diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7309 @@
+{
+  "best_metric": 1.1808723211288452,
+  "best_model_checkpoint": "t5-results/checkpoint-280",
+  "epoch": 17.0,
+  "global_step": 2380,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996428571428572e-05,
+      "loss": 6.6331,
+      "step": 2
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992857142857143e-05,
+      "loss": 2.8457,
+      "step": 4
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989285714285715e-05,
+      "loss": 2.425,
+      "step": 6
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.985714285714286e-05,
+      "loss": 2.7484,
+      "step": 8
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982142857142857e-05,
+      "loss": 3.3634,
+      "step": 10
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.978571428571429e-05,
+      "loss": 3.6255,
+      "step": 12
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.975e-05,
+      "loss": 1.0302,
+      "step": 14
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.971428571428572e-05,
+      "loss": 1.4391,
+      "step": 16
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.967857142857143e-05,
+      "loss": 0.9369,
+      "step": 18
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.964285714285715e-05,
+      "loss": 2.3108,
+      "step": 20
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.960714285714286e-05,
+      "loss": 1.2689,
+      "step": 22
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.957142857142857e-05,
+      "loss": 2.4004,
+      "step": 24
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.953571428571429e-05,
+      "loss": 0.9825,
+      "step": 26
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 1.2442,
+      "step": 28
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.946428571428572e-05,
+      "loss": 2.9793,
+      "step": 30
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.942857142857143e-05,
+      "loss": 3.1514,
+      "step": 32
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.939285714285714e-05,
+      "loss": 0.8941,
+      "step": 34
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.935714285714286e-05,
+      "loss": 0.9734,
+      "step": 36
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9321428571428574e-05,
+      "loss": 2.1279,
+      "step": 38
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.928571428571429e-05,
+      "loss": 0.9224,
+      "step": 40
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9250000000000004e-05,
+      "loss": 1.1355,
+      "step": 42
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.921428571428572e-05,
+      "loss": 0.938,
+      "step": 44
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.917857142857143e-05,
+      "loss": 0.9593,
+      "step": 46
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9142857142857144e-05,
+      "loss": 2.9186,
+      "step": 48
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.910714285714286e-05,
+      "loss": 0.6557,
+      "step": 50
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9071428571428574e-05,
+      "loss": 1.6295,
+      "step": 52
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.903571428571429e-05,
+      "loss": 1.048,
+      "step": 54
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9e-05,
+      "loss": 1.4319,
+      "step": 56
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.896428571428572e-05,
+      "loss": 1.0346,
+      "step": 58
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.892857142857143e-05,
+      "loss": 0.9637,
+      "step": 60
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8892857142857144e-05,
+      "loss": 1.7965,
+      "step": 62
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.885714285714286e-05,
+      "loss": 1.5955,
+      "step": 64
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8821428571428575e-05,
+      "loss": 1.081,
+      "step": 66
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.878571428571429e-05,
+      "loss": 1.1883,
+      "step": 68
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.875e-05,
+      "loss": 0.6773,
+      "step": 70
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8714285714285714e-05,
+      "loss": 1.3492,
+      "step": 72
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.867857142857143e-05,
+      "loss": 1.8582,
+      "step": 74
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8642857142857145e-05,
+      "loss": 0.9377,
+      "step": 76
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.860714285714286e-05,
+      "loss": 0.7285,
+      "step": 78
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8571428571428576e-05,
+      "loss": 1.1076,
+      "step": 80
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.853571428571429e-05,
+      "loss": 2.0815,
+      "step": 82
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.85e-05,
+      "loss": 0.4267,
+      "step": 84
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8464285714285715e-05,
+      "loss": 1.0915,
+      "step": 86
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.842857142857143e-05,
+      "loss": 1.1805,
+      "step": 88
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8392857142857146e-05,
+      "loss": 1.2944,
+      "step": 90
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.835714285714286e-05,
+      "loss": 1.2663,
+      "step": 92
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.832142857142857e-05,
+      "loss": 1.316,
+      "step": 94
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.828571428571429e-05,
+      "loss": 1.6691,
+      "step": 96
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.825e-05,
+      "loss": 1.3656,
+      "step": 98
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 1.4799,
+      "step": 100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.817857142857143e-05,
+      "loss": 1.6342,
+      "step": 102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8142857142857147e-05,
+      "loss": 1.2405,
+      "step": 104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.810714285714286e-05,
+      "loss": 1.283,
+      "step": 106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.807142857142857e-05,
+      "loss": 0.674,
+      "step": 108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.803571428571429e-05,
+      "loss": 1.1166,
+      "step": 110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8e-05,
+      "loss": 1.0384,
+      "step": 112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7964285714285717e-05,
+      "loss": 1.5797,
+      "step": 114
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7928571428571425e-05,
+      "loss": 0.792,
+      "step": 116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.789285714285715e-05,
+      "loss": 0.8006,
+      "step": 118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.785714285714286e-05,
+      "loss": 0.6637,
+      "step": 120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.782142857142857e-05,
+      "loss": 2.339,
+      "step": 122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7785714285714287e-05,
+      "loss": 0.869,
+      "step": 124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.775e-05,
+      "loss": 0.6961,
+      "step": 126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.771428571428572e-05,
+      "loss": 0.8871,
+      "step": 128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.767857142857143e-05,
+      "loss": 0.8566,
+      "step": 130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.764285714285715e-05,
+      "loss": 1.5061,
+      "step": 132
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.760714285714286e-05,
+      "loss": 1.288,
+      "step": 134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.757142857142857e-05,
+      "loss": 1.3533,
+      "step": 136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.753571428571429e-05,
+      "loss": 0.9773,
+      "step": 138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.75e-05,
+      "loss": 1.55,
+      "step": 140
+    },
+    {
+      "epoch": 1.0,
+      "eval_bleu": 0.057367108466817536,
+      "eval_loss": 1.2215405702590942,
+      "eval_runtime": 24.1171,
+      "eval_samples_per_second": 5.929,
+      "eval_steps_per_second": 1.493,
+      "step": 140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.746428571428572e-05,
+      "loss": 0.9627,
+      "step": 142
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.742857142857143e-05,
+      "loss": 1.9751,
+      "step": 144
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.739285714285714e-05,
+      "loss": 0.2509,
+      "step": 146
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.7357142857142864e-05,
+      "loss": 0.8852,
+      "step": 148
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.732142857142857e-05,
+      "loss": 0.9967,
+      "step": 150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.728571428571429e-05,
+      "loss": 0.4455,
+      "step": 152
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.7249999999999997e-05,
+      "loss": 1.1727,
+      "step": 154
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.721428571428572e-05,
+      "loss": 0.7705,
+      "step": 156
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.7178571428571434e-05,
+      "loss": 1.1398,
+      "step": 158
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.714285714285714e-05,
+      "loss": 0.945,
+      "step": 160
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.710714285714286e-05,
+      "loss": 0.547,
+      "step": 162
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.707142857142857e-05,
+      "loss": 0.5484,
+      "step": 164
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.703571428571429e-05,
+      "loss": 0.9512,
+      "step": 166
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7e-05,
+      "loss": 0.4727,
+      "step": 168
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.696428571428572e-05,
+      "loss": 0.8605,
+      "step": 170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.6928571428571435e-05,
+      "loss": 0.9351,
+      "step": 172
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.689285714285714e-05,
+      "loss": 1.2163,
+      "step": 174
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.685714285714286e-05,
+      "loss": 1.4119,
+      "step": 176
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.6821428571428574e-05,
+      "loss": 1.1117,
+      "step": 178
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.678571428571429e-05,
+      "loss": 0.6599,
+      "step": 180
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6750000000000005e-05,
+      "loss": 0.3745,
+      "step": 182
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.671428571428571e-05,
+      "loss": 0.6308,
+      "step": 184
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6678571428571435e-05,
+      "loss": 0.7839,
+      "step": 186
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6642857142857144e-05,
+      "loss": 0.8607,
+      "step": 188
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.660714285714286e-05,
+      "loss": 1.3602,
+      "step": 190
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6571428571428575e-05,
+      "loss": 1.7766,
+      "step": 192
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.653571428571429e-05,
+      "loss": 0.5767,
+      "step": 194
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 1.3011,
+      "step": 196
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6464285714285714e-05,
+      "loss": 2.1331,
+      "step": 198
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 1.1446,
+      "step": 200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.6392857142857145e-05,
+      "loss": 0.9878,
+      "step": 202
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.635714285714286e-05,
+      "loss": 0.5015,
+      "step": 204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.632142857142857e-05,
+      "loss": 0.6702,
+      "step": 206
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.628571428571429e-05,
+      "loss": 1.0258,
+      "step": 208
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 0.7996,
+      "step": 210
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.6214285714285715e-05,
+      "loss": 1.5507,
+      "step": 212
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.617857142857143e-05,
+      "loss": 0.6123,
+      "step": 214
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.6142857142857145e-05,
+      "loss": 1.2377,
+      "step": 216
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.610714285714286e-05,
+      "loss": 0.7282,
+      "step": 218
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.607142857142857e-05,
+      "loss": 0.6658,
+      "step": 220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.6035714285714285e-05,
+      "loss": 0.2666,
+      "step": 222
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.2779,
+      "step": 224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.5964285714285715e-05,
+      "loss": 1.0744,
+      "step": 226
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.592857142857143e-05,
+      "loss": 1.3369,
+      "step": 228
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.5892857142857146e-05,
+      "loss": 0.6118,
+      "step": 230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.585714285714286e-05,
+      "loss": 0.7165,
+      "step": 232
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.582142857142858e-05,
+      "loss": 0.8708,
+      "step": 234
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.5785714285714285e-05,
+      "loss": 0.8434,
+      "step": 236
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.575e-05,
+      "loss": 0.423,
+      "step": 238
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5714285714285716e-05,
+      "loss": 0.8223,
+      "step": 240
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.567857142857143e-05,
+      "loss": 0.4773,
+      "step": 242
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.564285714285714e-05,
+      "loss": 1.2649,
+      "step": 244
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.560714285714286e-05,
+      "loss": 0.5867,
+      "step": 246
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.557142857142858e-05,
+      "loss": 0.8064,
+      "step": 248
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.5535714285714286e-05,
+      "loss": 0.7751,
+      "step": 250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.55e-05,
+      "loss": 0.7568,
+      "step": 252
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.546428571428572e-05,
+      "loss": 1.3103,
+      "step": 254
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.542857142857143e-05,
+      "loss": 1.0422,
+      "step": 256
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.539285714285714e-05,
+      "loss": 1.0449,
+      "step": 258
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.5357142857142856e-05,
+      "loss": 0.1592,
+      "step": 260
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.532142857142858e-05,
+      "loss": 2.7095,
+      "step": 262
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.528571428571429e-05,
+      "loss": 0.8203,
+      "step": 264
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.525e-05,
+      "loss": 0.3886,
+      "step": 266
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.521428571428572e-05,
+      "loss": 1.0708,
+      "step": 268
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.517857142857143e-05,
+      "loss": 0.4944,
+      "step": 270
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.514285714285714e-05,
+      "loss": 1.2455,
+      "step": 272
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.510714285714286e-05,
+      "loss": 0.5608,
+      "step": 274
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.507142857142858e-05,
+      "loss": 1.7867,
+      "step": 276
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.503571428571429e-05,
+      "loss": 0.6859,
+      "step": 278
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.5e-05,
+      "loss": 1.5826,
+      "step": 280
+    },
+    {
+      "epoch": 2.0,
+      "eval_bleu": 0.040697863739456316,
+      "eval_loss": 1.1808723211288452,
+      "eval_runtime": 23.9524,
+      "eval_samples_per_second": 5.97,
+      "eval_steps_per_second": 1.503,
+      "step": 280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.496428571428571e-05,
+      "loss": 0.3028,
+      "step": 282
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.4928571428571434e-05,
+      "loss": 1.3228,
+      "step": 284
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.489285714285715e-05,
+      "loss": 0.9648,
+      "step": 286
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.485714285714286e-05,
+      "loss": 0.4298,
+      "step": 288
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.482142857142857e-05,
+      "loss": 0.7843,
+      "step": 290
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.478571428571429e-05,
+      "loss": 1.2349,
+      "step": 292
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4750000000000004e-05,
+      "loss": 0.8441,
+      "step": 294
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.471428571428571e-05,
+      "loss": 1.2609,
+      "step": 296
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.467857142857143e-05,
+      "loss": 0.6808,
+      "step": 298
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 0.2467,
+      "step": 300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.460714285714286e-05,
+      "loss": 0.7209,
+      "step": 302
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.4571428571428574e-05,
+      "loss": 0.4708,
+      "step": 304
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.453571428571429e-05,
+      "loss": 0.6394,
+      "step": 306
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 0.7535,
+      "step": 308
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.446428571428571e-05,
+      "loss": 0.4656,
+      "step": 310
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.442857142857143e-05,
+      "loss": 0.4037,
+      "step": 312
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.439285714285715e-05,
+      "loss": 1.2068,
+      "step": 314
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.435714285714286e-05,
+      "loss": 0.6649,
+      "step": 316
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.4321428571428574e-05,
+      "loss": 0.5143,
+      "step": 318
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.428571428571428e-05,
+      "loss": 0.3144,
+      "step": 320
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.4250000000000005e-05,
+      "loss": 0.6201,
+      "step": 322
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.4214285714285714e-05,
+      "loss": 1.8034,
+      "step": 324
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.417857142857143e-05,
+      "loss": 0.8533,
+      "step": 326
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4142857142857144e-05,
+      "loss": 1.0549,
+      "step": 328
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.410714285714286e-05,
+      "loss": 0.6378,
+      "step": 330
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.4071428571428575e-05,
+      "loss": 0.2709,
+      "step": 332
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.4035714285714284e-05,
+      "loss": 0.8126,
+      "step": 334
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.5166,
+      "step": 336
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.396428571428572e-05,
+      "loss": 1.1076,
+      "step": 338
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.392857142857143e-05,
+      "loss": 1.171,
+      "step": 340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.3892857142857145e-05,
+      "loss": 0.5459,
+      "step": 342
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.385714285714286e-05,
+      "loss": 0.2049,
+      "step": 344
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.3821428571428576e-05,
+      "loss": 1.0091,
+      "step": 346
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.3785714285714284e-05,
+      "loss": 0.1944,
+      "step": 348
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.375e-05,
+      "loss": 0.2146,
+      "step": 350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.371428571428572e-05,
+      "loss": 0.7686,
+      "step": 352
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.367857142857143e-05,
+      "loss": 0.6259,
+      "step": 354
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.3642857142857146e-05,
+      "loss": 1.0241,
+      "step": 356
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.3607142857142854e-05,
+      "loss": 1.3,
+      "step": 358
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.3571428571428576e-05,
+      "loss": 0.6116,
+      "step": 360
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.3535714285714285e-05,
+      "loss": 0.9921,
+      "step": 362
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.35e-05,
+      "loss": 0.804,
+      "step": 364
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3464285714285716e-05,
+      "loss": 0.4876,
+      "step": 366
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.342857142857143e-05,
+      "loss": 1.2895,
+      "step": 368
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.3392857142857146e-05,
+      "loss": 1.7439,
+      "step": 370
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.3357142857142855e-05,
+      "loss": 0.1348,
+      "step": 372
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.332142857142858e-05,
+      "loss": 0.3251,
+      "step": 374
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.328571428571429e-05,
+      "loss": 0.676,
+      "step": 376
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.325e-05,
+      "loss": 1.526,
+      "step": 378
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.3214285714285716e-05,
+      "loss": 1.3588,
+      "step": 380
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.317857142857143e-05,
+      "loss": 0.7616,
+      "step": 382
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.314285714285715e-05,
+      "loss": 0.6975,
+      "step": 384
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.3107142857142856e-05,
+      "loss": 0.6894,
+      "step": 386
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.307142857142857e-05,
+      "loss": 0.6453,
+      "step": 388
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.303571428571429e-05,
+      "loss": 0.5296,
+      "step": 390
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.3e-05,
+      "loss": 0.6604,
+      "step": 392
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.296428571428572e-05,
+      "loss": 0.5235,
+      "step": 394
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.292857142857143e-05,
+      "loss": 0.7616,
+      "step": 396
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.289285714285715e-05,
+      "loss": 0.9541,
+      "step": 398
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.3601,
+      "step": 400
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.282142857142857e-05,
+      "loss": 0.8298,
+      "step": 402
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.278571428571429e-05,
+      "loss": 0.6031,
+      "step": 404
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.275e-05,
+      "loss": 0.5512,
+      "step": 406
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.271428571428572e-05,
+      "loss": 0.3029,
+      "step": 408
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.2678571428571426e-05,
+      "loss": 0.6772,
+      "step": 410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.264285714285715e-05,
+      "loss": 0.2375,
+      "step": 412
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.260714285714286e-05,
+      "loss": 0.3395,
+      "step": 414
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.257142857142857e-05,
+      "loss": 0.2478,
+      "step": 416
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.253571428571429e-05,
+      "loss": 0.9358,
+      "step": 418
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.25e-05,
+      "loss": 1.0158,
+      "step": 420
+    },
+    {
+      "epoch": 3.0,
+      "eval_bleu": 0.05042323062396331,
+      "eval_loss": 1.191562533378601,
+      "eval_runtime": 24.0265,
+      "eval_samples_per_second": 5.952,
+      "eval_steps_per_second": 1.498,
+      "step": 420
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.246428571428572e-05,
+      "loss": 0.2094,
+      "step": 422
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.242857142857143e-05,
+      "loss": 0.4849,
+      "step": 424
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.239285714285714e-05,
+      "loss": 0.7713,
+      "step": 426
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.2357142857142864e-05,
+      "loss": 0.1941,
+      "step": 428
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.232142857142857e-05,
+      "loss": 0.4994,
+      "step": 430
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.228571428571429e-05,
+      "loss": 0.6792,
+      "step": 432
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.2250000000000004e-05,
+      "loss": 0.243,
+      "step": 434
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.221428571428572e-05,
+      "loss": 0.4933,
+      "step": 436
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.217857142857143e-05,
+      "loss": 0.6249,
+      "step": 438
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.214285714285714e-05,
+      "loss": 0.9461,
+      "step": 440
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.210714285714286e-05,
+      "loss": 0.2788,
+      "step": 442
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.2071428571428574e-05,
+      "loss": 1.0754,
+      "step": 444
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.203571428571429e-05,
+      "loss": 0.7201,
+      "step": 446
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.2e-05,
+      "loss": 0.5637,
+      "step": 448
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.196428571428572e-05,
+      "loss": 0.7228,
+      "step": 450
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.192857142857143e-05,
+      "loss": 0.9431,
+      "step": 452
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.1892857142857144e-05,
+      "loss": 1.06,
+      "step": 454
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.185714285714286e-05,
+      "loss": 0.4919,
+      "step": 456
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.1821428571428574e-05,
+      "loss": 0.2347,
+      "step": 458
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.178571428571429e-05,
+      "loss": 1.1592,
+      "step": 460
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.175e-05,
+      "loss": 1.1351,
+      "step": 462
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.1714285714285714e-05,
+      "loss": 0.1562,
+      "step": 464
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.167857142857143e-05,
+      "loss": 0.5529,
+      "step": 466
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.1642857142857144e-05,
+      "loss": 0.5436,
+      "step": 468
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.160714285714286e-05,
+      "loss": 0.4635,
+      "step": 470
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.1571428571428575e-05,
+      "loss": 1.2621,
+      "step": 472
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.153571428571429e-05,
+      "loss": 0.6432,
+      "step": 474
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.15e-05,
+      "loss": 0.8124,
+      "step": 476
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.1464285714285714e-05,
+      "loss": 0.6932,
+      "step": 478
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.1428571428571437e-05,
+      "loss": 0.2204,
+      "step": 480
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.1392857142857145e-05,
+      "loss": 1.2212,
+      "step": 482
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.135714285714286e-05,
+      "loss": 0.6411,
+      "step": 484
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.132142857142857e-05,
+      "loss": 0.557,
+      "step": 486
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.128571428571429e-05,
+      "loss": 0.2137,
+      "step": 488
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.125e-05,
+      "loss": 0.1436,
+      "step": 490
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.1214285714285715e-05,
+      "loss": 0.6621,
+      "step": 492
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.117857142857143e-05,
+      "loss": 0.6637,
+      "step": 494
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1142857142857146e-05,
+      "loss": 0.7444,
+      "step": 496
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.110714285714286e-05,
+      "loss": 0.6783,
+      "step": 498
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 0.6938,
+      "step": 500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.1035714285714285e-05,
+      "loss": 0.7404,
+      "step": 502
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.1e-05,
+      "loss": 0.3595,
+      "step": 504
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.0964285714285716e-05,
+      "loss": 1.0034,
+      "step": 506
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.092857142857143e-05,
+      "loss": 1.4737,
+      "step": 508
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.0892857142857147e-05,
+      "loss": 0.4108,
+      "step": 510
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.085714285714286e-05,
+      "loss": 0.822,
+      "step": 512
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.082142857142857e-05,
+      "loss": 0.8188,
+      "step": 514
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.0785714285714286e-05,
+      "loss": 0.6804,
+      "step": 516
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.075e-05,
+      "loss": 0.2107,
+      "step": 518
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.0714285714285717e-05,
+      "loss": 0.2722,
+      "step": 520
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.067857142857143e-05,
+      "loss": 0.508,
+      "step": 522
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.064285714285714e-05,
+      "loss": 0.7706,
+      "step": 524
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.060714285714286e-05,
+      "loss": 0.8016,
+      "step": 526
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.057142857142857e-05,
+      "loss": 0.7327,
+      "step": 528
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.0535714285714287e-05,
+      "loss": 1.1663,
+      "step": 530
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.05e-05,
+      "loss": 0.478,
+      "step": 532
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.046428571428572e-05,
+      "loss": 0.6164,
+      "step": 534
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.042857142857143e-05,
+      "loss": 0.3525,
+      "step": 536
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.039285714285714e-05,
+      "loss": 0.1825,
+      "step": 538
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.035714285714286e-05,
+      "loss": 0.2722,
+      "step": 540
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.032142857142857e-05,
+      "loss": 0.2387,
+      "step": 542
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.028571428571429e-05,
+      "loss": 0.5076,
+      "step": 544
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.025e-05,
+      "loss": 0.698,
+      "step": 546
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.021428571428572e-05,
+      "loss": 0.3564,
+      "step": 548
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.017857142857143e-05,
+      "loss": 0.4585,
+      "step": 550
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.014285714285714e-05,
+      "loss": 0.5317,
+      "step": 552
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.010714285714286e-05,
+      "loss": 0.3994,
+      "step": 554
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.007142857142857e-05,
+      "loss": 0.3532,
+      "step": 556
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.003571428571429e-05,
+      "loss": 0.231,
+      "step": 558
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4e-05,
+      "loss": 0.3985,
+      "step": 560
+    },
+    {
+      "epoch": 4.0,
+      "eval_bleu": 0.049467315645150965,
+      "eval_loss": 1.2375887632369995,
+      "eval_runtime": 24.1561,
+      "eval_samples_per_second": 5.92,
+      "eval_steps_per_second": 1.49,
+      "step": 560
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.996428571428571e-05,
+      "loss": 0.535,
+      "step": 562
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9928571428571434e-05,
+      "loss": 0.2703,
+      "step": 564
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.989285714285714e-05,
+      "loss": 0.2777,
+      "step": 566
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.985714285714286e-05,
+      "loss": 0.4745,
+      "step": 568
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.982142857142857e-05,
+      "loss": 0.1472,
+      "step": 570
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.978571428571429e-05,
+      "loss": 0.4268,
+      "step": 572
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.9750000000000004e-05,
+      "loss": 0.4603,
+      "step": 574
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.971428571428571e-05,
+      "loss": 0.2727,
+      "step": 576
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.9678571428571435e-05,
+      "loss": 0.4068,
+      "step": 578
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.964285714285714e-05,
+      "loss": 0.8799,
+      "step": 580
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.960714285714286e-05,
+      "loss": 1.206,
+      "step": 582
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.9571428571428574e-05,
+      "loss": 0.2441,
+      "step": 584
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.953571428571429e-05,
+      "loss": 0.3553,
+      "step": 586
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 0.53,
+      "step": 588
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.946428571428571e-05,
+      "loss": 1.0256,
+      "step": 590
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.942857142857143e-05,
+      "loss": 0.6089,
+      "step": 592
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.9392857142857144e-05,
+      "loss": 0.1547,
+      "step": 594
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.935714285714286e-05,
+      "loss": 0.215,
+      "step": 596
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.9321428571428575e-05,
+      "loss": 0.826,
+      "step": 598
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.2869,
+      "step": 600
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.9250000000000005e-05,
+      "loss": 0.4909,
+      "step": 602
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.9214285714285714e-05,
+      "loss": 0.6748,
+      "step": 604
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.917857142857143e-05,
+      "loss": 0.4753,
+      "step": 606
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.9142857142857145e-05,
+      "loss": 0.5192,
+      "step": 608
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.910714285714286e-05,
+      "loss": 0.4448,
+      "step": 610
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.9071428571428575e-05,
+      "loss": 0.6043,
+      "step": 612
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.9035714285714284e-05,
+      "loss": 0.9979,
+      "step": 614
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.8518,
+      "step": 616
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.8964285714285715e-05,
+      "loss": 0.6995,
+      "step": 618
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.892857142857143e-05,
+      "loss": 0.4926,
+      "step": 620
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.8892857142857145e-05,
+      "loss": 0.4764,
+      "step": 622
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.885714285714286e-05,
+      "loss": 0.2713,
+      "step": 624
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.8821428571428576e-05,
+      "loss": 0.3325,
+      "step": 626
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.8785714285714285e-05,
+      "loss": 1.1078,
+      "step": 628
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.875e-05,
+      "loss": 0.15,
+      "step": 630
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.8714285714285715e-05,
+      "loss": 0.5763,
+      "step": 632
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.867857142857143e-05,
+      "loss": 0.1159,
+      "step": 634
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.8642857142857146e-05,
+      "loss": 0.8402,
+      "step": 636
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.860714285714286e-05,
+      "loss": 0.91,
+      "step": 638
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.857142857142858e-05,
+      "loss": 0.7494,
+      "step": 640
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.8535714285714285e-05,
+      "loss": 1.9399,
+      "step": 642
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.85e-05,
+      "loss": 0.4523,
+      "step": 644
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.8464285714285716e-05,
+      "loss": 0.5572,
+      "step": 646
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.842857142857143e-05,
+      "loss": 0.1774,
+      "step": 648
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.839285714285715e-05,
+      "loss": 1.0661,
+      "step": 650
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.8357142857142855e-05,
+      "loss": 0.3178,
+      "step": 652
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.832142857142858e-05,
+      "loss": 0.2693,
+      "step": 654
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.8285714285714286e-05,
+      "loss": 0.1975,
+      "step": 656
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.825e-05,
+      "loss": 0.8066,
+      "step": 658
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.821428571428572e-05,
+      "loss": 0.3654,
+      "step": 660
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.817857142857143e-05,
+      "loss": 0.2776,
+      "step": 662
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.814285714285715e-05,
+      "loss": 0.3903,
+      "step": 664
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.8107142857142856e-05,
+      "loss": 0.1636,
+      "step": 666
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.807142857142857e-05,
+      "loss": 0.1759,
+      "step": 668
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.803571428571429e-05,
+      "loss": 0.87,
+      "step": 670
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.8e-05,
+      "loss": 0.5844,
+      "step": 672
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.796428571428571e-05,
+      "loss": 0.2741,
+      "step": 674
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.792857142857143e-05,
+      "loss": 0.1659,
+      "step": 676
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.789285714285715e-05,
+      "loss": 0.7734,
+      "step": 678
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.785714285714286e-05,
+      "loss": 0.0794,
+      "step": 680
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.782142857142857e-05,
+      "loss": 0.3805,
+      "step": 682
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.778571428571429e-05,
+      "loss": 0.0704,
+      "step": 684
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.775e-05,
+      "loss": 0.5585,
+      "step": 686
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.771428571428572e-05,
+      "loss": 0.1195,
+      "step": 688
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.767857142857143e-05,
+      "loss": 0.3252,
+      "step": 690
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.764285714285715e-05,
+      "loss": 0.5087,
+      "step": 692
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.760714285714286e-05,
+      "loss": 0.7044,
+      "step": 694
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.757142857142857e-05,
+      "loss": 0.3706,
+      "step": 696
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.753571428571429e-05,
+      "loss": 0.3268,
+      "step": 698
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.2792,
+      "step": 700
+    },
+    {
+      "epoch": 5.0,
+      "eval_bleu": 0.049270557108538345,
+      "eval_loss": 1.3052771091461182,
+      "eval_runtime": 24.2183,
+      "eval_samples_per_second": 5.905,
+      "eval_steps_per_second": 1.486,
+      "step": 700
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.746428571428572e-05,
+      "loss": 0.92,
+      "step": 702
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.742857142857143e-05,
+      "loss": 0.4549,
+      "step": 704
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.739285714285714e-05,
+      "loss": 0.8532,
+      "step": 706
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.735714285714286e-05,
+      "loss": 0.1456,
+      "step": 708
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.7321428571428573e-05,
+      "loss": 0.6135,
+      "step": 710
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.728571428571428e-05,
+      "loss": 0.2827,
+      "step": 712
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.7250000000000004e-05,
+      "loss": 1.2423,
+      "step": 714
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.721428571428572e-05,
+      "loss": 0.4182,
+      "step": 716
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.717857142857143e-05,
+      "loss": 0.4483,
+      "step": 718
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.7142857142857143e-05,
+      "loss": 0.2988,
+      "step": 720
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.710714285714286e-05,
+      "loss": 0.3521,
+      "step": 722
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.7071428571428574e-05,
+      "loss": 0.5617,
+      "step": 724
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.703571428571429e-05,
+      "loss": 0.4719,
+      "step": 726
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.7e-05,
+      "loss": 0.4596,
+      "step": 728
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.696428571428572e-05,
+      "loss": 0.3088,
+      "step": 730
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.692857142857143e-05,
+      "loss": 1.048,
+      "step": 732
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.6892857142857144e-05,
+      "loss": 0.365,
+      "step": 734
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.685714285714286e-05,
+      "loss": 1.3097,
+      "step": 736
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.6821428571428575e-05,
+      "loss": 0.3563,
+      "step": 738
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.678571428571429e-05,
+      "loss": 0.3085,
+      "step": 740
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.675e-05,
+      "loss": 0.1913,
+      "step": 742
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.671428571428572e-05,
+      "loss": 0.4925,
+      "step": 744
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.667857142857143e-05,
+      "loss": 0.5896,
+      "step": 746
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.6642857142857145e-05,
+      "loss": 0.7646,
+      "step": 748
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.6607142857142853e-05,
+      "loss": 0.0855,
+      "step": 750
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.6571428571428576e-05,
+      "loss": 0.2791,
+      "step": 752
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.653571428571429e-05,
+      "loss": 0.3042,
+      "step": 754
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.65e-05,
+      "loss": 0.3593,
+      "step": 756
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.6464285714285715e-05,
+      "loss": 0.2284,
+      "step": 758
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.642857142857143e-05,
+      "loss": 0.6942,
+      "step": 760
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.6392857142857146e-05,
+      "loss": 0.2439,
+      "step": 762
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.6357142857142854e-05,
+      "loss": 1.0947,
+      "step": 764
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.6321428571428576e-05,
+      "loss": 0.5959,
+      "step": 766
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.628571428571429e-05,
+      "loss": 0.6679,
+      "step": 768
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.625e-05,
+      "loss": 0.2515,
+      "step": 770
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.6214285714285716e-05,
+      "loss": 0.2723,
+      "step": 772
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.617857142857143e-05,
+      "loss": 0.4045,
+      "step": 774
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.6142857142857146e-05,
+      "loss": 0.1736,
+      "step": 776
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.610714285714286e-05,
+      "loss": 0.795,
+      "step": 778
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.607142857142857e-05,
+      "loss": 0.2802,
+      "step": 780
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.603571428571429e-05,
+      "loss": 0.1724,
+      "step": 782
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.6e-05,
+      "loss": 0.4669,
+      "step": 784
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.5964285714285716e-05,
+      "loss": 0.3666,
+      "step": 786
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.5928571428571425e-05,
+      "loss": 0.6533,
+      "step": 788
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.589285714285715e-05,
+      "loss": 0.6425,
+      "step": 790
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.585714285714286e-05,
+      "loss": 0.3825,
+      "step": 792
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.582142857142857e-05,
+      "loss": 0.2015,
+      "step": 794
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.5785714285714286e-05,
+      "loss": 0.0865,
+      "step": 796
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.575e-05,
+      "loss": 0.2579,
+      "step": 798
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.2077,
+      "step": 800
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.5678571428571426e-05,
+      "loss": 0.215,
+      "step": 802
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.564285714285715e-05,
+      "loss": 0.405,
+      "step": 804
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.560714285714286e-05,
+      "loss": 0.2575,
+      "step": 806
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.557142857142857e-05,
+      "loss": 0.2542,
+      "step": 808
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.553571428571429e-05,
+      "loss": 0.3022,
+      "step": 810
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.55e-05,
+      "loss": 0.2097,
+      "step": 812
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.546428571428572e-05,
+      "loss": 0.1469,
+      "step": 814
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.5428571428571426e-05,
+      "loss": 0.3287,
+      "step": 816
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.539285714285714e-05,
+      "loss": 0.2035,
+      "step": 818
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.5357142857142864e-05,
+      "loss": 0.5892,
+      "step": 820
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.532142857142857e-05,
+      "loss": 0.7407,
+      "step": 822
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.528571428571429e-05,
+      "loss": 0.8624,
+      "step": 824
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.525e-05,
+      "loss": 0.4122,
+      "step": 826
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.521428571428572e-05,
+      "loss": 0.1023,
+      "step": 828
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.5178571428571434e-05,
+      "loss": 0.2469,
+      "step": 830
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.514285714285714e-05,
+      "loss": 0.7497,
+      "step": 832
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.510714285714286e-05,
+      "loss": 0.56,
+      "step": 834
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.507142857142857e-05,
+      "loss": 0.2468,
+      "step": 836
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.503571428571429e-05,
+      "loss": 0.0324,
+      "step": 838
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.5e-05,
+      "loss": 0.1898,
+      "step": 840
+    },
+    {
+      "epoch": 6.0,
+      "eval_bleu": 0.05567447568279575,
+      "eval_loss": 1.3495681285858154,
+      "eval_runtime": 24.0884,
+      "eval_samples_per_second": 5.936,
+      "eval_steps_per_second": 1.494,
+      "step": 840
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.496428571428572e-05,
+      "loss": 0.5831,
+      "step": 842
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.4928571428571434e-05,
+      "loss": 0.2588,
+      "step": 844
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.489285714285714e-05,
+      "loss": 0.7811,
+      "step": 846
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.485714285714286e-05,
+      "loss": 0.3759,
+      "step": 848
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.4821428571428574e-05,
+      "loss": 0.1458,
+      "step": 850
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.478571428571429e-05,
+      "loss": 0.1662,
+      "step": 852
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.475e-05,
+      "loss": 0.2273,
+      "step": 854
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.471428571428571e-05,
+      "loss": 0.079,
+      "step": 856
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.4678571428571435e-05,
+      "loss": 0.5959,
+      "step": 858
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.4642857142857144e-05,
+      "loss": 0.2343,
+      "step": 860
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.460714285714286e-05,
+      "loss": 0.0758,
+      "step": 862
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.4571428571428574e-05,
+      "loss": 0.4292,
+      "step": 864
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.453571428571429e-05,
+      "loss": 0.1595,
+      "step": 866
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.45e-05,
+      "loss": 0.7137,
+      "step": 868
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.4464285714285714e-05,
+      "loss": 0.099,
+      "step": 870
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.442857142857143e-05,
+      "loss": 0.2587,
+      "step": 872
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.4392857142857144e-05,
+      "loss": 0.2536,
+      "step": 874
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.435714285714286e-05,
+      "loss": 0.2399,
+      "step": 876
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.432142857142857e-05,
+      "loss": 0.357,
+      "step": 878
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.428571428571429e-05,
+      "loss": 0.3433,
+      "step": 880
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.4250000000000006e-05,
+      "loss": 0.1991,
+      "step": 882
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.4214285714285714e-05,
+      "loss": 0.1974,
+      "step": 884
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.417857142857143e-05,
+      "loss": 0.198,
+      "step": 886
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.4142857142857145e-05,
+      "loss": 0.7005,
+      "step": 888
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.410714285714286e-05,
+      "loss": 0.7801,
+      "step": 890
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.407142857142857e-05,
+      "loss": 0.3049,
+      "step": 892
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.4035714285714284e-05,
+      "loss": 0.116,
+      "step": 894
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.3387,
+      "step": 896
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.3964285714285715e-05,
+      "loss": 0.6693,
+      "step": 898
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.9696,
+      "step": 900
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.3892857142857146e-05,
+      "loss": 0.306,
+      "step": 902
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.385714285714286e-05,
+      "loss": 0.2039,
+      "step": 904
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.382142857142857e-05,
+      "loss": 0.8002,
+      "step": 906
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.3785714285714285e-05,
+      "loss": 0.4355,
+      "step": 908
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.375000000000001e-05,
+      "loss": 0.2871,
+      "step": 910
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.3714285714285716e-05,
+      "loss": 0.096,
+      "step": 912
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.367857142857143e-05,
+      "loss": 0.1461,
+      "step": 914
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.364285714285714e-05,
+      "loss": 0.6932,
+      "step": 916
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.360714285714286e-05,
+      "loss": 0.1075,
+      "step": 918
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.357142857142857e-05,
+      "loss": 0.4688,
+      "step": 920
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.3535714285714286e-05,
+      "loss": 0.2818,
+      "step": 922
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.35e-05,
+      "loss": 0.1273,
+      "step": 924
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.3464285714285716e-05,
+      "loss": 0.3796,
+      "step": 926
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.342857142857143e-05,
+      "loss": 0.1439,
+      "step": 928
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.339285714285714e-05,
+      "loss": 0.3496,
+      "step": 930
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.3357142857142856e-05,
+      "loss": 0.5605,
+      "step": 932
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.332142857142858e-05,
+      "loss": 0.4569,
+      "step": 934
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.3285714285714286e-05,
+      "loss": 0.236,
+      "step": 936
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.325e-05,
+      "loss": 0.6998,
+      "step": 938
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.321428571428572e-05,
+      "loss": 0.1748,
+      "step": 940
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.317857142857143e-05,
+      "loss": 0.1404,
+      "step": 942
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.314285714285714e-05,
+      "loss": 0.1987,
+      "step": 944
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.3107142857142856e-05,
+      "loss": 0.2669,
+      "step": 946
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.307142857142858e-05,
+      "loss": 0.6321,
+      "step": 948
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.303571428571429e-05,
+      "loss": 0.3818,
+      "step": 950
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.3e-05,
+      "loss": 0.1348,
+      "step": 952
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.296428571428571e-05,
+      "loss": 1.1328,
+      "step": 954
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.292857142857143e-05,
+      "loss": 0.2837,
+      "step": 956
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.289285714285714e-05,
+      "loss": 0.6263,
+      "step": 958
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.285714285714286e-05,
+      "loss": 0.5201,
+      "step": 960
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.282142857142857e-05,
+      "loss": 0.3849,
+      "step": 962
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.278571428571429e-05,
+      "loss": 0.4097,
+      "step": 964
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.275e-05,
+      "loss": 0.4189,
+      "step": 966
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.271428571428571e-05,
+      "loss": 0.263,
+      "step": 968
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.2678571428571434e-05,
+      "loss": 0.0729,
+      "step": 970
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.264285714285714e-05,
+      "loss": 0.1987,
+      "step": 972
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.260714285714286e-05,
+      "loss": 0.2635,
+      "step": 974
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.257142857142857e-05,
+      "loss": 0.0468,
+      "step": 976
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.253571428571429e-05,
+      "loss": 0.4114,
+      "step": 978
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.4633,
+      "step": 980
+    },
+    {
+      "epoch": 7.0,
+      "eval_bleu": 0.05873296658447439,
+      "eval_loss": 1.4294265508651733,
+      "eval_runtime": 24.0892,
+      "eval_samples_per_second": 5.936,
+      "eval_steps_per_second": 1.494,
+      "step": 980
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.246428571428571e-05,
+      "loss": 0.3729,
+      "step": 982
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.242857142857143e-05,
+      "loss": 0.2422,
+      "step": 984
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.239285714285715e-05,
+      "loss": 0.7788,
+      "step": 986
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.235714285714286e-05,
+      "loss": 0.1707,
+      "step": 988
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.2321428571428574e-05,
+      "loss": 0.9828,
+      "step": 990
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.228571428571428e-05,
+      "loss": 0.3454,
+      "step": 992
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.2250000000000005e-05,
+      "loss": 0.6823,
+      "step": 994
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.221428571428571e-05,
+      "loss": 0.1005,
+      "step": 996
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.217857142857143e-05,
+      "loss": 0.084,
+      "step": 998
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.5005,
+      "step": 1000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.210714285714286e-05,
+      "loss": 0.2662,
+      "step": 1002
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.2071428571428575e-05,
+      "loss": 0.1154,
+      "step": 1004
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.203571428571428e-05,
+      "loss": 0.2506,
+      "step": 1006
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.4475,
+      "step": 1008
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.1964285714285714e-05,
+      "loss": 0.1707,
+      "step": 1010
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.192857142857143e-05,
+      "loss": 0.3593,
+      "step": 1012
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.1892857142857145e-05,
+      "loss": 0.7034,
+      "step": 1014
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.185714285714286e-05,
+      "loss": 0.1787,
+      "step": 1016
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.1821428571428575e-05,
+      "loss": 0.2277,
+      "step": 1018
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.1785714285714284e-05,
+      "loss": 0.2945,
+      "step": 1020
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.175e-05,
+      "loss": 0.1308,
+      "step": 1022
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.1714285714285715e-05,
+      "loss": 0.3174,
+      "step": 1024
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.167857142857143e-05,
+      "loss": 0.2672,
+      "step": 1026
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.1642857142857145e-05,
+      "loss": 0.1295,
+      "step": 1028
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.160714285714286e-05,
+      "loss": 0.3474,
+      "step": 1030
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.1571428571428576e-05,
+      "loss": 0.0681,
+      "step": 1032
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.1535714285714285e-05,
+      "loss": 0.4001,
+      "step": 1034
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.15e-05,
+      "loss": 0.3372,
+      "step": 1036
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.1464285714285715e-05,
+      "loss": 0.2988,
+      "step": 1038
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.142857142857143e-05,
+      "loss": 0.3184,
+      "step": 1040
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.1392857142857146e-05,
+      "loss": 0.2613,
+      "step": 1042
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.1357142857142855e-05,
+      "loss": 0.9791,
+      "step": 1044
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.132142857142858e-05,
+      "loss": 0.1734,
+      "step": 1046
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.1285714285714285e-05,
+      "loss": 0.4828,
+      "step": 1048
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.125e-05,
+      "loss": 0.1095,
+      "step": 1050
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.1214285714285716e-05,
+      "loss": 0.3117,
+      "step": 1052
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.117857142857143e-05,
+      "loss": 0.2029,
+      "step": 1054
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.114285714285715e-05,
+      "loss": 0.0962,
+      "step": 1056
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.1107142857142855e-05,
+      "loss": 0.195,
+      "step": 1058
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.107142857142857e-05,
+      "loss": 0.1416,
+      "step": 1060
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.1035714285714286e-05,
+      "loss": 0.0779,
+      "step": 1062
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.1e-05,
+      "loss": 0.2556,
+      "step": 1064
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.096428571428572e-05,
+      "loss": 0.355,
+      "step": 1066
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.092857142857143e-05,
+      "loss": 0.2322,
+      "step": 1068
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.089285714285715e-05,
+      "loss": 0.184,
+      "step": 1070
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.0857142857142856e-05,
+      "loss": 0.3164,
+      "step": 1072
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.082142857142857e-05,
+      "loss": 0.1381,
+      "step": 1074
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.078571428571429e-05,
+      "loss": 0.111,
+      "step": 1076
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.075e-05,
+      "loss": 0.1216,
+      "step": 1078
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.071428571428572e-05,
+      "loss": 0.3364,
+      "step": 1080
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.0678571428571426e-05,
+      "loss": 0.5024,
+      "step": 1082
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.064285714285715e-05,
+      "loss": 0.0239,
+      "step": 1084
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.060714285714286e-05,
+      "loss": 0.2569,
+      "step": 1086
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.057142857142857e-05,
+      "loss": 0.3466,
+      "step": 1088
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.053571428571429e-05,
+      "loss": 0.0734,
+      "step": 1090
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.05e-05,
+      "loss": 0.4764,
+      "step": 1092
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.0464285714285718e-05,
+      "loss": 0.4657,
+      "step": 1094
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.042857142857143e-05,
+      "loss": 0.5741,
+      "step": 1096
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.0392857142857145e-05,
+      "loss": 0.5995,
+      "step": 1098
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.6358,
+      "step": 1100
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.0321428571428573e-05,
+      "loss": 1.0986,
+      "step": 1102
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.0285714285714288e-05,
+      "loss": 0.0586,
+      "step": 1104
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.025e-05,
+      "loss": 0.2024,
+      "step": 1106
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.021428571428572e-05,
+      "loss": 0.2035,
+      "step": 1108
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.0178571428571427e-05,
+      "loss": 0.1783,
+      "step": 1110
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.0142857142857146e-05,
+      "loss": 0.2572,
+      "step": 1112
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.0107142857142855e-05,
+      "loss": 0.2595,
+      "step": 1114
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.0071428571428573e-05,
+      "loss": 0.2294,
+      "step": 1116
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.003571428571429e-05,
+      "loss": 0.3029,
+      "step": 1118
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3e-05,
+      "loss": 0.0778,
+      "step": 1120
+    },
+    {
+      "epoch": 8.0,
+      "eval_bleu": 0.05317877774724001,
+      "eval_loss": 1.5292445421218872,
+      "eval_runtime": 24.0525,
+      "eval_samples_per_second": 5.945,
+      "eval_steps_per_second": 1.497,
+      "step": 1120
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.9964285714285716e-05,
+      "loss": 0.3218,
+      "step": 1122
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.9928571428571428e-05,
+      "loss": 0.4281,
+      "step": 1124
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.9892857142857143e-05,
+      "loss": 0.1819,
+      "step": 1126
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.9857142857142862e-05,
+      "loss": 0.28,
+      "step": 1128
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.982142857142857e-05,
+      "loss": 0.087,
+      "step": 1130
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.978571428571429e-05,
+      "loss": 0.1645,
+      "step": 1132
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.975e-05,
+      "loss": 0.5545,
+      "step": 1134
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.9714285714285717e-05,
+      "loss": 0.7893,
+      "step": 1136
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.967857142857143e-05,
+      "loss": 0.2808,
+      "step": 1138
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.9642857142857144e-05,
+      "loss": 0.0783,
+      "step": 1140
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.960714285714286e-05,
+      "loss": 0.3364,
+      "step": 1142
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.957142857142857e-05,
+      "loss": 0.2115,
+      "step": 1144
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.953571428571429e-05,
+      "loss": 0.1666,
+      "step": 1146
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.95e-05,
+      "loss": 0.237,
+      "step": 1148
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.9464285714285718e-05,
+      "loss": 0.5135,
+      "step": 1150
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.9428571428571426e-05,
+      "loss": 0.2276,
+      "step": 1152
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.9392857142857145e-05,
+      "loss": 0.2953,
+      "step": 1154
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.935714285714286e-05,
+      "loss": 0.3382,
+      "step": 1156
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.9321428571428572e-05,
+      "loss": 0.197,
+      "step": 1158
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.9285714285714288e-05,
+      "loss": 0.1953,
+      "step": 1160
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.925e-05,
+      "loss": 0.3356,
+      "step": 1162
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.9214285714285715e-05,
+      "loss": 0.1067,
+      "step": 1164
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.9178571428571427e-05,
+      "loss": 0.0125,
+      "step": 1166
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.9142857142857146e-05,
+      "loss": 0.313,
+      "step": 1168
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.910714285714286e-05,
+      "loss": 0.3181,
+      "step": 1170
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.9071428571428573e-05,
+      "loss": 0.4426,
+      "step": 1172
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.9035714285714288e-05,
+      "loss": 0.2431,
+      "step": 1174
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.9e-05,
+      "loss": 0.0788,
+      "step": 1176
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.8964285714285716e-05,
+      "loss": 0.1517,
+      "step": 1178
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.8928571428571434e-05,
+      "loss": 0.3278,
+      "step": 1180
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.8892857142857143e-05,
+      "loss": 0.3431,
+      "step": 1182
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.885714285714286e-05,
+      "loss": 0.0974,
+      "step": 1184
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.882142857142857e-05,
+      "loss": 0.3486,
+      "step": 1186
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.878571428571429e-05,
+      "loss": 0.1011,
+      "step": 1188
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.8749999999999997e-05,
+      "loss": 0.1187,
+      "step": 1190
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.8714285714285716e-05,
+      "loss": 0.1652,
+      "step": 1192
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.867857142857143e-05,
+      "loss": 0.4376,
+      "step": 1194
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.8642857142857144e-05,
+      "loss": 0.0691,
+      "step": 1196
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.860714285714286e-05,
+      "loss": 0.5073,
+      "step": 1198
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.063,
+      "step": 1200
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.8535714285714286e-05,
+      "loss": 0.859,
+      "step": 1202
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 0.1316,
+      "step": 1204
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.8464285714285717e-05,
+      "loss": 0.0826,
+      "step": 1206
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.8428571428571432e-05,
+      "loss": 1.2901,
+      "step": 1208
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.8392857142857144e-05,
+      "loss": 0.132,
+      "step": 1210
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.835714285714286e-05,
+      "loss": 0.1185,
+      "step": 1212
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.832142857142857e-05,
+      "loss": 0.6,
+      "step": 1214
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.8285714285714287e-05,
+      "loss": 0.1606,
+      "step": 1216
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.825e-05,
+      "loss": 0.0629,
+      "step": 1218
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.8214285714285714e-05,
+      "loss": 0.5283,
+      "step": 1220
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.8178571428571433e-05,
+      "loss": 0.0899,
+      "step": 1222
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.814285714285714e-05,
+      "loss": 0.4774,
+      "step": 1224
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.810714285714286e-05,
+      "loss": 0.2202,
+      "step": 1226
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.8071428571428572e-05,
+      "loss": 0.1489,
+      "step": 1228
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.8035714285714288e-05,
+      "loss": 0.3167,
+      "step": 1230
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.2051,
+      "step": 1232
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.7964285714285715e-05,
+      "loss": 0.0927,
+      "step": 1234
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.792857142857143e-05,
+      "loss": 0.1964,
+      "step": 1236
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.7892857142857142e-05,
+      "loss": 0.3943,
+      "step": 1238
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.785714285714286e-05,
+      "loss": 0.1087,
+      "step": 1240
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.782142857142857e-05,
+      "loss": 0.3613,
+      "step": 1242
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.778571428571429e-05,
+      "loss": 0.029,
+      "step": 1244
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.7750000000000004e-05,
+      "loss": 0.1111,
+      "step": 1246
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.7714285714285716e-05,
+      "loss": 0.0562,
+      "step": 1248
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.767857142857143e-05,
+      "loss": 0.0307,
+      "step": 1250
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.7642857142857143e-05,
+      "loss": 0.4078,
+      "step": 1252
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.760714285714286e-05,
+      "loss": 0.19,
+      "step": 1254
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.757142857142857e-05,
+      "loss": 0.2553,
+      "step": 1256
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.7535714285714286e-05,
+      "loss": 0.0932,
+      "step": 1258
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.272,
+      "step": 1260
+    },
+    {
+      "epoch": 9.0,
+      "eval_bleu": 0.05657669082267582,
+      "eval_loss": 1.5999521017074585,
+      "eval_runtime": 24.2967,
+      "eval_samples_per_second": 5.886,
+      "eval_steps_per_second": 1.482,
+      "step": 1260
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.7464285714285713e-05,
+      "loss": 0.1858,
+      "step": 1262
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.742857142857143e-05,
+      "loss": 0.0688,
+      "step": 1264
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.7392857142857144e-05,
+      "loss": 1.1418,
+      "step": 1266
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.735714285714286e-05,
+      "loss": 0.1507,
+      "step": 1268
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.732142857142857e-05,
+      "loss": 0.2857,
+      "step": 1270
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7285714285714286e-05,
+      "loss": 0.2532,
+      "step": 1272
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.725e-05,
+      "loss": 0.1835,
+      "step": 1274
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.7214285714285714e-05,
+      "loss": 0.0759,
+      "step": 1276
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.7178571428571432e-05,
+      "loss": 0.0942,
+      "step": 1278
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.714285714285714e-05,
+      "loss": 0.177,
+      "step": 1280
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.710714285714286e-05,
+      "loss": 0.2376,
+      "step": 1282
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7071428571428575e-05,
+      "loss": 0.1879,
+      "step": 1284
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.7035714285714287e-05,
+      "loss": 0.0583,
+      "step": 1286
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.2176,
+      "step": 1288
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.6964285714285714e-05,
+      "loss": 0.3261,
+      "step": 1290
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.692857142857143e-05,
+      "loss": 0.1374,
+      "step": 1292
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.689285714285714e-05,
+      "loss": 0.3289,
+      "step": 1294
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.6857142857142857e-05,
+      "loss": 0.8159,
+      "step": 1296
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.6821428571428576e-05,
+      "loss": 0.0667,
+      "step": 1298
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 0.1363,
+      "step": 1300
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.6750000000000003e-05,
+      "loss": 0.1275,
+      "step": 1302
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.6714285714285715e-05,
+      "loss": 0.1353,
+      "step": 1304
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.667857142857143e-05,
+      "loss": 0.1045,
+      "step": 1306
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.6642857142857142e-05,
+      "loss": 0.5811,
+      "step": 1308
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.6607142857142858e-05,
+      "loss": 0.0586,
+      "step": 1310
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.6571428571428576e-05,
+      "loss": 0.2667,
+      "step": 1312
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.6535714285714285e-05,
+      "loss": 0.0402,
+      "step": 1314
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 0.1912,
+      "step": 1316
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.6464285714285712e-05,
+      "loss": 0.3051,
+      "step": 1318
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.642857142857143e-05,
+      "loss": 0.183,
+      "step": 1320
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.6392857142857143e-05,
+      "loss": 0.1257,
+      "step": 1322
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.635714285714286e-05,
+      "loss": 0.079,
+      "step": 1324
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.6321428571428574e-05,
+      "loss": 0.1727,
+      "step": 1326
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.6285714285714286e-05,
+      "loss": 0.1615,
+      "step": 1328
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.625e-05,
+      "loss": 0.0789,
+      "step": 1330
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.6214285714285713e-05,
+      "loss": 0.0782,
+      "step": 1332
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.617857142857143e-05,
+      "loss": 0.1698,
+      "step": 1334
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.6142857142857147e-05,
+      "loss": 0.5557,
+      "step": 1336
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.610714285714286e-05,
+      "loss": 0.0514,
+      "step": 1338
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.6071428571428574e-05,
+      "loss": 0.2548,
+      "step": 1340
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.6035714285714286e-05,
+      "loss": 0.0715,
+      "step": 1342
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.0556,
+      "step": 1344
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.5964285714285714e-05,
+      "loss": 0.1593,
+      "step": 1346
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.592857142857143e-05,
+      "loss": 0.1083,
+      "step": 1348
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.5892857142857148e-05,
+      "loss": 0.2346,
+      "step": 1350
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.5857142857142856e-05,
+      "loss": 0.1949,
+      "step": 1352
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.5821428571428575e-05,
+      "loss": 0.2366,
+      "step": 1354
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.5785714285714284e-05,
+      "loss": 0.4383,
+      "step": 1356
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.5750000000000002e-05,
+      "loss": 0.677,
+      "step": 1358
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.5714285714285714e-05,
+      "loss": 0.3252,
+      "step": 1360
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.567857142857143e-05,
+      "loss": 0.0591,
+      "step": 1362
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.5642857142857145e-05,
+      "loss": 0.177,
+      "step": 1364
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.5607142857142857e-05,
+      "loss": 0.3295,
+      "step": 1366
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.5571428571428572e-05,
+      "loss": 0.1593,
+      "step": 1368
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.5535714285714284e-05,
+      "loss": 0.1345,
+      "step": 1370
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 0.1458,
+      "step": 1372
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.5464285714285712e-05,
+      "loss": 0.0444,
+      "step": 1374
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.542857142857143e-05,
+      "loss": 0.4822,
+      "step": 1376
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.5392857142857146e-05,
+      "loss": 0.2425,
+      "step": 1378
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.5357142857142858e-05,
+      "loss": 0.206,
+      "step": 1380
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.5321428571428573e-05,
+      "loss": 0.1024,
+      "step": 1382
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.5285714285714285e-05,
+      "loss": 0.5527,
+      "step": 1384
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.525e-05,
+      "loss": 0.3467,
+      "step": 1386
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.521428571428572e-05,
+      "loss": 0.1362,
+      "step": 1388
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.5178571428571428e-05,
+      "loss": 0.0871,
+      "step": 1390
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.5142857142857147e-05,
+      "loss": 0.2885,
+      "step": 1392
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.510714285714286e-05,
+      "loss": 0.1643,
+      "step": 1394
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.5071428571428574e-05,
+      "loss": 0.1183,
+      "step": 1396
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.5035714285714286e-05,
+      "loss": 0.928,
+      "step": 1398
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.155,
+      "step": 1400
+    },
+    {
+      "epoch": 10.0,
+      "eval_bleu": 0.059813399926394645,
+      "eval_loss": 1.6736992597579956,
+      "eval_runtime": 24.1978,
+      "eval_samples_per_second": 5.91,
+      "eval_steps_per_second": 1.488,
+      "step": 1400
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.4964285714285717e-05,
+      "loss": 0.2578,
+      "step": 1402
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.492857142857143e-05,
+      "loss": 0.0821,
+      "step": 1404
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.4892857142857144e-05,
+      "loss": 0.1761,
+      "step": 1406
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 2.485714285714286e-05,
+      "loss": 0.105,
+      "step": 1408
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.4821428571428575e-05,
+      "loss": 0.0705,
+      "step": 1410
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.4785714285714287e-05,
+      "loss": 0.0522,
+      "step": 1412
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.4750000000000002e-05,
+      "loss": 0.1345,
+      "step": 1414
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.4714285714285714e-05,
+      "loss": 0.1109,
+      "step": 1416
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.467857142857143e-05,
+      "loss": 0.1266,
+      "step": 1418
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.4642857142857145e-05,
+      "loss": 0.347,
+      "step": 1420
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.460714285714286e-05,
+      "loss": 0.1939,
+      "step": 1422
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.4571428571428572e-05,
+      "loss": 0.3706,
+      "step": 1424
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.4535714285714287e-05,
+      "loss": 0.1594,
+      "step": 1426
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.45e-05,
+      "loss": 0.0764,
+      "step": 1428
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.4464285714285715e-05,
+      "loss": 0.2753,
+      "step": 1430
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.442857142857143e-05,
+      "loss": 0.2085,
+      "step": 1432
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.4392857142857145e-05,
+      "loss": 0.9499,
+      "step": 1434
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.4357142857142857e-05,
+      "loss": 0.1476,
+      "step": 1436
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.4321428571428573e-05,
+      "loss": 0.0417,
+      "step": 1438
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.4285714285714288e-05,
+      "loss": 0.434,
+      "step": 1440
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.425e-05,
+      "loss": 0.2271,
+      "step": 1442
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.4214285714285715e-05,
+      "loss": 0.1167,
+      "step": 1444
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.417857142857143e-05,
+      "loss": 0.0317,
+      "step": 1446
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4142857142857146e-05,
+      "loss": 0.0179,
+      "step": 1448
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.4107142857142858e-05,
+      "loss": 0.3724,
+      "step": 1450
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.4071428571428573e-05,
+      "loss": 0.0523,
+      "step": 1452
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.4035714285714285e-05,
+      "loss": 0.2323,
+      "step": 1454
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.4e-05,
+      "loss": 0.0689,
+      "step": 1456
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.3964285714285713e-05,
+      "loss": 0.0494,
+      "step": 1458
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.392857142857143e-05,
+      "loss": 0.1341,
+      "step": 1460
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.3892857142857143e-05,
+      "loss": 0.0609,
+      "step": 1462
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.385714285714286e-05,
+      "loss": 0.0551,
+      "step": 1464
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.3821428571428574e-05,
+      "loss": 0.8951,
+      "step": 1466
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.3785714285714286e-05,
+      "loss": 0.4101,
+      "step": 1468
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.375e-05,
+      "loss": 0.1412,
+      "step": 1470
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.3714285714285717e-05,
+      "loss": 0.0948,
+      "step": 1472
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.3678571428571432e-05,
+      "loss": 0.0951,
+      "step": 1474
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.3642857142857144e-05,
+      "loss": 0.2403,
+      "step": 1476
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.360714285714286e-05,
+      "loss": 0.1554,
+      "step": 1478
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.357142857142857e-05,
+      "loss": 0.0315,
+      "step": 1480
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.3535714285714287e-05,
+      "loss": 0.1061,
+      "step": 1482
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.35e-05,
+      "loss": 0.2534,
+      "step": 1484
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.3464285714285717e-05,
+      "loss": 0.5222,
+      "step": 1486
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.342857142857143e-05,
+      "loss": 0.2921,
+      "step": 1488
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.3392857142857145e-05,
+      "loss": 0.1921,
+      "step": 1490
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.3357142857142857e-05,
+      "loss": 0.0494,
+      "step": 1492
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.3321428571428572e-05,
+      "loss": 0.0756,
+      "step": 1494
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.3285714285714287e-05,
+      "loss": 0.3003,
+      "step": 1496
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.3250000000000003e-05,
+      "loss": 0.0587,
+      "step": 1498
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.3214285714285715e-05,
+      "loss": 0.259,
+      "step": 1500
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.317857142857143e-05,
+      "loss": 0.135,
+      "step": 1502
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.3142857142857145e-05,
+      "loss": 0.1344,
+      "step": 1504
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.3107142857142857e-05,
+      "loss": 0.5633,
+      "step": 1506
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.3071428571428573e-05,
+      "loss": 0.2419,
+      "step": 1508
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.3035714285714285e-05,
+      "loss": 0.0271,
+      "step": 1510
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.2679,
+      "step": 1512
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.2964285714285715e-05,
+      "loss": 0.0731,
+      "step": 1514
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.292857142857143e-05,
+      "loss": 0.4143,
+      "step": 1516
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.2892857142857143e-05,
+      "loss": 0.0471,
+      "step": 1518
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 0.0932,
+      "step": 1520
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.282142857142857e-05,
+      "loss": 0.1181,
+      "step": 1522
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.278571428571429e-05,
+      "loss": 0.1096,
+      "step": 1524
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.275e-05,
+      "loss": 0.3515,
+      "step": 1526
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.2714285714285716e-05,
+      "loss": 0.1167,
+      "step": 1528
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.2678571428571428e-05,
+      "loss": 0.0448,
+      "step": 1530
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.2642857142857143e-05,
+      "loss": 0.1904,
+      "step": 1532
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.260714285714286e-05,
+      "loss": 0.2376,
+      "step": 1534
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.257142857142857e-05,
+      "loss": 0.0849,
+      "step": 1536
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.253571428571429e-05,
+      "loss": 0.0372,
+      "step": 1538
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.25e-05,
+      "loss": 0.2339,
+      "step": 1540
+    },
+    {
+      "epoch": 11.0,
+      "eval_bleu": 0.056433679099699624,
+      "eval_loss": 1.719972848892212,
+      "eval_runtime": 24.2412,
+      "eval_samples_per_second": 5.899,
+      "eval_steps_per_second": 1.485,
+      "step": 1540
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.2464285714285717e-05,
+      "loss": 0.0698,
+      "step": 1542
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.242857142857143e-05,
+      "loss": 0.1026,
+      "step": 1544
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.2392857142857144e-05,
+      "loss": 0.1109,
+      "step": 1546
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.2357142857142856e-05,
+      "loss": 0.202,
+      "step": 1548
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.2321428571428575e-05,
+      "loss": 0.0851,
+      "step": 1550
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.2285714285714287e-05,
+      "loss": 0.2648,
+      "step": 1552
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.2250000000000002e-05,
+      "loss": 0.1442,
+      "step": 1554
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.2214285714285714e-05,
+      "loss": 0.2468,
+      "step": 1556
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.217857142857143e-05,
+      "loss": 0.0779,
+      "step": 1558
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.214285714285714e-05,
+      "loss": 0.148,
+      "step": 1560
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.2107142857142857e-05,
+      "loss": 0.0708,
+      "step": 1562
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.2071428571428572e-05,
+      "loss": 0.0334,
+      "step": 1564
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.2035714285714287e-05,
+      "loss": 0.9039,
+      "step": 1566
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.3214,
+      "step": 1568
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.1964285714285715e-05,
+      "loss": 0.1444,
+      "step": 1570
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.192857142857143e-05,
+      "loss": 0.1443,
+      "step": 1572
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.1892857142857142e-05,
+      "loss": 0.3402,
+      "step": 1574
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.185714285714286e-05,
+      "loss": 0.0442,
+      "step": 1576
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.1821428571428573e-05,
+      "loss": 0.2145,
+      "step": 1578
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.1785714285714288e-05,
+      "loss": 0.1515,
+      "step": 1580
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.175e-05,
+      "loss": 0.031,
+      "step": 1582
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.1714285714285715e-05,
+      "loss": 0.1018,
+      "step": 1584
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.1678571428571427e-05,
+      "loss": 0.0711,
+      "step": 1586
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.1642857142857146e-05,
+      "loss": 0.1052,
+      "step": 1588
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.1607142857142858e-05,
+      "loss": 0.0527,
+      "step": 1590
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.1571428571428574e-05,
+      "loss": 0.1146,
+      "step": 1592
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.1535714285714285e-05,
+      "loss": 0.2121,
+      "step": 1594
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.15e-05,
+      "loss": 0.2606,
+      "step": 1596
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.1464285714285716e-05,
+      "loss": 0.1558,
+      "step": 1598
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.0976,
+      "step": 1600
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.1392857142857143e-05,
+      "loss": 0.0381,
+      "step": 1602
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.135714285714286e-05,
+      "loss": 0.2439,
+      "step": 1604
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.1321428571428574e-05,
+      "loss": 0.0264,
+      "step": 1606
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.1285714285714286e-05,
+      "loss": 0.0418,
+      "step": 1608
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.125e-05,
+      "loss": 0.1194,
+      "step": 1610
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.1214285714285713e-05,
+      "loss": 0.3398,
+      "step": 1612
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.1178571428571432e-05,
+      "loss": 0.5066,
+      "step": 1614
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.1142857142857144e-05,
+      "loss": 0.1355,
+      "step": 1616
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.110714285714286e-05,
+      "loss": 0.0082,
+      "step": 1618
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.107142857142857e-05,
+      "loss": 0.1494,
+      "step": 1620
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.1035714285714287e-05,
+      "loss": 0.0588,
+      "step": 1622
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.1e-05,
+      "loss": 0.1003,
+      "step": 1624
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.0964285714285714e-05,
+      "loss": 0.0495,
+      "step": 1626
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.092857142857143e-05,
+      "loss": 0.0779,
+      "step": 1628
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.0892857142857145e-05,
+      "loss": 0.2657,
+      "step": 1630
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.0857142857142857e-05,
+      "loss": 0.3219,
+      "step": 1632
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.0821428571428572e-05,
+      "loss": 0.1618,
+      "step": 1634
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.0785714285714288e-05,
+      "loss": 0.2176,
+      "step": 1636
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.075e-05,
+      "loss": 0.083,
+      "step": 1638
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.0714285714285718e-05,
+      "loss": 0.2102,
+      "step": 1640
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.067857142857143e-05,
+      "loss": 0.0795,
+      "step": 1642
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.0642857142857146e-05,
+      "loss": 0.1542,
+      "step": 1644
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.0607142857142858e-05,
+      "loss": 0.199,
+      "step": 1646
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.0571428571428573e-05,
+      "loss": 0.2504,
+      "step": 1648
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.0535714285714285e-05,
+      "loss": 0.4486,
+      "step": 1650
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.05e-05,
+      "loss": 0.1149,
+      "step": 1652
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.0464285714285716e-05,
+      "loss": 0.1016,
+      "step": 1654
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.042857142857143e-05,
+      "loss": 0.105,
+      "step": 1656
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.0392857142857143e-05,
+      "loss": 1.0028,
+      "step": 1658
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.0357142857142858e-05,
+      "loss": 0.0936,
+      "step": 1660
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.032142857142857e-05,
+      "loss": 0.0676,
+      "step": 1662
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.0285714285714286e-05,
+      "loss": 0.1197,
+      "step": 1664
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.025e-05,
+      "loss": 0.2291,
+      "step": 1666
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.0214285714285716e-05,
+      "loss": 0.1547,
+      "step": 1668
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.017857142857143e-05,
+      "loss": 0.3888,
+      "step": 1670
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.0142857142857144e-05,
+      "loss": 0.0115,
+      "step": 1672
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.010714285714286e-05,
+      "loss": 0.0811,
+      "step": 1674
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.007142857142857e-05,
+      "loss": 0.0452,
+      "step": 1676
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.0035714285714286e-05,
+      "loss": 0.1913,
+      "step": 1678
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2e-05,
+      "loss": 0.2598,
+      "step": 1680
+    },
+    {
+      "epoch": 12.0,
+      "eval_bleu": 0.05236521898250645,
+      "eval_loss": 1.7801107168197632,
+      "eval_runtime": 24.1846,
+      "eval_samples_per_second": 5.913,
+      "eval_steps_per_second": 1.489,
+      "step": 1680
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.9964285714285717e-05,
+      "loss": 0.1087,
+      "step": 1682
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.992857142857143e-05,
+      "loss": 0.0528,
+      "step": 1684
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.9892857142857144e-05,
+      "loss": 0.1139,
+      "step": 1686
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.9857142857142856e-05,
+      "loss": 0.0691,
+      "step": 1688
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.982142857142857e-05,
+      "loss": 0.0266,
+      "step": 1690
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.9785714285714287e-05,
+      "loss": 0.1038,
+      "step": 1692
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 0.2035,
+      "step": 1694
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.9714285714285714e-05,
+      "loss": 0.1598,
+      "step": 1696
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.967857142857143e-05,
+      "loss": 0.077,
+      "step": 1698
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.3559,
+      "step": 1700
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.9607142857142857e-05,
+      "loss": 0.1641,
+      "step": 1702
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.9571428571428572e-05,
+      "loss": 0.0789,
+      "step": 1704
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.9535714285714288e-05,
+      "loss": 0.0121,
+      "step": 1706
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 0.7436,
+      "step": 1708
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.9464285714285715e-05,
+      "loss": 0.0932,
+      "step": 1710
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 0.159,
+      "step": 1712
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.9392857142857142e-05,
+      "loss": 0.0931,
+      "step": 1714
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.9357142857142858e-05,
+      "loss": 0.0474,
+      "step": 1716
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.9321428571428573e-05,
+      "loss": 0.0955,
+      "step": 1718
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.1552,
+      "step": 1720
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.925e-05,
+      "loss": 0.42,
+      "step": 1722
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.9214285714285716e-05,
+      "loss": 0.1122,
+      "step": 1724
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.9178571428571428e-05,
+      "loss": 0.0767,
+      "step": 1726
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.9142857142857143e-05,
+      "loss": 0.2043,
+      "step": 1728
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.910714285714286e-05,
+      "loss": 0.0451,
+      "step": 1730
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.9071428571428574e-05,
+      "loss": 0.0427,
+      "step": 1732
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.9035714285714286e-05,
+      "loss": 0.4838,
+      "step": 1734
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.9e-05,
+      "loss": 0.0312,
+      "step": 1736
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.8964285714285716e-05,
+      "loss": 0.0418,
+      "step": 1738
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.892857142857143e-05,
+      "loss": 0.3572,
+      "step": 1740
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.8892857142857144e-05,
+      "loss": 0.0803,
+      "step": 1742
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 0.1036,
+      "step": 1744
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.8821428571428574e-05,
+      "loss": 0.049,
+      "step": 1746
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.8785714285714286e-05,
+      "loss": 0.3821,
+      "step": 1748
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.3404,
+      "step": 1750
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.8714285714285714e-05,
+      "loss": 0.2504,
+      "step": 1752
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.867857142857143e-05,
+      "loss": 0.0391,
+      "step": 1754
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.864285714285714e-05,
+      "loss": 0.0462,
+      "step": 1756
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.860714285714286e-05,
+      "loss": 0.1358,
+      "step": 1758
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.8571428571428572e-05,
+      "loss": 0.0862,
+      "step": 1760
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.8535714285714287e-05,
+      "loss": 0.6036,
+      "step": 1762
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.85e-05,
+      "loss": 0.0517,
+      "step": 1764
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.8464285714285714e-05,
+      "loss": 0.036,
+      "step": 1766
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.842857142857143e-05,
+      "loss": 0.0821,
+      "step": 1768
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.8392857142857145e-05,
+      "loss": 0.0339,
+      "step": 1770
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.835714285714286e-05,
+      "loss": 0.1642,
+      "step": 1772
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.8321428571428572e-05,
+      "loss": 0.0753,
+      "step": 1774
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 0.0505,
+      "step": 1776
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.825e-05,
+      "loss": 0.1597,
+      "step": 1778
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.8214285714285715e-05,
+      "loss": 0.1481,
+      "step": 1780
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.8178571428571427e-05,
+      "loss": 0.0855,
+      "step": 1782
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.8142857142857146e-05,
+      "loss": 0.0269,
+      "step": 1784
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.8107142857142858e-05,
+      "loss": 0.1081,
+      "step": 1786
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.8071428571428573e-05,
+      "loss": 0.0153,
+      "step": 1788
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.8035714285714285e-05,
+      "loss": 0.1206,
+      "step": 1790
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.8e-05,
+      "loss": 0.0627,
+      "step": 1792
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.7964285714285712e-05,
+      "loss": 0.0295,
+      "step": 1794
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.792857142857143e-05,
+      "loss": 0.0765,
+      "step": 1796
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.7892857142857143e-05,
+      "loss": 0.1412,
+      "step": 1798
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.1499,
+      "step": 1800
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.7821428571428574e-05,
+      "loss": 0.0217,
+      "step": 1802
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.7785714285714286e-05,
+      "loss": 0.131,
+      "step": 1804
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.775e-05,
+      "loss": 0.1135,
+      "step": 1806
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.7714285714285713e-05,
+      "loss": 0.052,
+      "step": 1808
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.7678571428571432e-05,
+      "loss": 0.851,
+      "step": 1810
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.7642857142857144e-05,
+      "loss": 0.1207,
+      "step": 1812
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.760714285714286e-05,
+      "loss": 0.1081,
+      "step": 1814
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.757142857142857e-05,
+      "loss": 0.0295,
+      "step": 1816
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.7535714285714287e-05,
+      "loss": 0.1185,
+      "step": 1818
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.75e-05,
+      "loss": 0.0506,
+      "step": 1820
+    },
+    {
+      "epoch": 13.0,
+      "eval_bleu": 0.061011201542834875,
+      "eval_loss": 1.8598335981369019,
+      "eval_runtime": 24.0546,
+      "eval_samples_per_second": 5.945,
+      "eval_steps_per_second": 1.497,
+      "step": 1820
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.7464285714285717e-05,
+      "loss": 0.0403,
+      "step": 1822
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.742857142857143e-05,
+      "loss": 0.5271,
+      "step": 1824
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.7392857142857145e-05,
+      "loss": 0.1261,
+      "step": 1826
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.7357142857142856e-05,
+      "loss": 0.1601,
+      "step": 1828
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.7321428571428572e-05,
+      "loss": 0.1195,
+      "step": 1830
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.7285714285714287e-05,
+      "loss": 0.1902,
+      "step": 1832
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.725e-05,
+      "loss": 0.1603,
+      "step": 1834
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.7214285714285715e-05,
+      "loss": 0.2277,
+      "step": 1836
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.717857142857143e-05,
+      "loss": 0.6619,
+      "step": 1838
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.7142857142857145e-05,
+      "loss": 0.2079,
+      "step": 1840
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.7107142857142857e-05,
+      "loss": 0.0367,
+      "step": 1842
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.7071428571428573e-05,
+      "loss": 0.0216,
+      "step": 1844
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.7035714285714285e-05,
+      "loss": 0.0261,
+      "step": 1846
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.0616,
+      "step": 1848
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.6964285714285715e-05,
+      "loss": 0.1962,
+      "step": 1850
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.692857142857143e-05,
+      "loss": 0.0349,
+      "step": 1852
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.6892857142857143e-05,
+      "loss": 0.158,
+      "step": 1854
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.6857142857142858e-05,
+      "loss": 0.2398,
+      "step": 1856
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.682142857142857e-05,
+      "loss": 0.0433,
+      "step": 1858
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.6785714285714285e-05,
+      "loss": 0.1067,
+      "step": 1860
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.675e-05,
+      "loss": 0.0986,
+      "step": 1862
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.6714285714285716e-05,
+      "loss": 0.2894,
+      "step": 1864
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.6678571428571428e-05,
+      "loss": 0.0872,
+      "step": 1866
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.6642857142857143e-05,
+      "loss": 0.0762,
+      "step": 1868
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.660714285714286e-05,
+      "loss": 0.1113,
+      "step": 1870
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.657142857142857e-05,
+      "loss": 0.0527,
+      "step": 1872
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.653571428571429e-05,
+      "loss": 0.0784,
+      "step": 1874
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.65e-05,
+      "loss": 0.0554,
+      "step": 1876
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.6464285714285717e-05,
+      "loss": 0.1919,
+      "step": 1878
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 0.1355,
+      "step": 1880
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.6392857142857144e-05,
+      "loss": 0.0557,
+      "step": 1882
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.6357142857142856e-05,
+      "loss": 0.2345,
+      "step": 1884
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.632142857142857e-05,
+      "loss": 0.0032,
+      "step": 1886
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.6285714285714287e-05,
+      "loss": 0.1371,
+      "step": 1888
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 0.3181,
+      "step": 1890
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.6214285714285714e-05,
+      "loss": 0.0554,
+      "step": 1892
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.617857142857143e-05,
+      "loss": 0.4655,
+      "step": 1894
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.614285714285714e-05,
+      "loss": 0.0901,
+      "step": 1896
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.6107142857142857e-05,
+      "loss": 0.0824,
+      "step": 1898
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.0677,
+      "step": 1900
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.6035714285714287e-05,
+      "loss": 0.0293,
+      "step": 1902
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.0541,
+      "step": 1904
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.5964285714285715e-05,
+      "loss": 0.0308,
+      "step": 1906
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.592857142857143e-05,
+      "loss": 0.0323,
+      "step": 1908
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.5892857142857142e-05,
+      "loss": 0.0185,
+      "step": 1910
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.5857142857142857e-05,
+      "loss": 0.0834,
+      "step": 1912
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.5821428571428573e-05,
+      "loss": 0.0494,
+      "step": 1914
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.5785714285714288e-05,
+      "loss": 0.3638,
+      "step": 1916
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.575e-05,
+      "loss": 0.0987,
+      "step": 1918
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.1499,
+      "step": 1920
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.5678571428571427e-05,
+      "loss": 0.1923,
+      "step": 1922
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.5642857142857143e-05,
+      "loss": 0.1259,
+      "step": 1924
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.5607142857142858e-05,
+      "loss": 0.0653,
+      "step": 1926
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.5571428571428573e-05,
+      "loss": 0.0954,
+      "step": 1928
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.5535714285714285e-05,
+      "loss": 0.0128,
+      "step": 1930
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.55e-05,
+      "loss": 0.1522,
+      "step": 1932
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.5464285714285716e-05,
+      "loss": 0.0494,
+      "step": 1934
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.5428571428571428e-05,
+      "loss": 0.3574,
+      "step": 1936
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.5392857142857143e-05,
+      "loss": 0.0635,
+      "step": 1938
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.535714285714286e-05,
+      "loss": 0.1053,
+      "step": 1940
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.5321428571428574e-05,
+      "loss": 0.2044,
+      "step": 1942
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.5285714285714286e-05,
+      "loss": 0.4568,
+      "step": 1944
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.525e-05,
+      "loss": 0.125,
+      "step": 1946
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.5214285714285715e-05,
+      "loss": 0.1235,
+      "step": 1948
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.5178571428571429e-05,
+      "loss": 0.1872,
+      "step": 1950
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.5142857142857144e-05,
+      "loss": 0.1565,
+      "step": 1952
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.510714285714286e-05,
+      "loss": 0.1379,
+      "step": 1954
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.5071428571428573e-05,
+      "loss": 0.1156,
+      "step": 1956
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.5035714285714287e-05,
+      "loss": 0.046,
+      "step": 1958
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.0405,
+      "step": 1960
+    },
+    {
+      "epoch": 14.0,
+      "eval_bleu": 0.05121828550065784,
+      "eval_loss": 1.8986146450042725,
+      "eval_runtime": 24.2375,
+      "eval_samples_per_second": 5.9,
+      "eval_steps_per_second": 1.485,
+      "step": 1960
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.4964285714285714e-05,
+      "loss": 0.0217,
+      "step": 1962
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.4928571428571431e-05,
+      "loss": 0.1323,
+      "step": 1964
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4892857142857145e-05,
+      "loss": 0.0832,
+      "step": 1966
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 0.1244,
+      "step": 1968
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.4821428571428572e-05,
+      "loss": 0.051,
+      "step": 1970
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.4785714285714286e-05,
+      "loss": 0.1303,
+      "step": 1972
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.475e-05,
+      "loss": 0.229,
+      "step": 1974
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.4714285714285713e-05,
+      "loss": 0.1686,
+      "step": 1976
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.467857142857143e-05,
+      "loss": 0.5967,
+      "step": 1978
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.4642857142857144e-05,
+      "loss": 0.209,
+      "step": 1980
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.4607142857142857e-05,
+      "loss": 0.0374,
+      "step": 1982
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.4571428571428573e-05,
+      "loss": 0.1285,
+      "step": 1984
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.4535714285714286e-05,
+      "loss": 0.0955,
+      "step": 1986
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.45e-05,
+      "loss": 0.2161,
+      "step": 1988
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.4464285714285717e-05,
+      "loss": 0.0178,
+      "step": 1990
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.442857142857143e-05,
+      "loss": 0.1409,
+      "step": 1992
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.4392857142857144e-05,
+      "loss": 0.3652,
+      "step": 1994
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.4357142857142858e-05,
+      "loss": 0.1078,
+      "step": 1996
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.4321428571428572e-05,
+      "loss": 0.1511,
+      "step": 1998
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.0328,
+      "step": 2000
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 0.021,
+      "step": 2002
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.4214285714285716e-05,
+      "loss": 0.1506,
+      "step": 2004
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.417857142857143e-05,
+      "loss": 0.0671,
+      "step": 2006
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.4142857142857143e-05,
+      "loss": 0.1654,
+      "step": 2008
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.4107142857142857e-05,
+      "loss": 0.1088,
+      "step": 2010
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.407142857142857e-05,
+      "loss": 0.1767,
+      "step": 2012
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.4035714285714286e-05,
+      "loss": 0.0854,
+      "step": 2014
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.016,
+      "step": 2016
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.3964285714285715e-05,
+      "loss": 0.1581,
+      "step": 2018
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.392857142857143e-05,
+      "loss": 0.0147,
+      "step": 2020
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.3892857142857144e-05,
+      "loss": 0.4676,
+      "step": 2022
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.3857142857142858e-05,
+      "loss": 0.135,
+      "step": 2024
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.3821428571428571e-05,
+      "loss": 0.147,
+      "step": 2026
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.3785714285714285e-05,
+      "loss": 0.1849,
+      "step": 2028
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.3750000000000002e-05,
+      "loss": 0.1075,
+      "step": 2030
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 0.0314,
+      "step": 2032
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.367857142857143e-05,
+      "loss": 0.1343,
+      "step": 2034
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.3642857142857143e-05,
+      "loss": 0.0556,
+      "step": 2036
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.3607142857142857e-05,
+      "loss": 0.1728,
+      "step": 2038
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.357142857142857e-05,
+      "loss": 0.0605,
+      "step": 2040
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.3535714285714288e-05,
+      "loss": 0.0329,
+      "step": 2042
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.0633,
+      "step": 2044
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.3464285714285715e-05,
+      "loss": 0.077,
+      "step": 2046
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.3428571428571429e-05,
+      "loss": 0.2857,
+      "step": 2048
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.3392857142857144e-05,
+      "loss": 0.267,
+      "step": 2050
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.3357142857142858e-05,
+      "loss": 0.1483,
+      "step": 2052
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.3321428571428571e-05,
+      "loss": 0.1304,
+      "step": 2054
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.3285714285714288e-05,
+      "loss": 0.1087,
+      "step": 2056
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.3250000000000002e-05,
+      "loss": 0.1027,
+      "step": 2058
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.3214285714285716e-05,
+      "loss": 0.0551,
+      "step": 2060
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.317857142857143e-05,
+      "loss": 0.1644,
+      "step": 2062
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.3142857142857143e-05,
+      "loss": 0.0637,
+      "step": 2064
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.3107142857142857e-05,
+      "loss": 0.0579,
+      "step": 2066
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.3071428571428574e-05,
+      "loss": 0.4266,
+      "step": 2068
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.3035714285714287e-05,
+      "loss": 0.0466,
+      "step": 2070
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.074,
+      "step": 2072
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.2964285714285715e-05,
+      "loss": 0.1842,
+      "step": 2074
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.2928571428571428e-05,
+      "loss": 0.289,
+      "step": 2076
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.2892857142857142e-05,
+      "loss": 0.0131,
+      "step": 2078
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 0.6862,
+      "step": 2080
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.2821428571428573e-05,
+      "loss": 0.0031,
+      "step": 2082
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.2785714285714286e-05,
+      "loss": 0.4271,
+      "step": 2084
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.2750000000000002e-05,
+      "loss": 0.1194,
+      "step": 2086
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.2714285714285715e-05,
+      "loss": 0.0392,
+      "step": 2088
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.2678571428571429e-05,
+      "loss": 0.0211,
+      "step": 2090
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.2642857142857143e-05,
+      "loss": 0.1635,
+      "step": 2092
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.260714285714286e-05,
+      "loss": 0.1293,
+      "step": 2094
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.2571428571428573e-05,
+      "loss": 0.0308,
+      "step": 2096
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.2535714285714287e-05,
+      "loss": 0.0196,
+      "step": 2098
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0896,
+      "step": 2100
+    },
+    {
+      "epoch": 15.0,
+      "eval_bleu": 0.051168173950294785,
+      "eval_loss": 1.9366546869277954,
+      "eval_runtime": 24.0487,
+      "eval_samples_per_second": 5.946,
+      "eval_steps_per_second": 1.497,
+      "step": 2100
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.2464285714285714e-05,
+      "loss": 0.0391,
+      "step": 2102
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.242857142857143e-05,
+      "loss": 0.7486,
+      "step": 2104
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.2392857142857143e-05,
+      "loss": 0.0947,
+      "step": 2106
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 1.2357142857142857e-05,
+      "loss": 0.1021,
+      "step": 2108
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.2321428571428572e-05,
+      "loss": 0.0866,
+      "step": 2110
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.2285714285714286e-05,
+      "loss": 0.1501,
+      "step": 2112
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.225e-05,
+      "loss": 0.1316,
+      "step": 2114
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.2214285714285715e-05,
+      "loss": 0.04,
+      "step": 2116
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.2178571428571429e-05,
+      "loss": 0.1261,
+      "step": 2118
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.2142857142857144e-05,
+      "loss": 0.0111,
+      "step": 2120
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.2107142857142858e-05,
+      "loss": 0.0789,
+      "step": 2122
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.2071428571428573e-05,
+      "loss": 0.1307,
+      "step": 2124
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.2035714285714287e-05,
+      "loss": 0.0382,
+      "step": 2126
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.2e-05,
+      "loss": 0.1311,
+      "step": 2128
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.1964285714285716e-05,
+      "loss": 0.0964,
+      "step": 2130
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.192857142857143e-05,
+      "loss": 0.0497,
+      "step": 2132
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.1892857142857143e-05,
+      "loss": 0.0288,
+      "step": 2134
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.1857142857142858e-05,
+      "loss": 0.0139,
+      "step": 2136
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.1821428571428572e-05,
+      "loss": 0.0837,
+      "step": 2138
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.1785714285714286e-05,
+      "loss": 0.0677,
+      "step": 2140
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.175e-05,
+      "loss": 0.0752,
+      "step": 2142
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.1714285714285715e-05,
+      "loss": 0.0019,
+      "step": 2144
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 1.1678571428571428e-05,
+      "loss": 0.1102,
+      "step": 2146
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.1642857142857144e-05,
+      "loss": 0.1232,
+      "step": 2148
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.1607142857142857e-05,
+      "loss": 0.0402,
+      "step": 2150
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.1571428571428573e-05,
+      "loss": 0.1241,
+      "step": 2152
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.1535714285714286e-05,
+      "loss": 0.0722,
+      "step": 2154
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 0.0421,
+      "step": 2156
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.1464285714285715e-05,
+      "loss": 0.9368,
+      "step": 2158
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.1967,
+      "step": 2160
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.1392857142857144e-05,
+      "loss": 0.2906,
+      "step": 2162
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 1.1357142857142858e-05,
+      "loss": 0.0229,
+      "step": 2164
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.1321428571428572e-05,
+      "loss": 0.0711,
+      "step": 2166
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.1285714285714285e-05,
+      "loss": 0.0274,
+      "step": 2168
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 1.125e-05,
+      "loss": 0.1014,
+      "step": 2170
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.1214285714285714e-05,
+      "loss": 0.0447,
+      "step": 2172
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.1178571428571428e-05,
+      "loss": 0.1368,
+      "step": 2174
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.1142857142857143e-05,
+      "loss": 0.0401,
+      "step": 2176
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.1107142857142857e-05,
+      "loss": 0.0983,
+      "step": 2178
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.107142857142857e-05,
+      "loss": 0.0698,
+      "step": 2180
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.1035714285714286e-05,
+      "loss": 0.0559,
+      "step": 2182
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.0629,
+      "step": 2184
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.0964285714285715e-05,
+      "loss": 0.0415,
+      "step": 2186
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.092857142857143e-05,
+      "loss": 0.2813,
+      "step": 2188
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.0892857142857144e-05,
+      "loss": 0.5957,
+      "step": 2190
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 0.1439,
+      "step": 2192
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.0821428571428573e-05,
+      "loss": 0.1383,
+      "step": 2194
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.0785714285714287e-05,
+      "loss": 0.0243,
+      "step": 2196
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.075e-05,
+      "loss": 0.1043,
+      "step": 2198
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.247,
+      "step": 2200
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.067857142857143e-05,
+      "loss": 0.0148,
+      "step": 2202
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.0642857142857143e-05,
+      "loss": 0.2422,
+      "step": 2204
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.0607142857142857e-05,
+      "loss": 0.4656,
+      "step": 2206
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.0571428571428572e-05,
+      "loss": 0.2611,
+      "step": 2208
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.0535714285714286e-05,
+      "loss": 0.1034,
+      "step": 2210
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 1.05e-05,
+      "loss": 0.0696,
+      "step": 2212
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.0464285714285715e-05,
+      "loss": 0.0463,
+      "step": 2214
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.0428571428571428e-05,
+      "loss": 0.069,
+      "step": 2216
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.0392857142857144e-05,
+      "loss": 0.1243,
+      "step": 2218
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.0357142857142859e-05,
+      "loss": 0.0912,
+      "step": 2220
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 1.0321428571428573e-05,
+      "loss": 0.0374,
+      "step": 2222
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 1.0285714285714286e-05,
+      "loss": 0.0458,
+      "step": 2224
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.025e-05,
+      "loss": 0.0456,
+      "step": 2226
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.0214285714285715e-05,
+      "loss": 0.0712,
+      "step": 2228
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.0178571428571429e-05,
+      "loss": 0.1019,
+      "step": 2230
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.0142857142857143e-05,
+      "loss": 0.0707,
+      "step": 2232
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.0107142857142858e-05,
+      "loss": 0.0196,
+      "step": 2234
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.0071428571428572e-05,
+      "loss": 0.016,
+      "step": 2236
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.0035714285714285e-05,
+      "loss": 0.1188,
+      "step": 2238
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 1e-05,
+      "loss": 0.1058,
+      "step": 2240
+    },
+    {
+      "epoch": 16.0,
+      "eval_bleu": 0.05832891852668863,
+      "eval_loss": 1.96096932888031,
+      "eval_runtime": 24.094,
+      "eval_samples_per_second": 5.935,
+      "eval_steps_per_second": 1.494,
+      "step": 2240
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 9.964285714285714e-06,
+      "loss": 0.0488,
+      "step": 2242
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.928571428571428e-06,
+      "loss": 0.0818,
+      "step": 2244
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.892857142857143e-06,
+      "loss": 0.1164,
+      "step": 2246
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.857142857142857e-06,
+      "loss": 0.0771,
+      "step": 2248
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.821428571428573e-06,
+      "loss": 0.036,
+      "step": 2250
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.785714285714286e-06,
+      "loss": 0.5959,
+      "step": 2252
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.750000000000002e-06,
+      "loss": 0.1131,
+      "step": 2254
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 0.18,
+      "step": 2256
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.678571428571429e-06,
+      "loss": 0.0078,
+      "step": 2258
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.642857142857144e-06,
+      "loss": 0.0812,
+      "step": 2260
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.607142857142858e-06,
+      "loss": 0.2596,
+      "step": 2262
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.571428571428572e-06,
+      "loss": 0.0091,
+      "step": 2264
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.535714285714287e-06,
+      "loss": 0.0533,
+      "step": 2266
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.5e-06,
+      "loss": 0.0814,
+      "step": 2268
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.464285714285714e-06,
+      "loss": 0.2831,
+      "step": 2270
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.42857142857143e-06,
+      "loss": 0.2188,
+      "step": 2272
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 9.392857142857143e-06,
+      "loss": 0.3086,
+      "step": 2274
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 9.357142857142857e-06,
+      "loss": 0.0402,
+      "step": 2276
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 9.32142857142857e-06,
+      "loss": 0.0286,
+      "step": 2278
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 9.285714285714286e-06,
+      "loss": 0.1817,
+      "step": 2280
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 9.25e-06,
+      "loss": 0.2277,
+      "step": 2282
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 9.214285714285715e-06,
+      "loss": 0.0168,
+      "step": 2284
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 9.17857142857143e-06,
+      "loss": 0.0107,
+      "step": 2286
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 0.0172,
+      "step": 2288
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.107142857142858e-06,
+      "loss": 0.1477,
+      "step": 2290
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 9.071428571428573e-06,
+      "loss": 0.2319,
+      "step": 2292
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 9.035714285714287e-06,
+      "loss": 0.0038,
+      "step": 2294
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 9e-06,
+      "loss": 0.121,
+      "step": 2296
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 8.964285714285716e-06,
+      "loss": 0.0091,
+      "step": 2298
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.1473,
+      "step": 2300
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 8.892857142857143e-06,
+      "loss": 0.0162,
+      "step": 2302
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 8.857142857142857e-06,
+      "loss": 0.0326,
+      "step": 2304
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 8.821428571428572e-06,
+      "loss": 0.0076,
+      "step": 2306
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 8.785714285714286e-06,
+      "loss": 0.0074,
+      "step": 2308
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 8.75e-06,
+      "loss": 0.1026,
+      "step": 2310
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 8.714285714285715e-06,
+      "loss": 0.0213,
+      "step": 2312
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 8.678571428571428e-06,
+      "loss": 0.5917,
+      "step": 2314
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 8.642857142857144e-06,
+      "loss": 0.1254,
+      "step": 2316
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 8.607142857142857e-06,
+      "loss": 0.0164,
+      "step": 2318
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 8.571428571428573e-06,
+      "loss": 0.242,
+      "step": 2320
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 8.535714285714286e-06,
+      "loss": 0.3301,
+      "step": 2322
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 0.2119,
+      "step": 2324
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 8.464285714285715e-06,
+      "loss": 0.0233,
+      "step": 2326
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 8.428571428571429e-06,
+      "loss": 0.0795,
+      "step": 2328
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 8.392857142857143e-06,
+      "loss": 0.1141,
+      "step": 2330
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 8.357142857142858e-06,
+      "loss": 0.1092,
+      "step": 2332
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 8.321428571428572e-06,
+      "loss": 0.148,
+      "step": 2334
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 8.285714285714285e-06,
+      "loss": 0.0406,
+      "step": 2336
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 8.25e-06,
+      "loss": 0.016,
+      "step": 2338
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 8.214285714285714e-06,
+      "loss": 0.0063,
+      "step": 2340
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 8.178571428571428e-06,
+      "loss": 0.0488,
+      "step": 2342
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 8.142857142857143e-06,
+      "loss": 0.4284,
+      "step": 2344
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 8.107142857142857e-06,
+      "loss": 0.0494,
+      "step": 2346
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.07142857142857e-06,
+      "loss": 0.0051,
+      "step": 2348
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.035714285714286e-06,
+      "loss": 0.0966,
+      "step": 2350
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.0238,
+      "step": 2352
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 7.964285714285715e-06,
+      "loss": 0.0934,
+      "step": 2354
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 7.928571428571429e-06,
+      "loss": 0.0754,
+      "step": 2356
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 7.892857142857144e-06,
+      "loss": 0.0603,
+      "step": 2358
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.0936,
+      "step": 2360
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 7.821428571428571e-06,
+      "loss": 0.057,
+      "step": 2362
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 7.785714285714287e-06,
+      "loss": 0.0663,
+      "step": 2364
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 7.75e-06,
+      "loss": 0.0238,
+      "step": 2366
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 7.714285714285714e-06,
+      "loss": 0.0517,
+      "step": 2368
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 7.67857142857143e-06,
+      "loss": 0.2118,
+      "step": 2370
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 7.642857142857143e-06,
+      "loss": 0.2074,
+      "step": 2372
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 7.6071428571428575e-06,
+      "loss": 0.1105,
+      "step": 2374
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 7.571428571428572e-06,
+      "loss": 0.0714,
+      "step": 2376
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 7.5357142857142865e-06,
+      "loss": 0.1295,
+      "step": 2378
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 7.5e-06,
+      "loss": 0.093,
+      "step": 2380
+    },
+    {
+      "epoch": 17.0,
+      "eval_bleu": 0.06161404659326856,
+      "eval_loss": 2.0050530433654785,
+      "eval_runtime": 24.2381,
+      "eval_samples_per_second": 5.9,
+      "eval_steps_per_second": 1.485,
+      "step": 2380
+    }
+  ],
+  "max_steps": 2800,
+  "num_train_epochs": 20,
+  "total_flos": 2.0545863292416e+16,
+  "trial_name": null,
+  "trial_params": null
+}