1623
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1425332955511477e-05,
+      "loss": 0.329,
+      "step": 1624
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.141966562765656e-05,
+      "loss": 0.3642,
+      "step": 1625
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1413998299801646e-05,
+      "loss": 0.3104,
+      "step": 1626
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1408330971946728e-05,
+      "loss": 0.3462,
+      "step": 1627
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.140266364409181e-05,
+      "loss": 0.3549,
+      "step": 1628
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1396996316236895e-05,
+      "loss": 0.3192,
+      "step": 1629
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1391328988381977e-05,
+      "loss": 0.328,
+      "step": 1630
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1385661660527063e-05,
+      "loss": 0.3302,
+      "step": 1631
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1379994332672146e-05,
+      "loss": 0.3069,
+      "step": 1632
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.137432700481723e-05,
+      "loss": 0.3528,
+      "step": 1633
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1368659676962312e-05,
+      "loss": 0.3196,
+      "step": 1634
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1362992349107398e-05,
+      "loss": 0.303,
+      "step": 1635
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.135732502125248e-05,
+      "loss": 0.3673,
+      "step": 1636
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1351657693397565e-05,
+      "loss": 0.3175,
+      "step": 1637
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1345990365542647e-05,
+      "loss": 0.3052,
+      "step": 1638
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.134032303768773e-05,
+      "loss": 0.3201,
+      "step": 1639
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1334655709832816e-05,
+      "loss": 0.3151,
+      "step": 1640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1328988381977898e-05,
+      "loss": 0.3484,
+      "step": 1641
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1323321054122982e-05,
+      "loss": 0.3389,
+      "step": 1642
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1317653726268065e-05,
+      "loss": 0.3399,
+      "step": 1643
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1311986398413149e-05,
+      "loss": 0.3539,
+      "step": 1644
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1306319070558233e-05,
+      "loss": 0.3232,
+      "step": 1645
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1300651742703317e-05,
+      "loss": 0.3203,
+      "step": 1646
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.12949844148484e-05,
+      "loss": 0.3235,
+      "step": 1647
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1289317086993482e-05,
+      "loss": 0.3508,
+      "step": 1648
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1283649759138567e-05,
+      "loss": 0.3281,
+      "step": 1649
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1277982431283652e-05,
+      "loss": 0.3225,
+      "step": 1650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1272315103428735e-05,
+      "loss": 0.3401,
+      "step": 1651
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1266647775573817e-05,
+      "loss": 0.3518,
+      "step": 1652
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1260980447718902e-05,
+      "loss": 0.3265,
+      "step": 1653
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1255313119863984e-05,
+      "loss": 0.3064,
+      "step": 1654
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.124964579200907e-05,
+      "loss": 0.3234,
+      "step": 1655
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1243978464154152e-05,
+      "loss": 0.3363,
+      "step": 1656
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1238311136299237e-05,
+      "loss": 0.3431,
+      "step": 1657
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1232643808444319e-05,
+      "loss": 0.3309,
+      "step": 1658
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1226976480589402e-05,
+      "loss": 0.363,
+      "step": 1659
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1221309152734487e-05,
+      "loss": 0.3091,
+      "step": 1660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.121564182487957e-05,
+      "loss": 0.3445,
+      "step": 1661
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1209974497024654e-05,
+      "loss": 0.3436,
+      "step": 1662
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1204307169169737e-05,
+      "loss": 0.3467,
+      "step": 1663
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.119863984131482e-05,
+      "loss": 0.3191,
+      "step": 1664
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1192972513459905e-05,
+      "loss": 0.3143,
+      "step": 1665
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1187305185604989e-05,
+      "loss": 0.3366,
+      "step": 1666
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1181637857750072e-05,
+      "loss": 0.361,
+      "step": 1667
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1175970529895154e-05,
+      "loss": 0.3302,
+      "step": 1668
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1170303202040238e-05,
+      "loss": 0.3415,
+      "step": 1669
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1164635874185324e-05,
+      "loss": 0.3116,
+      "step": 1670
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1158968546330407e-05,
+      "loss": 0.3939,
+      "step": 1671
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1153301218475489e-05,
+      "loss": 0.3396,
+      "step": 1672
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1147633890620573e-05,
+      "loss": 0.3387,
+      "step": 1673
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1141966562765656e-05,
+      "loss": 0.3509,
+      "step": 1674
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1136299234910742e-05,
+      "loss": 0.3272,
+      "step": 1675
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1130631907055824e-05,
+      "loss": 0.3443,
+      "step": 1676
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1124964579200908e-05,
+      "loss": 0.3216,
+      "step": 1677
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.111929725134599e-05,
+      "loss": 0.3535,
+      "step": 1678
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1113629923491073e-05,
+      "loss": 0.3399,
+      "step": 1679
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.110796259563616e-05,
+      "loss": 0.3901,
+      "step": 1680
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1102295267781242e-05,
+      "loss": 0.348,
+      "step": 1681
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1096627939926326e-05,
+      "loss": 0.3118,
+      "step": 1682
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1090960612071408e-05,
+      "loss": 0.3388,
+      "step": 1683
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1085293284216492e-05,
+      "loss": 0.354,
+      "step": 1684
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1079625956361577e-05,
+      "loss": 0.3329,
+      "step": 1685
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.107395862850666e-05,
+      "loss": 0.3193,
+      "step": 1686
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1068291300651743e-05,
+      "loss": 0.3416,
+      "step": 1687
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1062623972796826e-05,
+      "loss": 0.3614,
+      "step": 1688
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1056956644941912e-05,
+      "loss": 0.3493,
+      "step": 1689
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1051289317086996e-05,
+      "loss": 0.3498,
+      "step": 1690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1045621989232078e-05,
+      "loss": 0.3228,
+      "step": 1691
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.103995466137716e-05,
+      "loss": 0.3452,
+      "step": 1692
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1034287333522245e-05,
+      "loss": 0.3445,
+      "step": 1693
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.102862000566733e-05,
+      "loss": 0.3447,
+      "step": 1694
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1022952677812413e-05,
+      "loss": 0.3206,
+      "step": 1695
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1017285349957496e-05,
+      "loss": 0.3573,
+      "step": 1696
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.101161802210258e-05,
+      "loss": 0.3689,
+      "step": 1697
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1005950694247663e-05,
+      "loss": 0.3147,
+      "step": 1698
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1000283366392748e-05,
+      "loss": 0.3565,
+      "step": 1699
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0994616038537831e-05,
+      "loss": 0.3446,
+      "step": 1700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0988948710682913e-05,
+      "loss": 0.3153,
+      "step": 1701
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0983281382827998e-05,
+      "loss": 0.3172,
+      "step": 1702
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.097761405497308e-05,
+      "loss": 0.3343,
+      "step": 1703
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0971946727118166e-05,
+      "loss": 0.3356,
+      "step": 1704
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0966279399263248e-05,
+      "loss": 0.3373,
+      "step": 1705
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0960612071408333e-05,
+      "loss": 0.3453,
+      "step": 1706
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0954944743553415e-05,
+      "loss": 0.3432,
+      "step": 1707
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0949277415698498e-05,
+      "loss": 0.315,
+      "step": 1708
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0943610087843583e-05,
+      "loss": 0.3522,
+      "step": 1709
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0937942759988668e-05,
+      "loss": 0.292,
+      "step": 1710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.093227543213375e-05,
+      "loss": 0.3177,
+      "step": 1711
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0926608104278833e-05,
+      "loss": 0.3404,
+      "step": 1712
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0920940776423917e-05,
+      "loss": 0.3672,
+      "step": 1713
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0915273448569001e-05,
+      "loss": 0.346,
+      "step": 1714
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0909606120714085e-05,
+      "loss": 0.3452,
+      "step": 1715
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0903938792859168e-05,
+      "loss": 0.336,
+      "step": 1716
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0898271465004252e-05,
+      "loss": 0.3421,
+      "step": 1717
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0892604137149334e-05,
+      "loss": 0.3241,
+      "step": 1718
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.088693680929442e-05,
+      "loss": 0.3202,
+      "step": 1719
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0881269481439503e-05,
+      "loss": 0.3051,
+      "step": 1720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0875602153584585e-05,
+      "loss": 0.3363,
+      "step": 1721
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.086993482572967e-05,
+      "loss": 0.3325,
+      "step": 1722
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0864267497874752e-05,
+      "loss": 0.3125,
+      "step": 1723
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0858600170019838e-05,
+      "loss": 0.3279,
+      "step": 1724
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.085293284216492e-05,
+      "loss": 0.3266,
+      "step": 1725
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0847265514310004e-05,
+      "loss": 0.3369,
+      "step": 1726
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0841598186455087e-05,
+      "loss": 0.3497,
+      "step": 1727
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.083593085860017e-05,
+      "loss": 0.3356,
+      "step": 1728
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0830263530745255e-05,
+      "loss": 0.3646,
+      "step": 1729
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0824596202890338e-05,
+      "loss": 0.3225,
+      "step": 1730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0818928875035422e-05,
+      "loss": 0.3452,
+      "step": 1731
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0813261547180504e-05,
+      "loss": 0.3415,
+      "step": 1732
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0807594219325588e-05,
+      "loss": 0.3213,
+      "step": 1733
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0801926891470673e-05,
+      "loss": 0.3451,
+      "step": 1734
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0796259563615757e-05,
+      "loss": 0.3597,
+      "step": 1735
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.079059223576084e-05,
+      "loss": 0.3368,
+      "step": 1736
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0784924907905923e-05,
+      "loss": 0.3243,
+      "step": 1737
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0779257580051006e-05,
+      "loss": 0.3488,
+      "step": 1738
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0773590252196092e-05,
+      "loss": 0.3486,
+      "step": 1739
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0767922924341174e-05,
+      "loss": 0.3338,
+      "step": 1740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0762255596486257e-05,
+      "loss": 0.3669,
+      "step": 1741
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0756588268631341e-05,
+      "loss": 0.3408,
+      "step": 1742
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0750920940776425e-05,
+      "loss": 0.3412,
+      "step": 1743
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.074525361292151e-05,
+      "loss": 0.3482,
+      "step": 1744
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0739586285066592e-05,
+      "loss": 0.321,
+      "step": 1745
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0733918957211676e-05,
+      "loss": 0.3228,
+      "step": 1746
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0728251629356758e-05,
+      "loss": 0.3271,
+      "step": 1747
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0722584301501844e-05,
+      "loss": 0.3594,
+      "step": 1748
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0716916973646927e-05,
+      "loss": 0.34,
+      "step": 1749
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.071124964579201e-05,
+      "loss": 0.331,
+      "step": 1750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0705582317937093e-05,
+      "loss": 0.3737,
+      "step": 1751
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0699914990082176e-05,
+      "loss": 0.3486,
+      "step": 1752
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0694247662227262e-05,
+      "loss": 0.3071,
+      "step": 1753
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0688580334372344e-05,
+      "loss": 0.3402,
+      "step": 1754
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0682913006517428e-05,
+      "loss": 0.3604,
+      "step": 1755
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0677245678662511e-05,
+      "loss": 0.3314,
+      "step": 1756
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0671578350807593e-05,
+      "loss": 0.3254,
+      "step": 1757
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.066591102295268e-05,
+      "loss": 0.3089,
+      "step": 1758
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0660243695097764e-05,
+      "loss": 0.3389,
+      "step": 1759
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0654576367242846e-05,
+      "loss": 0.3166,
+      "step": 1760
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0648909039387928e-05,
+      "loss": 0.3402,
+      "step": 1761
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0643241711533013e-05,
+      "loss": 0.2869,
+      "step": 1762
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0637574383678097e-05,
+      "loss": 0.3561,
+      "step": 1763
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0631907055823181e-05,
+      "loss": 0.3655,
+      "step": 1764
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0626239727968263e-05,
+      "loss": 0.3488,
+      "step": 1765
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0620572400113348e-05,
+      "loss": 0.3185,
+      "step": 1766
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.061490507225843e-05,
+      "loss": 0.3519,
+      "step": 1767
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0609237744403516e-05,
+      "loss": 0.3293,
+      "step": 1768
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0603570416548599e-05,
+      "loss": 0.3163,
+      "step": 1769
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0597903088693681e-05,
+      "loss": 0.3222,
+      "step": 1770
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0592235760838765e-05,
+      "loss": 0.3054,
+      "step": 1771
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0586568432983848e-05,
+      "loss": 0.3506,
+      "step": 1772
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0580901105128934e-05,
+      "loss": 0.3202,
+      "step": 1773
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0575233777274016e-05,
+      "loss": 0.3503,
+      "step": 1774
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.05695664494191e-05,
+      "loss": 0.3578,
+      "step": 1775
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0563899121564183e-05,
+      "loss": 0.3266,
+      "step": 1776
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0558231793709265e-05,
+      "loss": 0.3368,
+      "step": 1777
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0552564465854351e-05,
+      "loss": 0.3555,
+      "step": 1778
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0546897137999435e-05,
+      "loss": 0.3424,
+      "step": 1779
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0541229810144518e-05,
+      "loss": 0.3099,
+      "step": 1780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.05355624822896e-05,
+      "loss": 0.3378,
+      "step": 1781
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0529895154434684e-05,
+      "loss": 0.3406,
+      "step": 1782
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0524227826579769e-05,
+      "loss": 0.3617,
+      "step": 1783
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0518560498724853e-05,
+      "loss": 0.36,
+      "step": 1784
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0512893170869935e-05,
+      "loss": 0.3444,
+      "step": 1785
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.050722584301502e-05,
+      "loss": 0.3279,
+      "step": 1786
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0501558515160102e-05,
+      "loss": 0.3118,
+      "step": 1787
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0495891187305188e-05,
+      "loss": 0.3225,
+      "step": 1788
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.049022385945027e-05,
+      "loss": 0.3447,
+      "step": 1789
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0484556531595353e-05,
+      "loss": 0.368,
+      "step": 1790
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0478889203740437e-05,
+      "loss": 0.3158,
+      "step": 1791
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.047322187588552e-05,
+      "loss": 0.3246,
+      "step": 1792
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0467554548030605e-05,
+      "loss": 0.35,
+      "step": 1793
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0461887220175688e-05,
+      "loss": 0.3431,
+      "step": 1794
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0456219892320772e-05,
+      "loss": 0.3329,
+      "step": 1795
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0450552564465854e-05,
+      "loss": 0.3102,
+      "step": 1796
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.044488523661094e-05,
+      "loss": 0.3547,
+      "step": 1797
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0439217908756023e-05,
+      "loss": 0.3189,
+      "step": 1798
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0433550580901107e-05,
+      "loss": 0.3257,
+      "step": 1799
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.042788325304619e-05,
+      "loss": 0.3436,
+      "step": 1800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0422215925191272e-05,
+      "loss": 0.3048,
+      "step": 1801
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0416548597336358e-05,
+      "loss": 0.3349,
+      "step": 1802
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.041088126948144e-05,
+      "loss": 0.3514,
+      "step": 1803
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0405213941626524e-05,
+      "loss": 0.3109,
+      "step": 1804
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0399546613771607e-05,
+      "loss": 0.3116,
+      "step": 1805
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0393879285916691e-05,
+      "loss": 0.3163,
+      "step": 1806
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0388211958061775e-05,
+      "loss": 0.3366,
+      "step": 1807
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.038254463020686e-05,
+      "loss": 0.3499,
+      "step": 1808
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0376877302351942e-05,
+      "loss": 0.3079,
+      "step": 1809
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0371209974497024e-05,
+      "loss": 0.322,
+      "step": 1810
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0365542646642109e-05,
+      "loss": 0.2955,
+      "step": 1811
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0359875318787194e-05,
+      "loss": 0.3664,
+      "step": 1812
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0354207990932277e-05,
+      "loss": 0.3477,
+      "step": 1813
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.034854066307736e-05,
+      "loss": 0.3395,
+      "step": 1814
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0342873335222444e-05,
+      "loss": 0.3247,
+      "step": 1815
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0337206007367526e-05,
+      "loss": 0.3278,
+      "step": 1816
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0331538679512612e-05,
+      "loss": 0.3364,
+      "step": 1817
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0325871351657694e-05,
+      "loss": 0.3305,
+      "step": 1818
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0320204023802779e-05,
+      "loss": 0.2918,
+      "step": 1819
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0314536695947861e-05,
+      "loss": 0.3158,
+      "step": 1820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0308869368092944e-05,
+      "loss": 0.3242,
+      "step": 1821
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.030320204023803e-05,
+      "loss": 0.3409,
+      "step": 1822
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0297534712383112e-05,
+      "loss": 0.3274,
+      "step": 1823
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0291867384528196e-05,
+      "loss": 0.3512,
+      "step": 1824
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0286200056673279e-05,
+      "loss": 0.3405,
+      "step": 1825
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0280532728818363e-05,
+      "loss": 0.3107,
+      "step": 1826
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0274865400963447e-05,
+      "loss": 0.3296,
+      "step": 1827
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0269198073108531e-05,
+      "loss": 0.3406,
+      "step": 1828
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0263530745253614e-05,
+      "loss": 0.3293,
+      "step": 1829
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0257863417398696e-05,
+      "loss": 0.3189,
+      "step": 1830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.025219608954378e-05,
+      "loss": 0.3367,
+      "step": 1831
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0246528761688866e-05,
+      "loss": 0.338,
+      "step": 1832
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0240861433833949e-05,
+      "loss": 0.3315,
+      "step": 1833
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0235194105979031e-05,
+      "loss": 0.32,
+      "step": 1834
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0229526778124115e-05,
+      "loss": 0.3579,
+      "step": 1835
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0223859450269198e-05,
+      "loss": 0.3421,
+      "step": 1836
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0218192122414284e-05,
+      "loss": 0.3399,
+      "step": 1837
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0212524794559366e-05,
+      "loss": 0.3344,
+      "step": 1838
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.020685746670445e-05,
+      "loss": 0.2972,
+      "step": 1839
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0201190138849533e-05,
+      "loss": 0.3327,
+      "step": 1840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0195522810994615e-05,
+      "loss": 0.3357,
+      "step": 1841
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0189855483139701e-05,
+      "loss": 0.3213,
+      "step": 1842
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0184188155284784e-05,
+      "loss": 0.3091,
+      "step": 1843
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0178520827429868e-05,
+      "loss": 0.3694,
+      "step": 1844
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.017285349957495e-05,
+      "loss": 0.3193,
+      "step": 1845
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0167186171720034e-05,
+      "loss": 0.309,
+      "step": 1846
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0161518843865119e-05,
+      "loss": 0.3313,
+      "step": 1847
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0155851516010203e-05,
+      "loss": 0.3328,
+      "step": 1848
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0150184188155285e-05,
+      "loss": 0.3905,
+      "step": 1849
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0144516860300368e-05,
+      "loss": 0.3127,
+      "step": 1850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0138849532445454e-05,
+      "loss": 0.3425,
+      "step": 1851
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0133182204590538e-05,
+      "loss": 0.3278,
+      "step": 1852
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.012751487673562e-05,
+      "loss": 0.3336,
+      "step": 1853
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0121847548880703e-05,
+      "loss": 0.3398,
+      "step": 1854
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0116180221025787e-05,
+      "loss": 0.3104,
+      "step": 1855
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0110512893170871e-05,
+      "loss": 0.3226,
+      "step": 1856
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0104845565315955e-05,
+      "loss": 0.3321,
+      "step": 1857
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0099178237461038e-05,
+      "loss": 0.3681,
+      "step": 1858
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0093510909606122e-05,
+      "loss": 0.3496,
+      "step": 1859
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0087843581751205e-05,
+      "loss": 0.3097,
+      "step": 1860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.008217625389629e-05,
+      "loss": 0.3433,
+      "step": 1861
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0076508926041373e-05,
+      "loss": 0.3363,
+      "step": 1862
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0070841598186455e-05,
+      "loss": 0.3126,
+      "step": 1863
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006517427033154e-05,
+      "loss": 0.315,
+      "step": 1864
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0059506942476622e-05,
+      "loss": 0.3079,
+      "step": 1865
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0053839614621708e-05,
+      "loss": 0.3155,
+      "step": 1866
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004817228676679e-05,
+      "loss": 0.3373,
+      "step": 1867
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0042504958911875e-05,
+      "loss": 0.3333,
+      "step": 1868
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0036837631056957e-05,
+      "loss": 0.3399,
+      "step": 1869
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003117030320204e-05,
+      "loss": 0.3603,
+      "step": 1870
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0025502975347125e-05,
+      "loss": 0.3055,
+      "step": 1871
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0019835647492208e-05,
+      "loss": 0.3274,
+      "step": 1872
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0014168319637292e-05,
+      "loss": 0.343,
+      "step": 1873
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0008500991782375e-05,
+      "loss": 0.3372,
+      "step": 1874
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0002833663927459e-05,
+      "loss": 0.3403,
+      "step": 1875
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.997166336072543e-06,
+      "loss": 0.3766,
+      "step": 1876
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.991499008217627e-06,
+      "loss": 0.3421,
+      "step": 1877
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98583168036271e-06,
+      "loss": 0.323,
+      "step": 1878
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980164352507794e-06,
+      "loss": 0.3575,
+      "step": 1879
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.974497024652878e-06,
+      "loss": 0.3182,
+      "step": 1880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96882969679796e-06,
+      "loss": 0.3251,
+      "step": 1881
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.963162368943045e-06,
+      "loss": 0.3342,
+      "step": 1882
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.957495041088127e-06,
+      "loss": 0.3545,
+      "step": 1883
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.951827713233211e-06,
+      "loss": 0.3543,
+      "step": 1884
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946160385378295e-06,
+      "loss": 0.3189,
+      "step": 1885
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940493057523378e-06,
+      "loss": 0.3251,
+      "step": 1886
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.934825729668462e-06,
+      "loss": 0.3388,
+      "step": 1887
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.929158401813546e-06,
+      "loss": 0.3131,
+      "step": 1888
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.92349107395863e-06,
+      "loss": 0.3506,
+      "step": 1889
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.917823746103713e-06,
+      "loss": 0.328,
+      "step": 1890
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.912156418248795e-06,
+      "loss": 0.3098,
+      "step": 1891
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.90648909039388e-06,
+      "loss": 0.314,
+      "step": 1892
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.900821762538964e-06,
+      "loss": 0.3367,
+      "step": 1893
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.895154434684048e-06,
+      "loss": 0.3746,
+      "step": 1894
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88948710682913e-06,
+      "loss": 0.3096,
+      "step": 1895
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.883819778974215e-06,
+      "loss": 0.3502,
+      "step": 1896
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.878152451119299e-06,
+      "loss": 0.3296,
+      "step": 1897
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.872485123264381e-06,
+      "loss": 0.342,
+      "step": 1898
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.866817795409465e-06,
+      "loss": 0.3301,
+      "step": 1899
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.86115046755455e-06,
+      "loss": 0.3241,
+      "step": 1900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.855483139699632e-06,
+      "loss": 0.3734,
+      "step": 1901
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.849815811844716e-06,
+      "loss": 0.2937,
+      "step": 1902
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.844148483989799e-06,
+      "loss": 0.3326,
+      "step": 1903
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.838481156134883e-06,
+      "loss": 0.3253,
+      "step": 1904
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.832813828279967e-06,
+      "loss": 0.352,
+      "step": 1905
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.82714650042505e-06,
+      "loss": 0.3383,
+      "step": 1906
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.821479172570134e-06,
+      "loss": 0.3396,
+      "step": 1907
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.815811844715218e-06,
+      "loss": 0.3573,
+      "step": 1908
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.810144516860302e-06,
+      "loss": 0.3296,
+      "step": 1909
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.804477189005385e-06,
+      "loss": 0.3193,
+      "step": 1910
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.798809861150467e-06,
+      "loss": 0.3057,
+      "step": 1911
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.793142533295551e-06,
+      "loss": 0.3463,
+      "step": 1912
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.787475205440635e-06,
+      "loss": 0.3434,
+      "step": 1913
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.78180787758572e-06,
+      "loss": 0.2955,
+      "step": 1914
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.776140549730802e-06,
+      "loss": 0.3285,
+      "step": 1915
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.770473221875886e-06,
+      "loss": 0.3419,
+      "step": 1916
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.76480589402097e-06,
+      "loss": 0.3054,
+      "step": 1917
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.759138566166053e-06,
+      "loss": 0.3413,
+      "step": 1918
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.753471238311137e-06,
+      "loss": 0.3126,
+      "step": 1919
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.747803910456221e-06,
+      "loss": 0.3377,
+      "step": 1920
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.742136582601306e-06,
+      "loss": 0.3528,
+      "step": 1921
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.736469254746388e-06,
+      "loss": 0.345,
+      "step": 1922
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.73080192689147e-06,
+      "loss": 0.306,
+      "step": 1923
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.725134599036555e-06,
+      "loss": 0.3258,
+      "step": 1924
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.719467271181639e-06,
+      "loss": 0.3196,
+      "step": 1925
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.713799943326723e-06,
+      "loss": 0.3382,
+      "step": 1926
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.708132615471806e-06,
+      "loss": 0.3331,
+      "step": 1927
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.70246528761689e-06,
+      "loss": 0.3728,
+      "step": 1928
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.696797959761974e-06,
+      "loss": 0.3256,
+      "step": 1929
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.691130631907056e-06,
+      "loss": 0.3373,
+      "step": 1930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.68546330405214e-06,
+      "loss": 0.3295,
+      "step": 1931
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.679795976197223e-06,
+      "loss": 0.3237,
+      "step": 1932
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.674128648342307e-06,
+      "loss": 0.3306,
+      "step": 1933
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.668461320487391e-06,
+      "loss": 0.325,
+      "step": 1934
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.662793992632474e-06,
+      "loss": 0.3026,
+      "step": 1935
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.657126664777558e-06,
+      "loss": 0.3488,
+      "step": 1936
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.651459336922642e-06,
+      "loss": 0.3332,
+      "step": 1937
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.645792009067725e-06,
+      "loss": 0.3426,
+      "step": 1938
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.640124681212809e-06,
+      "loss": 0.3357,
+      "step": 1939
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.634457353357893e-06,
+      "loss": 0.3114,
+      "step": 1940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.628790025502977e-06,
+      "loss": 0.3052,
+      "step": 1941
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.62312269764806e-06,
+      "loss": 0.3237,
+      "step": 1942
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.617455369793144e-06,
+      "loss": 0.3099,
+      "step": 1943
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.611788041938226e-06,
+      "loss": 0.3337,
+      "step": 1944
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.60612071408331e-06,
+      "loss": 0.3275,
+      "step": 1945
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.600453386228395e-06,
+      "loss": 0.3306,
+      "step": 1946
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.594786058373477e-06,
+      "loss": 0.3256,
+      "step": 1947
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.589118730518561e-06,
+      "loss": 0.3535,
+      "step": 1948
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.583451402663646e-06,
+      "loss": 0.3358,
+      "step": 1949
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.577784074808728e-06,
+      "loss": 0.3558,
+      "step": 1950
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.572116746953812e-06,
+      "loss": 0.3424,
+      "step": 1951
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.566449419098895e-06,
+      "loss": 0.3422,
+      "step": 1952
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.560782091243979e-06,
+      "loss": 0.332,
+      "step": 1953
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.555114763389063e-06,
+      "loss": 0.3336,
+      "step": 1954
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.549447435534146e-06,
+      "loss": 0.3194,
+      "step": 1955
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.54378010767923e-06,
+      "loss": 0.328,
+      "step": 1956
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.538112779824314e-06,
+      "loss": 0.3292,
+      "step": 1957
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.532445451969398e-06,
+      "loss": 0.3165,
+      "step": 1958
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.52677812411448e-06,
+      "loss": 0.3341,
+      "step": 1959
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.521110796259565e-06,
+      "loss": 0.3184,
+      "step": 1960
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.515443468404649e-06,
+      "loss": 0.335,
+      "step": 1961
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.509776140549731e-06,
+      "loss": 0.3587,
+      "step": 1962
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.504108812694816e-06,
+      "loss": 0.3038,
+      "step": 1963
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.498441484839898e-06,
+      "loss": 0.3258,
+      "step": 1964
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.492774156984982e-06,
+      "loss": 0.3056,
+      "step": 1965
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.487106829130066e-06,
+      "loss": 0.3289,
+      "step": 1966
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.481439501275149e-06,
+      "loss": 0.3435,
+      "step": 1967
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.475772173420233e-06,
+      "loss": 0.3226,
+      "step": 1968
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.470104845565317e-06,
+      "loss": 0.3291,
+      "step": 1969
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.464437517710401e-06,
+      "loss": 0.3186,
+      "step": 1970
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.458770189855484e-06,
+      "loss": 0.3709,
+      "step": 1971
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.453102862000566e-06,
+      "loss": 0.3538,
+      "step": 1972
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.44743553414565e-06,
+      "loss": 0.3445,
+      "step": 1973
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.441768206290735e-06,
+      "loss": 0.321,
+      "step": 1974
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.436100878435819e-06,
+      "loss": 0.3372,
+      "step": 1975
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.430433550580901e-06,
+      "loss": 0.3468,
+      "step": 1976
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.424766222725986e-06,
+      "loss": 0.3385,
+      "step": 1977
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.41909889487107e-06,
+      "loss": 0.3328,
+      "step": 1978
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.413431567016152e-06,
+      "loss": 0.3332,
+      "step": 1979
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.407764239161236e-06,
+      "loss": 0.332,
+      "step": 1980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.40209691130632e-06,
+      "loss": 0.313,
+      "step": 1981
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.396429583451403e-06,
+      "loss": 0.3385,
+      "step": 1982
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.390762255596487e-06,
+      "loss": 0.3407,
+      "step": 1983
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.38509492774157e-06,
+      "loss": 0.3537,
+      "step": 1984
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.379427599886654e-06,
+      "loss": 0.3492,
+      "step": 1985
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.373760272031738e-06,
+      "loss": 0.318,
+      "step": 1986
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.36809294417682e-06,
+      "loss": 0.345,
+      "step": 1987
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.362425616321905e-06,
+      "loss": 0.3331,
+      "step": 1988
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.356758288466989e-06,
+      "loss": 0.3078,
+      "step": 1989
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.351090960612073e-06,
+      "loss": 0.3358,
+      "step": 1990
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.345423632757156e-06,
+      "loss": 0.3075,
+      "step": 1991
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.339756304902238e-06,
+      "loss": 0.344,
+      "step": 1992
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.334088977047322e-06,
+      "loss": 0.3142,
+      "step": 1993
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.328421649192406e-06,
+      "loss": 0.3521,
+      "step": 1994
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.32275432133749e-06,
+      "loss": 0.3224,
+      "step": 1995
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.317086993482573e-06,
+      "loss": 0.3823,
+      "step": 1996
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.311419665627657e-06,
+      "loss": 0.3537,
+      "step": 1997
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.305752337772742e-06,
+      "loss": 0.3243,
+      "step": 1998
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.300085009917824e-06,
+      "loss": 0.3227,
+      "step": 1999
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.294417682062908e-06,
+      "loss": 0.3201,
+      "step": 2000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.288750354207992e-06,
+      "loss": 0.3527,
+      "step": 2001
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.283083026353077e-06,
+      "loss": 0.3047,
+      "step": 2002
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.277415698498159e-06,
+      "loss": 0.3385,
+      "step": 2003
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.271748370643241e-06,
+      "loss": 0.3006,
+      "step": 2004
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.266081042788326e-06,
+      "loss": 0.3521,
+      "step": 2005
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.26041371493341e-06,
+      "loss": 0.3544,
+      "step": 2006
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.254746387078494e-06,
+      "loss": 0.3335,
+      "step": 2007
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.249079059223577e-06,
+      "loss": 0.3187,
+      "step": 2008
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.24341173136866e-06,
+      "loss": 0.3367,
+      "step": 2009
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.237744403513745e-06,
+      "loss": 0.3275,
+      "step": 2010
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.232077075658827e-06,
+      "loss": 0.352,
+      "step": 2011
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.226409747803912e-06,
+      "loss": 0.3135,
+      "step": 2012
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.220742419948994e-06,
+      "loss": 0.3408,
+      "step": 2013
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.215075092094078e-06,
+      "loss": 0.3245,
+      "step": 2014
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.209407764239162e-06,
+      "loss": 0.3618,
+      "step": 2015
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.203740436384245e-06,
+      "loss": 0.3257,
+      "step": 2016
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.198073108529329e-06,
+      "loss": 0.3787,
+      "step": 2017
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.192405780674413e-06,
+      "loss": 0.3354,
+      "step": 2018
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.186738452819496e-06,
+      "loss": 0.3397,
+      "step": 2019
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.18107112496458e-06,
+      "loss": 0.3015,
+      "step": 2020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.175403797109664e-06,
+      "loss": 0.3175,
+      "step": 2021
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.169736469254748e-06,
+      "loss": 0.3418,
+      "step": 2022
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.16406914139983e-06,
+      "loss": 0.3028,
+      "step": 2023
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.158401813544915e-06,
+      "loss": 0.362,
+      "step": 2024
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.152734485689997e-06,
+      "loss": 0.3431,
+      "step": 2025
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.147067157835082e-06,
+      "loss": 0.3314,
+      "step": 2026
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.141399829980166e-06,
+      "loss": 0.3151,
+      "step": 2027
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.135732502125248e-06,
+      "loss": 0.3113,
+      "step": 2028
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.130065174270332e-06,
+      "loss": 0.3352,
+      "step": 2029
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.124397846415417e-06,
+      "loss": 0.3195,
+      "step": 2030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.118730518560499e-06,
+      "loss": 0.3364,
+      "step": 2031
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.113063190705583e-06,
+      "loss": 0.3497,
+      "step": 2032
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.107395862850666e-06,
+      "loss": 0.3437,
+      "step": 2033
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.10172853499575e-06,
+      "loss": 0.3622,
+      "step": 2034
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.096061207140834e-06,
+      "loss": 0.3712,
+      "step": 2035
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.090393879285917e-06,
+      "loss": 0.3315,
+      "step": 2036
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.084726551431e-06,
+      "loss": 0.3353,
+      "step": 2037
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.079059223576085e-06,
+      "loss": 0.3353,
+      "step": 2038
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.073391895721169e-06,
+      "loss": 0.3453,
+      "step": 2039
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.067724567866252e-06,
+      "loss": 0.2996,
+      "step": 2040
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.062057240011334e-06,
+      "loss": 0.3423,
+      "step": 2041
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.05638991215642e-06,
+      "loss": 0.3339,
+      "step": 2042
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.050722584301502e-06,
+      "loss": 0.3608,
+      "step": 2043
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.045055256446587e-06,
+      "loss": 0.3563,
+      "step": 2044
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.039387928591669e-06,
+      "loss": 0.3078,
+      "step": 2045
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.033720600736753e-06,
+      "loss": 0.3483,
+      "step": 2046
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.028053272881837e-06,
+      "loss": 0.3256,
+      "step": 2047
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.02238594502692e-06,
+      "loss": 0.3365,
+      "step": 2048
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.016718617172004e-06,
+      "loss": 0.3474,
+      "step": 2049
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.011051289317088e-06,
+      "loss": 0.3258,
+      "step": 2050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.005383961462172e-06,
+      "loss": 0.3267,
+      "step": 2051
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.999716633607255e-06,
+      "loss": 0.3578,
+      "step": 2052
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.994049305752337e-06,
+      "loss": 0.3328,
+      "step": 2053
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.988381977897422e-06,
+      "loss": 0.3501,
+      "step": 2054
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.982714650042506e-06,
+      "loss": 0.3385,
+      "step": 2055
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.97704732218759e-06,
+      "loss": 0.3191,
+      "step": 2056
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.971379994332672e-06,
+      "loss": 0.3216,
+      "step": 2057
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.965712666477757e-06,
+      "loss": 0.3431,
+      "step": 2058
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.96004533862284e-06,
+      "loss": 0.3315,
+      "step": 2059
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.954378010767923e-06,
+      "loss": 0.3158,
+      "step": 2060
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.948710682913007e-06,
+      "loss": 0.2993,
+      "step": 2061
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.943043355058092e-06,
+      "loss": 0.3232,
+      "step": 2062
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.937376027203174e-06,
+      "loss": 0.339,
+      "step": 2063
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.931708699348258e-06,
+      "loss": 0.3198,
+      "step": 2064
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.92604137149334e-06,
+      "loss": 0.3808,
+      "step": 2065
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.920374043638425e-06,
+      "loss": 0.3404,
+      "step": 2066
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.914706715783509e-06,
+      "loss": 0.3297,
+      "step": 2067
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.909039387928592e-06,
+      "loss": 0.3296,
+      "step": 2068
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.903372060073676e-06,
+      "loss": 0.318,
+      "step": 2069
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.89770473221876e-06,
+      "loss": 0.3278,
+      "step": 2070
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.892037404363844e-06,
+      "loss": 0.3406,
+      "step": 2071
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.886370076508927e-06,
+      "loss": 0.3551,
+      "step": 2072
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.880702748654009e-06,
+      "loss": 0.3349,
+      "step": 2073
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.875035420799093e-06,
+      "loss": 0.3511,
+      "step": 2074
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.869368092944177e-06,
+      "loss": 0.3197,
+      "step": 2075
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.863700765089262e-06,
+      "loss": 0.3309,
+      "step": 2076
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.858033437234344e-06,
+      "loss": 0.3251,
+      "step": 2077
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.852366109379428e-06,
+      "loss": 0.2851,
+      "step": 2078
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.846698781524513e-06,
+      "loss": 0.3278,
+      "step": 2079
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.841031453669595e-06,
+      "loss": 0.3252,
+      "step": 2080
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.83536412581468e-06,
+      "loss": 0.3331,
+      "step": 2081
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.829696797959763e-06,
+      "loss": 0.3193,
+      "step": 2082
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.824029470104848e-06,
+      "loss": 0.3633,
+      "step": 2083
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.81836214224993e-06,
+      "loss": 0.3301,
+      "step": 2084
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.812694814395012e-06,
+      "loss": 0.3469,
+      "step": 2085
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.807027486540097e-06,
+      "loss": 0.3278,
+      "step": 2086
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.801360158685181e-06,
+      "loss": 0.3273,
+      "step": 2087
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.795692830830265e-06,
+      "loss": 0.3496,
+      "step": 2088
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.790025502975348e-06,
+      "loss": 0.3464,
+      "step": 2089
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.784358175120432e-06,
+      "loss": 0.3515,
+      "step": 2090
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.778690847265516e-06,
+      "loss": 0.3131,
+      "step": 2091
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.773023519410598e-06,
+      "loss": 0.3637,
+      "step": 2092
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.767356191555683e-06,
+      "loss": 0.3289,
+      "step": 2093
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.761688863700765e-06,
+      "loss": 0.3228,
+      "step": 2094
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.75602153584585e-06,
+      "loss": 0.3285,
+      "step": 2095
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.750354207990933e-06,
+      "loss": 0.3164,
+      "step": 2096
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.744686880136016e-06,
+      "loss": 0.338,
+      "step": 2097
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.7390195522811e-06,
+      "loss": 0.2962,
+      "step": 2098
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.733352224426184e-06,
+      "loss": 0.3284,
+      "step": 2099
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.727684896571267e-06,
+      "loss": 0.322,
+      "step": 2100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.722017568716351e-06,
+      "loss": 0.346,
+      "step": 2101
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.716350240861435e-06,
+      "loss": 0.3169,
+      "step": 2102
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.71068291300652e-06,
+      "loss": 0.3379,
+      "step": 2103
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.705015585151602e-06,
+      "loss": 0.3337,
+      "step": 2104
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.699348257296686e-06,
+      "loss": 0.3371,
+      "step": 2105
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.693680929441768e-06,
+      "loss": 0.3404,
+      "step": 2106
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.688013601586853e-06,
+      "loss": 0.3157,
+      "step": 2107
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.682346273731937e-06,
+      "loss": 0.3566,
+      "step": 2108
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.67667894587702e-06,
+      "loss": 0.3163,
+      "step": 2109
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.671011618022103e-06,
+      "loss": 0.3232,
+      "step": 2110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.665344290167188e-06,
+      "loss": 0.311,
+      "step": 2111
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.65967696231227e-06,
+      "loss": 0.3368,
+      "step": 2112
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.654009634457354e-06,
+      "loss": 0.3034,
+      "step": 2113
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.648342306602437e-06,
+      "loss": 0.3341,
+      "step": 2114
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.642674978747521e-06,
+      "loss": 0.3475,
+      "step": 2115
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.637007650892605e-06,
+      "loss": 0.3536,
+      "step": 2116
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.631340323037688e-06,
+      "loss": 0.331,
+      "step": 2117
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.625672995182772e-06,
+      "loss": 0.3421,
+      "step": 2118
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.620005667327856e-06,
+      "loss": 0.3386,
+      "step": 2119
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.61433833947294e-06,
+      "loss": 0.3469,
+      "step": 2120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.608671011618023e-06,
+      "loss": 0.3478,
+      "step": 2121
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.603003683763105e-06,
+      "loss": 0.3257,
+      "step": 2122
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.597336355908191e-06,
+      "loss": 0.3482,
+      "step": 2123
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.591669028053273e-06,
+      "loss": 0.294,
+      "step": 2124
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.586001700198358e-06,
+      "loss": 0.3382,
+      "step": 2125
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.58033437234344e-06,
+      "loss": 0.3409,
+      "step": 2126
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.574667044488524e-06,
+      "loss": 0.346,
+      "step": 2127
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.568999716633608e-06,
+      "loss": 0.3189,
+      "step": 2128
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.563332388778691e-06,
+      "loss": 0.3304,
+      "step": 2129
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.557665060923775e-06,
+      "loss": 0.318,
+      "step": 2130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.55199773306886e-06,
+      "loss": 0.3043,
+      "step": 2131
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.546330405213943e-06,
+      "loss": 0.337,
+      "step": 2132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.540663077359026e-06,
+      "loss": 0.298,
+      "step": 2133
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.534995749504108e-06,
+      "loss": 0.2875,
+      "step": 2134
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.529328421649193e-06,
+      "loss": 0.3466,
+      "step": 2135
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.523661093794277e-06,
+      "loss": 0.3513,
+      "step": 2136
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.517993765939361e-06,
+      "loss": 0.3467,
+      "step": 2137
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.512326438084443e-06,
+      "loss": 0.3581,
+      "step": 2138
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.506659110229528e-06,
+      "loss": 0.3193,
+      "step": 2139
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.500991782374612e-06,
+      "loss": 0.3438,
+      "step": 2140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.495324454519694e-06,
+      "loss": 0.3026,
+      "step": 2141
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.489657126664778e-06,
+      "loss": 0.3488,
+      "step": 2142
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.483989798809863e-06,
+      "loss": 0.3205,
+      "step": 2143
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.478322470954945e-06,
+      "loss": 0.3745,
+      "step": 2144
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.47265514310003e-06,
+      "loss": 0.3586,
+      "step": 2145
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.466987815245112e-06,
+      "loss": 0.3123,
+      "step": 2146
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.461320487390196e-06,
+      "loss": 0.3234,
+      "step": 2147
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.45565315953528e-06,
+      "loss": 0.3462,
+      "step": 2148
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.449985831680363e-06,
+      "loss": 0.3377,
+      "step": 2149
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.444318503825447e-06,
+      "loss": 0.3191,
+      "step": 2150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.438651175970531e-06,
+      "loss": 0.3271,
+      "step": 2151
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.432983848115615e-06,
+      "loss": 0.3225,
+      "step": 2152
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.427316520260698e-06,
+      "loss": 0.3187,
+      "step": 2153
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.42164919240578e-06,
+      "loss": 0.3261,
+      "step": 2154
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.415981864550864e-06,
+      "loss": 0.3612,
+      "step": 2155
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.410314536695948e-06,
+      "loss": 0.343,
+      "step": 2156
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.404647208841033e-06,
+      "loss": 0.3116,
+      "step": 2157
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.398979880986115e-06,
+      "loss": 0.3371,
+      "step": 2158
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.3933125531312e-06,
+      "loss": 0.3101,
+      "step": 2159
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.387645225276284e-06,
+      "loss": 0.3359,
+      "step": 2160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.381977897421366e-06,
+      "loss": 0.3223,
+      "step": 2161
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.37631056956645e-06,
+      "loss": 0.3242,
+      "step": 2162
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.370643241711534e-06,
+      "loss": 0.3424,
+      "step": 2163
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.364975913856619e-06,
+      "loss": 0.3301,
+      "step": 2164
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.359308586001701e-06,
+      "loss": 0.3161,
+      "step": 2165
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.353641258146784e-06,
+      "loss": 0.3319,
+      "step": 2166
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.347973930291868e-06,
+      "loss": 0.3559,
+      "step": 2167
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.342306602436952e-06,
+      "loss": 0.3042,
+      "step": 2168
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.336639274582036e-06,
+      "loss": 0.3344,
+      "step": 2169
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.330971946727119e-06,
+      "loss": 0.3229,
+      "step": 2170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.325304618872203e-06,
+      "loss": 0.308,
+      "step": 2171
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.319637291017287e-06,
+      "loss": 0.3413,
+      "step": 2172
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.31396996316237e-06,
+      "loss": 0.3431,
+      "step": 2173
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.308302635307454e-06,
+      "loss": 0.3377,
+      "step": 2174
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.302635307452536e-06,
+      "loss": 0.3183,
+      "step": 2175
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.29696797959762e-06,
+      "loss": 0.3252,
+      "step": 2176
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.291300651742704e-06,
+      "loss": 0.3171,
+      "step": 2177
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.285633323887787e-06,
+      "loss": 0.3139,
+      "step": 2178
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.279965996032871e-06,
+      "loss": 0.3192,
+      "step": 2179
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.274298668177955e-06,
+      "loss": 0.3426,
+      "step": 2180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.268631340323038e-06,
+      "loss": 0.3382,
+      "step": 2181
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.262964012468122e-06,
+      "loss": 0.3425,
+      "step": 2182
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.257296684613204e-06,
+      "loss": 0.3346,
+      "step": 2183
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.25162935675829e-06,
+      "loss": 0.3235,
+      "step": 2184
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.245962028903373e-06,
+      "loss": 0.3209,
+      "step": 2185
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.240294701048457e-06,
+      "loss": 0.3288,
+      "step": 2186
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.23462737319354e-06,
+      "loss": 0.3048,
+      "step": 2187
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.228960045338624e-06,
+      "loss": 0.3559,
+      "step": 2188
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.223292717483708e-06,
+      "loss": 0.3523,
+      "step": 2189
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.21762538962879e-06,
+      "loss": 0.325,
+      "step": 2190
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.211958061773874e-06,
+      "loss": 0.3163,
+      "step": 2191
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.206290733918959e-06,
+      "loss": 0.3229,
+      "step": 2192
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.200623406064041e-06,
+      "loss": 0.3148,
+      "step": 2193
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.194956078209125e-06,
+      "loss": 0.3458,
+      "step": 2194
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.189288750354208e-06,
+      "loss": 0.3168,
+      "step": 2195
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.183621422499292e-06,
+      "loss": 0.3235,
+      "step": 2196
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.177954094644376e-06,
+      "loss": 0.3231,
+      "step": 2197
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.172286766789459e-06,
+      "loss": 0.3312,
+      "step": 2198
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.166619438934543e-06,
+      "loss": 0.3524,
+      "step": 2199
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.160952111079627e-06,
+      "loss": 0.316,
+      "step": 2200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.155284783224711e-06,
+      "loss": 0.2921,
+      "step": 2201
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.149617455369794e-06,
+      "loss": 0.3347,
+      "step": 2202
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.143950127514876e-06,
+      "loss": 0.3405,
+      "step": 2203
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.138282799659962e-06,
+      "loss": 0.3412,
+      "step": 2204
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.132615471805044e-06,
+      "loss": 0.3324,
+      "step": 2205
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.126948143950129e-06,
+      "loss": 0.3224,
+      "step": 2206
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.121280816095211e-06,
+      "loss": 0.2976,
+      "step": 2207
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.115613488240295e-06,
+      "loss": 0.2989,
+      "step": 2208
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.10994616038538e-06,
+      "loss": 0.2943,
+      "step": 2209
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.104278832530462e-06,
+      "loss": 0.3136,
+      "step": 2210
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.098611504675546e-06,
+      "loss": 0.3399,
+      "step": 2211
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.09294417682063e-06,
+      "loss": 0.3247,
+      "step": 2212
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.087276848965714e-06,
+      "loss": 0.3347,
+      "step": 2213
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.081609521110797e-06,
+      "loss": 0.3626,
+      "step": 2214
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.07594219325588e-06,
+      "loss": 0.3472,
+      "step": 2215
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.070274865400964e-06,
+      "loss": 0.3174,
+      "step": 2216
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.064607537546048e-06,
+      "loss": 0.3576,
+      "step": 2217
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.058940209691132e-06,
+      "loss": 0.3307,
+      "step": 2218
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.053272881836214e-06,
+      "loss": 0.3362,
+      "step": 2219
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.047605553981299e-06,
+      "loss": 0.327,
+      "step": 2220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.041938226126383e-06,
+      "loss": 0.307,
+      "step": 2221
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.036270898271465e-06,
+      "loss": 0.3396,
+      "step": 2222
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.03060357041655e-06,
+      "loss": 0.3357,
+      "step": 2223
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.024936242561634e-06,
+      "loss": 0.3122,
+      "step": 2224
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.019268914706716e-06,
+      "loss": 0.3371,
+      "step": 2225
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.0136015868518e-06,
+      "loss": 0.3283,
+      "step": 2226
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.007934258996883e-06,
+      "loss": 0.3427,
+      "step": 2227
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.002266931141967e-06,
+      "loss": 0.2895,
+      "step": 2228
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.996599603287051e-06,
+      "loss": 0.3339,
+      "step": 2229
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.990932275432134e-06,
+      "loss": 0.3276,
+      "step": 2230
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.985264947577218e-06,
+      "loss": 0.3276,
+      "step": 2231
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.979597619722302e-06,
+      "loss": 0.3463,
+      "step": 2232
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.973930291867386e-06,
+      "loss": 0.3322,
+      "step": 2233
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.968262964012469e-06,
+      "loss": 0.321,
+      "step": 2234
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.962595636157551e-06,
+      "loss": 0.3224,
+      "step": 2235
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.956928308302635e-06,
+      "loss": 0.3225,
+      "step": 2236
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.95126098044772e-06,
+      "loss": 0.3426,
+      "step": 2237
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.945593652592804e-06,
+      "loss": 0.3057,
+      "step": 2238
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.939926324737886e-06,
+      "loss": 0.3543,
+      "step": 2239
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.93425899688297e-06,
+      "loss": 0.3457,
+      "step": 2240
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.928591669028055e-06,
+      "loss": 0.3231,
+      "step": 2241
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.922924341173137e-06,
+      "loss": 0.3014,
+      "step": 2242
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.917257013318221e-06,
+      "loss": 0.3311,
+      "step": 2243
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.911589685463305e-06,
+      "loss": 0.3443,
+      "step": 2244
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.90592235760839e-06,
+      "loss": 0.319,
+      "step": 2245
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.900255029753472e-06,
+      "loss": 0.3346,
+      "step": 2246
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.894587701898555e-06,
+      "loss": 0.3121,
+      "step": 2247
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.888920374043639e-06,
+      "loss": 0.3198,
+      "step": 2248
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.883253046188723e-06,
+      "loss": 0.3166,
+      "step": 2249
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.877585718333807e-06,
+      "loss": 0.3027,
+      "step": 2250
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.87191839047889e-06,
+      "loss": 0.34,
+      "step": 2251
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.866251062623974e-06,
+      "loss": 0.3139,
+      "step": 2252
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.860583734769058e-06,
+      "loss": 0.3377,
+      "step": 2253
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.85491640691414e-06,
+      "loss": 0.2972,
+      "step": 2254
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.849249079059225e-06,
+      "loss": 0.3469,
+      "step": 2255
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.843581751204307e-06,
+      "loss": 0.3305,
+      "step": 2256
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.837914423349391e-06,
+      "loss": 0.3451,
+      "step": 2257
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.832247095494475e-06,
+      "loss": 0.3658,
+      "step": 2258
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.826579767639558e-06,
+      "loss": 0.3028,
+      "step": 2259
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.820912439784642e-06,
+      "loss": 0.3255,
+      "step": 2260
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.815245111929726e-06,
+      "loss": 0.3308,
+      "step": 2261
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.809577784074809e-06,
+      "loss": 0.3304,
+      "step": 2262
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.803910456219893e-06,
+      "loss": 0.3298,
+      "step": 2263
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.798243128364975e-06,
+      "loss": 0.352,
+      "step": 2264
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.792575800510061e-06,
+      "loss": 0.3385,
+      "step": 2265
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.786908472655144e-06,
+      "loss": 0.3483,
+      "step": 2266
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.781241144800228e-06,
+      "loss": 0.3442,
+      "step": 2267
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.77557381694531e-06,
+      "loss": 0.3237,
+      "step": 2268
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.769906489090395e-06,
+      "loss": 0.3372,
+      "step": 2269
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.764239161235479e-06,
+      "loss": 0.3494,
+      "step": 2270
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.758571833380561e-06,
+      "loss": 0.3083,
+      "step": 2271
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.752904505525645e-06,
+      "loss": 0.3367,
+      "step": 2272
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.74723717767073e-06,
+      "loss": 0.3259,
+      "step": 2273
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.741569849815812e-06,
+      "loss": 0.3307,
+      "step": 2274
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.735902521960896e-06,
+      "loss": 0.353,
+      "step": 2275
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.730235194105979e-06,
+      "loss": 0.3252,
+      "step": 2276
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.724567866251063e-06,
+      "loss": 0.3298,
+      "step": 2277
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.718900538396147e-06,
+      "loss": 0.317,
+      "step": 2278
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.71323321054123e-06,
+      "loss": 0.3228,
+      "step": 2279
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.707565882686314e-06,
+      "loss": 0.3336,
+      "step": 2280
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.701898554831398e-06,
+      "loss": 0.3148,
+      "step": 2281
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.696231226976482e-06,
+      "loss": 0.347,
+      "step": 2282
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.690563899121565e-06,
+      "loss": 0.3088,
+      "step": 2283
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.684896571266647e-06,
+      "loss": 0.3169,
+      "step": 2284
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.679229243411733e-06,
+      "loss": 0.3322,
+      "step": 2285
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.673561915556815e-06,
+      "loss": 0.3258,
+      "step": 2286
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.6678945877019e-06,
+      "loss": 0.2803,
+      "step": 2287
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.662227259846982e-06,
+      "loss": 0.3555,
+      "step": 2288
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.656559931992066e-06,
+      "loss": 0.3228,
+      "step": 2289
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.65089260413715e-06,
+      "loss": 0.3137,
+      "step": 2290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.645225276282233e-06,
+      "loss": 0.3164,
+      "step": 2291
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.639557948427317e-06,
+      "loss": 0.3128,
+      "step": 2292
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.633890620572401e-06,
+      "loss": 0.3071,
+      "step": 2293
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.628223292717485e-06,
+      "loss": 0.3391,
+      "step": 2294
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.622555964862568e-06,
+      "loss": 0.3324,
+      "step": 2295
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.616888637007651e-06,
+      "loss": 0.2929,
+      "step": 2296
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.6112213091527355e-06,
+      "loss": 0.3105,
+      "step": 2297
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.605553981297819e-06,
+      "loss": 0.3104,
+      "step": 2298
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.599886653442903e-06,
+      "loss": 0.3481,
+      "step": 2299
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.594219325587986e-06,
+      "loss": 0.3194,
+      "step": 2300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.588551997733069e-06,
+      "loss": 0.344,
+      "step": 2301
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.582884669878153e-06,
+      "loss": 0.3131,
+      "step": 2302
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.577217342023236e-06,
+      "loss": 0.3437,
+      "step": 2303
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.5715500141683205e-06,
+      "loss": 0.3612,
+      "step": 2304
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.565882686313404e-06,
+      "loss": 0.3378,
+      "step": 2305
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.560215358458487e-06,
+      "loss": 0.2831,
+      "step": 2306
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.554548030603571e-06,
+      "loss": 0.3371,
+      "step": 2307
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.548880702748655e-06,
+      "loss": 0.3117,
+      "step": 2308
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.543213374893739e-06,
+      "loss": 0.3013,
+      "step": 2309
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.537546047038822e-06,
+      "loss": 0.3259,
+      "step": 2310
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.531878719183905e-06,
+      "loss": 0.3216,
+      "step": 2311
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.526211391328989e-06,
+      "loss": 0.3494,
+      "step": 2312
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.520544063474072e-06,
+      "loss": 0.3117,
+      "step": 2313
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.514876735619156e-06,
+      "loss": 0.31,
+      "step": 2314
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.50920940776424e-06,
+      "loss": 0.313,
+      "step": 2315
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.503542079909323e-06,
+      "loss": 0.3419,
+      "step": 2316
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.497874752054407e-06,
+      "loss": 0.3165,
+      "step": 2317
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.4922074241994905e-06,
+      "loss": 0.3323,
+      "step": 2318
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.486540096344575e-06,
+      "loss": 0.3275,
+      "step": 2319
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.480872768489658e-06,
+      "loss": 0.3253,
+      "step": 2320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.475205440634742e-06,
+      "loss": 0.3228,
+      "step": 2321
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.469538112779825e-06,
+      "loss": 0.3232,
+      "step": 2322
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.463870784924908e-06,
+      "loss": 0.3266,
+      "step": 2323
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.458203457069992e-06,
+      "loss": 0.3401,
+      "step": 2324
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.4525361292150755e-06,
+      "loss": 0.3214,
+      "step": 2325
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.44686880136016e-06,
+      "loss": 0.3232,
+      "step": 2326
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.441201473505243e-06,
+      "loss": 0.3223,
+      "step": 2327
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.435534145650326e-06,
+      "loss": 0.3392,
+      "step": 2328
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.4298668177954105e-06,
+      "loss": 0.341,
+      "step": 2329
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.424199489940493e-06,
+      "loss": 0.3367,
+      "step": 2330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.418532162085578e-06,
+      "loss": 0.2957,
+      "step": 2331
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.4128648342306605e-06,
+      "loss": 0.3474,
+      "step": 2332
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.407197506375744e-06,
+      "loss": 0.3249,
+      "step": 2333
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.401530178520828e-06,
+      "loss": 0.3192,
+      "step": 2334
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.395862850665911e-06,
+      "loss": 0.2981,
+      "step": 2335
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3901955228109956e-06,
+      "loss": 0.3489,
+      "step": 2336
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.384528194956079e-06,
+      "loss": 0.3332,
+      "step": 2337
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.378860867101162e-06,
+      "loss": 0.321,
+      "step": 2338
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.373193539246246e-06,
+      "loss": 0.3234,
+      "step": 2339
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.367526211391329e-06,
+      "loss": 0.3273,
+      "step": 2340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.361858883536414e-06,
+      "loss": 0.3245,
+      "step": 2341
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.356191555681496e-06,
+      "loss": 0.3149,
+      "step": 2342
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.35052422782658e-06,
+      "loss": 0.3144,
+      "step": 2343
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.344856899971664e-06,
+      "loss": 0.3532,
+      "step": 2344
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.339189572116747e-06,
+      "loss": 0.3642,
+      "step": 2345
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.333522244261831e-06,
+      "loss": 0.3226,
+      "step": 2346
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.327854916406915e-06,
+      "loss": 0.3238,
+      "step": 2347
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.322187588551999e-06,
+      "loss": 0.3105,
+      "step": 2348
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.316520260697082e-06,
+      "loss": 0.317,
+      "step": 2349
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.310852932842165e-06,
+      "loss": 0.3214,
+      "step": 2350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.30518560498725e-06,
+      "loss": 0.3365,
+      "step": 2351
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.299518277132332e-06,
+      "loss": 0.3483,
+      "step": 2352
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.293850949277416e-06,
+      "loss": 0.3321,
+      "step": 2353
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.2881836214225e-06,
+      "loss": 0.3414,
+      "step": 2354
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.282516293567583e-06,
+      "loss": 0.3028,
+      "step": 2355
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.276848965712667e-06,
+      "loss": 0.3235,
+      "step": 2356
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.271181637857751e-06,
+      "loss": 0.3093,
+      "step": 2357
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.265514310002835e-06,
+      "loss": 0.2966,
+      "step": 2358
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.259846982147918e-06,
+      "loss": 0.3136,
+      "step": 2359
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.254179654293001e-06,
+      "loss": 0.3355,
+      "step": 2360
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.248512326438086e-06,
+      "loss": 0.3176,
+      "step": 2361
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.242844998583168e-06,
+      "loss": 0.3415,
+      "step": 2362
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.237177670728252e-06,
+      "loss": 0.328,
+      "step": 2363
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.231510342873336e-06,
+      "loss": 0.3136,
+      "step": 2364
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.225843015018419e-06,
+      "loss": 0.3402,
+      "step": 2365
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.220175687163503e-06,
+      "loss": 0.3226,
+      "step": 2366
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.2145083593085864e-06,
+      "loss": 0.3326,
+      "step": 2367
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.208841031453671e-06,
+      "loss": 0.3287,
+      "step": 2368
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.203173703598754e-06,
+      "loss": 0.3117,
+      "step": 2369
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.1975063757438364e-06,
+      "loss": 0.2966,
+      "step": 2370
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.1918390478889215e-06,
+      "loss": 0.3328,
+      "step": 2371
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.186171720034004e-06,
+      "loss": 0.2938,
+      "step": 2372
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.180504392179088e-06,
+      "loss": 0.3303,
+      "step": 2373
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1748370643241715e-06,
+      "loss": 0.3397,
+      "step": 2374
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.169169736469256e-06,
+      "loss": 0.3223,
+      "step": 2375
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.163502408614339e-06,
+      "loss": 0.3312,
+      "step": 2376
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.157835080759422e-06,
+      "loss": 0.3276,
+      "step": 2377
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1521677529045065e-06,
+      "loss": 0.3152,
+      "step": 2378
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.14650042504959e-06,
+      "loss": 0.2909,
+      "step": 2379
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.140833097194674e-06,
+      "loss": 0.3245,
+      "step": 2380
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.135165769339757e-06,
+      "loss": 0.3251,
+      "step": 2381
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.12949844148484e-06,
+      "loss": 0.3142,
+      "step": 2382
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.123831113629924e-06,
+      "loss": 0.3407,
+      "step": 2383
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.118163785775007e-06,
+      "loss": 0.3136,
+      "step": 2384
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.1124964579200915e-06,
+      "loss": 0.3314,
+      "step": 2385
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.106829130065175e-06,
+      "loss": 0.2896,
+      "step": 2386
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.101161802210258e-06,
+      "loss": 0.3546,
+      "step": 2387
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.095494474355342e-06,
+      "loss": 0.3369,
+      "step": 2388
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.089827146500426e-06,
+      "loss": 0.3158,
+      "step": 2389
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.08415981864551e-06,
+      "loss": 0.3095,
+      "step": 2390
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.078492490790593e-06,
+      "loss": 0.3149,
+      "step": 2391
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.072825162935676e-06,
+      "loss": 0.3177,
+      "step": 2392
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.06715783508076e-06,
+      "loss": 0.2859,
+      "step": 2393
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.061490507225843e-06,
+      "loss": 0.3211,
+      "step": 2394
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.055823179370927e-06,
+      "loss": 0.3283,
+      "step": 2395
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.050155851516011e-06,
+      "loss": 0.3204,
+      "step": 2396
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.044488523661095e-06,
+      "loss": 0.3046,
+      "step": 2397
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.038821195806178e-06,
+      "loss": 0.3328,
+      "step": 2398
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.0331538679512615e-06,
+      "loss": 0.3432,
+      "step": 2399
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.027486540096346e-06,
+      "loss": 0.3202,
+      "step": 2400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.021819212241428e-06,
+      "loss": 0.3596,
+      "step": 2401
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.016151884386513e-06,
+      "loss": 0.3016,
+      "step": 2402
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.010484556531596e-06,
+      "loss": 0.3076,
+      "step": 2403
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.004817228676679e-06,
+      "loss": 0.3191,
+      "step": 2404
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.999149900821763e-06,
+      "loss": 0.3025,
+      "step": 2405
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.9934825729668465e-06,
+      "loss": 0.329,
+      "step": 2406
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.987815245111931e-06,
+      "loss": 0.3376,
+      "step": 2407
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.982147917257014e-06,
+      "loss": 0.3205,
+      "step": 2408
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.976480589402097e-06,
+      "loss": 0.3326,
+      "step": 2409
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9708132615471815e-06,
+      "loss": 0.3175,
+      "step": 2410
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.965145933692264e-06,
+      "loss": 0.3203,
+      "step": 2411
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.959478605837349e-06,
+      "loss": 0.3212,
+      "step": 2412
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9538112779824315e-06,
+      "loss": 0.3114,
+      "step": 2413
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.948143950127515e-06,
+      "loss": 0.3388,
+      "step": 2414
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.942476622272599e-06,
+      "loss": 0.2873,
+      "step": 2415
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.936809294417682e-06,
+      "loss": 0.3582,
+      "step": 2416
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9311419665627666e-06,
+      "loss": 0.3093,
+      "step": 2417
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.92547463870785e-06,
+      "loss": 0.3194,
+      "step": 2418
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.919807310852933e-06,
+      "loss": 0.3129,
+      "step": 2419
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.914139982998017e-06,
+      "loss": 0.3222,
+      "step": 2420
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.9084726551431e-06,
+      "loss": 0.3087,
+      "step": 2421
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.902805327288185e-06,
+      "loss": 0.3229,
+      "step": 2422
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.897137999433267e-06,
+      "loss": 0.3392,
+      "step": 2423
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.8914706715783516e-06,
+      "loss": 0.3425,
+      "step": 2424
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.885803343723435e-06,
+      "loss": 0.2918,
+      "step": 2425
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.880136015868518e-06,
+      "loss": 0.334,
+      "step": 2426
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.874468688013602e-06,
+      "loss": 0.288,
+      "step": 2427
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.868801360158686e-06,
+      "loss": 0.1311,
+      "step": 2428
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.86313403230377e-06,
+      "loss": 0.1189,
+      "step": 2429
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.857466704448853e-06,
+      "loss": 0.1179,
+      "step": 2430
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.851799376593936e-06,
+      "loss": 0.1287,
+      "step": 2431
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.846132048739021e-06,
+      "loss": 0.1184,
+      "step": 2432
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.840464720884103e-06,
+      "loss": 0.1179,
+      "step": 2433
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.8347973930291874e-06,
+      "loss": 0.0959,
+      "step": 2434
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.829130065174271e-06,
+      "loss": 0.0945,
+      "step": 2435
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.823462737319354e-06,
+      "loss": 0.1016,
+      "step": 2436
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.817795409464438e-06,
+      "loss": 0.1005,
+      "step": 2437
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.812128081609522e-06,
+      "loss": 0.0948,
+      "step": 2438
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.806460753754606e-06,
+      "loss": 0.0828,
+      "step": 2439
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.800793425899689e-06,
+      "loss": 0.0966,
+      "step": 2440
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.795126098044772e-06,
+      "loss": 0.0977,
+      "step": 2441
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.789458770189857e-06,
+      "loss": 0.1107,
+      "step": 2442
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.783791442334939e-06,
+      "loss": 0.1038,
+      "step": 2443
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.778124114480023e-06,
+      "loss": 0.1126,
+      "step": 2444
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.772456786625107e-06,
+      "loss": 0.1019,
+      "step": 2445
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.76678945877019e-06,
+      "loss": 0.1085,
+      "step": 2446
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.761122130915274e-06,
+      "loss": 0.104,
+      "step": 2447
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7554548030603575e-06,
+      "loss": 0.0911,
+      "step": 2448
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.749787475205442e-06,
+      "loss": 0.1125,
+      "step": 2449
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.744120147350525e-06,
+      "loss": 0.1074,
+      "step": 2450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.738452819495609e-06,
+      "loss": 0.0977,
+      "step": 2451
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7327854916406925e-06,
+      "loss": 0.0986,
+      "step": 2452
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.727118163785775e-06,
+      "loss": 0.0954,
+      "step": 2453
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.721450835930859e-06,
+      "loss": 0.1038,
+      "step": 2454
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7157835080759425e-06,
+      "loss": 0.1035,
+      "step": 2455
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.710116180221027e-06,
+      "loss": 0.0933,
+      "step": 2456
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.70444885236611e-06,
+      "loss": 0.0901,
+      "step": 2457
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.698781524511193e-06,
+      "loss": 0.0864,
+      "step": 2458
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6931141966562775e-06,
+      "loss": 0.096,
+      "step": 2459
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.687446868801361e-06,
+      "loss": 0.0957,
+      "step": 2460
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.681779540946445e-06,
+      "loss": 0.0996,
+      "step": 2461
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.676112213091528e-06,
+      "loss": 0.0967,
+      "step": 2462
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.670444885236611e-06,
+      "loss": 0.0898,
+      "step": 2463
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.664777557381695e-06,
+      "loss": 0.1032,
+      "step": 2464
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.659110229526778e-06,
+      "loss": 0.0926,
+      "step": 2465
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6534429016718625e-06,
+      "loss": 0.0936,
+      "step": 2466
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.647775573816946e-06,
+      "loss": 0.0996,
+      "step": 2467
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.642108245962029e-06,
+      "loss": 0.0901,
+      "step": 2468
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.636440918107113e-06,
+      "loss": 0.1154,
+      "step": 2469
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.630773590252197e-06,
+      "loss": 0.1075,
+      "step": 2470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.625106262397281e-06,
+      "loss": 0.0952,
+      "step": 2471
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.619438934542363e-06,
+      "loss": 0.0986,
+      "step": 2472
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.613771606687447e-06,
+      "loss": 0.0932,
+      "step": 2473
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.608104278832531e-06,
+      "loss": 0.0967,
+      "step": 2474
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.602436950977614e-06,
+      "loss": 0.0857,
+      "step": 2475
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.596769623122698e-06,
+      "loss": 0.0965,
+      "step": 2476
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.591102295267782e-06,
+      "loss": 0.0821,
+      "step": 2477
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.585434967412866e-06,
+      "loss": 0.0883,
+      "step": 2478
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.579767639557949e-06,
+      "loss": 0.1053,
+      "step": 2479
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.5741003117030325e-06,
+      "loss": 0.0917,
+      "step": 2480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.568432983848117e-06,
+      "loss": 0.095,
+      "step": 2481
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.562765655993199e-06,
+      "loss": 0.0982,
+      "step": 2482
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.557098328138284e-06,
+      "loss": 0.1028,
+      "step": 2483
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.551431000283367e-06,
+      "loss": 0.1089,
+      "step": 2484
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.54576367242845e-06,
+      "loss": 0.0803,
+      "step": 2485
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.540096344573534e-06,
+      "loss": 0.1145,
+      "step": 2486
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5344290167186175e-06,
+      "loss": 0.1117,
+      "step": 2487
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.528761688863702e-06,
+      "loss": 0.096,
+      "step": 2488
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.523094361008785e-06,
+      "loss": 0.1006,
+      "step": 2489
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.517427033153868e-06,
+      "loss": 0.0976,
+      "step": 2490
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5117597052989526e-06,
+      "loss": 0.0914,
+      "step": 2491
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.506092377444035e-06,
+      "loss": 0.1082,
+      "step": 2492
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.50042504958912e-06,
+      "loss": 0.0997,
+      "step": 2493
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4947577217342025e-06,
+      "loss": 0.0889,
+      "step": 2494
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.489090393879286e-06,
+      "loss": 0.091,
+      "step": 2495
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.48342306602437e-06,
+      "loss": 0.1036,
+      "step": 2496
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.477755738169453e-06,
+      "loss": 0.0911,
+      "step": 2497
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4720884103145376e-06,
+      "loss": 0.0883,
+      "step": 2498
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.466421082459621e-06,
+      "loss": 0.093,
+      "step": 2499
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.460753754604704e-06,
+      "loss": 0.0869,
+      "step": 2500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.455086426749788e-06,
+      "loss": 0.0821,
+      "step": 2501
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.449419098894871e-06,
+      "loss": 0.0943,
+      "step": 2502
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.443751771039956e-06,
+      "loss": 0.0957,
+      "step": 2503
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.438084443185038e-06,
+      "loss": 0.0885,
+      "step": 2504
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.432417115330123e-06,
+      "loss": 0.0852,
+      "step": 2505
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.426749787475206e-06,
+      "loss": 0.0796,
+      "step": 2506
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.421082459620289e-06,
+      "loss": 0.0811,
+      "step": 2507
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.415415131765373e-06,
+      "loss": 0.0921,
+      "step": 2508
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.409747803910457e-06,
+      "loss": 0.0997,
+      "step": 2509
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.404080476055541e-06,
+      "loss": 0.1184,
+      "step": 2510
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.398413148200624e-06,
+      "loss": 0.0853,
+      "step": 2511
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.392745820345707e-06,
+      "loss": 0.0813,
+      "step": 2512
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.387078492490792e-06,
+      "loss": 0.1007,
+      "step": 2513
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.381411164635874e-06,
+      "loss": 0.1042,
+      "step": 2514
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.3757438367809584e-06,
+      "loss": 0.1019,
+      "step": 2515
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.370076508926042e-06,
+      "loss": 0.0949,
+      "step": 2516
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.364409181071125e-06,
+      "loss": 0.0976,
+      "step": 2517
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.358741853216209e-06,
+      "loss": 0.0995,
+      "step": 2518
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.353074525361293e-06,
+      "loss": 0.0955,
+      "step": 2519
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.347407197506377e-06,
+      "loss": 0.0955,
+      "step": 2520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.34173986965146e-06,
+      "loss": 0.0926,
+      "step": 2521
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.336072541796543e-06,
+      "loss": 0.0951,
+      "step": 2522
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.330405213941628e-06,
+      "loss": 0.0844,
+      "step": 2523
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.32473788608671e-06,
+      "loss": 0.1072,
+      "step": 2524
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.319070558231794e-06,
+      "loss": 0.091,
+      "step": 2525
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.313403230376878e-06,
+      "loss": 0.1049,
+      "step": 2526
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.307735902521961e-06,
+      "loss": 0.0984,
+      "step": 2527
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.302068574667045e-06,
+      "loss": 0.0979,
+      "step": 2528
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.2964012468121285e-06,
+      "loss": 0.0956,
+      "step": 2529
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.290733918957213e-06,
+      "loss": 0.0899,
+      "step": 2530
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.285066591102296e-06,
+      "loss": 0.0925,
+      "step": 2531
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.27939926324738e-06,
+      "loss": 0.1073,
+      "step": 2532
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2737319353924635e-06,
+      "loss": 0.0943,
+      "step": 2533
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.268064607537546e-06,
+      "loss": 0.0942,
+      "step": 2534
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.26239727968263e-06,
+      "loss": 0.0889,
+      "step": 2535
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2567299518277135e-06,
+      "loss": 0.0888,
+      "step": 2536
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.251062623972798e-06,
+      "loss": 0.0891,
+      "step": 2537
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.245395296117881e-06,
+      "loss": 0.0919,
+      "step": 2538
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.239727968262964e-06,
+      "loss": 0.0808,
+      "step": 2539
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2340606404080485e-06,
+      "loss": 0.1024,
+      "step": 2540
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.228393312553132e-06,
+      "loss": 0.0863,
+      "step": 2541
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.222725984698216e-06,
+      "loss": 0.0913,
+      "step": 2542
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.2170586568432985e-06,
+      "loss": 0.0982,
+      "step": 2543
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.211391328988382e-06,
+      "loss": 0.0962,
+      "step": 2544
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.205724001133466e-06,
+      "loss": 0.0962,
+      "step": 2545
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.200056673278549e-06,
+      "loss": 0.0939,
+      "step": 2546
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.1943893454236335e-06,
+      "loss": 0.0863,
+      "step": 2547
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.188722017568717e-06,
+      "loss": 0.0962,
+      "step": 2548
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.1830546897138e-06,
+      "loss": 0.0983,
+      "step": 2549
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.177387361858884e-06,
+      "loss": 0.0828,
+      "step": 2550
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.171720034003968e-06,
+      "loss": 0.0899,
+      "step": 2551
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.166052706149052e-06,
+      "loss": 0.0829,
+      "step": 2552
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.160385378294134e-06,
+      "loss": 0.0933,
+      "step": 2553
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.154718050439218e-06,
+      "loss": 0.0893,
+      "step": 2554
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.149050722584302e-06,
+      "loss": 0.0971,
+      "step": 2555
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.143383394729385e-06,
+      "loss": 0.1024,
+      "step": 2556
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.137716066874469e-06,
+      "loss": 0.098,
+      "step": 2557
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.132048739019553e-06,
+      "loss": 0.0994,
+      "step": 2558
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.126381411164637e-06,
+      "loss": 0.0925,
+      "step": 2559
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.12071408330972e-06,
+      "loss": 0.1026,
+      "step": 2560
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.1150467554548035e-06,
+      "loss": 0.1004,
+      "step": 2561
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.109379427599888e-06,
+      "loss": 0.0877,
+      "step": 2562
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.10371209974497e-06,
+      "loss": 0.0962,
+      "step": 2563
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.098044771890055e-06,
+      "loss": 0.0944,
+      "step": 2564
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.092377444035138e-06,
+      "loss": 0.0868,
+      "step": 2565
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.086710116180221e-06,
+      "loss": 0.0874,
+      "step": 2566
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.081042788325305e-06,
+      "loss": 0.0983,
+      "step": 2567
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0753754604703885e-06,
+      "loss": 0.1035,
+      "step": 2568
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.069708132615473e-06,
+      "loss": 0.0897,
+      "step": 2569
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.064040804760556e-06,
+      "loss": 0.0948,
+      "step": 2570
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.058373476905639e-06,
+      "loss": 0.0921,
+      "step": 2571
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0527061490507236e-06,
+      "loss": 0.0931,
+      "step": 2572
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.047038821195806e-06,
+      "loss": 0.0865,
+      "step": 2573
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.041371493340891e-06,
+      "loss": 0.0977,
+      "step": 2574
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0357041654859736e-06,
+      "loss": 0.1029,
+      "step": 2575
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.030036837631057e-06,
+      "loss": 0.0962,
+      "step": 2576
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.024369509776141e-06,
+      "loss": 0.0866,
+      "step": 2577
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.018702181921224e-06,
+      "loss": 0.0876,
+      "step": 2578
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.0130348540663086e-06,
+      "loss": 0.0963,
+      "step": 2579
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.007367526211392e-06,
+      "loss": 0.0942,
+      "step": 2580
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.001700198356475e-06,
+      "loss": 0.0881,
+      "step": 2581
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.996032870501559e-06,
+      "loss": 0.0876,
+      "step": 2582
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.990365542646642e-06,
+      "loss": 0.0919,
+      "step": 2583
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.984698214791727e-06,
+      "loss": 0.0895,
+      "step": 2584
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.979030886936809e-06,
+      "loss": 0.092,
+      "step": 2585
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.973363559081894e-06,
+      "loss": 0.09,
+      "step": 2586
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.967696231226977e-06,
+      "loss": 0.1096,
+      "step": 2587
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.96202890337206e-06,
+      "loss": 0.0872,
+      "step": 2588
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9563615755171444e-06,
+      "loss": 0.0878,
+      "step": 2589
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.950694247662228e-06,
+      "loss": 0.095,
+      "step": 2590
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.945026919807312e-06,
+      "loss": 0.0921,
+      "step": 2591
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.939359591952395e-06,
+      "loss": 0.0978,
+      "step": 2592
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.933692264097478e-06,
+      "loss": 0.1015,
+      "step": 2593
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.928024936242563e-06,
+      "loss": 0.1029,
+      "step": 2594
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.922357608387645e-06,
+      "loss": 0.0951,
+      "step": 2595
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.9166902805327294e-06,
+      "loss": 0.0979,
+      "step": 2596
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.911022952677813e-06,
+      "loss": 0.0989,
+      "step": 2597
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.905355624822896e-06,
+      "loss": 0.0924,
+      "step": 2598
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.89968829696798e-06,
+      "loss": 0.0873,
+      "step": 2599
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.894020969113064e-06,
+      "loss": 0.0989,
+      "step": 2600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.888353641258148e-06,
+      "loss": 0.1012,
+      "step": 2601
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.882686313403231e-06,
+      "loss": 0.0989,
+      "step": 2602
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.877018985548314e-06,
+      "loss": 0.0867,
+      "step": 2603
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.871351657693399e-06,
+      "loss": 0.0921,
+      "step": 2604
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.865684329838481e-06,
+      "loss": 0.1006,
+      "step": 2605
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.860017001983565e-06,
+      "loss": 0.1012,
+      "step": 2606
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.854349674128649e-06,
+      "loss": 0.0937,
+      "step": 2607
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.848682346273732e-06,
+      "loss": 0.0848,
+      "step": 2608
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.843015018418816e-06,
+      "loss": 0.0819,
+      "step": 2609
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8373476905638995e-06,
+      "loss": 0.093,
+      "step": 2610
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.831680362708984e-06,
+      "loss": 0.0976,
+      "step": 2611
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.826013034854067e-06,
+      "loss": 0.0858,
+      "step": 2612
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.820345706999151e-06,
+      "loss": 0.0911,
+      "step": 2613
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.814678379144234e-06,
+      "loss": 0.0963,
+      "step": 2614
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.809011051289317e-06,
+      "loss": 0.1005,
+      "step": 2615
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.803343723434401e-06,
+      "loss": 0.0851,
+      "step": 2616
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7976763955794845e-06,
+      "loss": 0.0854,
+      "step": 2617
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.792009067724569e-06,
+      "loss": 0.0933,
+      "step": 2618
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.786341739869652e-06,
+      "loss": 0.1026,
+      "step": 2619
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.780674412014735e-06,
+      "loss": 0.1001,
+      "step": 2620
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7750070841598195e-06,
+      "loss": 0.0967,
+      "step": 2621
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.769339756304903e-06,
+      "loss": 0.0987,
+      "step": 2622
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.763672428449987e-06,
+      "loss": 0.0926,
+      "step": 2623
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7580051005950695e-06,
+      "loss": 0.0952,
+      "step": 2624
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.752337772740153e-06,
+      "loss": 0.0766,
+      "step": 2625
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.746670444885237e-06,
+      "loss": 0.0904,
+      "step": 2626
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.74100311703032e-06,
+      "loss": 0.0941,
+      "step": 2627
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7353357891754045e-06,
+      "loss": 0.0923,
+      "step": 2628
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.729668461320488e-06,
+      "loss": 0.0942,
+      "step": 2629
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.724001133465571e-06,
+      "loss": 0.0998,
+      "step": 2630
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.718333805610655e-06,
+      "loss": 0.0956,
+      "step": 2631
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.712666477755739e-06,
+      "loss": 0.084,
+      "step": 2632
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.706999149900823e-06,
+      "loss": 0.0979,
+      "step": 2633
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.701331822045905e-06,
+      "loss": 0.0938,
+      "step": 2634
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.695664494190989e-06,
+      "loss": 0.09,
+      "step": 2635
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.689997166336073e-06,
+      "loss": 0.1021,
+      "step": 2636
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.684329838481156e-06,
+      "loss": 0.0971,
+      "step": 2637
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.67866251062624e-06,
+      "loss": 0.0966,
+      "step": 2638
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.672995182771324e-06,
+      "loss": 0.0929,
+      "step": 2639
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.667327854916408e-06,
+      "loss": 0.088,
+      "step": 2640
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.661660527061491e-06,
+      "loss": 0.083,
+      "step": 2641
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6559931992065745e-06,
+      "loss": 0.1052,
+      "step": 2642
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.650325871351659e-06,
+      "loss": 0.1014,
+      "step": 2643
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.644658543496741e-06,
+      "loss": 0.0969,
+      "step": 2644
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.638991215641826e-06,
+      "loss": 0.0905,
+      "step": 2645
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.633323887786909e-06,
+      "loss": 0.1036,
+      "step": 2646
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.627656559931992e-06,
+      "loss": 0.0986,
+      "step": 2647
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.621989232077076e-06,
+      "loss": 0.0816,
+      "step": 2648
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6163219042221595e-06,
+      "loss": 0.0907,
+      "step": 2649
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.610654576367244e-06,
+      "loss": 0.0923,
+      "step": 2650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.604987248512327e-06,
+      "loss": 0.1026,
+      "step": 2651
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.59931992065741e-06,
+      "loss": 0.0966,
+      "step": 2652
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.5936525928024946e-06,
+      "loss": 0.0975,
+      "step": 2653
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.587985264947577e-06,
+      "loss": 0.1059,
+      "step": 2654
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.582317937092662e-06,
+      "loss": 0.0931,
+      "step": 2655
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.5766506092377446e-06,
+      "loss": 0.0934,
+      "step": 2656
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.570983281382828e-06,
+      "loss": 0.0809,
+      "step": 2657
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.565315953527912e-06,
+      "loss": 0.0888,
+      "step": 2658
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.559648625672995e-06,
+      "loss": 0.0906,
+      "step": 2659
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.55398129781808e-06,
+      "loss": 0.0867,
+      "step": 2660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.548313969963163e-06,
+      "loss": 0.1078,
+      "step": 2661
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.542646642108246e-06,
+      "loss": 0.0899,
+      "step": 2662
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.53697931425333e-06,
+      "loss": 0.0882,
+      "step": 2663
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.531311986398413e-06,
+      "loss": 0.1019,
+      "step": 2664
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.525644658543498e-06,
+      "loss": 0.0896,
+      "step": 2665
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.51997733068858e-06,
+      "loss": 0.0893,
+      "step": 2666
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.514310002833665e-06,
+      "loss": 0.0953,
+      "step": 2667
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.508642674978748e-06,
+      "loss": 0.0884,
+      "step": 2668
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.502975347123831e-06,
+      "loss": 0.0988,
+      "step": 2669
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.4973080192689154e-06,
+      "loss": 0.0911,
+      "step": 2670
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.491640691413999e-06,
+      "loss": 0.0977,
+      "step": 2671
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.485973363559083e-06,
+      "loss": 0.0984,
+      "step": 2672
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.480306035704166e-06,
+      "loss": 0.0855,
+      "step": 2673
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.474638707849249e-06,
+      "loss": 0.1,
+      "step": 2674
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.468971379994334e-06,
+      "loss": 0.0878,
+      "step": 2675
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.463304052139416e-06,
+      "loss": 0.0873,
+      "step": 2676
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.4576367242845004e-06,
+      "loss": 0.0958,
+      "step": 2677
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.451969396429584e-06,
+      "loss": 0.0997,
+      "step": 2678
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.446302068574667e-06,
+      "loss": 0.0996,
+      "step": 2679
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.440634740719751e-06,
+      "loss": 0.1023,
+      "step": 2680
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.434967412864835e-06,
+      "loss": 0.1035,
+      "step": 2681
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.429300085009919e-06,
+      "loss": 0.0903,
+      "step": 2682
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.423632757155002e-06,
+      "loss": 0.0963,
+      "step": 2683
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.417965429300085e-06,
+      "loss": 0.092,
+      "step": 2684
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.412298101445169e-06,
+      "loss": 0.0899,
+      "step": 2685
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.406630773590252e-06,
+      "loss": 0.0953,
+      "step": 2686
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.400963445735336e-06,
+      "loss": 0.0877,
+      "step": 2687
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.39529611788042e-06,
+      "loss": 0.1051,
+      "step": 2688
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.389628790025503e-06,
+      "loss": 0.0918,
+      "step": 2689
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.383961462170587e-06,
+      "loss": 0.0918,
+      "step": 2690
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3782941343156705e-06,
+      "loss": 0.0916,
+      "step": 2691
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.372626806460755e-06,
+      "loss": 0.1006,
+      "step": 2692
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.366959478605838e-06,
+      "loss": 0.0863,
+      "step": 2693
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.361292150750922e-06,
+      "loss": 0.0907,
+      "step": 2694
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.355624822896005e-06,
+      "loss": 0.0928,
+      "step": 2695
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.349957495041088e-06,
+      "loss": 0.0944,
+      "step": 2696
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.344290167186172e-06,
+      "loss": 0.0835,
+      "step": 2697
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3386228393312555e-06,
+      "loss": 0.0817,
+      "step": 2698
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.33295551147634e-06,
+      "loss": 0.0896,
+      "step": 2699
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.327288183621423e-06,
+      "loss": 0.0924,
+      "step": 2700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.321620855766506e-06,
+      "loss": 0.0876,
+      "step": 2701
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.3159535279115905e-06,
+      "loss": 0.0854,
+      "step": 2702
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.310286200056674e-06,
+      "loss": 0.0868,
+      "step": 2703
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.304618872201758e-06,
+      "loss": 0.0969,
+      "step": 2704
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.2989515443468405e-06,
+      "loss": 0.0829,
+      "step": 2705
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.293284216491924e-06,
+      "loss": 0.0862,
+      "step": 2706
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.287616888637008e-06,
+      "loss": 0.1093,
+      "step": 2707
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.281949560782091e-06,
+      "loss": 0.0829,
+      "step": 2708
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.2762822329271755e-06,
+      "loss": 0.0854,
+      "step": 2709
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.270614905072259e-06,
+      "loss": 0.0829,
+      "step": 2710
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.264947577217342e-06,
+      "loss": 0.0897,
+      "step": 2711
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.259280249362426e-06,
+      "loss": 0.096,
+      "step": 2712
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.25361292150751e-06,
+      "loss": 0.092,
+      "step": 2713
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.247945593652594e-06,
+      "loss": 0.1037,
+      "step": 2714
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.242278265797676e-06,
+      "loss": 0.096,
+      "step": 2715
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.23661093794276e-06,
+      "loss": 0.098,
+      "step": 2716
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.230943610087844e-06,
+      "loss": 0.0819,
+      "step": 2717
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.225276282232927e-06,
+      "loss": 0.0952,
+      "step": 2718
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.219608954378011e-06,
+      "loss": 0.0908,
+      "step": 2719
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.213941626523095e-06,
+      "loss": 0.0949,
+      "step": 2720
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.208274298668179e-06,
+      "loss": 0.0813,
+      "step": 2721
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.202606970813262e-06,
+      "loss": 0.078,
+      "step": 2722
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.1969396429583455e-06,
+      "loss": 0.1046,
+      "step": 2723
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.19127231510343e-06,
+      "loss": 0.0849,
+      "step": 2724
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.185604987248512e-06,
+      "loss": 0.092,
+      "step": 2725
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.179937659393597e-06,
+      "loss": 0.0861,
+      "step": 2726
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.17427033153868e-06,
+      "loss": 0.0873,
+      "step": 2727
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.168603003683763e-06,
+      "loss": 0.0913,
+      "step": 2728
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.162935675828847e-06,
+      "loss": 0.0849,
+      "step": 2729
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.1572683479739306e-06,
+      "loss": 0.0923,
+      "step": 2730
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.151601020119015e-06,
+      "loss": 0.0854,
+      "step": 2731
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.145933692264098e-06,
+      "loss": 0.0944,
+      "step": 2732
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.140266364409181e-06,
+      "loss": 0.0846,
+      "step": 2733
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.1345990365542656e-06,
+      "loss": 0.0899,
+      "step": 2734
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.128931708699348e-06,
+      "loss": 0.1071,
+      "step": 2735
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.123264380844433e-06,
+      "loss": 0.0937,
+      "step": 2736
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.1175970529895156e-06,
+      "loss": 0.0881,
+      "step": 2737
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.111929725134599e-06,
+      "loss": 0.0839,
+      "step": 2738
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.106262397279683e-06,
+      "loss": 0.0898,
+      "step": 2739
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.100595069424766e-06,
+      "loss": 0.0844,
+      "step": 2740
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.094927741569851e-06,
+      "loss": 0.0909,
+      "step": 2741
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.089260413714934e-06,
+      "loss": 0.0874,
+      "step": 2742
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.083593085860017e-06,
+      "loss": 0.0857,
+      "step": 2743
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0779257580051014e-06,
+      "loss": 0.0891,
+      "step": 2744
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.072258430150184e-06,
+      "loss": 0.1043,
+      "step": 2745
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.066591102295269e-06,
+      "loss": 0.0899,
+      "step": 2746
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.060923774440351e-06,
+      "loss": 0.0992,
+      "step": 2747
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.055256446585436e-06,
+      "loss": 0.0958,
+      "step": 2748
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.049589118730519e-06,
+      "loss": 0.1186,
+      "step": 2749
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.043921790875602e-06,
+      "loss": 0.0842,
+      "step": 2750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.0382544630206864e-06,
+      "loss": 0.1008,
+      "step": 2751
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.03258713516577e-06,
+      "loss": 0.0937,
+      "step": 2752
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.026919807310854e-06,
+      "loss": 0.0831,
+      "step": 2753
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.021252479455937e-06,
+      "loss": 0.0828,
+      "step": 2754
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.01558515160102e-06,
+      "loss": 0.0982,
+      "step": 2755
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.009917823746104e-06,
+      "loss": 0.0819,
+      "step": 2756
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.004250495891187e-06,
+      "loss": 0.0882,
+      "step": 2757
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.9985831680362715e-06,
+      "loss": 0.0992,
+      "step": 2758
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.992915840181355e-06,
+      "loss": 0.0862,
+      "step": 2759
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.987248512326439e-06,
+      "loss": 0.0875,
+      "step": 2760
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.981581184471522e-06,
+      "loss": 0.0922,
+      "step": 2761
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.975913856616606e-06,
+      "loss": 0.0862,
+      "step": 2762
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.970246528761689e-06,
+      "loss": 0.0848,
+      "step": 2763
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.964579200906773e-06,
+      "loss": 0.0849,
+      "step": 2764
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9589118730518565e-06,
+      "loss": 0.0988,
+      "step": 2765
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.95324454519694e-06,
+      "loss": 0.0961,
+      "step": 2766
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.947577217342024e-06,
+      "loss": 0.0922,
+      "step": 2767
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.941909889487107e-06,
+      "loss": 0.0992,
+      "step": 2768
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.936242561632191e-06,
+      "loss": 0.0862,
+      "step": 2769
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.930575233777275e-06,
+      "loss": 0.0875,
+      "step": 2770
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.924907905922358e-06,
+      "loss": 0.0966,
+      "step": 2771
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9192405780674415e-06,
+      "loss": 0.0939,
+      "step": 2772
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.913573250212525e-06,
+      "loss": 0.0894,
+      "step": 2773
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.907905922357609e-06,
+      "loss": 0.0825,
+      "step": 2774
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.902238594502692e-06,
+      "loss": 0.1006,
+      "step": 2775
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.896571266647776e-06,
+      "loss": 0.0877,
+      "step": 2776
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.89090393879286e-06,
+      "loss": 0.0907,
+      "step": 2777
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.885236610937943e-06,
+      "loss": 0.0994,
+      "step": 2778
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8795692830830265e-06,
+      "loss": 0.0871,
+      "step": 2779
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.873901955228111e-06,
+      "loss": 0.0976,
+      "step": 2780
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.868234627373194e-06,
+      "loss": 0.1051,
+      "step": 2781
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.862567299518277e-06,
+      "loss": 0.0867,
+      "step": 2782
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8568999716633615e-06,
+      "loss": 0.0887,
+      "step": 2783
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.851232643808445e-06,
+      "loss": 0.1026,
+      "step": 2784
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.845565315953528e-06,
+      "loss": 0.1121,
+      "step": 2785
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8398979880986115e-06,
+      "loss": 0.093,
+      "step": 2786
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.834230660243696e-06,
+      "loss": 0.0929,
+      "step": 2787
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.828563332388779e-06,
+      "loss": 0.0998,
+      "step": 2788
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.822896004533862e-06,
+      "loss": 0.0883,
+      "step": 2789
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8172286766789465e-06,
+      "loss": 0.1055,
+      "step": 2790
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.81156134882403e-06,
+      "loss": 0.1019,
+      "step": 2791
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.805894020969113e-06,
+      "loss": 0.0945,
+      "step": 2792
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.800226693114197e-06,
+      "loss": 0.0991,
+      "step": 2793
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.794559365259281e-06,
+      "loss": 0.0826,
+      "step": 2794
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.788892037404364e-06,
+      "loss": 0.0942,
+      "step": 2795
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.783224709549447e-06,
+      "loss": 0.0954,
+      "step": 2796
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7775573816945315e-06,
+      "loss": 0.1003,
+      "step": 2797
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.771890053839615e-06,
+      "loss": 0.0971,
+      "step": 2798
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.766222725984699e-06,
+      "loss": 0.1094,
+      "step": 2799
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.760555398129782e-06,
+      "loss": 0.1034,
+      "step": 2800
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.754888070274866e-06,
+      "loss": 0.1003,
+      "step": 2801
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.749220742419949e-06,
+      "loss": 0.1063,
+      "step": 2802
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.743553414565033e-06,
+      "loss": 0.1041,
+      "step": 2803
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7378860867101165e-06,
+      "loss": 0.0895,
+      "step": 2804
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.732218758855201e-06,
+      "loss": 0.0985,
+      "step": 2805
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.726551431000283e-06,
+      "loss": 0.0886,
+      "step": 2806
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.720884103145367e-06,
+      "loss": 0.0887,
+      "step": 2807
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.715216775290451e-06,
+      "loss": 0.0989,
+      "step": 2808
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.709549447435535e-06,
+      "loss": 0.1044,
+      "step": 2809
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.703882119580618e-06,
+      "loss": 0.0918,
+      "step": 2810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6982147917257016e-06,
+      "loss": 0.0948,
+      "step": 2811
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.692547463870785e-06,
+      "loss": 0.094,
+      "step": 2812
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.686880136015869e-06,
+      "loss": 0.0918,
+      "step": 2813
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.681212808160952e-06,
+      "loss": 0.0911,
+      "step": 2814
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.675545480306037e-06,
+      "loss": 0.081,
+      "step": 2815
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.669878152451119e-06,
+      "loss": 0.102,
+      "step": 2816
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.664210824596203e-06,
+      "loss": 0.0921,
+      "step": 2817
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6585434967412866e-06,
+      "loss": 0.0981,
+      "step": 2818
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.652876168886371e-06,
+      "loss": 0.0914,
+      "step": 2819
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.647208841031454e-06,
+      "loss": 0.0811,
+      "step": 2820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.641541513176538e-06,
+      "loss": 0.0982,
+      "step": 2821
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.635874185321621e-06,
+      "loss": 0.0773,
+      "step": 2822
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.630206857466705e-06,
+      "loss": 0.0958,
+      "step": 2823
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.624539529611788e-06,
+      "loss": 0.0857,
+      "step": 2824
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6188722017568724e-06,
+      "loss": 0.0946,
+      "step": 2825
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.613204873901956e-06,
+      "loss": 0.0811,
+      "step": 2826
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.607537546047039e-06,
+      "loss": 0.0991,
+      "step": 2827
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6018702181921224e-06,
+      "loss": 0.0851,
+      "step": 2828
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.596202890337207e-06,
+      "loss": 0.0866,
+      "step": 2829
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.59053556248229e-06,
+      "loss": 0.0861,
+      "step": 2830
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.584868234627374e-06,
+      "loss": 0.0851,
+      "step": 2831
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.5792009067724574e-06,
+      "loss": 0.1003,
+      "step": 2832
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.573533578917541e-06,
+      "loss": 0.0872,
+      "step": 2833
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.567866251062624e-06,
+      "loss": 0.0908,
+      "step": 2834
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.562198923207708e-06,
+      "loss": 0.0892,
+      "step": 2835
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.556531595352792e-06,
+      "loss": 0.0822,
+      "step": 2836
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.550864267497875e-06,
+      "loss": 0.0888,
+      "step": 2837
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.545196939642958e-06,
+      "loss": 0.0919,
+      "step": 2838
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5395296117880425e-06,
+      "loss": 0.0843,
+      "step": 2839
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.533862283933126e-06,
+      "loss": 0.0849,
+      "step": 2840
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.52819495607821e-06,
+      "loss": 0.082,
+      "step": 2841
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.522527628223293e-06,
+      "loss": 0.0943,
+      "step": 2842
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.516860300368377e-06,
+      "loss": 0.0957,
+      "step": 2843
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.51119297251346e-06,
+      "loss": 0.097,
+      "step": 2844
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.505525644658544e-06,
+      "loss": 0.0906,
+      "step": 2845
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4998583168036275e-06,
+      "loss": 0.0863,
+      "step": 2846
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.494190988948711e-06,
+      "loss": 0.0916,
+      "step": 2847
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.488523661093795e-06,
+      "loss": 0.105,
+      "step": 2848
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.482856333238878e-06,
+      "loss": 0.097,
+      "step": 2849
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.477189005383962e-06,
+      "loss": 0.0941,
+      "step": 2850
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.471521677529046e-06,
+      "loss": 0.0901,
+      "step": 2851
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.465854349674129e-06,
+      "loss": 0.0893,
+      "step": 2852
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4601870218192125e-06,
+      "loss": 0.0921,
+      "step": 2853
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.454519693964296e-06,
+      "loss": 0.0835,
+      "step": 2854
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.44885236610938e-06,
+      "loss": 0.0897,
+      "step": 2855
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.443185038254463e-06,
+      "loss": 0.0985,
+      "step": 2856
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.437517710399547e-06,
+      "loss": 0.0914,
+      "step": 2857
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.431850382544631e-06,
+      "loss": 0.0989,
+      "step": 2858
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.426183054689714e-06,
+      "loss": 0.0847,
+      "step": 2859
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4205157268347975e-06,
+      "loss": 0.0944,
+      "step": 2860
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.414848398979882e-06,
+      "loss": 0.0846,
+      "step": 2861
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.409181071124965e-06,
+      "loss": 0.0898,
+      "step": 2862
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.403513743270048e-06,
+      "loss": 0.0916,
+      "step": 2863
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.3978464154151325e-06,
+      "loss": 0.0948,
+      "step": 2864
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.392179087560216e-06,
+      "loss": 0.0879,
+      "step": 2865
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.386511759705299e-06,
+      "loss": 0.0859,
+      "step": 2866
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.3808444318503825e-06,
+      "loss": 0.0823,
+      "step": 2867
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.375177103995467e-06,
+      "loss": 0.0939,
+      "step": 2868
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.36950977614055e-06,
+      "loss": 0.078,
+      "step": 2869
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.363842448285633e-06,
+      "loss": 0.0855,
+      "step": 2870
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3581751204307175e-06,
+      "loss": 0.0934,
+      "step": 2871
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.352507792575801e-06,
+      "loss": 0.0956,
+      "step": 2872
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.346840464720884e-06,
+      "loss": 0.0867,
+      "step": 2873
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.341173136865968e-06,
+      "loss": 0.0983,
+      "step": 2874
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.335505809011052e-06,
+      "loss": 0.0739,
+      "step": 2875
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.329838481156135e-06,
+      "loss": 0.1013,
+      "step": 2876
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.324171153301218e-06,
+      "loss": 0.1008,
+      "step": 2877
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3185038254463025e-06,
+      "loss": 0.0923,
+      "step": 2878
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.312836497591386e-06,
+      "loss": 0.0794,
+      "step": 2879
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.30716916973647e-06,
+      "loss": 0.0867,
+      "step": 2880
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3015018418815525e-06,
+      "loss": 0.0848,
+      "step": 2881
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.295834514026637e-06,
+      "loss": 0.09,
+      "step": 2882
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.29016718617172e-06,
+      "loss": 0.0874,
+      "step": 2883
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.284499858316804e-06,
+      "loss": 0.0966,
+      "step": 2884
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2788325304618876e-06,
+      "loss": 0.087,
+      "step": 2885
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.273165202606972e-06,
+      "loss": 0.0913,
+      "step": 2886
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.267497874752054e-06,
+      "loss": 0.0879,
+      "step": 2887
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.261830546897138e-06,
+      "loss": 0.0991,
+      "step": 2888
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.256163219042222e-06,
+      "loss": 0.0967,
+      "step": 2889
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.250495891187306e-06,
+      "loss": 0.0851,
+      "step": 2890
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.244828563332389e-06,
+      "loss": 0.0767,
+      "step": 2891
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2391612354774726e-06,
+      "loss": 0.0946,
+      "step": 2892
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.233493907622556e-06,
+      "loss": 0.0764,
+      "step": 2893
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.22782657976764e-06,
+      "loss": 0.0969,
+      "step": 2894
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.222159251912723e-06,
+      "loss": 0.0936,
+      "step": 2895
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.216491924057808e-06,
+      "loss": 0.0893,
+      "step": 2896
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.21082459620289e-06,
+      "loss": 0.1064,
+      "step": 2897
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.205157268347974e-06,
+      "loss": 0.0986,
+      "step": 2898
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.199489940493058e-06,
+      "loss": 0.0853,
+      "step": 2899
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.193822612638142e-06,
+      "loss": 0.0981,
+      "step": 2900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.188155284783225e-06,
+      "loss": 0.0931,
+      "step": 2901
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.182487956928309e-06,
+      "loss": 0.1014,
+      "step": 2902
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.176820629073392e-06,
+      "loss": 0.0803,
+      "step": 2903
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.171153301218476e-06,
+      "loss": 0.091,
+      "step": 2904
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.165485973363559e-06,
+      "loss": 0.0922,
+      "step": 2905
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1598186455086434e-06,
+      "loss": 0.1031,
+      "step": 2906
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.154151317653727e-06,
+      "loss": 0.0895,
+      "step": 2907
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.14848398979881e-06,
+      "loss": 0.0755,
+      "step": 2908
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1428166619438934e-06,
+      "loss": 0.0901,
+      "step": 2909
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.137149334088978e-06,
+      "loss": 0.0904,
+      "step": 2910
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.131482006234061e-06,
+      "loss": 0.0816,
+      "step": 2911
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.125814678379145e-06,
+      "loss": 0.0801,
+      "step": 2912
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1201473505242285e-06,
+      "loss": 0.1107,
+      "step": 2913
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.114480022669312e-06,
+      "loss": 0.1014,
+      "step": 2914
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.108812694814395e-06,
+      "loss": 0.095,
+      "step": 2915
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.103145366959479e-06,
+      "loss": 0.0857,
+      "step": 2916
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.097478039104563e-06,
+      "loss": 0.0857,
+      "step": 2917
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.091810711249646e-06,
+      "loss": 0.0839,
+      "step": 2918
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.086143383394729e-06,
+      "loss": 0.0858,
+      "step": 2919
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0804760555398135e-06,
+      "loss": 0.0844,
+      "step": 2920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.074808727684897e-06,
+      "loss": 0.0947,
+      "step": 2921
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.069141399829981e-06,
+      "loss": 0.0788,
+      "step": 2922
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.063474071975064e-06,
+      "loss": 0.0888,
+      "step": 2923
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.057806744120148e-06,
+      "loss": 0.1015,
+      "step": 2924
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.052139416265231e-06,
+      "loss": 0.1039,
+      "step": 2925
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.046472088410315e-06,
+      "loss": 0.0883,
+      "step": 2926
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0408047605553985e-06,
+      "loss": 0.0832,
+      "step": 2927
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.035137432700482e-06,
+      "loss": 0.1001,
+      "step": 2928
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.029470104845566e-06,
+      "loss": 0.1048,
+      "step": 2929
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.023802776990649e-06,
+      "loss": 0.0887,
+      "step": 2930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.018135449135733e-06,
+      "loss": 0.0837,
+      "step": 2931
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.012468121280817e-06,
+      "loss": 0.0902,
+      "step": 2932
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.0068007934259e-06,
+      "loss": 0.0951,
+      "step": 2933
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.0011334655709835e-06,
+      "loss": 0.0884,
+      "step": 2934
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.995466137716067e-06,
+      "loss": 0.093,
+      "step": 2935
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.989798809861151e-06,
+      "loss": 0.0884,
+      "step": 2936
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.984131482006234e-06,
+      "loss": 0.0998,
+      "step": 2937
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.978464154151318e-06,
+      "loss": 0.0955,
+      "step": 2938
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.972796826296402e-06,
+      "loss": 0.0979,
+      "step": 2939
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.967129498441485e-06,
+      "loss": 0.0935,
+      "step": 2940
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9614621705865685e-06,
+      "loss": 0.0998,
+      "step": 2941
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.955794842731653e-06,
+      "loss": 0.0932,
+      "step": 2942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.950127514876736e-06,
+      "loss": 0.0877,
+      "step": 2943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.944460187021819e-06,
+      "loss": 0.085,
+      "step": 2944
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9387928591669035e-06,
+      "loss": 0.0867,
+      "step": 2945
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.933125531311987e-06,
+      "loss": 0.0938,
+      "step": 2946
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.92745820345707e-06,
+      "loss": 0.0904,
+      "step": 2947
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9217908756021535e-06,
+      "loss": 0.0869,
+      "step": 2948
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.916123547747238e-06,
+      "loss": 0.1024,
+      "step": 2949
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.910456219892321e-06,
+      "loss": 0.0992,
+      "step": 2950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.904788892037404e-06,
+      "loss": 0.0883,
+      "step": 2951
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.899121564182488e-06,
+      "loss": 0.0813,
+      "step": 2952
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.893454236327572e-06,
+      "loss": 0.09,
+      "step": 2953
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.887786908472655e-06,
+      "loss": 0.1051,
+      "step": 2954
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.882119580617739e-06,
+      "loss": 0.1034,
+      "step": 2955
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.876452252762823e-06,
+      "loss": 0.0793,
+      "step": 2956
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.870784924907906e-06,
+      "loss": 0.0968,
+      "step": 2957
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.865117597052989e-06,
+      "loss": 0.0953,
+      "step": 2958
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8594502691980735e-06,
+      "loss": 0.0965,
+      "step": 2959
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.853782941343157e-06,
+      "loss": 0.102,
+      "step": 2960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.848115613488241e-06,
+      "loss": 0.0906,
+      "step": 2961
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8424482856333235e-06,
+      "loss": 0.0891,
+      "step": 2962
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.836780957778408e-06,
+      "loss": 0.1087,
+      "step": 2963
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.831113629923491e-06,
+      "loss": 0.0836,
+      "step": 2964
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.825446302068575e-06,
+      "loss": 0.0844,
+      "step": 2965
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8197789742136586e-06,
+      "loss": 0.084,
+      "step": 2966
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8141116463587423e-06,
+      "loss": 0.0991,
+      "step": 2967
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8084443185038256e-06,
+      "loss": 0.081,
+      "step": 2968
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8027769906489094e-06,
+      "loss": 0.0945,
+      "step": 2969
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.797109662793993e-06,
+      "loss": 0.0965,
+      "step": 2970
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7914423349390765e-06,
+      "loss": 0.0883,
+      "step": 2971
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7857750070841602e-06,
+      "loss": 0.0916,
+      "step": 2972
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7801076792292436e-06,
+      "loss": 0.0985,
+      "step": 2973
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7744403513743273e-06,
+      "loss": 0.092,
+      "step": 2974
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.768773023519411e-06,
+      "loss": 0.0869,
+      "step": 2975
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7631056956644944e-06,
+      "loss": 0.0953,
+      "step": 2976
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.757438367809578e-06,
+      "loss": 0.0914,
+      "step": 2977
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7517710399546615e-06,
+      "loss": 0.0903,
+      "step": 2978
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7461037120997453e-06,
+      "loss": 0.0875,
+      "step": 2979
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.740436384244829e-06,
+      "loss": 0.108,
+      "step": 2980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7347690563899123e-06,
+      "loss": 0.0838,
+      "step": 2981
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.729101728534996e-06,
+      "loss": 0.0902,
+      "step": 2982
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.72343440068008e-06,
+      "loss": 0.09,
+      "step": 2983
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.717767072825163e-06,
+      "loss": 0.0861,
+      "step": 2984
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7120997449702465e-06,
+      "loss": 0.0817,
+      "step": 2985
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7064324171153303e-06,
+      "loss": 0.0969,
+      "step": 2986
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.700765089260414e-06,
+      "loss": 0.1084,
+      "step": 2987
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6950977614054978e-06,
+      "loss": 0.0948,
+      "step": 2988
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.689430433550581e-06,
+      "loss": 0.0928,
+      "step": 2989
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6837631056956644e-06,
+      "loss": 0.073,
+      "step": 2990
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.678095777840748e-06,
+      "loss": 0.0888,
+      "step": 2991
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.672428449985832e-06,
+      "loss": 0.1054,
+      "step": 2992
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6667611221309157e-06,
+      "loss": 0.095,
+      "step": 2993
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6610937942759995e-06,
+      "loss": 0.0991,
+      "step": 2994
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6554264664210824e-06,
+      "loss": 0.0921,
+      "step": 2995
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.649759138566166e-06,
+      "loss": 0.094,
+      "step": 2996
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.64409181071125e-06,
+      "loss": 0.0868,
+      "step": 2997
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6384244828563336e-06,
+      "loss": 0.0901,
+      "step": 2998
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6327571550014174e-06,
+      "loss": 0.0854,
+      "step": 2999
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6270898271465003e-06,
+      "loss": 0.0862,
+      "step": 3000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.621422499291584e-06,
+      "loss": 0.0824,
+      "step": 3001
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.615755171436668e-06,
+      "loss": 0.0882,
+      "step": 3002
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6100878435817516e-06,
+      "loss": 0.0891,
+      "step": 3003
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6044205157268353e-06,
+      "loss": 0.0837,
+      "step": 3004
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5987531878719182e-06,
+      "loss": 0.0828,
+      "step": 3005
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.593085860017002e-06,
+      "loss": 0.0956,
+      "step": 3006
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5874185321620857e-06,
+      "loss": 0.0822,
+      "step": 3007
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5817512043071695e-06,
+      "loss": 0.0957,
+      "step": 3008
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5760838764522532e-06,
+      "loss": 0.0792,
+      "step": 3009
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.570416548597337e-06,
+      "loss": 0.0949,
+      "step": 3010
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.56474922074242e-06,
+      "loss": 0.0829,
+      "step": 3011
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5590818928875037e-06,
+      "loss": 0.0977,
+      "step": 3012
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5534145650325874e-06,
+      "loss": 0.1063,
+      "step": 3013
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.547747237177671e-06,
+      "loss": 0.0834,
+      "step": 3014
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.542079909322755e-06,
+      "loss": 0.089,
+      "step": 3015
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.536412581467838e-06,
+      "loss": 0.0969,
+      "step": 3016
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5307452536129216e-06,
+      "loss": 0.0866,
+      "step": 3017
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5250779257580053e-06,
+      "loss": 0.0953,
+      "step": 3018
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.519410597903089e-06,
+      "loss": 0.0881,
+      "step": 3019
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.513743270048173e-06,
+      "loss": 0.0855,
+      "step": 3020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5080759421932566e-06,
+      "loss": 0.0822,
+      "step": 3021
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5024086143383395e-06,
+      "loss": 0.1101,
+      "step": 3022
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4967412864834233e-06,
+      "loss": 0.0997,
+      "step": 3023
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.491073958628507e-06,
+      "loss": 0.1059,
+      "step": 3024
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4854066307735908e-06,
+      "loss": 0.0837,
+      "step": 3025
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4797393029186745e-06,
+      "loss": 0.0902,
+      "step": 3026
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4740719750637574e-06,
+      "loss": 0.0953,
+      "step": 3027
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.468404647208841e-06,
+      "loss": 0.0924,
+      "step": 3028
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.462737319353925e-06,
+      "loss": 0.0848,
+      "step": 3029
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4570699914990087e-06,
+      "loss": 0.0916,
+      "step": 3030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4514026636440925e-06,
+      "loss": 0.0968,
+      "step": 3031
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4457353357891758e-06,
+      "loss": 0.0944,
+      "step": 3032
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.440068007934259e-06,
+      "loss": 0.0856,
+      "step": 3033
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.434400680079343e-06,
+      "loss": 0.0925,
+      "step": 3034
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4287333522244266e-06,
+      "loss": 0.0806,
+      "step": 3035
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4230660243695104e-06,
+      "loss": 0.0821,
+      "step": 3036
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4173986965145937e-06,
+      "loss": 0.1052,
+      "step": 3037
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.411731368659677e-06,
+      "loss": 0.0849,
+      "step": 3038
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.406064040804761e-06,
+      "loss": 0.089,
+      "step": 3039
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.4003967129498446e-06,
+      "loss": 0.0968,
+      "step": 3040
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3947293850949283e-06,
+      "loss": 0.0777,
+      "step": 3041
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3890620572400116e-06,
+      "loss": 0.1012,
+      "step": 3042
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.383394729385095e-06,
+      "loss": 0.0936,
+      "step": 3043
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3777274015301787e-06,
+      "loss": 0.0854,
+      "step": 3044
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3720600736752625e-06,
+      "loss": 0.0926,
+      "step": 3045
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3663927458203462e-06,
+      "loss": 0.0887,
+      "step": 3046
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3607254179654296e-06,
+      "loss": 0.1022,
+      "step": 3047
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3550580901105133e-06,
+      "loss": 0.0897,
+      "step": 3048
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3493907622555967e-06,
+      "loss": 0.0867,
+      "step": 3049
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3437234344006804e-06,
+      "loss": 0.0905,
+      "step": 3050
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.338056106545764e-06,
+      "loss": 0.0859,
+      "step": 3051
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3323887786908475e-06,
+      "loss": 0.1015,
+      "step": 3052
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3267214508359312e-06,
+      "loss": 0.1032,
+      "step": 3053
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3210541229810146e-06,
+      "loss": 0.0903,
+      "step": 3054
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3153867951260983e-06,
+      "loss": 0.094,
+      "step": 3055
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3097194672711817e-06,
+      "loss": 0.0738,
+      "step": 3056
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3040521394162654e-06,
+      "loss": 0.0965,
+      "step": 3057
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.298384811561349e-06,
+      "loss": 0.1157,
+      "step": 3058
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.292717483706433e-06,
+      "loss": 0.0919,
+      "step": 3059
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2870501558515163e-06,
+      "loss": 0.0946,
+      "step": 3060
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2813828279965996e-06,
+      "loss": 0.1017,
+      "step": 3061
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2757155001416833e-06,
+      "loss": 0.083,
+      "step": 3062
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.270048172286767e-06,
+      "loss": 0.1026,
+      "step": 3063
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.264380844431851e-06,
+      "loss": 0.0699,
+      "step": 3064
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.258713516576934e-06,
+      "loss": 0.0944,
+      "step": 3065
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2530461887220175e-06,
+      "loss": 0.084,
+      "step": 3066
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2473788608671013e-06,
+      "loss": 0.0966,
+      "step": 3067
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.241711533012185e-06,
+      "loss": 0.0862,
+      "step": 3068
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2360442051572688e-06,
+      "loss": 0.0913,
+      "step": 3069
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.230376877302352e-06,
+      "loss": 0.0881,
+      "step": 3070
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2247095494474354e-06,
+      "loss": 0.0899,
+      "step": 3071
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.219042221592519e-06,
+      "loss": 0.0827,
+      "step": 3072
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.213374893737603e-06,
+      "loss": 0.0901,
+      "step": 3073
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2077075658826867e-06,
+      "loss": 0.0868,
+      "step": 3074
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2020402380277705e-06,
+      "loss": 0.0982,
+      "step": 3075
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1963729101728534e-06,
+      "loss": 0.085,
+      "step": 3076
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.190705582317937e-06,
+      "loss": 0.0877,
+      "step": 3077
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.185038254463021e-06,
+      "loss": 0.0815,
+      "step": 3078
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1793709266081046e-06,
+      "loss": 0.0809,
+      "step": 3079
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1737035987531884e-06,
+      "loss": 0.0992,
+      "step": 3080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1680362708982713e-06,
+      "loss": 0.0831,
+      "step": 3081
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.162368943043355e-06,
+      "loss": 0.0806,
+      "step": 3082
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.156701615188439e-06,
+      "loss": 0.0996,
+      "step": 3083
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1510342873335226e-06,
+      "loss": 0.0939,
+      "step": 3084
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1453669594786063e-06,
+      "loss": 0.094,
+      "step": 3085
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.13969963162369e-06,
+      "loss": 0.1008,
+      "step": 3086
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.134032303768773e-06,
+      "loss": 0.0869,
+      "step": 3087
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1283649759138567e-06,
+      "loss": 0.0939,
+      "step": 3088
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1226976480589405e-06,
+      "loss": 0.0819,
+      "step": 3089
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1170303202040242e-06,
+      "loss": 0.0884,
+      "step": 3090
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.111362992349108e-06,
+      "loss": 0.0918,
+      "step": 3091
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.105695664494191e-06,
+      "loss": 0.0914,
+      "step": 3092
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1000283366392747e-06,
+      "loss": 0.0814,
+      "step": 3093
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0943610087843584e-06,
+      "loss": 0.0926,
+      "step": 3094
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.088693680929442e-06,
+      "loss": 0.0856,
+      "step": 3095
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.083026353074526e-06,
+      "loss": 0.0909,
+      "step": 3096
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.077359025219609e-06,
+      "loss": 0.0785,
+      "step": 3097
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0716916973646926e-06,
+      "loss": 0.0776,
+      "step": 3098
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0660243695097763e-06,
+      "loss": 0.087,
+      "step": 3099
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.06035704165486e-06,
+      "loss": 0.0978,
+      "step": 3100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.054689713799944e-06,
+      "loss": 0.0941,
+      "step": 3101
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0490223859450276e-06,
+      "loss": 0.0706,
+      "step": 3102
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0433550580901105e-06,
+      "loss": 0.0798,
+      "step": 3103
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0376877302351943e-06,
+      "loss": 0.0919,
+      "step": 3104
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.032020402380278e-06,
+      "loss": 0.0843,
+      "step": 3105
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0263530745253618e-06,
+      "loss": 0.0986,
+      "step": 3106
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0206857466704455e-06,
+      "loss": 0.0849,
+      "step": 3107
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0150184188155284e-06,
+      "loss": 0.0836,
+      "step": 3108
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.009351090960612e-06,
+      "loss": 0.0903,
+      "step": 3109
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.003683763105696e-06,
+      "loss": 0.0801,
+      "step": 3110
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9980164352507797e-06,
+      "loss": 0.0973,
+      "step": 3111
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9923491073958635e-06,
+      "loss": 0.085,
+      "step": 3112
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.986681779540947e-06,
+      "loss": 0.099,
+      "step": 3113
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.98101445168603e-06,
+      "loss": 0.0912,
+      "step": 3114
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.975347123831114e-06,
+      "loss": 0.0852,
+      "step": 3115
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9696797959761976e-06,
+      "loss": 0.0858,
+      "step": 3116
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9640124681212814e-06,
+      "loss": 0.0905,
+      "step": 3117
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9583451402663647e-06,
+      "loss": 0.0821,
+      "step": 3118
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.952677812411448e-06,
+      "loss": 0.0893,
+      "step": 3119
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.947010484556532e-06,
+      "loss": 0.0931,
+      "step": 3120
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9413431567016156e-06,
+      "loss": 0.0911,
+      "step": 3121
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9356758288466993e-06,
+      "loss": 0.0802,
+      "step": 3122
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9300085009917826e-06,
+      "loss": 0.0913,
+      "step": 3123
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.924341173136866e-06,
+      "loss": 0.0854,
+      "step": 3124
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9186738452819497e-06,
+      "loss": 0.0955,
+      "step": 3125
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9130065174270335e-06,
+      "loss": 0.0983,
+      "step": 3126
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.907339189572117e-06,
+      "loss": 0.095,
+      "step": 3127
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9016718617172006e-06,
+      "loss": 0.095,
+      "step": 3128
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8960045338622843e-06,
+      "loss": 0.08,
+      "step": 3129
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8903372060073677e-06,
+      "loss": 0.1156,
+      "step": 3130
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8846698781524514e-06,
+      "loss": 0.0885,
+      "step": 3131
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8790025502975347e-06,
+      "loss": 0.0971,
+      "step": 3132
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8733352224426185e-06,
+      "loss": 0.0899,
+      "step": 3133
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8676678945877023e-06,
+      "loss": 0.0917,
+      "step": 3134
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8620005667327856e-06,
+      "loss": 0.0965,
+      "step": 3135
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8563332388778693e-06,
+      "loss": 0.0903,
+      "step": 3136
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8506659110229527e-06,
+      "loss": 0.0878,
+      "step": 3137
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8449985831680364e-06,
+      "loss": 0.0802,
+      "step": 3138
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.83933125531312e-06,
+      "loss": 0.0861,
+      "step": 3139
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.833663927458204e-06,
+      "loss": 0.0787,
+      "step": 3140
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8279965996032873e-06,
+      "loss": 0.0971,
+      "step": 3141
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8223292717483706e-06,
+      "loss": 0.1033,
+      "step": 3142
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8166619438934544e-06,
+      "loss": 0.0817,
+      "step": 3143
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.810994616038538e-06,
+      "loss": 0.0916,
+      "step": 3144
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.805327288183622e-06,
+      "loss": 0.0778,
+      "step": 3145
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.799659960328705e-06,
+      "loss": 0.0962,
+      "step": 3146
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7939926324737885e-06,
+      "loss": 0.0745,
+      "step": 3147
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7883253046188723e-06,
+      "loss": 0.0939,
+      "step": 3148
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.782657976763956e-06,
+      "loss": 0.0907,
+      "step": 3149
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.77699064890904e-06,
+      "loss": 0.0873,
+      "step": 3150
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.771323321054123e-06,
+      "loss": 0.0936,
+      "step": 3151
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7656559931992065e-06,
+      "loss": 0.0874,
+      "step": 3152
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.75998866534429e-06,
+      "loss": 0.0846,
+      "step": 3153
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.754321337489374e-06,
+      "loss": 0.0939,
+      "step": 3154
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7486540096344577e-06,
+      "loss": 0.0958,
+      "step": 3155
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7429866817795415e-06,
+      "loss": 0.0867,
+      "step": 3156
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7373193539246244e-06,
+      "loss": 0.0827,
+      "step": 3157
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.731652026069708e-06,
+      "loss": 0.0951,
+      "step": 3158
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.725984698214792e-06,
+      "loss": 0.0761,
+      "step": 3159
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7203173703598756e-06,
+      "loss": 0.0921,
+      "step": 3160
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7146500425049594e-06,
+      "loss": 0.0843,
+      "step": 3161
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7089827146500423e-06,
+      "loss": 0.1012,
+      "step": 3162
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.703315386795126e-06,
+      "loss": 0.0998,
+      "step": 3163
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.69764805894021e-06,
+      "loss": 0.0742,
+      "step": 3164
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6919807310852936e-06,
+      "loss": 0.0851,
+      "step": 3165
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6863134032303773e-06,
+      "loss": 0.0999,
+      "step": 3166
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.680646075375461e-06,
+      "loss": 0.08,
+      "step": 3167
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.674978747520544e-06,
+      "loss": 0.0878,
+      "step": 3168
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6693114196656277e-06,
+      "loss": 0.0804,
+      "step": 3169
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6636440918107115e-06,
+      "loss": 0.0895,
+      "step": 3170
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6579767639557953e-06,
+      "loss": 0.0942,
+      "step": 3171
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.652309436100879e-06,
+      "loss": 0.0786,
+      "step": 3172
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.646642108245962e-06,
+      "loss": 0.0847,
+      "step": 3173
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6409747803910457e-06,
+      "loss": 0.0913,
+      "step": 3174
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6353074525361294e-06,
+      "loss": 0.094,
+      "step": 3175
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.629640124681213e-06,
+      "loss": 0.0849,
+      "step": 3176
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.623972796826297e-06,
+      "loss": 0.0898,
+      "step": 3177
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.61830546897138e-06,
+      "loss": 0.0934,
+      "step": 3178
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6126381411164636e-06,
+      "loss": 0.0925,
+      "step": 3179
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6069708132615474e-06,
+      "loss": 0.085,
+      "step": 3180
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.601303485406631e-06,
+      "loss": 0.0846,
+      "step": 3181
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.595636157551715e-06,
+      "loss": 0.0766,
+      "step": 3182
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5899688296967986e-06,
+      "loss": 0.0757,
+      "step": 3183
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5843015018418815e-06,
+      "loss": 0.0861,
+      "step": 3184
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5786341739869653e-06,
+      "loss": 0.0873,
+      "step": 3185
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.572966846132049e-06,
+      "loss": 0.0973,
+      "step": 3186
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5672995182771328e-06,
+      "loss": 0.0796,
+      "step": 3187
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5616321904222165e-06,
+      "loss": 0.0912,
+      "step": 3188
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5559648625672994e-06,
+      "loss": 0.0784,
+      "step": 3189
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.550297534712383e-06,
+      "loss": 0.0849,
+      "step": 3190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.544630206857467e-06,
+      "loss": 0.0834,
+      "step": 3191
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5389628790025507e-06,
+      "loss": 0.0851,
+      "step": 3192
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5332955511476345e-06,
+      "loss": 0.0795,
+      "step": 3193
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.527628223292718e-06,
+      "loss": 0.1028,
+      "step": 3194
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.521960895437801e-06,
+      "loss": 0.1025,
+      "step": 3195
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.516293567582885e-06,
+      "loss": 0.0867,
+      "step": 3196
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5106262397279686e-06,
+      "loss": 0.0916,
+      "step": 3197
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.504958911873052e-06,
+      "loss": 0.0892,
+      "step": 3198
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4992915840181357e-06,
+      "loss": 0.0818,
+      "step": 3199
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4936242561632195e-06,
+      "loss": 0.1023,
+      "step": 3200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.487956928308303e-06,
+      "loss": 0.0962,
+      "step": 3201
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4822896004533866e-06,
+      "loss": 0.0936,
+      "step": 3202
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.47662227259847e-06,
+      "loss": 0.0773,
+      "step": 3203
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4709549447435537e-06,
+      "loss": 0.0852,
+      "step": 3204
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4652876168886374e-06,
+      "loss": 0.0892,
+      "step": 3205
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4596202890337207e-06,
+      "loss": 0.0843,
+      "step": 3206
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4539529611788045e-06,
+      "loss": 0.1021,
+      "step": 3207
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.448285633323888e-06,
+      "loss": 0.0886,
+      "step": 3208
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4426183054689716e-06,
+      "loss": 0.0944,
+      "step": 3209
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4369509776140553e-06,
+      "loss": 0.0859,
+      "step": 3210
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4312836497591387e-06,
+      "loss": 0.078,
+      "step": 3211
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4256163219042224e-06,
+      "loss": 0.0933,
+      "step": 3212
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4199489940493058e-06,
+      "loss": 0.0878,
+      "step": 3213
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4142816661943895e-06,
+      "loss": 0.0825,
+      "step": 3214
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4086143383394733e-06,
+      "loss": 0.0744,
+      "step": 3215
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4029470104845566e-06,
+      "loss": 0.0985,
+      "step": 3216
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3972796826296403e-06,
+      "loss": 0.089,
+      "step": 3217
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3916123547747237e-06,
+      "loss": 0.0746,
+      "step": 3218
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3859450269198074e-06,
+      "loss": 0.0928,
+      "step": 3219
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.380277699064891e-06,
+      "loss": 0.0929,
+      "step": 3220
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3746103712099745e-06,
+      "loss": 0.0803,
+      "step": 3221
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3689430433550583e-06,
+      "loss": 0.0934,
+      "step": 3222
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3632757155001416e-06,
+      "loss": 0.0833,
+      "step": 3223
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3576083876452254e-06,
+      "loss": 0.0949,
+      "step": 3224
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.351941059790309e-06,
+      "loss": 0.0886,
+      "step": 3225
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3462737319353924e-06,
+      "loss": 0.0848,
+      "step": 3226
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.340606404080476e-06,
+      "loss": 0.098,
+      "step": 3227
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3349390762255595e-06,
+      "loss": 0.0806,
+      "step": 3228
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3292717483706433e-06,
+      "loss": 0.0849,
+      "step": 3229
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.323604420515727e-06,
+      "loss": 0.0987,
+      "step": 3230
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3179370926608104e-06,
+      "loss": 0.0957,
+      "step": 3231
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.312269764805894e-06,
+      "loss": 0.0921,
+      "step": 3232
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.306602436950978e-06,
+      "loss": 0.0909,
+      "step": 3233
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3009351090960612e-06,
+      "loss": 0.0831,
+      "step": 3234
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.295267781241145e-06,
+      "loss": 0.0799,
+      "step": 3235
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2896004533862287e-06,
+      "loss": 0.0773,
+      "step": 3236
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.283933125531312e-06,
+      "loss": 0.0812,
+      "step": 3237
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.278265797676396e-06,
+      "loss": 0.0785,
+      "step": 3238
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.272598469821479e-06,
+      "loss": 0.0816,
+      "step": 3239
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.266931141966563e-06,
+      "loss": 0.093,
+      "step": 3240
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2612638141116467e-06,
+      "loss": 0.0936,
+      "step": 3241
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.25559648625673e-06,
+      "loss": 0.0779,
+      "step": 3242
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2499291584018137e-06,
+      "loss": 0.0881,
+      "step": 3243
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2442618305468975e-06,
+      "loss": 0.0896,
+      "step": 3244
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.238594502691981e-06,
+      "loss": 0.089,
+      "step": 3245
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2329271748370646e-06,
+      "loss": 0.0735,
+      "step": 3246
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.227259846982148e-06,
+      "loss": 0.0831,
+      "step": 3247
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2215925191272317e-06,
+      "loss": 0.0824,
+      "step": 3248
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2159251912723154e-06,
+      "loss": 0.0777,
+      "step": 3249
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2102578634173987e-06,
+      "loss": 0.0796,
+      "step": 3250
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2045905355624825e-06,
+      "loss": 0.087,
+      "step": 3251
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1989232077075663e-06,
+      "loss": 0.0888,
+      "step": 3252
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1932558798526496e-06,
+      "loss": 0.0756,
+      "step": 3253
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1875885519977333e-06,
+      "loss": 0.0871,
+      "step": 3254
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1819212241428167e-06,
+      "loss": 0.0897,
+      "step": 3255
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1762538962879004e-06,
+      "loss": 0.0963,
+      "step": 3256
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.170586568432984e-06,
+      "loss": 0.0832,
+      "step": 3257
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1649192405780675e-06,
+      "loss": 0.0793,
+      "step": 3258
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1592519127231513e-06,
+      "loss": 0.0955,
+      "step": 3259
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.153584584868235e-06,
+      "loss": 0.084,
+      "step": 3260
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1479172570133184e-06,
+      "loss": 0.0941,
+      "step": 3261
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.142249929158402e-06,
+      "loss": 0.0941,
+      "step": 3262
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.136582601303486e-06,
+      "loss": 0.0979,
+      "step": 3263
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.130915273448569e-06,
+      "loss": 0.0897,
+      "step": 3264
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.125247945593653e-06,
+      "loss": 0.0919,
+      "step": 3265
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1195806177387363e-06,
+      "loss": 0.0703,
+      "step": 3266
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.11391328988382e-06,
+      "loss": 0.0852,
+      "step": 3267
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.108245962028904e-06,
+      "loss": 0.0911,
+      "step": 3268
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.102578634173987e-06,
+      "loss": 0.0954,
+      "step": 3269
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.096911306319071e-06,
+      "loss": 0.0915,
+      "step": 3270
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0912439784641546e-06,
+      "loss": 0.0807,
+      "step": 3271
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.085576650609238e-06,
+      "loss": 0.0845,
+      "step": 3272
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0799093227543217e-06,
+      "loss": 0.0872,
+      "step": 3273
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.074241994899405e-06,
+      "loss": 0.0936,
+      "step": 3274
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.068574667044489e-06,
+      "loss": 0.0858,
+      "step": 3275
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0629073391895726e-06,
+      "loss": 0.0847,
+      "step": 3276
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.057240011334656e-06,
+      "loss": 0.0968,
+      "step": 3277
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0515726834797396e-06,
+      "loss": 0.0899,
+      "step": 3278
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.045905355624823e-06,
+      "loss": 0.0957,
+      "step": 3279
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0402380277699067e-06,
+      "loss": 0.0891,
+      "step": 3280
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0345706999149905e-06,
+      "loss": 0.0767,
+      "step": 3281
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.028903372060074e-06,
+      "loss": 0.0814,
+      "step": 3282
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0232360442051576e-06,
+      "loss": 0.0857,
+      "step": 3283
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.017568716350241e-06,
+      "loss": 0.0921,
+      "step": 3284
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0119013884953247e-06,
+      "loss": 0.0836,
+      "step": 3285
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0062340606404084e-06,
+      "loss": 0.077,
+      "step": 3286
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0005667327854917e-06,
+      "loss": 0.0896,
+      "step": 3287
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9948994049305755e-06,
+      "loss": 0.0874,
+      "step": 3288
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.989232077075659e-06,
+      "loss": 0.0833,
+      "step": 3289
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9835647492207426e-06,
+      "loss": 0.0917,
+      "step": 3290
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9778974213658263e-06,
+      "loss": 0.0827,
+      "step": 3291
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9722300935109097e-06,
+      "loss": 0.09,
+      "step": 3292
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9665627656559934e-06,
+      "loss": 0.097,
+      "step": 3293
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9608954378010768e-06,
+      "loss": 0.0961,
+      "step": 3294
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9552281099461605e-06,
+      "loss": 0.0781,
+      "step": 3295
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.949560782091244e-06,
+      "loss": 0.079,
+      "step": 3296
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9438934542363276e-06,
+      "loss": 0.0794,
+      "step": 3297
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9382261263814114e-06,
+      "loss": 0.098,
+      "step": 3298
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9325587985264947e-06,
+      "loss": 0.0905,
+      "step": 3299
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9268914706715784e-06,
+      "loss": 0.0924,
+      "step": 3300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9212241428166618e-06,
+      "loss": 0.0883,
+      "step": 3301
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9155568149617455e-06,
+      "loss": 0.0875,
+      "step": 3302
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9098894871068293e-06,
+      "loss": 0.0872,
+      "step": 3303
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9042221592519128e-06,
+      "loss": 0.0782,
+      "step": 3304
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8985548313969966e-06,
+      "loss": 0.0781,
+      "step": 3305
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8928875035420801e-06,
+      "loss": 0.0928,
+      "step": 3306
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8872201756871637e-06,
+      "loss": 0.0801,
+      "step": 3307
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8815528478322472e-06,
+      "loss": 0.0849,
+      "step": 3308
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8758855199773308e-06,
+      "loss": 0.0919,
+      "step": 3309
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8702181921224145e-06,
+      "loss": 0.0884,
+      "step": 3310
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.864550864267498e-06,
+      "loss": 0.0896,
+      "step": 3311
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8588835364125816e-06,
+      "loss": 0.0747,
+      "step": 3312
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8532162085576651e-06,
+      "loss": 0.0767,
+      "step": 3313
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8475488807027489e-06,
+      "loss": 0.0827,
+      "step": 3314
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8418815528478322e-06,
+      "loss": 0.0823,
+      "step": 3315
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.836214224992916e-06,
+      "loss": 0.0925,
+      "step": 3316
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8305468971379997e-06,
+      "loss": 0.0819,
+      "step": 3317
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.824879569283083e-06,
+      "loss": 0.0879,
+      "step": 3318
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8192122414281668e-06,
+      "loss": 0.0779,
+      "step": 3319
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8135449135732501e-06,
+      "loss": 0.0989,
+      "step": 3320
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.807877585718334e-06,
+      "loss": 0.1062,
+      "step": 3321
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8022102578634177e-06,
+      "loss": 0.0842,
+      "step": 3322
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.796542930008501e-06,
+      "loss": 0.079,
+      "step": 3323
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7908756021535847e-06,
+      "loss": 0.0747,
+      "step": 3324
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7852082742986685e-06,
+      "loss": 0.0921,
+      "step": 3325
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7795409464437518e-06,
+      "loss": 0.0895,
+      "step": 3326
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7738736185888356e-06,
+      "loss": 0.0909,
+      "step": 3327
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.768206290733919e-06,
+      "loss": 0.0794,
+      "step": 3328
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7625389628790027e-06,
+      "loss": 0.0855,
+      "step": 3329
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7568716350240864e-06,
+      "loss": 0.0882,
+      "step": 3330
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7512043071691698e-06,
+      "loss": 0.0817,
+      "step": 3331
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7455369793142535e-06,
+      "loss": 0.0779,
+      "step": 3332
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7398696514593373e-06,
+      "loss": 0.0781,
+      "step": 3333
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7342023236044206e-06,
+      "loss": 0.0823,
+      "step": 3334
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7285349957495044e-06,
+      "loss": 0.0969,
+      "step": 3335
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7228676678945879e-06,
+      "loss": 0.0752,
+      "step": 3336
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7172003400396714e-06,
+      "loss": 0.0899,
+      "step": 3337
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7115330121847552e-06,
+      "loss": 0.0927,
+      "step": 3338
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7058656843298385e-06,
+      "loss": 0.078,
+      "step": 3339
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7001983564749223e-06,
+      "loss": 0.0799,
+      "step": 3340
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6945310286200058e-06,
+      "loss": 0.0895,
+      "step": 3341
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6888637007650894e-06,
+      "loss": 0.085,
+      "step": 3342
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6831963729101731e-06,
+      "loss": 0.0831,
+      "step": 3343
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6775290450552567e-06,
+      "loss": 0.0915,
+      "step": 3344
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6718617172003402e-06,
+      "loss": 0.0857,
+      "step": 3345
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6661943893454237e-06,
+      "loss": 0.0973,
+      "step": 3346
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6605270614905073e-06,
+      "loss": 0.086,
+      "step": 3347
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6548597336355908e-06,
+      "loss": 0.0851,
+      "step": 3348
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6491924057806746e-06,
+      "loss": 0.0875,
+      "step": 3349
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6435250779257581e-06,
+      "loss": 0.0814,
+      "step": 3350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6378577500708417e-06,
+      "loss": 0.0813,
+      "step": 3351
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6321904222159254e-06,
+      "loss": 0.0871,
+      "step": 3352
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6265230943610088e-06,
+      "loss": 0.0794,
+      "step": 3353
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6208557665060925e-06,
+      "loss": 0.0823,
+      "step": 3354
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.615188438651176e-06,
+      "loss": 0.0814,
+      "step": 3355
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6095211107962596e-06,
+      "loss": 0.0956,
+      "step": 3356
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6038537829413434e-06,
+      "loss": 0.0886,
+      "step": 3357
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5981864550864267e-06,
+      "loss": 0.0825,
+      "step": 3358
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5925191272315104e-06,
+      "loss": 0.0859,
+      "step": 3359
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5868517993765942e-06,
+      "loss": 0.071,
+      "step": 3360
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5811844715216775e-06,
+      "loss": 0.0737,
+      "step": 3361
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5755171436667613e-06,
+      "loss": 0.0834,
+      "step": 3362
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.569849815811845e-06,
+      "loss": 0.0723,
+      "step": 3363
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5641824879569284e-06,
+      "loss": 0.0954,
+      "step": 3364
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5585151601020121e-06,
+      "loss": 0.0803,
+      "step": 3365
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5528478322470955e-06,
+      "loss": 0.1014,
+      "step": 3366
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5471805043921792e-06,
+      "loss": 0.0806,
+      "step": 3367
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.541513176537263e-06,
+      "loss": 0.0816,
+      "step": 3368
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5358458486823463e-06,
+      "loss": 0.0755,
+      "step": 3369
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.53017852082743e-06,
+      "loss": 0.0728,
+      "step": 3370
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5245111929725138e-06,
+      "loss": 0.0787,
+      "step": 3371
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5188438651175971e-06,
+      "loss": 0.0733,
+      "step": 3372
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5131765372626809e-06,
+      "loss": 0.1004,
+      "step": 3373
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5075092094077642e-06,
+      "loss": 0.0951,
+      "step": 3374
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.501841881552848e-06,
+      "loss": 0.092,
+      "step": 3375
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4961745536979317e-06,
+      "loss": 0.0888,
+      "step": 3376
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.490507225843015e-06,
+      "loss": 0.0875,
+      "step": 3377
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4848398979880988e-06,
+      "loss": 0.0739,
+      "step": 3378
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4791725701331824e-06,
+      "loss": 0.0771,
+      "step": 3379
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.473505242278266e-06,
+      "loss": 0.0758,
+      "step": 3380
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4678379144233497e-06,
+      "loss": 0.1048,
+      "step": 3381
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.462170586568433e-06,
+      "loss": 0.0824,
+      "step": 3382
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4565032587135167e-06,
+      "loss": 0.0884,
+      "step": 3383
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4508359308586003e-06,
+      "loss": 0.0925,
+      "step": 3384
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4451686030036838e-06,
+      "loss": 0.0852,
+      "step": 3385
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4395012751487674e-06,
+      "loss": 0.0909,
+      "step": 3386
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4338339472938511e-06,
+      "loss": 0.0914,
+      "step": 3387
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4281666194389347e-06,
+      "loss": 0.0817,
+      "step": 3388
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4224992915840182e-06,
+      "loss": 0.0876,
+      "step": 3389
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.416831963729102e-06,
+      "loss": 0.0817,
+      "step": 3390
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4111646358741853e-06,
+      "loss": 0.0916,
+      "step": 3391
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.405497308019269e-06,
+      "loss": 0.0819,
+      "step": 3392
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3998299801643526e-06,
+      "loss": 0.0901,
+      "step": 3393
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3941626523094361e-06,
+      "loss": 0.0982,
+      "step": 3394
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.38849532445452e-06,
+      "loss": 0.0988,
+      "step": 3395
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3828279965996032e-06,
+      "loss": 0.0875,
+      "step": 3396
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.377160668744687e-06,
+      "loss": 0.0846,
+      "step": 3397
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3714933408897707e-06,
+      "loss": 0.0831,
+      "step": 3398
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.365826013034854e-06,
+      "loss": 0.0733,
+      "step": 3399
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3601586851799378e-06,
+      "loss": 0.0758,
+      "step": 3400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3544913573250212e-06,
+      "loss": 0.0772,
+      "step": 3401
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.348824029470105e-06,
+      "loss": 0.0772,
+      "step": 3402
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3431567016151887e-06,
+      "loss": 0.0822,
+      "step": 3403
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.337489373760272e-06,
+      "loss": 0.0824,
+      "step": 3404
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3318220459053557e-06,
+      "loss": 0.0837,
+      "step": 3405
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3261547180504395e-06,
+      "loss": 0.0807,
+      "step": 3406
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3204873901955228e-06,
+      "loss": 0.0793,
+      "step": 3407
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3148200623406066e-06,
+      "loss": 0.0752,
+      "step": 3408
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.30915273448569e-06,
+      "loss": 0.0847,
+      "step": 3409
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3034854066307737e-06,
+      "loss": 0.0853,
+      "step": 3410
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2978180787758574e-06,
+      "loss": 0.0757,
+      "step": 3411
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2921507509209408e-06,
+      "loss": 0.095,
+      "step": 3412
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2864834230660245e-06,
+      "loss": 0.1061,
+      "step": 3413
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2808160952111083e-06,
+      "loss": 0.0837,
+      "step": 3414
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2751487673561916e-06,
+      "loss": 0.0843,
+      "step": 3415
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2694814395012754e-06,
+      "loss": 0.0976,
+      "step": 3416
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.263814111646359e-06,
+      "loss": 0.0965,
+      "step": 3417
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2581467837914424e-06,
+      "loss": 0.0934,
+      "step": 3418
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.252479455936526e-06,
+      "loss": 0.0772,
+      "step": 3419
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2468121280816097e-06,
+      "loss": 0.0857,
+      "step": 3420
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2411448002266933e-06,
+      "loss": 0.0902,
+      "step": 3421
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2354774723717768e-06,
+      "loss": 0.0747,
+      "step": 3422
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2298101445168604e-06,
+      "loss": 0.0872,
+      "step": 3423
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.224142816661944e-06,
+      "loss": 0.0933,
+      "step": 3424
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2184754888070277e-06,
+      "loss": 0.097,
+      "step": 3425
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2128081609521112e-06,
+      "loss": 0.0911,
+      "step": 3426
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2071408330971948e-06,
+      "loss": 0.0917,
+      "step": 3427
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.2014735052422783e-06,
+      "loss": 0.0841,
+      "step": 3428
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1958061773873618e-06,
+      "loss": 0.0844,
+      "step": 3429
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1901388495324456e-06,
+      "loss": 0.0783,
+      "step": 3430
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1844715216775291e-06,
+      "loss": 0.0876,
+      "step": 3431
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1788041938226127e-06,
+      "loss": 0.0881,
+      "step": 3432
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1731368659676962e-06,
+      "loss": 0.1,
+      "step": 3433
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1674695381127798e-06,
+      "loss": 0.0757,
+      "step": 3434
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1618022102578635e-06,
+      "loss": 0.0837,
+      "step": 3435
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.156134882402947e-06,
+      "loss": 0.083,
+      "step": 3436
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1504675545480306e-06,
+      "loss": 0.0923,
+      "step": 3437
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1448002266931144e-06,
+      "loss": 0.0811,
+      "step": 3438
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.139132898838198e-06,
+      "loss": 0.086,
+      "step": 3439
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1334655709832814e-06,
+      "loss": 0.0937,
+      "step": 3440
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.127798243128365e-06,
+      "loss": 0.0819,
+      "step": 3441
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1221309152734487e-06,
+      "loss": 0.083,
+      "step": 3442
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1164635874185323e-06,
+      "loss": 0.0909,
+      "step": 3443
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1107962595636158e-06,
+      "loss": 0.1005,
+      "step": 3444
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1051289317086994e-06,
+      "loss": 0.0896,
+      "step": 3445
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0994616038537831e-06,
+      "loss": 0.1017,
+      "step": 3446
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0937942759988667e-06,
+      "loss": 0.0965,
+      "step": 3447
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0881269481439502e-06,
+      "loss": 0.0854,
+      "step": 3448
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0824596202890338e-06,
+      "loss": 0.083,
+      "step": 3449
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0767922924341175e-06,
+      "loss": 0.0769,
+      "step": 3450
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.071124964579201e-06,
+      "loss": 0.0713,
+      "step": 3451
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0654576367242846e-06,
+      "loss": 0.0945,
+      "step": 3452
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0597903088693681e-06,
+      "loss": 0.0819,
+      "step": 3453
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.054122981014452e-06,
+      "loss": 0.0806,
+      "step": 3454
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0484556531595354e-06,
+      "loss": 0.0836,
+      "step": 3455
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.042788325304619e-06,
+      "loss": 0.087,
+      "step": 3456
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0371209974497025e-06,
+      "loss": 0.0682,
+      "step": 3457
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0314536695947863e-06,
+      "loss": 0.0886,
+      "step": 3458
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0257863417398698e-06,
+      "loss": 0.0731,
+      "step": 3459
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0201190138849534e-06,
+      "loss": 0.069,
+      "step": 3460
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.014451686030037e-06,
+      "loss": 0.0802,
+      "step": 3461
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0087843581751205e-06,
+      "loss": 0.1054,
+      "step": 3462
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0031170303202042e-06,
+      "loss": 0.0841,
+      "step": 3463
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.974497024652878e-07,
+      "loss": 0.1012,
+      "step": 3464
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.917823746103713e-07,
+      "loss": 0.102,
+      "step": 3465
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.861150467554548e-07,
+      "loss": 0.0875,
+      "step": 3466
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.804477189005384e-07,
+      "loss": 0.0887,
+      "step": 3467
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.74780391045622e-07,
+      "loss": 0.0802,
+      "step": 3468
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.691130631907057e-07,
+      "loss": 0.071,
+      "step": 3469
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.634457353357892e-07,
+      "loss": 0.0847,
+      "step": 3470
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.577784074808728e-07,
+      "loss": 0.0715,
+      "step": 3471
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.521110796259564e-07,
+      "loss": 0.0842,
+      "step": 3472
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.464437517710401e-07,
+      "loss": 0.0852,
+      "step": 3473
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.407764239161236e-07,
+      "loss": 0.0785,
+      "step": 3474
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.351090960612073e-07,
+      "loss": 0.0785,
+      "step": 3475
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.294417682062908e-07,
+      "loss": 0.0893,
+      "step": 3476
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.237744403513744e-07,
+      "loss": 0.091,
+      "step": 3477
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.18107112496458e-07,
+      "loss": 0.0815,
+      "step": 3478
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.124397846415415e-07,
+      "loss": 0.0909,
+      "step": 3479
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.067724567866251e-07,
+      "loss": 0.0836,
+      "step": 3480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.011051289317088e-07,
+      "loss": 0.0845,
+      "step": 3481
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.954378010767924e-07,
+      "loss": 0.0826,
+      "step": 3482
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.897704732218759e-07,
+      "loss": 0.0865,
+      "step": 3483
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.841031453669595e-07,
+      "loss": 0.0913,
+      "step": 3484
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.784358175120432e-07,
+      "loss": 0.0861,
+      "step": 3485
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.727684896571268e-07,
+      "loss": 0.0807,
+      "step": 3486
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.671011618022103e-07,
+      "loss": 0.0935,
+      "step": 3487
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.614338339472939e-07,
+      "loss": 0.099,
+      "step": 3488
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.557665060923776e-07,
+      "loss": 0.0792,
+      "step": 3489
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.500991782374611e-07,
+      "loss": 0.0837,
+      "step": 3490
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.444318503825447e-07,
+      "loss": 0.0868,
+      "step": 3491
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.387645225276283e-07,
+      "loss": 0.0963,
+      "step": 3492
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.330971946727119e-07,
+      "loss": 0.0829,
+      "step": 3493
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.274298668177954e-07,
+      "loss": 0.0941,
+      "step": 3494
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.217625389628791e-07,
+      "loss": 0.078,
+      "step": 3495
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.160952111079627e-07,
+      "loss": 0.0906,
+      "step": 3496
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.104278832530463e-07,
+      "loss": 0.0855,
+      "step": 3497
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.047605553981298e-07,
+      "loss": 0.0847,
+      "step": 3498
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.990932275432133e-07,
+      "loss": 0.08,
+      "step": 3499
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.934258996882971e-07,
+      "loss": 0.0895,
+      "step": 3500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.877585718333806e-07,
+      "loss": 0.0897,
+      "step": 3501
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.820912439784642e-07,
+      "loss": 0.0909,
+      "step": 3502
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.764239161235477e-07,
+      "loss": 0.0773,
+      "step": 3503
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.707565882686315e-07,
+      "loss": 0.0857,
+      "step": 3504
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.65089260413715e-07,
+      "loss": 0.088,
+      "step": 3505
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.594219325587986e-07,
+      "loss": 0.0869,
+      "step": 3506
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.537546047038821e-07,
+      "loss": 0.083,
+      "step": 3507
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.480872768489659e-07,
+      "loss": 0.0786,
+      "step": 3508
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.424199489940494e-07,
+      "loss": 0.0776,
+      "step": 3509
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.36752621139133e-07,
+      "loss": 0.082,
+      "step": 3510
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.310852932842165e-07,
+      "loss": 0.0917,
+      "step": 3511
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.254179654293001e-07,
+      "loss": 0.0895,
+      "step": 3512
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.197506375743837e-07,
+      "loss": 0.1003,
+      "step": 3513
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.140833097194673e-07,
+      "loss": 0.0872,
+      "step": 3514
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.08415981864551e-07,
+      "loss": 0.0835,
+      "step": 3515
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.027486540096345e-07,
+      "loss": 0.0856,
+      "step": 3516
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.970813261547181e-07,
+      "loss": 0.0788,
+      "step": 3517
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.914139982998016e-07,
+      "loss": 0.0888,
+      "step": 3518
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.857466704448854e-07,
+      "loss": 0.0951,
+      "step": 3519
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.800793425899689e-07,
+      "loss": 0.0812,
+      "step": 3520
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.744120147350525e-07,
+      "loss": 0.0996,
+      "step": 3521
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.68744686880136e-07,
+      "loss": 0.0816,
+      "step": 3522
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.630773590252198e-07,
+      "loss": 0.0881,
+      "step": 3523
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.574100311703033e-07,
+      "loss": 0.088,
+      "step": 3524
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.517427033153868e-07,
+      "loss": 0.0785,
+      "step": 3525
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.460753754604704e-07,
+      "loss": 0.0739,
+      "step": 3526
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.404080476055541e-07,
+      "loss": 0.0832,
+      "step": 3527
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.347407197506377e-07,
+      "loss": 0.081,
+      "step": 3528
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.290733918957212e-07,
+      "loss": 0.076,
+      "step": 3529
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.234060640408049e-07,
+      "loss": 0.0779,
+      "step": 3530
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.177387361858884e-07,
+      "loss": 0.0939,
+      "step": 3531
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.12071408330972e-07,
+      "loss": 0.0851,
+      "step": 3532
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.064040804760556e-07,
+      "loss": 0.0786,
+      "step": 3533
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.007367526211391e-07,
+      "loss": 0.0859,
+      "step": 3534
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.950694247662228e-07,
+      "loss": 0.0855,
+      "step": 3535
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.894020969113063e-07,
+      "loss": 0.09,
+      "step": 3536
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.837347690563899e-07,
+      "loss": 0.0925,
+      "step": 3537
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.780674412014735e-07,
+      "loss": 0.0764,
+      "step": 3538
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.724001133465572e-07,
+      "loss": 0.0989,
+      "step": 3539
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.667327854916407e-07,
+      "loss": 0.086,
+      "step": 3540
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.610654576367244e-07,
+      "loss": 0.0936,
+      "step": 3541
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.553981297818079e-07,
+      "loss": 0.0804,
+      "step": 3542
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.497308019268916e-07,
+      "loss": 0.0961,
+      "step": 3543
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.440634740719751e-07,
+      "loss": 0.0833,
+      "step": 3544
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.383961462170588e-07,
+      "loss": 0.0882,
+      "step": 3545
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.327288183621423e-07,
+      "loss": 0.0798,
+      "step": 3546
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.27061490507226e-07,
+      "loss": 0.0778,
+      "step": 3547
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.213941626523095e-07,
+      "loss": 0.0835,
+      "step": 3548
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.157268347973931e-07,
+      "loss": 0.0793,
+      "step": 3549
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.100595069424767e-07,
+      "loss": 0.0919,
+      "step": 3550
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.043921790875602e-07,
+      "loss": 0.0912,
+      "step": 3551
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.987248512326439e-07,
+      "loss": 0.0869,
+      "step": 3552
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.930575233777274e-07,
+      "loss": 0.0857,
+      "step": 3553
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.87390195522811e-07,
+      "loss": 0.0819,
+      "step": 3554
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.817228676678946e-07,
+      "loss": 0.0837,
+      "step": 3555
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.760555398129782e-07,
+      "loss": 0.0874,
+      "step": 3556
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.703882119580618e-07,
+      "loss": 0.0842,
+      "step": 3557
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.647208841031454e-07,
+      "loss": 0.0825,
+      "step": 3558
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.59053556248229e-07,
+      "loss": 0.0879,
+      "step": 3559
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5338622839331254e-07,
+      "loss": 0.0709,
+      "step": 3560
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.477189005383962e-07,
+      "loss": 0.0778,
+      "step": 3561
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4205157268347973e-07,
+      "loss": 0.0867,
+      "step": 3562
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.363842448285634e-07,
+      "loss": 0.0796,
+      "step": 3563
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.3071691697364697e-07,
+      "loss": 0.0873,
+      "step": 3564
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2504958911873057e-07,
+      "loss": 0.0764,
+      "step": 3565
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1938226126381417e-07,
+      "loss": 0.077,
+      "step": 3566
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.137149334088977e-07,
+      "loss": 0.0933,
+      "step": 3567
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0804760555398136e-07,
+      "loss": 0.0891,
+      "step": 3568
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.023802776990649e-07,
+      "loss": 0.0833,
+      "step": 3569
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9671294984414855e-07,
+      "loss": 0.0832,
+      "step": 3570
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.910456219892321e-07,
+      "loss": 0.0851,
+      "step": 3571
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8537829413431574e-07,
+      "loss": 0.0774,
+      "step": 3572
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.797109662793993e-07,
+      "loss": 0.092,
+      "step": 3573
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.7404363842448293e-07,
+      "loss": 0.0848,
+      "step": 3574
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.683763105695665e-07,
+      "loss": 0.0798,
+      "step": 3575
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6270898271465007e-07,
+      "loss": 0.0825,
+      "step": 3576
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5704165485973367e-07,
+      "loss": 0.0788,
+      "step": 3577
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5137432700481726e-07,
+      "loss": 0.084,
+      "step": 3578
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.457069991499008e-07,
+      "loss": 0.0823,
+      "step": 3579
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4003967129498446e-07,
+      "loss": 0.088,
+      "step": 3580
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.34372343440068e-07,
+      "loss": 0.0834,
+      "step": 3581
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2870501558515165e-07,
+      "loss": 0.0852,
+      "step": 3582
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.230376877302352e-07,
+      "loss": 0.0859,
+      "step": 3583
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1737035987531884e-07,
+      "loss": 0.1048,
+      "step": 3584
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1170303202040244e-07,
+      "loss": 0.0873,
+      "step": 3585
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.06035704165486e-07,
+      "loss": 0.0815,
+      "step": 3586
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.003683763105696e-07,
+      "loss": 0.0829,
+      "step": 3587
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9470104845565317e-07,
+      "loss": 0.0789,
+      "step": 3588
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.8903372060073677e-07,
+      "loss": 0.0943,
+      "step": 3589
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.8336639274582036e-07,
+      "loss": 0.0865,
+      "step": 3590
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7769906489090396e-07,
+      "loss": 0.0846,
+      "step": 3591
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7203173703598755e-07,
+      "loss": 0.0742,
+      "step": 3592
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.6636440918107115e-07,
+      "loss": 0.0761,
+      "step": 3593
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.6069708132615475e-07,
+      "loss": 0.0782,
+      "step": 3594
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.5502975347123834e-07,
+      "loss": 0.0805,
+      "step": 3595
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4936242561632194e-07,
+      "loss": 0.0919,
+      "step": 3596
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.436950977614055e-07,
+      "loss": 0.0762,
+      "step": 3597
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.380277699064891e-07,
+      "loss": 0.0778,
+      "step": 3598
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.323604420515727e-07,
+      "loss": 0.0919,
+      "step": 3599
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2669311419665627e-07,
+      "loss": 0.0874,
+      "step": 3600
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2102578634173986e-07,
+      "loss": 0.0872,
+      "step": 3601
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1535845848682349e-07,
+      "loss": 0.0861,
+      "step": 3602
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0969113063190708e-07,
+      "loss": 0.0929,
+      "step": 3603
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0402380277699068e-07,
+      "loss": 0.0786,
+      "step": 3604
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9835647492207427e-07,
+      "loss": 0.0935,
+      "step": 3605
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9268914706715787e-07,
+      "loss": 0.08,
+      "step": 3606
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8702181921224147e-07,
+      "loss": 0.0772,
+      "step": 3607
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8135449135732504e-07,
+      "loss": 0.0891,
+      "step": 3608
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.7568716350240863e-07,
+      "loss": 0.0922,
+      "step": 3609
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7001983564749223e-07,
+      "loss": 0.0817,
+      "step": 3610
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6435250779257582e-07,
+      "loss": 0.0998,
+      "step": 3611
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5868517993765942e-07,
+      "loss": 0.0796,
+      "step": 3612
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.53017852082743e-07,
+      "loss": 0.0773,
+      "step": 3613
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4735052422782659e-07,
+      "loss": 0.0792,
+      "step": 3614
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4168319637291018e-07,
+      "loss": 0.0844,
+      "step": 3615
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3601586851799378e-07,
+      "loss": 0.0734,
+      "step": 3616
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3034854066307737e-07,
+      "loss": 0.0746,
+      "step": 3617
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2468121280816097e-07,
+      "loss": 0.0711,
+      "step": 3618
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1901388495324455e-07,
+      "loss": 0.0782,
+      "step": 3619
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1334655709832813e-07,
+      "loss": 0.0899,
+      "step": 3620
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0767922924341174e-07,
+      "loss": 0.084,
+      "step": 3621
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0201190138849534e-07,
+      "loss": 0.0868,
+      "step": 3622
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.634457353357894e-08,
+      "loss": 0.0809,
+      "step": 3623
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.067724567866252e-08,
+      "loss": 0.0909,
+      "step": 3624
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.500991782374611e-08,
+      "loss": 0.0783,
+      "step": 3625
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.934258996882971e-08,
+      "loss": 0.0831,
+      "step": 3626
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.367526211391329e-08,
+      "loss": 0.0851,
+      "step": 3627
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.800793425899689e-08,
+      "loss": 0.0879,
+      "step": 3628
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.234060640408048e-08,
+      "loss": 0.0894,
+      "step": 3629
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.667327854916407e-08,
+      "loss": 0.0921,
+      "step": 3630
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.100595069424767e-08,
+      "loss": 0.0878,
+      "step": 3631
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.533862283933126e-08,
+      "loss": 0.0863,
+      "step": 3632
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.9671294984414855e-08,
+      "loss": 0.0916,
+      "step": 3633
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.4003967129498444e-08,
+      "loss": 0.0783,
+      "step": 3634
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8336639274582034e-08,
+      "loss": 0.0959,
+      "step": 3635
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.266931141966563e-08,
+      "loss": 0.0791,
+      "step": 3636
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.7001983564749222e-08,
+      "loss": 0.0853,
+      "step": 3637
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1334655709832815e-08,
+      "loss": 0.0894,
+      "step": 3638
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.667327854916407e-09,
+      "loss": 0.0797,
+      "step": 3639
+    },
+    {
+      "epoch": 3.0,
+      "step": 3639,
+      "total_flos": 238259409223680.0,
+      "train_loss": 0.3776975158324573,
+      "train_runtime": 169779.6674,
+      "train_samples_per_second": 5.489,
+      "train_steps_per_second": 0.021
+    }
+  ],
+  "max_steps": 3639,
+  "num_train_epochs": 3,
+  "total_flos": 238259409223680.0,
+  "trial_name": null,
+  "trial_params": null
+}