diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18736 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 12.48,
+  "global_step": 1560000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998e-05,
+      "loss": 167.4813,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999600000000001e-05,
+      "loss": 339.0435,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994e-05,
+      "loss": 460.3145,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9992e-05,
+      "loss": 391.4744,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999e-05,
+      "loss": 429.6304,
+      "step": 2500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9988e-05,
+      "loss": 402.7589,
+      "step": 3000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9986000000000006e-05,
+      "loss": 346.8141,
+      "step": 3500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9984e-05,
+      "loss": 269.2469,
+      "step": 4000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9982000000000004e-05,
+      "loss": 249.9555,
+      "step": 4500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9980000000000006e-05,
+      "loss": 282.5226,
+      "step": 5000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9978e-05,
+      "loss": 276.7962,
+      "step": 5500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9976000000000004e-05,
+      "loss": 239.1784,
+      "step": 6000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9974000000000006e-05,
+      "loss": 214.4642,
+      "step": 6500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9972e-05,
+      "loss": 206.2337,
+      "step": 7000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.997e-05,
+      "loss": 202.7909,
+      "step": 7500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.996800000000001e-05,
+      "loss": 207.2598,
+      "step": 8000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9966e-05,
+      "loss": 198.4287,
+      "step": 8500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9964e-05,
+      "loss": 202.6692,
+      "step": 9000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9962e-05,
+      "loss": 172.0607,
+      "step": 9500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.996e-05,
+      "loss": 152.3532,
+      "step": 10000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9958000000000005e-05,
+      "loss": 141.7099,
+      "step": 10500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9956e-05,
+      "loss": 131.85,
+      "step": 11000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9954e-05,
+      "loss": 131.9644,
+      "step": 11500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9952000000000006e-05,
+      "loss": 120.3319,
+      "step": 12000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.995e-05,
+      "loss": 125.1385,
+      "step": 12500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9948000000000004e-05,
+      "loss": 102.6195,
+      "step": 13000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9946000000000006e-05,
+      "loss": 97.9456,
+      "step": 13500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9944e-05,
+      "loss": 90.2301,
+      "step": 14000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9942e-05,
+      "loss": 85.6502,
+      "step": 14500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9940000000000006e-05,
+      "loss": 71.9912,
+      "step": 15000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9938e-05,
+      "loss": 66.2195,
+      "step": 15500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9936000000000004e-05,
+      "loss": 59.0722,
+      "step": 16000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9934e-05,
+      "loss": 52.456,
+      "step": 16500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9932e-05,
+      "loss": 44.7411,
+      "step": 17000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9930000000000005e-05,
+      "loss": 37.4041,
+      "step": 17500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9928e-05,
+      "loss": 33.0052,
+      "step": 18000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9926e-05,
+      "loss": 27.9028,
+      "step": 18500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9924000000000005e-05,
+      "loss": 23.4986,
+      "step": 19000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9922e-05,
+      "loss": 22.3736,
+      "step": 19500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.992e-05,
+      "loss": 19.143,
+      "step": 20000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9918000000000006e-05,
+      "loss": 16.728,
+      "step": 20500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9916e-05,
+      "loss": 14.6861,
+      "step": 21000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9914e-05,
+      "loss": 13.893,
+      "step": 21500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9912000000000006e-05,
+      "loss": 13.4733,
+      "step": 22000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.991e-05,
+      "loss": 12.3987,
+      "step": 22500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9908000000000004e-05,
+      "loss": 11.5524,
+      "step": 23000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9906e-05,
+      "loss": 11.2799,
+      "step": 23500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9904e-05,
+      "loss": 10.9152,
+      "step": 24000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9902000000000004e-05,
+      "loss": 10.5949,
+      "step": 24500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.99e-05,
+      "loss": 10.8094,
+      "step": 25000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9898e-05,
+      "loss": 10.2322,
+      "step": 25500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9896000000000005e-05,
+      "loss": 10.0505,
+      "step": 26000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9894e-05,
+      "loss": 10.1343,
+      "step": 26500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9892e-05,
+      "loss": 9.9442,
+      "step": 27000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9890000000000005e-05,
+      "loss": 9.7331,
+      "step": 27500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9888e-05,
+      "loss": 9.7229,
+      "step": 28000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9886e-05,
+      "loss": 9.7828,
+      "step": 28500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9884000000000006e-05,
+      "loss": 9.7042,
+      "step": 29000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9882e-05,
+      "loss": 9.6642,
+      "step": 29500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9880000000000004e-05,
+      "loss": 9.8011,
+      "step": 30000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9878e-05,
+      "loss": 9.7558,
+      "step": 30500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9876e-05,
+      "loss": 9.8933,
+      "step": 31000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9874000000000004e-05,
+      "loss": 9.6691,
+      "step": 31500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9872e-05,
+      "loss": 9.5504,
+      "step": 32000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.987e-05,
+      "loss": 9.4497,
+      "step": 32500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9868000000000004e-05,
+      "loss": 9.4204,
+      "step": 33000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9866e-05,
+      "loss": 9.3652,
+      "step": 33500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9864e-05,
+      "loss": 9.7497,
+      "step": 34000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9862000000000005e-05,
+      "loss": 9.5389,
+      "step": 34500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.986e-05,
+      "loss": 9.6565,
+      "step": 35000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9858e-05,
+      "loss": 9.3094,
+      "step": 35500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9856000000000005e-05,
+      "loss": 9.3693,
+      "step": 36000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9854e-05,
+      "loss": 9.3462,
+      "step": 36500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9852e-05,
+      "loss": 9.4076,
+      "step": 37000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9850000000000006e-05,
+      "loss": 9.2531,
+      "step": 37500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9848e-05,
+      "loss": 9.227,
+      "step": 38000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9846000000000004e-05,
+      "loss": 9.3106,
+      "step": 38500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9844e-05,
+      "loss": 9.2394,
+      "step": 39000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9842e-05,
+      "loss": 9.2606,
+      "step": 39500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9840000000000004e-05,
+      "loss": 9.2406,
+      "step": 40000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9838e-05,
+      "loss": 9.2116,
+      "step": 40500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9836e-05,
+      "loss": 9.2192,
+      "step": 41000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9834000000000004e-05,
+      "loss": 9.2842,
+      "step": 41500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9832e-05,
+      "loss": 9.2024,
+      "step": 42000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.983e-05,
+      "loss": 9.3669,
+      "step": 42500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9828000000000005e-05,
+      "loss": 9.3505,
+      "step": 43000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9826e-05,
+      "loss": 9.2114,
+      "step": 43500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9824e-05,
+      "loss": 9.1553,
+      "step": 44000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9822000000000005e-05,
+      "loss": 9.1217,
+      "step": 44500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.982e-05,
+      "loss": 9.2315,
+      "step": 45000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9818e-05,
+      "loss": 9.3219,
+      "step": 45500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9816e-05,
+      "loss": 9.0737,
+      "step": 46000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.981400000000001e-05,
+      "loss": 9.1281,
+      "step": 46500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9812000000000004e-05,
+      "loss": 9.1292,
+      "step": 47000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.981e-05,
+      "loss": 9.2538,
+      "step": 47500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9808e-05,
+      "loss": 9.1272,
+      "step": 48000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9806000000000004e-05,
+      "loss": 9.225,
+      "step": 48500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9804e-05,
+      "loss": 9.0919,
+      "step": 49000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9802e-05,
+      "loss": 9.122,
+      "step": 49500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 9.1869,
+      "step": 50000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9798e-05,
+      "loss": 9.0186,
+      "step": 50500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9796e-05,
+      "loss": 9.1333,
+      "step": 51000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9794000000000005e-05,
+      "loss": 9.1123,
+      "step": 51500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9792e-05,
+      "loss": 9.1951,
+      "step": 52000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.979e-05,
+      "loss": 9.1012,
+      "step": 52500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9788e-05,
+      "loss": 8.9948,
+      "step": 53000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.978600000000001e-05,
+      "loss": 9.2808,
+      "step": 53500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9784e-05,
+      "loss": 9.0797,
+      "step": 54000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9782e-05,
+      "loss": 9.0534,
+      "step": 54500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.978e-05,
+      "loss": 8.995,
+      "step": 55000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9778000000000004e-05,
+      "loss": 9.0096,
+      "step": 55500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9776e-05,
+      "loss": 8.9898,
+      "step": 56000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9774e-05,
+      "loss": 8.9447,
+      "step": 56500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9772000000000004e-05,
+      "loss": 9.2217,
+      "step": 57000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.977e-05,
+      "loss": 9.0623,
+      "step": 57500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9768e-05,
+      "loss": 9.0159,
+      "step": 58000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9766000000000004e-05,
+      "loss": 9.0159,
+      "step": 58500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.976400000000001e-05,
+      "loss": 8.938,
+      "step": 59000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9762e-05,
+      "loss": 8.9805,
+      "step": 59500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.976e-05,
+      "loss": 8.9252,
+      "step": 60000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.975800000000001e-05,
+      "loss": 9.0771,
+      "step": 60500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9756e-05,
+      "loss": 9.0443,
+      "step": 61000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9754e-05,
+      "loss": 9.1127,
+      "step": 61500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.975200000000001e-05,
+      "loss": 8.9619,
+      "step": 62000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.975e-05,
+      "loss": 8.9725,
+      "step": 62500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9748e-05,
+      "loss": 8.9863,
+      "step": 63000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9746e-05,
+      "loss": 8.9595,
+      "step": 63500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9744000000000003e-05,
+      "loss": 8.8928,
+      "step": 64000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9742e-05,
+      "loss": 9.0314,
+      "step": 64500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974e-05,
+      "loss": 8.9132,
+      "step": 65000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9738000000000004e-05,
+      "loss": 8.9137,
+      "step": 65500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9736000000000006e-05,
+      "loss": 8.9657,
+      "step": 66000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9734e-05,
+      "loss": 8.9497,
+      "step": 66500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9732e-05,
+      "loss": 8.8762,
+      "step": 67000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.973000000000001e-05,
+      "loss": 8.8903,
+      "step": 67500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9728e-05,
+      "loss": 8.8748,
+      "step": 68000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9726e-05,
+      "loss": 8.8706,
+      "step": 68500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.972400000000001e-05,
+      "loss": 8.8593,
+      "step": 69000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9722e-05,
+      "loss": 8.9197,
+      "step": 69500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.972e-05,
+      "loss": 8.8718,
+      "step": 70000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9718e-05,
+      "loss": 8.8917,
+      "step": 70500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9716e-05,
+      "loss": 8.859,
+      "step": 71000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9714000000000005e-05,
+      "loss": 9.1695,
+      "step": 71500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9712e-05,
+      "loss": 8.8216,
+      "step": 72000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9710000000000003e-05,
+      "loss": 8.8896,
+      "step": 72500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9708000000000006e-05,
+      "loss": 8.857,
+      "step": 73000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9706e-05,
+      "loss": 8.9091,
+      "step": 73500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9704000000000004e-05,
+      "loss": 8.9185,
+      "step": 74000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9702000000000006e-05,
+      "loss": 8.8406,
+      "step": 74500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.97e-05,
+      "loss": 8.8969,
+      "step": 75000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9698e-05,
+      "loss": 8.8779,
+      "step": 75500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.969600000000001e-05,
+      "loss": 8.8665,
+      "step": 76000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.9694e-05,
+      "loss": 8.8258,
+      "step": 76500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9692e-05,
+      "loss": 8.9677,
+      "step": 77000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.969e-05,
+      "loss": 8.8426,
+      "step": 77500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9688e-05,
+      "loss": 8.8894,
+      "step": 78000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.9686000000000005e-05,
+      "loss": 8.81,
+      "step": 78500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.9684e-05,
+      "loss": 8.8793,
+      "step": 79000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9682e-05,
+      "loss": 8.8209,
+      "step": 79500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9680000000000005e-05,
+      "loss": 8.8762,
+      "step": 80000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9678e-05,
+      "loss": 8.8526,
+      "step": 80500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9676000000000003e-05,
+      "loss": 8.8207,
+      "step": 81000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9674000000000006e-05,
+      "loss": 8.8084,
+      "step": 81500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.9672e-05,
+      "loss": 8.8289,
+      "step": 82000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.967e-05,
+      "loss": 8.8631,
+      "step": 82500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.9668000000000006e-05,
+      "loss": 8.7972,
+      "step": 83000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9666e-05,
+      "loss": 8.8412,
+      "step": 83500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9664000000000004e-05,
+      "loss": 8.8133,
+      "step": 84000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9662e-05,
+      "loss": 8.8066,
+      "step": 84500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966e-05,
+      "loss": 8.8591,
+      "step": 85000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9658000000000005e-05,
+      "loss": 8.8006,
+      "step": 85500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9656e-05,
+      "loss": 8.8801,
+      "step": 86000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9654e-05,
+      "loss": 8.9259,
+      "step": 86500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9652000000000005e-05,
+      "loss": 8.8784,
+      "step": 87000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.965e-05,
+      "loss": 8.8027,
+      "step": 87500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9648e-05,
+      "loss": 8.7856,
+      "step": 88000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9646000000000005e-05,
+      "loss": 8.8136,
+      "step": 88500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9644e-05,
+      "loss": 8.7518,
+      "step": 89000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9642e-05,
+      "loss": 9.0135,
+      "step": 89500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9640000000000006e-05,
+      "loss": 8.8534,
+      "step": 90000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9638e-05,
+      "loss": 8.7789,
+      "step": 90500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9636000000000004e-05,
+      "loss": 8.8706,
+      "step": 91000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9634e-05,
+      "loss": 8.7578,
+      "step": 91500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9632e-05,
+      "loss": 8.8256,
+      "step": 92000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9630000000000004e-05,
+      "loss": 8.8399,
+      "step": 92500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9628e-05,
+      "loss": 8.9445,
+      "step": 93000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9626e-05,
+      "loss": 8.8409,
+      "step": 93500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9624000000000005e-05,
+      "loss": 8.8985,
+      "step": 94000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9622e-05,
+      "loss": 8.8067,
+      "step": 94500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.962e-05,
+      "loss": 8.7973,
+      "step": 95000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9618000000000005e-05,
+      "loss": 8.8158,
+      "step": 95500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9616e-05,
+      "loss": 8.7935,
+      "step": 96000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9614e-05,
+      "loss": 8.8262,
+      "step": 96500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9612000000000005e-05,
+      "loss": 8.7424,
+      "step": 97000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.961e-05,
+      "loss": 8.8062,
+      "step": 97500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9608000000000003e-05,
+      "loss": 8.821,
+      "step": 98000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9606000000000006e-05,
+      "loss": 8.7761,
+      "step": 98500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9604e-05,
+      "loss": 8.7752,
+      "step": 99000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9602000000000004e-05,
+      "loss": 8.7686,
+      "step": 99500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.96e-05,
+      "loss": 8.8078,
+      "step": 100000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9598e-05,
+      "loss": 8.755,
+      "step": 100500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9596000000000004e-05,
+      "loss": 8.8267,
+      "step": 101000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9594e-05,
+      "loss": 8.7747,
+      "step": 101500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9592e-05,
+      "loss": 8.8316,
+      "step": 102000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9590000000000005e-05,
+      "loss": 8.7557,
+      "step": 102500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9588e-05,
+      "loss": 8.7278,
+      "step": 103000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9586e-05,
+      "loss": 8.7551,
+      "step": 103500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9584000000000005e-05,
+      "loss": 8.7773,
+      "step": 104000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9582e-05,
+      "loss": 8.7889,
+      "step": 104500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.958e-05,
+      "loss": 8.769,
+      "step": 105000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9578000000000005e-05,
+      "loss": 8.7733,
+      "step": 105500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9576e-05,
+      "loss": 8.772,
+      "step": 106000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9574000000000003e-05,
+      "loss": 8.7841,
+      "step": 106500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9572e-05,
+      "loss": 8.763,
+      "step": 107000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.957e-05,
+      "loss": 8.7616,
+      "step": 107500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9568000000000004e-05,
+      "loss": 8.8227,
+      "step": 108000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9566e-05,
+      "loss": 8.7499,
+      "step": 108500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9564e-05,
+      "loss": 8.7877,
+      "step": 109000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9562000000000004e-05,
+      "loss": 8.7573,
+      "step": 109500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.956e-05,
+      "loss": 8.7705,
+      "step": 110000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9558e-05,
+      "loss": 8.7416,
+      "step": 110500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9556000000000005e-05,
+      "loss": 8.7569,
+      "step": 111000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9554e-05,
+      "loss": 8.786,
+      "step": 111500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9552e-05,
+      "loss": 8.7313,
+      "step": 112000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9550000000000005e-05,
+      "loss": 8.7426,
+      "step": 112500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9548e-05,
+      "loss": 8.7644,
+      "step": 113000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9546e-05,
+      "loss": 8.777,
+      "step": 113500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9544e-05,
+      "loss": 8.7138,
+      "step": 114000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.954200000000001e-05,
+      "loss": 8.745,
+      "step": 114500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9540000000000003e-05,
+      "loss": 8.7519,
+      "step": 115000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9538e-05,
+      "loss": 8.7538,
+      "step": 115500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9536e-05,
+      "loss": 8.7966,
+      "step": 116000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9534000000000004e-05,
+      "loss": 8.7846,
+      "step": 116500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9532e-05,
+      "loss": 8.739,
+      "step": 117000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.953e-05,
+      "loss": 8.7326,
+      "step": 117500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9528000000000004e-05,
+      "loss": 8.7237,
+      "step": 118000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9526e-05,
+      "loss": 8.7634,
+      "step": 118500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9524e-05,
+      "loss": 8.7499,
+      "step": 119000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9522000000000005e-05,
+      "loss": 8.7802,
+      "step": 119500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.952e-05,
+      "loss": 8.7407,
+      "step": 120000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9518e-05,
+      "loss": 8.7336,
+      "step": 120500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9516e-05,
+      "loss": 8.7446,
+      "step": 121000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.951400000000001e-05,
+      "loss": 8.7434,
+      "step": 121500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9512e-05,
+      "loss": 8.7396,
+      "step": 122000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.951e-05,
+      "loss": 8.7272,
+      "step": 122500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9508e-05,
+      "loss": 8.751,
+      "step": 123000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9506000000000003e-05,
+      "loss": 8.7366,
+      "step": 123500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9504e-05,
+      "loss": 8.761,
+      "step": 124000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9502e-05,
+      "loss": 8.7528,
+      "step": 124500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 8.77,
+      "step": 125000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9498e-05,
+      "loss": 8.7479,
+      "step": 125500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.9496e-05,
+      "loss": 8.7402,
+      "step": 126000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.9494000000000004e-05,
+      "loss": 8.7317,
+      "step": 126500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.9492000000000007e-05,
+      "loss": 8.7582,
+      "step": 127000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.949e-05,
+      "loss": 8.785,
+      "step": 127500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.9488e-05,
+      "loss": 8.7487,
+      "step": 128000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.948600000000001e-05,
+      "loss": 8.7275,
+      "step": 128500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.9484e-05,
+      "loss": 8.7491,
+      "step": 129000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.9482e-05,
+      "loss": 8.7415,
+      "step": 129500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.948000000000001e-05,
+      "loss": 8.7348,
+      "step": 130000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.9478e-05,
+      "loss": 8.7247,
+      "step": 130500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.9476e-05,
+      "loss": 8.7298,
+      "step": 131000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.9474e-05,
+      "loss": 8.7433,
+      "step": 131500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.9472e-05,
+      "loss": 8.7733,
+      "step": 132000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.947e-05,
+      "loss": 8.7228,
+      "step": 132500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.9468e-05,
+      "loss": 8.7441,
+      "step": 133000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.9466000000000004e-05,
+      "loss": 8.7274,
+      "step": 133500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.9464000000000006e-05,
+      "loss": 8.7556,
+      "step": 134000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.9462e-05,
+      "loss": 8.7592,
+      "step": 134500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.946e-05,
+      "loss": 8.7349,
+      "step": 135000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.9458000000000007e-05,
+      "loss": 8.7432,
+      "step": 135500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.9456e-05,
+      "loss": 8.7744,
+      "step": 136000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.9454e-05,
+      "loss": 8.7506,
+      "step": 136500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.945200000000001e-05,
+      "loss": 8.7566,
+      "step": 137000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.945e-05,
+      "loss": 8.7312,
+      "step": 137500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.9448e-05,
+      "loss": 8.7084,
+      "step": 138000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.9446e-05,
+      "loss": 8.7298,
+      "step": 138500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.9444e-05,
+      "loss": 8.7195,
+      "step": 139000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.9442000000000005e-05,
+      "loss": 8.7922,
+      "step": 139500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.944e-05,
+      "loss": 8.7165,
+      "step": 140000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.9438e-05,
+      "loss": 8.7332,
+      "step": 140500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.9436000000000006e-05,
+      "loss": 8.7391,
+      "step": 141000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.9434e-05,
+      "loss": 8.7532,
+      "step": 141500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9432000000000004e-05,
+      "loss": 8.7782,
+      "step": 142000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9430000000000006e-05,
+      "loss": 8.7489,
+      "step": 142500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9428e-05,
+      "loss": 8.7545,
+      "step": 143000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.9426e-05,
+      "loss": 8.7656,
+      "step": 143500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.9424000000000007e-05,
+      "loss": 8.7302,
+      "step": 144000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.9422e-05,
+      "loss": 8.7296,
+      "step": 144500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.942e-05,
+      "loss": 8.7566,
+      "step": 145000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.9418e-05,
+      "loss": 8.7108,
+      "step": 145500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.9416e-05,
+      "loss": 8.7323,
+      "step": 146000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.9414000000000005e-05,
+      "loss": 8.7436,
+      "step": 146500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.9412e-05,
+      "loss": 8.6866,
+      "step": 147000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.941e-05,
+      "loss": 8.7285,
+      "step": 147500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.9408000000000005e-05,
+      "loss": 8.7386,
+      "step": 148000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.9406e-05,
+      "loss": 8.7394,
+      "step": 148500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.9404e-05,
+      "loss": 8.7729,
+      "step": 149000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.9402000000000006e-05,
+      "loss": 8.7637,
+      "step": 149500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.94e-05,
+      "loss": 8.7381,
+      "step": 150000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.9398e-05,
+      "loss": 8.7235,
+      "step": 150500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.9396000000000006e-05,
+      "loss": 8.7198,
+      "step": 151000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.9394e-05,
+      "loss": 8.7512,
+      "step": 151500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.9392000000000004e-05,
+      "loss": 8.7068,
+      "step": 152000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.939e-05,
+      "loss": 8.7282,
+      "step": 152500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.9388e-05,
+      "loss": 8.732,
+      "step": 153000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.9386000000000005e-05,
+      "loss": 8.7478,
+      "step": 153500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.9384e-05,
+      "loss": 8.7374,
+      "step": 154000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.9382e-05,
+      "loss": 8.7333,
+      "step": 154500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.9380000000000005e-05,
+      "loss": 8.7092,
+      "step": 155000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.9378e-05,
+      "loss": 8.7646,
+      "step": 155500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.9376e-05,
+      "loss": 8.7004,
+      "step": 156000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.9374000000000005e-05,
+      "loss": 8.7022,
+      "step": 156500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.9372e-05,
+      "loss": 8.7469,
+      "step": 157000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.937e-05,
+      "loss": 8.7383,
+      "step": 157500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.9368000000000006e-05,
+      "loss": 8.7374,
+      "step": 158000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9366e-05,
+      "loss": 8.7537,
+      "step": 158500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9364000000000004e-05,
+      "loss": 8.7513,
+      "step": 159000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9362e-05,
+      "loss": 8.7303,
+      "step": 159500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.936e-05,
+      "loss": 8.734,
+      "step": 160000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9358000000000004e-05,
+      "loss": 8.7497,
+      "step": 160500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9356e-05,
+      "loss": 8.7377,
+      "step": 161000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9354e-05,
+      "loss": 8.7308,
+      "step": 161500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9352000000000005e-05,
+      "loss": 8.7586,
+      "step": 162000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.935e-05,
+      "loss": 8.744,
+      "step": 162500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9348e-05,
+      "loss": 8.7429,
+      "step": 163000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.9346000000000005e-05,
+      "loss": 8.7083,
+      "step": 163500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.9344e-05,
+      "loss": 8.7558,
+      "step": 164000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.9342e-05,
+      "loss": 8.7276,
+      "step": 164500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.9340000000000005e-05,
+      "loss": 8.7388,
+      "step": 165000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.9338e-05,
+      "loss": 8.7236,
+      "step": 165500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.9336e-05,
+      "loss": 8.6968,
+      "step": 166000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.9334000000000006e-05,
+      "loss": 8.738,
+      "step": 166500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.9332e-05,
+      "loss": 8.6997,
+      "step": 167000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.9330000000000004e-05,
+      "loss": 8.7297,
+      "step": 167500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.9328e-05,
+      "loss": 8.7298,
+      "step": 168000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9326e-05,
+      "loss": 8.7035,
+      "step": 168500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9324000000000004e-05,
+      "loss": 8.7204,
+      "step": 169000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9322e-05,
+      "loss": 8.7452,
+      "step": 169500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.932e-05,
+      "loss": 8.6998,
+      "step": 170000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9318000000000005e-05,
+      "loss": 8.7377,
+      "step": 170500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.9316e-05,
+      "loss": 8.7331,
+      "step": 171000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.9314e-05,
+      "loss": 8.7163,
+      "step": 171500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.9312000000000005e-05,
+      "loss": 8.7268,
+      "step": 172000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.931e-05,
+      "loss": 8.6967,
+      "step": 172500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.9308e-05,
+      "loss": 8.7252,
+      "step": 173000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.9306000000000005e-05,
+      "loss": 8.7349,
+      "step": 173500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.9304e-05,
+      "loss": 8.7176,
+      "step": 174000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.9302e-05,
+      "loss": 8.7122,
+      "step": 174500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.93e-05,
+      "loss": 8.7241,
+      "step": 175000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.9298e-05,
+      "loss": 8.7116,
+      "step": 175500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9296000000000004e-05,
+      "loss": 8.7077,
+      "step": 176000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9294e-05,
+      "loss": 8.7184,
+      "step": 176500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.9292e-05,
+      "loss": 8.7507,
+      "step": 177000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.9290000000000004e-05,
+      "loss": 8.7461,
+      "step": 177500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.9288e-05,
+      "loss": 8.7107,
+      "step": 178000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.9286e-05,
+      "loss": 8.7484,
+      "step": 178500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.9284000000000005e-05,
+      "loss": 8.7338,
+      "step": 179000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.9282e-05,
+      "loss": 8.7505,
+      "step": 179500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.928e-05,
+      "loss": 8.6856,
+      "step": 180000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.9278000000000005e-05,
+      "loss": 8.7148,
+      "step": 180500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.9276e-05,
+      "loss": 8.727,
+      "step": 181000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.9274e-05,
+      "loss": 8.7365,
+      "step": 181500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.9272e-05,
+      "loss": 8.7123,
+      "step": 182000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.927000000000001e-05,
+      "loss": 8.7501,
+      "step": 182500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.9268e-05,
+      "loss": 8.7205,
+      "step": 183000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.9266e-05,
+      "loss": 8.7338,
+      "step": 183500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.9264e-05,
+      "loss": 8.725,
+      "step": 184000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9262000000000004e-05,
+      "loss": 8.7439,
+      "step": 184500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.926e-05,
+      "loss": 8.7272,
+      "step": 185000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9258e-05,
+      "loss": 8.711,
+      "step": 185500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.9256000000000004e-05,
+      "loss": 8.7415,
+      "step": 186000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.9254e-05,
+      "loss": 8.7222,
+      "step": 186500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9252e-05,
+      "loss": 8.6877,
+      "step": 187000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9250000000000004e-05,
+      "loss": 8.718,
+      "step": 187500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9248e-05,
+      "loss": 8.7498,
+      "step": 188000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9246e-05,
+      "loss": 8.7576,
+      "step": 188500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9244e-05,
+      "loss": 8.7261,
+      "step": 189000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.924200000000001e-05,
+      "loss": 8.6984,
+      "step": 189500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.924e-05,
+      "loss": 8.7172,
+      "step": 190000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.9238e-05,
+      "loss": 8.7044,
+      "step": 190500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.923600000000001e-05,
+      "loss": 8.7316,
+      "step": 191000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9234e-05,
+      "loss": 8.7253,
+      "step": 191500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.9232e-05,
+      "loss": 8.7278,
+      "step": 192000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.923e-05,
+      "loss": 8.7171,
+      "step": 192500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.9228000000000004e-05,
+      "loss": 8.7442,
+      "step": 193000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.9226e-05,
+      "loss": 8.7353,
+      "step": 193500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.9224e-05,
+      "loss": 8.7023,
+      "step": 194000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.9222000000000004e-05,
+      "loss": 8.6957,
+      "step": 194500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.9220000000000006e-05,
+      "loss": 8.7123,
+      "step": 195000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.9218e-05,
+      "loss": 8.7047,
+      "step": 195500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.9216e-05,
+      "loss": 8.7271,
+      "step": 196000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.921400000000001e-05,
+      "loss": 8.7182,
+      "step": 196500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.9212e-05,
+      "loss": 8.6991,
+      "step": 197000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.921e-05,
+      "loss": 8.695,
+      "step": 197500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.920800000000001e-05,
+      "loss": 8.7277,
+      "step": 198000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.9206e-05,
+      "loss": 8.7258,
+      "step": 198500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.9204e-05,
+      "loss": 8.7348,
+      "step": 199000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.9202e-05,
+      "loss": 8.7168,
+      "step": 199500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.92e-05,
+      "loss": 8.7074,
+      "step": 200000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.9198e-05,
+      "loss": 8.7199,
+      "step": 200500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.9196e-05,
+      "loss": 8.7036,
+      "step": 201000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.9194000000000004e-05,
+      "loss": 8.6906,
+      "step": 201500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.9192000000000006e-05,
+      "loss": 8.7086,
+      "step": 202000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.919e-05,
+      "loss": 8.708,
+      "step": 202500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.9188000000000004e-05,
+      "loss": 8.7123,
+      "step": 203000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.9186000000000006e-05,
+      "loss": 8.7107,
+      "step": 203500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.9184e-05,
+      "loss": 8.7279,
+      "step": 204000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.9182e-05,
+      "loss": 8.7128,
+      "step": 204500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.918000000000001e-05,
+      "loss": 8.7156,
+      "step": 205000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.9178e-05,
+      "loss": 8.7054,
+      "step": 205500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9176e-05,
+      "loss": 8.6972,
+      "step": 206000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9174e-05,
+      "loss": 8.7417,
+      "step": 206500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9172e-05,
+      "loss": 8.7225,
+      "step": 207000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9170000000000005e-05,
+      "loss": 8.6858,
+      "step": 207500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9168e-05,
+      "loss": 8.7073,
+      "step": 208000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9166e-05,
+      "loss": 8.7324,
+      "step": 208500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9164000000000006e-05,
+      "loss": 8.7009,
+      "step": 209000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.9162e-05,
+      "loss": 8.7222,
+      "step": 209500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.9160000000000004e-05,
+      "loss": 8.7189,
+      "step": 210000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.9158000000000006e-05,
+      "loss": 8.6845,
+      "step": 210500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.9156e-05,
+      "loss": 8.7141,
+      "step": 211000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.9154e-05,
+      "loss": 8.6976,
+      "step": 211500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.9152000000000006e-05,
+      "loss": 8.7004,
+      "step": 212000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.915e-05,
+      "loss": 8.7042,
+      "step": 212500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.9148e-05,
+      "loss": 8.7278,
+      "step": 213000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.9146e-05,
+      "loss": 8.7039,
+      "step": 213500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.9144e-05,
+      "loss": 8.7104,
+      "step": 214000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.9142000000000005e-05,
+      "loss": 8.6993,
+      "step": 214500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.914e-05,
+      "loss": 8.7215,
+      "step": 215000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.9138e-05,
+      "loss": 8.7066,
+      "step": 215500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.9136000000000005e-05,
+      "loss": 8.6924,
+      "step": 216000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.9134e-05,
+      "loss": 8.6996,
+      "step": 216500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.9132e-05,
+      "loss": 8.7252,
+      "step": 217000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.9130000000000006e-05,
+      "loss": 8.698,
+      "step": 217500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.9128e-05,
+      "loss": 8.7188,
+      "step": 218000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.9126e-05,
+      "loss": 8.7053,
+      "step": 218500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.9124000000000006e-05,
+      "loss": 8.7161,
+      "step": 219000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.9122e-05,
+      "loss": 8.711,
+      "step": 219500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.9120000000000004e-05,
+      "loss": 8.6989,
+      "step": 220000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.9118e-05,
+      "loss": 8.6944,
+      "step": 220500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.9116e-05,
+      "loss": 8.7141,
+      "step": 221000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.9114000000000004e-05,
+      "loss": 8.7241,
+      "step": 221500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.9112e-05,
+      "loss": 8.736,
+      "step": 222000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.911e-05,
+      "loss": 8.6733,
+      "step": 222500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.9108000000000005e-05,
+      "loss": 8.7222,
+      "step": 223000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.9106e-05,
+      "loss": 8.6865,
+      "step": 223500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.9104e-05,
+      "loss": 8.7143,
+      "step": 224000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.9102000000000005e-05,
+      "loss": 8.7387,
+      "step": 224500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.91e-05,
+      "loss": 8.6927,
+      "step": 225000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.9098e-05,
+      "loss": 8.7262,
+      "step": 225500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.9096000000000006e-05,
+      "loss": 8.7573,
+      "step": 226000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.9094e-05,
+      "loss": 8.7058,
+      "step": 226500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.9092000000000004e-05,
+      "loss": 8.7144,
+      "step": 227000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.9090000000000006e-05,
+      "loss": 8.6902,
+      "step": 227500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.9088e-05,
+      "loss": 8.7049,
+      "step": 228000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.9086000000000004e-05,
+      "loss": 8.7139,
+      "step": 228500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.9084e-05,
+      "loss": 8.7206,
+      "step": 229000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.9082e-05,
+      "loss": 8.7224,
+      "step": 229500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.9080000000000004e-05,
+      "loss": 8.7396,
+      "step": 230000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.9078e-05,
+      "loss": 8.7352,
+      "step": 230500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9076e-05,
+      "loss": 8.7382,
+      "step": 231000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9074000000000005e-05,
+      "loss": 8.7159,
+      "step": 231500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9072e-05,
+      "loss": 8.6957,
+      "step": 232000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.907e-05,
+      "loss": 8.7114,
+      "step": 232500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9068000000000005e-05,
+      "loss": 8.7356,
+      "step": 233000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9066e-05,
+      "loss": 8.692,
+      "step": 233500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9064e-05,
+      "loss": 8.7049,
+      "step": 234000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.9062000000000006e-05,
+      "loss": 8.7047,
+      "step": 234500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.906e-05,
+      "loss": 8.7047,
+      "step": 235000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.9058000000000004e-05,
+      "loss": 8.6655,
+      "step": 235500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.9056e-05,
+      "loss": 8.7262,
+      "step": 236000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.9054e-05,
+      "loss": 8.7033,
+      "step": 236500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.9052000000000004e-05,
+      "loss": 8.7357,
+      "step": 237000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.905e-05,
+      "loss": 8.719,
+      "step": 237500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.9048e-05,
+      "loss": 8.6847,
+      "step": 238000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.9046000000000004e-05,
+      "loss": 8.7089,
+      "step": 238500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.9044e-05,
+      "loss": 8.7355,
+      "step": 239000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.9042e-05,
+      "loss": 8.7264,
+      "step": 239500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.9040000000000005e-05,
+      "loss": 8.7056,
+      "step": 240000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.9038e-05,
+      "loss": 8.7343,
+      "step": 240500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.9036e-05,
+      "loss": 8.682,
+      "step": 241000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.9034000000000005e-05,
+      "loss": 8.717,
+      "step": 241500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9032e-05,
+      "loss": 8.7235,
+      "step": 242000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.903e-05,
+      "loss": 8.7069,
+      "step": 242500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9028e-05,
+      "loss": 8.7021,
+      "step": 243000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.9026e-05,
+      "loss": 8.6757,
+      "step": 243500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.9024000000000004e-05,
+      "loss": 8.7079,
+      "step": 244000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.9022e-05,
+      "loss": 8.6958,
+      "step": 244500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.902e-05,
+      "loss": 8.6831,
+      "step": 245000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.9018000000000004e-05,
+      "loss": 8.7316,
+      "step": 245500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.9016e-05,
+      "loss": 8.691,
+      "step": 246000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.9014e-05,
+      "loss": 8.7349,
+      "step": 246500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9012000000000004e-05,
+      "loss": 8.7,
+      "step": 247000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.901e-05,
+      "loss": 8.7117,
+      "step": 247500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9008e-05,
+      "loss": 8.7021,
+      "step": 248000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9006000000000005e-05,
+      "loss": 8.7323,
+      "step": 248500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9004e-05,
+      "loss": 8.7232,
+      "step": 249000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.9002e-05,
+      "loss": 8.6874,
+      "step": 249500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.9e-05,
+      "loss": 8.7074,
+      "step": 250000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.899800000000001e-05,
+      "loss": 8.7001,
+      "step": 250500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.8996e-05,
+      "loss": 8.7197,
+      "step": 251000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.8994e-05,
+      "loss": 8.6888,
+      "step": 251500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.8992e-05,
+      "loss": 8.6867,
+      "step": 252000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.8990000000000004e-05,
+      "loss": 8.7181,
+      "step": 252500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.8988e-05,
+      "loss": 8.7285,
+      "step": 253000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.8986e-05,
+      "loss": 8.6933,
+      "step": 253500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.8984000000000004e-05,
+      "loss": 8.7063,
+      "step": 254000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8982e-05,
+      "loss": 8.6891,
+      "step": 254500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.898e-05,
+      "loss": 8.7265,
+      "step": 255000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8978000000000004e-05,
+      "loss": 8.7078,
+      "step": 255500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8976e-05,
+      "loss": 8.7243,
+      "step": 256000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8974e-05,
+      "loss": 8.7045,
+      "step": 256500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.8972e-05,
+      "loss": 8.681,
+      "step": 257000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.897000000000001e-05,
+      "loss": 8.7018,
+      "step": 257500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.8968e-05,
+      "loss": 8.6847,
+      "step": 258000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.8966e-05,
+      "loss": 8.7208,
+      "step": 258500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.896400000000001e-05,
+      "loss": 8.7038,
+      "step": 259000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.8962e-05,
+      "loss": 8.7077,
+      "step": 259500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.896e-05,
+      "loss": 8.7176,
+      "step": 260000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.8958e-05,
+      "loss": 8.71,
+      "step": 260500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.8956000000000004e-05,
+      "loss": 8.6918,
+      "step": 261000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.8954e-05,
+      "loss": 8.6844,
+      "step": 261500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.8952e-05,
+      "loss": 8.7114,
+      "step": 262000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.8950000000000004e-05,
+      "loss": 8.6925,
+      "step": 262500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.8948000000000006e-05,
+      "loss": 8.7158,
+      "step": 263000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.8946e-05,
+      "loss": 8.7094,
+      "step": 263500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.8944e-05,
+      "loss": 8.6932,
+      "step": 264000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.894200000000001e-05,
+      "loss": 8.7112,
+      "step": 264500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.894e-05,
+      "loss": 8.7103,
+      "step": 265000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.8938e-05,
+      "loss": 8.7032,
+      "step": 265500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.893600000000001e-05,
+      "loss": 8.7108,
+      "step": 266000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.8934e-05,
+      "loss": 8.7083,
+      "step": 266500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.8932e-05,
+      "loss": 8.734,
+      "step": 267000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.893e-05,
+      "loss": 8.7262,
+      "step": 267500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.8928e-05,
+      "loss": 8.7284,
+      "step": 268000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.8926e-05,
+      "loss": 8.6903,
+      "step": 268500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.8924e-05,
+      "loss": 8.7144,
+      "step": 269000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.8922000000000004e-05,
+      "loss": 8.6904,
+      "step": 269500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.8920000000000006e-05,
+      "loss": 8.6849,
+      "step": 270000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.8918e-05,
+      "loss": 8.6964,
+      "step": 270500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.8916000000000004e-05,
+      "loss": 8.7546,
+      "step": 271000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.8914000000000006e-05,
+      "loss": 8.72,
+      "step": 271500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.8912e-05,
+      "loss": 8.7185,
+      "step": 272000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.891e-05,
+      "loss": 8.7199,
+      "step": 272500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.890800000000001e-05,
+      "loss": 8.7157,
+      "step": 273000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.8906e-05,
+      "loss": 8.7014,
+      "step": 273500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.8904e-05,
+      "loss": 8.7041,
+      "step": 274000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.8902e-05,
+      "loss": 8.695,
+      "step": 274500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.89e-05,
+      "loss": 8.7169,
+      "step": 275000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.8898000000000005e-05,
+      "loss": 8.7167,
+      "step": 275500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.8896e-05,
+      "loss": 8.6953,
+      "step": 276000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.8894e-05,
+      "loss": 8.7152,
+      "step": 276500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.8892000000000006e-05,
+      "loss": 8.71,
+      "step": 277000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.889e-05,
+      "loss": 8.7259,
+      "step": 277500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.8888000000000004e-05,
+      "loss": 8.698,
+      "step": 278000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.8886000000000006e-05,
+      "loss": 8.6828,
+      "step": 278500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.8884e-05,
+      "loss": 8.6984,
+      "step": 279000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.8882e-05,
+      "loss": 8.7121,
+      "step": 279500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.8880000000000006e-05,
+      "loss": 8.6799,
+      "step": 280000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.8878e-05,
+      "loss": 8.7074,
+      "step": 280500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.8876e-05,
+      "loss": 8.6898,
+      "step": 281000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.8874e-05,
+      "loss": 8.7204,
+      "step": 281500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.8872e-05,
+      "loss": 8.6998,
+      "step": 282000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.8870000000000005e-05,
+      "loss": 8.7008,
+      "step": 282500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.8868e-05,
+      "loss": 8.6955,
+      "step": 283000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8866e-05,
+      "loss": 8.6972,
+      "step": 283500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8864000000000005e-05,
+      "loss": 8.7064,
+      "step": 284000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8862e-05,
+      "loss": 8.6976,
+      "step": 284500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.886e-05,
+      "loss": 8.6878,
+      "step": 285000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8858000000000006e-05,
+      "loss": 8.7264,
+      "step": 285500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8856e-05,
+      "loss": 8.6855,
+      "step": 286000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8854e-05,
+      "loss": 8.7167,
+      "step": 286500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8852000000000006e-05,
+      "loss": 8.7011,
+      "step": 287000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.885e-05,
+      "loss": 8.6987,
+      "step": 287500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8848000000000004e-05,
+      "loss": 8.678,
+      "step": 288000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.8846e-05,
+      "loss": 8.7051,
+      "step": 288500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.8844e-05,
+      "loss": 8.7251,
+      "step": 289000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.8842000000000004e-05,
+      "loss": 8.6853,
+      "step": 289500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.884e-05,
+      "loss": 8.6958,
+      "step": 290000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.8838e-05,
+      "loss": 8.7028,
+      "step": 290500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.8836000000000005e-05,
+      "loss": 8.698,
+      "step": 291000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.8834e-05,
+      "loss": 8.7032,
+      "step": 291500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8832e-05,
+      "loss": 8.7055,
+      "step": 292000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8830000000000005e-05,
+      "loss": 8.7102,
+      "step": 292500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8828e-05,
+      "loss": 8.6988,
+      "step": 293000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.8826e-05,
+      "loss": 8.7191,
+      "step": 293500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.8824000000000006e-05,
+      "loss": 8.7392,
+      "step": 294000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.8822e-05,
+      "loss": 8.7151,
+      "step": 294500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.8820000000000004e-05,
+      "loss": 8.6793,
+      "step": 295000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.8818000000000006e-05,
+      "loss": 8.677,
+      "step": 295500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.8816e-05,
+      "loss": 8.7137,
+      "step": 296000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.8814000000000004e-05,
+      "loss": 8.6906,
+      "step": 296500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.8812e-05,
+      "loss": 8.6967,
+      "step": 297000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.881e-05,
+      "loss": 8.6804,
+      "step": 297500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.8808000000000004e-05,
+      "loss": 8.7055,
+      "step": 298000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8806e-05,
+      "loss": 8.7177,
+      "step": 298500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8804e-05,
+      "loss": 8.7065,
+      "step": 299000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.8802000000000005e-05,
+      "loss": 8.7076,
+      "step": 299500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.88e-05,
+      "loss": 8.7121,
+      "step": 300000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.8798e-05,
+      "loss": 8.7141,
+      "step": 300500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.8796000000000005e-05,
+      "loss": 8.6999,
+      "step": 301000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.8794e-05,
+      "loss": 8.6949,
+      "step": 301500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.8792e-05,
+      "loss": 8.6949,
+      "step": 302000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.8790000000000006e-05,
+      "loss": 8.7159,
+      "step": 302500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.8788e-05,
+      "loss": 8.7052,
+      "step": 303000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.8786000000000004e-05,
+      "loss": 8.6986,
+      "step": 303500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.8784e-05,
+      "loss": 8.7041,
+      "step": 304000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.8782e-05,
+      "loss": 8.7078,
+      "step": 304500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.8780000000000004e-05,
+      "loss": 8.6994,
+      "step": 305000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.8778e-05,
+      "loss": 8.6767,
+      "step": 305500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.8776e-05,
+      "loss": 8.6984,
+      "step": 306000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.8774000000000004e-05,
+      "loss": 8.7268,
+      "step": 306500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.8772e-05,
+      "loss": 8.7017,
+      "step": 307000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.877e-05,
+      "loss": 8.7056,
+      "step": 307500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.8768000000000005e-05,
+      "loss": 8.7253,
+      "step": 308000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.8766e-05,
+      "loss": 8.7167,
+      "step": 308500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.8764e-05,
+      "loss": 8.6715,
+      "step": 309000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.8762000000000005e-05,
+      "loss": 8.7112,
+      "step": 309500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.876e-05,
+      "loss": 8.7181,
+      "step": 310000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.8758e-05,
+      "loss": 8.6818,
+      "step": 310500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.8756e-05,
+      "loss": 8.6989,
+      "step": 311000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.8754e-05,
+      "loss": 8.7037,
+      "step": 311500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.8752000000000004e-05,
+      "loss": 8.6931,
+      "step": 312000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.875e-05,
+      "loss": 8.7028,
+      "step": 312500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.8748e-05,
+      "loss": 8.6934,
+      "step": 313000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.8746000000000004e-05,
+      "loss": 8.7059,
+      "step": 313500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.8744e-05,
+      "loss": 8.6945,
+      "step": 314000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.8742e-05,
+      "loss": 8.67,
+      "step": 314500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.8740000000000004e-05,
+      "loss": 8.6978,
+      "step": 315000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.8738e-05,
+      "loss": 8.7081,
+      "step": 315500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8736e-05,
+      "loss": 8.6743,
+      "step": 316000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8734000000000005e-05,
+      "loss": 8.7259,
+      "step": 316500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8732e-05,
+      "loss": 8.7153,
+      "step": 317000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.873e-05,
+      "loss": 8.7164,
+      "step": 317500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8728e-05,
+      "loss": 8.6748,
+      "step": 318000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.872600000000001e-05,
+      "loss": 8.6927,
+      "step": 318500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.8724e-05,
+      "loss": 8.7285,
+      "step": 319000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.8722e-05,
+      "loss": 8.6986,
+      "step": 319500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.872000000000001e-05,
+      "loss": 8.7152,
+      "step": 320000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.8718000000000003e-05,
+      "loss": 8.7072,
+      "step": 320500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8716e-05,
+      "loss": 8.6971,
+      "step": 321000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8714e-05,
+      "loss": 8.714,
+      "step": 321500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8712000000000004e-05,
+      "loss": 8.703,
+      "step": 322000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.871e-05,
+      "loss": 8.7072,
+      "step": 322500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8708e-05,
+      "loss": 8.6667,
+      "step": 323000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.8706000000000004e-05,
+      "loss": 8.7107,
+      "step": 323500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.8704e-05,
+      "loss": 8.701,
+      "step": 324000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.8702e-05,
+      "loss": 8.7082,
+      "step": 324500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.87e-05,
+      "loss": 8.701,
+      "step": 325000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.869800000000001e-05,
+      "loss": 8.6732,
+      "step": 325500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.8696e-05,
+      "loss": 8.7238,
+      "step": 326000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.8694e-05,
+      "loss": 8.7107,
+      "step": 326500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.869200000000001e-05,
+      "loss": 8.7347,
+      "step": 327000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.869e-05,
+      "loss": 8.6986,
+      "step": 327500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.8688e-05,
+      "loss": 8.6949,
+      "step": 328000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8686e-05,
+      "loss": 8.6919,
+      "step": 328500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8684000000000003e-05,
+      "loss": 8.7026,
+      "step": 329000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.8682e-05,
+      "loss": 8.6933,
+      "step": 329500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.868e-05,
+      "loss": 8.7021,
+      "step": 330000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.8678000000000004e-05,
+      "loss": 8.7339,
+      "step": 330500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8676000000000006e-05,
+      "loss": 8.7103,
+      "step": 331000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8674e-05,
+      "loss": 8.683,
+      "step": 331500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8672000000000004e-05,
+      "loss": 8.707,
+      "step": 332000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.867000000000001e-05,
+      "loss": 8.713,
+      "step": 332500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8668e-05,
+      "loss": 8.7049,
+      "step": 333000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8666e-05,
+      "loss": 8.6956,
+      "step": 333500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.866400000000001e-05,
+      "loss": 8.6998,
+      "step": 334000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.8662e-05,
+      "loss": 8.7074,
+      "step": 334500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.866e-05,
+      "loss": 8.7095,
+      "step": 335000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.8658e-05,
+      "loss": 8.7078,
+      "step": 335500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.8656e-05,
+      "loss": 8.6895,
+      "step": 336000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.8654e-05,
+      "loss": 8.6883,
+      "step": 336500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.8652e-05,
+      "loss": 8.7053,
+      "step": 337000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.8650000000000003e-05,
+      "loss": 8.6985,
+      "step": 337500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.8648000000000006e-05,
+      "loss": 8.7011,
+      "step": 338000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8646e-05,
+      "loss": 8.6913,
+      "step": 338500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8644000000000004e-05,
+      "loss": 8.699,
+      "step": 339000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.8642000000000006e-05,
+      "loss": 8.7051,
+      "step": 339500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.864e-05,
+      "loss": 8.6863,
+      "step": 340000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.8638e-05,
+      "loss": 8.69,
+      "step": 340500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.863600000000001e-05,
+      "loss": 8.6958,
+      "step": 341000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8634e-05,
+      "loss": 8.6927,
+      "step": 341500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.8632e-05,
+      "loss": 8.6934,
+      "step": 342000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.863e-05,
+      "loss": 8.7063,
+      "step": 342500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.8628e-05,
+      "loss": 8.7023,
+      "step": 343000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.8626000000000005e-05,
+      "loss": 8.7075,
+      "step": 343500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.8624e-05,
+      "loss": 8.6927,
+      "step": 344000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8622e-05,
+      "loss": 8.7182,
+      "step": 344500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8620000000000005e-05,
+      "loss": 8.6924,
+      "step": 345000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8618e-05,
+      "loss": 8.6963,
+      "step": 345500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.8616000000000003e-05,
+      "loss": 8.6809,
+      "step": 346000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.8614000000000006e-05,
+      "loss": 8.6995,
+      "step": 346500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.8612e-05,
+      "loss": 8.7175,
+      "step": 347000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.861e-05,
+      "loss": 8.7025,
+      "step": 347500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.8608000000000006e-05,
+      "loss": 8.7097,
+      "step": 348000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.8606e-05,
+      "loss": 8.699,
+      "step": 348500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.8604000000000004e-05,
+      "loss": 8.7232,
+      "step": 349000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.8602e-05,
+      "loss": 8.713,
+      "step": 349500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.86e-05,
+      "loss": 8.7031,
+      "step": 350000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.8598000000000005e-05,
+      "loss": 8.6899,
+      "step": 350500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.8596e-05,
+      "loss": 8.6683,
+      "step": 351000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.8594e-05,
+      "loss": 8.6702,
+      "step": 351500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.8592000000000005e-05,
+      "loss": 8.6919,
+      "step": 352000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.859e-05,
+      "loss": 8.6994,
+      "step": 352500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.8588e-05,
+      "loss": 8.7184,
+      "step": 353000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.8586000000000005e-05,
+      "loss": 8.7062,
+      "step": 353500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.8584e-05,
+      "loss": 8.7013,
+      "step": 354000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.8582e-05,
+      "loss": 8.7126,
+      "step": 354500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.8580000000000006e-05,
+      "loss": 8.6769,
+      "step": 355000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.8578e-05,
+      "loss": 8.6851,
+      "step": 355500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.8576000000000004e-05,
+      "loss": 8.6912,
+      "step": 356000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.8574000000000006e-05,
+      "loss": 8.711,
+      "step": 356500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.8572e-05,
+      "loss": 8.7133,
+      "step": 357000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.8570000000000004e-05,
+      "loss": 8.7251,
+      "step": 357500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.8568e-05,
+      "loss": 8.6906,
+      "step": 358000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.8566e-05,
+      "loss": 8.688,
+      "step": 358500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.8564000000000005e-05,
+      "loss": 8.7029,
+      "step": 359000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.8562e-05,
+      "loss": 8.7185,
+      "step": 359500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.856e-05,
+      "loss": 8.7108,
+      "step": 360000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.8558000000000005e-05,
+      "loss": 8.7167,
+      "step": 360500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.8556e-05,
+      "loss": 8.6894,
+      "step": 361000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.8554e-05,
+      "loss": 8.6856,
+      "step": 361500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.8552000000000005e-05,
+      "loss": 8.675,
+      "step": 362000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.855e-05,
+      "loss": 8.6954,
+      "step": 362500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.8548000000000003e-05,
+      "loss": 8.6984,
+      "step": 363000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8546000000000006e-05,
+      "loss": 8.7377,
+      "step": 363500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8544e-05,
+      "loss": 8.6971,
+      "step": 364000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.8542000000000004e-05,
+      "loss": 8.6765,
+      "step": 364500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.854e-05,
+      "loss": 8.7073,
+      "step": 365000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.8538e-05,
+      "loss": 8.6889,
+      "step": 365500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.8536000000000004e-05,
+      "loss": 8.7079,
+      "step": 366000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.8534e-05,
+      "loss": 8.6962,
+      "step": 366500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.8532e-05,
+      "loss": 8.6914,
+      "step": 367000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.8530000000000005e-05,
+      "loss": 8.7295,
+      "step": 367500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.8528e-05,
+      "loss": 8.7068,
+      "step": 368000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.8526e-05,
+      "loss": 8.6933,
+      "step": 368500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.8524000000000005e-05,
+      "loss": 8.6943,
+      "step": 369000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.8522e-05,
+      "loss": 8.6945,
+      "step": 369500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.852e-05,
+      "loss": 8.7121,
+      "step": 370000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.8518000000000005e-05,
+      "loss": 8.686,
+      "step": 370500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8516e-05,
+      "loss": 8.7053,
+      "step": 371000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8514000000000003e-05,
+      "loss": 8.6727,
+      "step": 371500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.8512e-05,
+      "loss": 8.6801,
+      "step": 372000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.851e-05,
+      "loss": 8.6943,
+      "step": 372500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.8508000000000004e-05,
+      "loss": 8.687,
+      "step": 373000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8506e-05,
+      "loss": 8.7085,
+      "step": 373500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8504e-05,
+      "loss": 8.7113,
+      "step": 374000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.8502000000000004e-05,
+      "loss": 8.7011,
+      "step": 374500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.85e-05,
+      "loss": 8.699,
+      "step": 375000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.8498e-05,
+      "loss": 8.6982,
+      "step": 375500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.8496000000000005e-05,
+      "loss": 8.7074,
+      "step": 376000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.8494e-05,
+      "loss": 8.6757,
+      "step": 376500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.8492e-05,
+      "loss": 8.7023,
+      "step": 377000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.8490000000000005e-05,
+      "loss": 8.6954,
+      "step": 377500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.8488e-05,
+      "loss": 8.6936,
+      "step": 378000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.8486e-05,
+      "loss": 8.7095,
+      "step": 378500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.8484e-05,
+      "loss": 8.7091,
+      "step": 379000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.8482e-05,
+      "loss": 8.7081,
+      "step": 379500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.8480000000000003e-05,
+      "loss": 8.6958,
+      "step": 380000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.8478e-05,
+      "loss": 8.6931,
+      "step": 380500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.8476e-05,
+      "loss": 8.7034,
+      "step": 381000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.8474000000000004e-05,
+      "loss": 8.6621,
+      "step": 381500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.8472e-05,
+      "loss": 8.6779,
+      "step": 382000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.847e-05,
+      "loss": 8.7087,
+      "step": 382500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.8468000000000004e-05,
+      "loss": 8.7243,
+      "step": 383000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.8466e-05,
+      "loss": 8.6768,
+      "step": 383500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.8464e-05,
+      "loss": 8.6821,
+      "step": 384000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.8462000000000005e-05,
+      "loss": 8.7114,
+      "step": 384500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.846e-05,
+      "loss": 8.6725,
+      "step": 385000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.8458e-05,
+      "loss": 8.7012,
+      "step": 385500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.8456e-05,
+      "loss": 8.6869,
+      "step": 386000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.845400000000001e-05,
+      "loss": 8.7117,
+      "step": 386500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.8452e-05,
+      "loss": 8.715,
+      "step": 387000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.845e-05,
+      "loss": 8.6688,
+      "step": 387500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.844800000000001e-05,
+      "loss": 8.6967,
+      "step": 388000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.8446e-05,
+      "loss": 8.6825,
+      "step": 388500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.8444e-05,
+      "loss": 8.6951,
+      "step": 389000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.8442e-05,
+      "loss": 8.6929,
+      "step": 389500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.8440000000000004e-05,
+      "loss": 8.7185,
+      "step": 390000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.8438e-05,
+      "loss": 8.6915,
+      "step": 390500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.8436e-05,
+      "loss": 8.7071,
+      "step": 391000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.8434000000000004e-05,
+      "loss": 8.6841,
+      "step": 391500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.8432e-05,
+      "loss": 8.6782,
+      "step": 392000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.843e-05,
+      "loss": 8.6894,
+      "step": 392500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.8428e-05,
+      "loss": 8.6945,
+      "step": 393000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.842600000000001e-05,
+      "loss": 8.684,
+      "step": 393500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.8424e-05,
+      "loss": 8.654,
+      "step": 394000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.8422e-05,
+      "loss": 8.6775,
+      "step": 394500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.842000000000001e-05,
+      "loss": 8.6904,
+      "step": 395000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.8418e-05,
+      "loss": 8.7069,
+      "step": 395500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.8416e-05,
+      "loss": 8.691,
+      "step": 396000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.8414e-05,
+      "loss": 8.6932,
+      "step": 396500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.8412e-05,
+      "loss": 8.6928,
+      "step": 397000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.841e-05,
+      "loss": 8.7049,
+      "step": 397500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.8408e-05,
+      "loss": 8.7108,
+      "step": 398000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.8406000000000004e-05,
+      "loss": 8.7185,
+      "step": 398500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.8404000000000006e-05,
+      "loss": 8.718,
+      "step": 399000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.8402e-05,
+      "loss": 8.678,
+      "step": 399500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 8.6971,
+      "step": 400000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.8398000000000007e-05,
+      "loss": 8.6769,
+      "step": 400500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.8396e-05,
+      "loss": 8.7164,
+      "step": 401000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.8394e-05,
+      "loss": 8.6986,
+      "step": 401500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.839200000000001e-05,
+      "loss": 8.6726,
+      "step": 402000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.839e-05,
+      "loss": 8.6929,
+      "step": 402500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.8388e-05,
+      "loss": 8.7256,
+      "step": 403000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.8386e-05,
+      "loss": 8.6922,
+      "step": 403500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.8384e-05,
+      "loss": 8.6959,
+      "step": 404000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.8382e-05,
+      "loss": 8.6909,
+      "step": 404500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.838e-05,
+      "loss": 8.6923,
+      "step": 405000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.8378e-05,
+      "loss": 8.6927,
+      "step": 405500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.8376000000000006e-05,
+      "loss": 8.7172,
+      "step": 406000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.8374e-05,
+      "loss": 8.6758,
+      "step": 406500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.8372000000000004e-05,
+      "loss": 8.7046,
+      "step": 407000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.8370000000000006e-05,
+      "loss": 8.7108,
+      "step": 407500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.8368e-05,
+      "loss": 8.692,
+      "step": 408000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.8366e-05,
+      "loss": 8.689,
+      "step": 408500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.8364000000000007e-05,
+      "loss": 8.6759,
+      "step": 409000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.8362e-05,
+      "loss": 8.6801,
+      "step": 409500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.836e-05,
+      "loss": 8.6988,
+      "step": 410000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.8358e-05,
+      "loss": 8.6677,
+      "step": 410500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.8356e-05,
+      "loss": 8.6791,
+      "step": 411000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.8354000000000005e-05,
+      "loss": 8.6931,
+      "step": 411500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.8352e-05,
+      "loss": 8.668,
+      "step": 412000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.835e-05,
+      "loss": 8.694,
+      "step": 412500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.8348000000000005e-05,
+      "loss": 8.6952,
+      "step": 413000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.8346e-05,
+      "loss": 8.6957,
+      "step": 413500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.8344e-05,
+      "loss": 8.6833,
+      "step": 414000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.8342000000000006e-05,
+      "loss": 8.7016,
+      "step": 414500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.834e-05,
+      "loss": 8.7104,
+      "step": 415000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.8338e-05,
+      "loss": 8.6991,
+      "step": 415500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.8336000000000006e-05,
+      "loss": 8.6973,
+      "step": 416000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.8334e-05,
+      "loss": 8.7134,
+      "step": 416500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.8332000000000004e-05,
+      "loss": 8.6838,
+      "step": 417000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.833e-05,
+      "loss": 8.7218,
+      "step": 417500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.8328e-05,
+      "loss": 8.7094,
+      "step": 418000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.8326000000000005e-05,
+      "loss": 8.684,
+      "step": 418500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.8324e-05,
+      "loss": 8.7029,
+      "step": 419000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.8322e-05,
+      "loss": 8.6928,
+      "step": 419500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.8320000000000005e-05,
+      "loss": 8.6705,
+      "step": 420000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.8318e-05,
+      "loss": 8.6614,
+      "step": 420500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.8316e-05,
+      "loss": 8.6882,
+      "step": 421000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.8314000000000005e-05,
+      "loss": 8.6703,
+      "step": 421500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.8312e-05,
+      "loss": 8.6985,
+      "step": 422000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.8309999999999997e-05,
+      "loss": 8.686,
+      "step": 422500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.8308000000000006e-05,
+      "loss": 8.6813,
+      "step": 423000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.8306e-05,
+      "loss": 8.7127,
+      "step": 423500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.8304000000000004e-05,
+      "loss": 8.6972,
+      "step": 424000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8302000000000006e-05,
+      "loss": 8.7183,
+      "step": 424500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.83e-05,
+      "loss": 8.7121,
+      "step": 425000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8298000000000004e-05,
+      "loss": 8.7204,
+      "step": 425500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8296e-05,
+      "loss": 8.7153,
+      "step": 426000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8294e-05,
+      "loss": 8.6959,
+      "step": 426500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8292000000000005e-05,
+      "loss": 8.7169,
+      "step": 427000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.829e-05,
+      "loss": 8.6833,
+      "step": 427500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8288e-05,
+      "loss": 8.707,
+      "step": 428000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.8286000000000005e-05,
+      "loss": 8.6847,
+      "step": 428500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.8284e-05,
+      "loss": 8.6911,
+      "step": 429000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.8282e-05,
+      "loss": 8.7043,
+      "step": 429500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.8280000000000005e-05,
+      "loss": 8.7095,
+      "step": 430000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.8278e-05,
+      "loss": 8.6738,
+      "step": 430500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.8276e-05,
+      "loss": 8.7012,
+      "step": 431000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.8274000000000006e-05,
+      "loss": 8.6963,
+      "step": 431500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.8272e-05,
+      "loss": 8.7061,
+      "step": 432000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.8270000000000004e-05,
+      "loss": 8.6929,
+      "step": 432500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.8268e-05,
+      "loss": 8.6864,
+      "step": 433000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.8266e-05,
+      "loss": 8.701,
+      "step": 433500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.8264000000000004e-05,
+      "loss": 8.7049,
+      "step": 434000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.8262e-05,
+      "loss": 8.6915,
+      "step": 434500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.826e-05,
+      "loss": 8.7128,
+      "step": 435000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.8258000000000005e-05,
+      "loss": 8.7074,
+      "step": 435500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.8256e-05,
+      "loss": 8.6953,
+      "step": 436000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.8254e-05,
+      "loss": 8.7117,
+      "step": 436500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.8252000000000005e-05,
+      "loss": 8.6716,
+      "step": 437000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.825e-05,
+      "loss": 8.7095,
+      "step": 437500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.8248e-05,
+      "loss": 8.7017,
+      "step": 438000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.8246000000000005e-05,
+      "loss": 8.6966,
+      "step": 438500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.8244e-05,
+      "loss": 8.674,
+      "step": 439000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.8242e-05,
+      "loss": 8.6855,
+      "step": 439500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.824e-05,
+      "loss": 8.6747,
+      "step": 440000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.8238e-05,
+      "loss": 8.701,
+      "step": 440500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.8236000000000004e-05,
+      "loss": 8.6833,
+      "step": 441000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.8234e-05,
+      "loss": 8.6943,
+      "step": 441500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.8232e-05,
+      "loss": 8.689,
+      "step": 442000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.8230000000000004e-05,
+      "loss": 8.6998,
+      "step": 442500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.8228e-05,
+      "loss": 8.7034,
+      "step": 443000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.8226e-05,
+      "loss": 8.7114,
+      "step": 443500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.8224000000000004e-05,
+      "loss": 8.6846,
+      "step": 444000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.8222e-05,
+      "loss": 8.6767,
+      "step": 444500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.822e-05,
+      "loss": 8.726,
+      "step": 445000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.8218000000000005e-05,
+      "loss": 8.6712,
+      "step": 445500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.8216e-05,
+      "loss": 8.6972,
+      "step": 446000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.8214e-05,
+      "loss": 8.7112,
+      "step": 446500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.8212e-05,
+      "loss": 8.6705,
+      "step": 447000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.821e-05,
+      "loss": 8.7263,
+      "step": 447500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.8208e-05,
+      "loss": 8.6963,
+      "step": 448000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.8206e-05,
+      "loss": 8.7303,
+      "step": 448500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.820400000000001e-05,
+      "loss": 8.707,
+      "step": 449000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.8202000000000004e-05,
+      "loss": 8.6949,
+      "step": 449500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.82e-05,
+      "loss": 8.7131,
+      "step": 450000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.8198e-05,
+      "loss": 8.683,
+      "step": 450500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.8196000000000004e-05,
+      "loss": 8.6739,
+      "step": 451000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.8194e-05,
+      "loss": 8.693,
+      "step": 451500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.8192e-05,
+      "loss": 8.7077,
+      "step": 452000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.8190000000000004e-05,
+      "loss": 8.6969,
+      "step": 452500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.8188e-05,
+      "loss": 8.6799,
+      "step": 453000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.8186e-05,
+      "loss": 8.7057,
+      "step": 453500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.8184e-05,
+      "loss": 8.7014,
+      "step": 454000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.818200000000001e-05,
+      "loss": 8.7065,
+      "step": 454500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.818e-05,
+      "loss": 8.7073,
+      "step": 455000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.8178e-05,
+      "loss": 8.7097,
+      "step": 455500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.817600000000001e-05,
+      "loss": 8.7036,
+      "step": 456000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.8174e-05,
+      "loss": 8.6856,
+      "step": 456500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.8172e-05,
+      "loss": 8.7143,
+      "step": 457000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.817e-05,
+      "loss": 8.6936,
+      "step": 457500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.8168000000000004e-05,
+      "loss": 8.7067,
+      "step": 458000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.8166e-05,
+      "loss": 8.6913,
+      "step": 458500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.8164e-05,
+      "loss": 8.6758,
+      "step": 459000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.8162000000000004e-05,
+      "loss": 8.6876,
+      "step": 459500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.816e-05,
+      "loss": 8.6895,
+      "step": 460000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.8158e-05,
+      "loss": 8.7215,
+      "step": 460500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8156000000000004e-05,
+      "loss": 8.6853,
+      "step": 461000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.815400000000001e-05,
+      "loss": 8.6857,
+      "step": 461500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8152e-05,
+      "loss": 8.7186,
+      "step": 462000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.815e-05,
+      "loss": 8.6972,
+      "step": 462500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.814800000000001e-05,
+      "loss": 8.705,
+      "step": 463000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.8146e-05,
+      "loss": 8.7002,
+      "step": 463500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.8144e-05,
+      "loss": 8.6838,
+      "step": 464000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.8142e-05,
+      "loss": 8.6851,
+      "step": 464500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.814e-05,
+      "loss": 8.7308,
+      "step": 465000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.8138e-05,
+      "loss": 8.6843,
+      "step": 465500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.8136e-05,
+      "loss": 8.678,
+      "step": 466000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.8134000000000004e-05,
+      "loss": 8.661,
+      "step": 466500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.8132000000000006e-05,
+      "loss": 8.6863,
+      "step": 467000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.813e-05,
+      "loss": 8.6986,
+      "step": 467500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.8128000000000004e-05,
+      "loss": 8.7165,
+      "step": 468000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.8126000000000006e-05,
+      "loss": 8.6645,
+      "step": 468500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.8124e-05,
+      "loss": 8.7061,
+      "step": 469000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.8122e-05,
+      "loss": 8.6844,
+      "step": 469500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.812000000000001e-05,
+      "loss": 8.6871,
+      "step": 470000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.8118e-05,
+      "loss": 8.711,
+      "step": 470500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.8116e-05,
+      "loss": 8.6687,
+      "step": 471000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.8114e-05,
+      "loss": 8.6982,
+      "step": 471500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.8112e-05,
+      "loss": 8.6865,
+      "step": 472000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.8110000000000005e-05,
+      "loss": 8.6927,
+      "step": 472500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.8108e-05,
+      "loss": 8.6943,
+      "step": 473000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8106e-05,
+      "loss": 8.6945,
+      "step": 473500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8104000000000006e-05,
+      "loss": 8.6925,
+      "step": 474000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8102e-05,
+      "loss": 8.6788,
+      "step": 474500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8100000000000004e-05,
+      "loss": 8.7,
+      "step": 475000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8098000000000006e-05,
+      "loss": 8.6685,
+      "step": 475500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.8096e-05,
+      "loss": 8.6779,
+      "step": 476000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.8094e-05,
+      "loss": 8.6893,
+      "step": 476500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.8092000000000006e-05,
+      "loss": 8.6897,
+      "step": 477000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.809e-05,
+      "loss": 8.7191,
+      "step": 477500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.8088e-05,
+      "loss": 8.663,
+      "step": 478000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.8086e-05,
+      "loss": 8.6669,
+      "step": 478500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.8084e-05,
+      "loss": 8.6691,
+      "step": 479000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.8082000000000005e-05,
+      "loss": 8.6918,
+      "step": 479500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.808e-05,
+      "loss": 8.6969,
+      "step": 480000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.8078e-05,
+      "loss": 8.6683,
+      "step": 480500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.8076000000000005e-05,
+      "loss": 8.7101,
+      "step": 481000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.8074e-05,
+      "loss": 8.6696,
+      "step": 481500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.8072e-05,
+      "loss": 8.6914,
+      "step": 482000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.8070000000000006e-05,
+      "loss": 8.6875,
+      "step": 482500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.8068e-05,
+      "loss": 8.7111,
+      "step": 483000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.8066e-05,
+      "loss": 8.708,
+      "step": 483500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.8064000000000006e-05,
+      "loss": 8.6654,
+      "step": 484000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.8062e-05,
+      "loss": 8.6937,
+      "step": 484500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.8060000000000004e-05,
+      "loss": 8.6961,
+      "step": 485000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.8058e-05,
+      "loss": 8.7011,
+      "step": 485500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.8056e-05,
+      "loss": 8.7058,
+      "step": 486000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.8054000000000004e-05,
+      "loss": 8.6886,
+      "step": 486500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.8052e-05,
+      "loss": 8.7001,
+      "step": 487000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.805e-05,
+      "loss": 8.7077,
+      "step": 487500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.8048000000000005e-05,
+      "loss": 8.6888,
+      "step": 488000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.8046e-05,
+      "loss": 8.7076,
+      "step": 488500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.8044e-05,
+      "loss": 8.6638,
+      "step": 489000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.8042000000000005e-05,
+      "loss": 8.6892,
+      "step": 489500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.804e-05,
+      "loss": 8.676,
+      "step": 490000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.8037999999999996e-05,
+      "loss": 8.7038,
+      "step": 490500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.8036000000000006e-05,
+      "loss": 8.6922,
+      "step": 491000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.8034e-05,
+      "loss": 8.6947,
+      "step": 491500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.8032000000000004e-05,
+      "loss": 8.6718,
+      "step": 492000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.8030000000000006e-05,
+      "loss": 8.7013,
+      "step": 492500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.8028e-05,
+      "loss": 8.6745,
+      "step": 493000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.8026000000000004e-05,
+      "loss": 8.7123,
+      "step": 493500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.8024e-05,
+      "loss": 8.7301,
+      "step": 494000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.8022e-05,
+      "loss": 8.7264,
+      "step": 494500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.8020000000000004e-05,
+      "loss": 8.6981,
+      "step": 495000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.8018e-05,
+      "loss": 8.697,
+      "step": 495500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.8016e-05,
+      "loss": 8.6812,
+      "step": 496000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.8014000000000005e-05,
+      "loss": 8.7023,
+      "step": 496500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.8012e-05,
+      "loss": 8.6937,
+      "step": 497000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.801e-05,
+      "loss": 8.6924,
+      "step": 497500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.8008000000000005e-05,
+      "loss": 8.7007,
+      "step": 498000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.8006e-05,
+      "loss": 8.7113,
+      "step": 498500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.8004e-05,
+      "loss": 8.6855,
+      "step": 499000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.8002000000000006e-05,
+      "loss": 8.6865,
+      "step": 499500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.8e-05,
+      "loss": 8.7076,
+      "step": 500000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.7998000000000004e-05,
+      "loss": 8.6981,
+      "step": 500500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.7996e-05,
+      "loss": 8.6832,
+      "step": 501000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.7994e-05,
+      "loss": 8.6853,
+      "step": 501500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.7992000000000004e-05,
+      "loss": 8.6963,
+      "step": 502000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.799e-05,
+      "loss": 8.6834,
+      "step": 502500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.7988e-05,
+      "loss": 8.6712,
+      "step": 503000
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.7986000000000004e-05,
+      "loss": 8.7123,
+      "step": 503500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.7984e-05,
+      "loss": 8.6588,
+      "step": 504000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.7982e-05,
+      "loss": 8.6922,
+      "step": 504500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.7980000000000005e-05,
+      "loss": 8.702,
+      "step": 505000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.7978e-05,
+      "loss": 8.7003,
+      "step": 505500
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.7976e-05,
+      "loss": 8.6942,
+      "step": 506000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.7974000000000005e-05,
+      "loss": 8.7065,
+      "step": 506500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.7972e-05,
+      "loss": 8.6901,
+      "step": 507000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.797e-05,
+      "loss": 8.6761,
+      "step": 507500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.7968e-05,
+      "loss": 8.6732,
+      "step": 508000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.7966e-05,
+      "loss": 8.6666,
+      "step": 508500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.7964000000000004e-05,
+      "loss": 8.7143,
+      "step": 509000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.7962e-05,
+      "loss": 8.7032,
+      "step": 509500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.796e-05,
+      "loss": 8.6971,
+      "step": 510000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.7958000000000004e-05,
+      "loss": 8.7349,
+      "step": 510500
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.7956e-05,
+      "loss": 8.7171,
+      "step": 511000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.7954e-05,
+      "loss": 8.6922,
+      "step": 511500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.7952000000000004e-05,
+      "loss": 8.6794,
+      "step": 512000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.795e-05,
+      "loss": 8.7029,
+      "step": 512500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.7948e-05,
+      "loss": 8.6932,
+      "step": 513000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.7946000000000005e-05,
+      "loss": 8.7035,
+      "step": 513500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.7944e-05,
+      "loss": 8.6815,
+      "step": 514000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.7942e-05,
+      "loss": 8.6972,
+      "step": 514500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.794e-05,
+      "loss": 8.6814,
+      "step": 515000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.7938e-05,
+      "loss": 8.6591,
+      "step": 515500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.7936e-05,
+      "loss": 8.6983,
+      "step": 516000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.7934e-05,
+      "loss": 8.7041,
+      "step": 516500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.793200000000001e-05,
+      "loss": 8.6969,
+      "step": 517000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.7930000000000004e-05,
+      "loss": 8.7102,
+      "step": 517500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.7928e-05,
+      "loss": 8.6954,
+      "step": 518000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.7926e-05,
+      "loss": 8.6718,
+      "step": 518500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.7924000000000004e-05,
+      "loss": 8.6878,
+      "step": 519000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.7922e-05,
+      "loss": 8.6994,
+      "step": 519500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.792e-05,
+      "loss": 8.6711,
+      "step": 520000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.7918000000000004e-05,
+      "loss": 8.7005,
+      "step": 520500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.7916e-05,
+      "loss": 8.6749,
+      "step": 521000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.7914e-05,
+      "loss": 8.6821,
+      "step": 521500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.7912e-05,
+      "loss": 8.6771,
+      "step": 522000
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.791000000000001e-05,
+      "loss": 8.6896,
+      "step": 522500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.7908e-05,
+      "loss": 8.7081,
+      "step": 523000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.7906e-05,
+      "loss": 8.6892,
+      "step": 523500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.790400000000001e-05,
+      "loss": 8.6998,
+      "step": 524000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.7902e-05,
+      "loss": 8.6664,
+      "step": 524500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.79e-05,
+      "loss": 8.7092,
+      "step": 525000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.7898e-05,
+      "loss": 8.6845,
+      "step": 525500
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.7896000000000004e-05,
+      "loss": 8.7126,
+      "step": 526000
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.7894e-05,
+      "loss": 8.6845,
+      "step": 526500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.7892e-05,
+      "loss": 8.721,
+      "step": 527000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.7890000000000004e-05,
+      "loss": 8.7022,
+      "step": 527500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.7888e-05,
+      "loss": 8.6862,
+      "step": 528000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.7886e-05,
+      "loss": 8.6964,
+      "step": 528500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.7884000000000004e-05,
+      "loss": 8.6632,
+      "step": 529000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.788200000000001e-05,
+      "loss": 8.7079,
+      "step": 529500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.788e-05,
+      "loss": 8.7429,
+      "step": 530000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.7878e-05,
+      "loss": 8.7043,
+      "step": 530500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.787600000000001e-05,
+      "loss": 8.7092,
+      "step": 531000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.7874e-05,
+      "loss": 8.6881,
+      "step": 531500
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.7872e-05,
+      "loss": 8.6922,
+      "step": 532000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.787e-05,
+      "loss": 8.6983,
+      "step": 532500
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.7868e-05,
+      "loss": 8.7213,
+      "step": 533000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.7866e-05,
+      "loss": 8.6881,
+      "step": 533500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.7864e-05,
+      "loss": 8.6939,
+      "step": 534000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.7862000000000004e-05,
+      "loss": 8.7001,
+      "step": 534500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.7860000000000006e-05,
+      "loss": 8.6931,
+      "step": 535000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.7858e-05,
+      "loss": 8.7165,
+      "step": 535500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.7856000000000004e-05,
+      "loss": 8.6589,
+      "step": 536000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.7854000000000006e-05,
+      "loss": 8.6572,
+      "step": 536500
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.7852e-05,
+      "loss": 8.6591,
+      "step": 537000
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.785e-05,
+      "loss": 8.6989,
+      "step": 537500
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.784800000000001e-05,
+      "loss": 8.6687,
+      "step": 538000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.7846e-05,
+      "loss": 8.696,
+      "step": 538500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.7844e-05,
+      "loss": 8.6808,
+      "step": 539000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.7842e-05,
+      "loss": 8.6872,
+      "step": 539500
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.784e-05,
+      "loss": 8.6958,
+      "step": 540000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.7838000000000005e-05,
+      "loss": 8.6808,
+      "step": 540500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.7836e-05,
+      "loss": 8.7062,
+      "step": 541000
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.7834e-05,
+      "loss": 8.7068,
+      "step": 541500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.7832000000000006e-05,
+      "loss": 8.6692,
+      "step": 542000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.783e-05,
+      "loss": 8.6858,
+      "step": 542500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.7828000000000004e-05,
+      "loss": 8.7149,
+      "step": 543000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.7826000000000006e-05,
+      "loss": 8.6997,
+      "step": 543500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.7824e-05,
+      "loss": 8.6814,
+      "step": 544000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.7822e-05,
+      "loss": 8.7158,
+      "step": 544500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.7820000000000006e-05,
+      "loss": 8.6842,
+      "step": 545000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.7818e-05,
+      "loss": 8.6719,
+      "step": 545500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.7816e-05,
+      "loss": 8.6848,
+      "step": 546000
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.7814e-05,
+      "loss": 8.6839,
+      "step": 546500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.7812e-05,
+      "loss": 8.6997,
+      "step": 547000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.7810000000000005e-05,
+      "loss": 8.6884,
+      "step": 547500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.7808e-05,
+      "loss": 8.6709,
+      "step": 548000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.7806e-05,
+      "loss": 8.7005,
+      "step": 548500
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.7804000000000005e-05,
+      "loss": 8.688,
+      "step": 549000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.7802e-05,
+      "loss": 8.6819,
+      "step": 549500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.78e-05,
+      "loss": 8.6875,
+      "step": 550000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.7798000000000006e-05,
+      "loss": 8.6997,
+      "step": 550500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.7796e-05,
+      "loss": 8.7242,
+      "step": 551000
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.7794e-05,
+      "loss": 8.7325,
+      "step": 551500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.7792000000000006e-05,
+      "loss": 8.6911,
+      "step": 552000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.779e-05,
+      "loss": 8.6963,
+      "step": 552500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.7788000000000004e-05,
+      "loss": 8.685,
+      "step": 553000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.7786000000000006e-05,
+      "loss": 8.6815,
+      "step": 553500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.7784e-05,
+      "loss": 8.7159,
+      "step": 554000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.7782000000000004e-05,
+      "loss": 8.6886,
+      "step": 554500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.778e-05,
+      "loss": 8.6965,
+      "step": 555000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.7778e-05,
+      "loss": 8.6883,
+      "step": 555500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.7776000000000005e-05,
+      "loss": 8.7095,
+      "step": 556000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.7774e-05,
+      "loss": 8.686,
+      "step": 556500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.7772e-05,
+      "loss": 8.6664,
+      "step": 557000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.7770000000000005e-05,
+      "loss": 8.7078,
+      "step": 557500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.7768e-05,
+      "loss": 8.6967,
+      "step": 558000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.7765999999999996e-05,
+      "loss": 8.6748,
+      "step": 558500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.7764000000000006e-05,
+      "loss": 8.7147,
+      "step": 559000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.7762e-05,
+      "loss": 8.7062,
+      "step": 559500
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.7760000000000004e-05,
+      "loss": 8.6903,
+      "step": 560000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.7758000000000006e-05,
+      "loss": 8.725,
+      "step": 560500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.7756e-05,
+      "loss": 8.695,
+      "step": 561000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.7754000000000004e-05,
+      "loss": 8.6866,
+      "step": 561500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.7752e-05,
+      "loss": 8.6951,
+      "step": 562000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.775e-05,
+      "loss": 8.6861,
+      "step": 562500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.7748000000000004e-05,
+      "loss": 8.6742,
+      "step": 563000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.7746e-05,
+      "loss": 8.7243,
+      "step": 563500
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.7744e-05,
+      "loss": 8.6688,
+      "step": 564000
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7742000000000005e-05,
+      "loss": 8.6872,
+      "step": 564500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.774e-05,
+      "loss": 8.694,
+      "step": 565000
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7738e-05,
+      "loss": 8.6958,
+      "step": 565500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.7736000000000005e-05,
+      "loss": 8.6991,
+      "step": 566000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.7734e-05,
+      "loss": 8.6851,
+      "step": 566500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.7732e-05,
+      "loss": 8.6653,
+      "step": 567000
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.7730000000000005e-05,
+      "loss": 8.7074,
+      "step": 567500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.7728e-05,
+      "loss": 8.7233,
+      "step": 568000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.7726000000000004e-05,
+      "loss": 8.6912,
+      "step": 568500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.7724e-05,
+      "loss": 8.6986,
+      "step": 569000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.7722e-05,
+      "loss": 8.6708,
+      "step": 569500
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.7720000000000004e-05,
+      "loss": 8.6864,
+      "step": 570000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.7718e-05,
+      "loss": 8.6781,
+      "step": 570500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.7716e-05,
+      "loss": 8.6778,
+      "step": 571000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.7714000000000004e-05,
+      "loss": 8.6937,
+      "step": 571500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.7712e-05,
+      "loss": 8.722,
+      "step": 572000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.771e-05,
+      "loss": 8.7032,
+      "step": 572500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.7708000000000005e-05,
+      "loss": 8.6986,
+      "step": 573000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.7706e-05,
+      "loss": 8.713,
+      "step": 573500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.7704e-05,
+      "loss": 8.6979,
+      "step": 574000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.7702000000000005e-05,
+      "loss": 8.6977,
+      "step": 574500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.77e-05,
+      "loss": 8.7149,
+      "step": 575000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.7698e-05,
+      "loss": 8.7074,
+      "step": 575500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.7696e-05,
+      "loss": 8.6984,
+      "step": 576000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.7694e-05,
+      "loss": 8.7013,
+      "step": 576500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.7692000000000003e-05,
+      "loss": 8.7002,
+      "step": 577000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.769e-05,
+      "loss": 8.6604,
+      "step": 577500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.768800000000001e-05,
+      "loss": 8.6784,
+      "step": 578000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.7686000000000004e-05,
+      "loss": 8.6893,
+      "step": 578500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.7684e-05,
+      "loss": 8.7328,
+      "step": 579000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.7682e-05,
+      "loss": 8.6774,
+      "step": 579500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.7680000000000004e-05,
+      "loss": 8.7057,
+      "step": 580000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.7678e-05,
+      "loss": 8.6948,
+      "step": 580500
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.7676e-05,
+      "loss": 8.7093,
+      "step": 581000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.7674000000000005e-05,
+      "loss": 8.6769,
+      "step": 581500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.7672e-05,
+      "loss": 8.6765,
+      "step": 582000
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.767e-05,
+      "loss": 8.6831,
+      "step": 582500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.7668e-05,
+      "loss": 8.6693,
+      "step": 583000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.7666e-05,
+      "loss": 8.6931,
+      "step": 583500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.7664e-05,
+      "loss": 8.7038,
+      "step": 584000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.7662e-05,
+      "loss": 8.7228,
+      "step": 584500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.766000000000001e-05,
+      "loss": 8.7059,
+      "step": 585000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.7658000000000003e-05,
+      "loss": 8.7001,
+      "step": 585500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.7656e-05,
+      "loss": 8.6872,
+      "step": 586000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.7654e-05,
+      "loss": 8.6756,
+      "step": 586500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7652000000000004e-05,
+      "loss": 8.6775,
+      "step": 587000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.765e-05,
+      "loss": 8.6848,
+      "step": 587500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7648e-05,
+      "loss": 8.675,
+      "step": 588000
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7646000000000004e-05,
+      "loss": 8.6876,
+      "step": 588500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7644e-05,
+      "loss": 8.6964,
+      "step": 589000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.7642e-05,
+      "loss": 8.704,
+      "step": 589500
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.7640000000000005e-05,
+      "loss": 8.6812,
+      "step": 590000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.763800000000001e-05,
+      "loss": 8.6891,
+      "step": 590500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.7636e-05,
+      "loss": 8.6733,
+      "step": 591000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.7634e-05,
+      "loss": 8.6834,
+      "step": 591500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.763200000000001e-05,
+      "loss": 8.6956,
+      "step": 592000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.763e-05,
+      "loss": 8.6648,
+      "step": 592500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.7628e-05,
+      "loss": 8.6825,
+      "step": 593000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.7626e-05,
+      "loss": 8.6911,
+      "step": 593500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.7624000000000003e-05,
+      "loss": 8.6713,
+      "step": 594000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.7622e-05,
+      "loss": 8.6945,
+      "step": 594500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.762e-05,
+      "loss": 8.6941,
+      "step": 595000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.7618000000000004e-05,
+      "loss": 8.7096,
+      "step": 595500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.7616000000000006e-05,
+      "loss": 8.7098,
+      "step": 596000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.7614e-05,
+      "loss": 8.7069,
+      "step": 596500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.7612000000000004e-05,
+      "loss": 8.6807,
+      "step": 597000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.761000000000001e-05,
+      "loss": 8.7137,
+      "step": 597500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.7608e-05,
+      "loss": 8.6549,
+      "step": 598000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.7606e-05,
+      "loss": 8.6963,
+      "step": 598500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.760400000000001e-05,
+      "loss": 8.683,
+      "step": 599000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.7602e-05,
+      "loss": 8.672,
+      "step": 599500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.76e-05,
+      "loss": 8.6915,
+      "step": 600000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.7598e-05,
+      "loss": 8.6956,
+      "step": 600500
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.7596e-05,
+      "loss": 8.7001,
+      "step": 601000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.7594e-05,
+      "loss": 8.7155,
+      "step": 601500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.7592e-05,
+      "loss": 8.6802,
+      "step": 602000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.7590000000000003e-05,
+      "loss": 8.7002,
+      "step": 602500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.7588000000000006e-05,
+      "loss": 8.6859,
+      "step": 603000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.7586e-05,
+      "loss": 8.692,
+      "step": 603500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.7584000000000004e-05,
+      "loss": 8.6832,
+      "step": 604000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.7582000000000006e-05,
+      "loss": 8.6765,
+      "step": 604500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.758e-05,
+      "loss": 8.679,
+      "step": 605000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.7578e-05,
+      "loss": 8.7049,
+      "step": 605500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.757600000000001e-05,
+      "loss": 8.7157,
+      "step": 606000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7574e-05,
+      "loss": 8.7064,
+      "step": 606500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.7572e-05,
+      "loss": 8.6896,
+      "step": 607000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.757e-05,
+      "loss": 8.7057,
+      "step": 607500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.7568e-05,
+      "loss": 8.6853,
+      "step": 608000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.7566000000000005e-05,
+      "loss": 8.6964,
+      "step": 608500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.7564e-05,
+      "loss": 8.7125,
+      "step": 609000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.7562e-05,
+      "loss": 8.7113,
+      "step": 609500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.7560000000000005e-05,
+      "loss": 8.6883,
+      "step": 610000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.7558e-05,
+      "loss": 8.7056,
+      "step": 610500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.7556000000000003e-05,
+      "loss": 8.6759,
+      "step": 611000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.7554000000000006e-05,
+      "loss": 8.6787,
+      "step": 611500
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.7552e-05,
+      "loss": 8.6963,
+      "step": 612000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.755e-05,
+      "loss": 8.7135,
+      "step": 612500
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.7548000000000006e-05,
+      "loss": 8.7052,
+      "step": 613000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.7546e-05,
+      "loss": 8.6763,
+      "step": 613500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.7544e-05,
+      "loss": 8.6894,
+      "step": 614000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.7542e-05,
+      "loss": 8.684,
+      "step": 614500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.754e-05,
+      "loss": 8.7096,
+      "step": 615000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.7538000000000005e-05,
+      "loss": 8.7041,
+      "step": 615500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.7536e-05,
+      "loss": 8.7029,
+      "step": 616000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.7534e-05,
+      "loss": 8.6898,
+      "step": 616500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.7532000000000005e-05,
+      "loss": 8.701,
+      "step": 617000
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.753e-05,
+      "loss": 8.7274,
+      "step": 617500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.7528e-05,
+      "loss": 8.7009,
+      "step": 618000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.7526000000000005e-05,
+      "loss": 8.6851,
+      "step": 618500
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.7524e-05,
+      "loss": 8.6986,
+      "step": 619000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.7522e-05,
+      "loss": 8.6776,
+      "step": 619500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.7520000000000006e-05,
+      "loss": 8.6817,
+      "step": 620000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.7518e-05,
+      "loss": 8.6879,
+      "step": 620500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.7516000000000004e-05,
+      "loss": 8.6722,
+      "step": 621000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.7514000000000006e-05,
+      "loss": 8.695,
+      "step": 621500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.7512e-05,
+      "loss": 8.673,
+      "step": 622000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.7510000000000004e-05,
+      "loss": 8.699,
+      "step": 622500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.7508e-05,
+      "loss": 8.6841,
+      "step": 623000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.7506e-05,
+      "loss": 8.6652,
+      "step": 623500
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.7504000000000005e-05,
+      "loss": 8.7034,
+      "step": 624000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.7502e-05,
+      "loss": 8.6708,
+      "step": 624500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.75e-05,
+      "loss": 8.6996,
+      "step": 625000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.7498000000000005e-05,
+      "loss": 8.6895,
+      "step": 625500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.7496e-05,
+      "loss": 8.6687,
+      "step": 626000
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.7493999999999996e-05,
+      "loss": 8.6684,
+      "step": 626500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.7492000000000005e-05,
+      "loss": 8.6835,
+      "step": 627000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.749e-05,
+      "loss": 8.6745,
+      "step": 627500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.7488000000000003e-05,
+      "loss": 8.6795,
+      "step": 628000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.7486000000000006e-05,
+      "loss": 8.6914,
+      "step": 628500
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.7484e-05,
+      "loss": 8.6906,
+      "step": 629000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.7482000000000004e-05,
+      "loss": 8.7079,
+      "step": 629500
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.748e-05,
+      "loss": 8.6805,
+      "step": 630000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.7478e-05,
+      "loss": 8.7025,
+      "step": 630500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.7476000000000004e-05,
+      "loss": 8.6981,
+      "step": 631000
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.7474e-05,
+      "loss": 8.6627,
+      "step": 631500
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.7472e-05,
+      "loss": 8.6893,
+      "step": 632000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.7470000000000005e-05,
+      "loss": 8.6852,
+      "step": 632500
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.7468e-05,
+      "loss": 8.663,
+      "step": 633000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.7466e-05,
+      "loss": 8.6674,
+      "step": 633500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.7464000000000005e-05,
+      "loss": 8.6806,
+      "step": 634000
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.7462e-05,
+      "loss": 8.6666,
+      "step": 634500
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.746e-05,
+      "loss": 8.6728,
+      "step": 635000
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.7458000000000005e-05,
+      "loss": 8.7154,
+      "step": 635500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.7456e-05,
+      "loss": 8.7059,
+      "step": 636000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.7454000000000003e-05,
+      "loss": 8.6599,
+      "step": 636500
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.7452e-05,
+      "loss": 8.671,
+      "step": 637000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.745e-05,
+      "loss": 8.6911,
+      "step": 637500
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.7448000000000004e-05,
+      "loss": 8.7119,
+      "step": 638000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.7446e-05,
+      "loss": 8.7202,
+      "step": 638500
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.7444e-05,
+      "loss": 8.6662,
+      "step": 639000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.7442000000000004e-05,
+      "loss": 8.7054,
+      "step": 639500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.744e-05,
+      "loss": 8.673,
+      "step": 640000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.7438e-05,
+      "loss": 8.6985,
+      "step": 640500
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.7436000000000005e-05,
+      "loss": 8.6946,
+      "step": 641000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.7434e-05,
+      "loss": 8.6743,
+      "step": 641500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.7432e-05,
+      "loss": 8.6483,
+      "step": 642000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.7430000000000005e-05,
+      "loss": 8.6967,
+      "step": 642500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.7428e-05,
+      "loss": 8.6999,
+      "step": 643000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.7426e-05,
+      "loss": 8.6907,
+      "step": 643500
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.7424e-05,
+      "loss": 8.6941,
+      "step": 644000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.7422e-05,
+      "loss": 8.6523,
+      "step": 644500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.742e-05,
+      "loss": 8.7051,
+      "step": 645000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.7418e-05,
+      "loss": 8.6798,
+      "step": 645500
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.741600000000001e-05,
+      "loss": 8.65,
+      "step": 646000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.7414000000000004e-05,
+      "loss": 8.6642,
+      "step": 646500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.7412e-05,
+      "loss": 8.6939,
+      "step": 647000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.741e-05,
+      "loss": 8.706,
+      "step": 647500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.7408000000000004e-05,
+      "loss": 8.6656,
+      "step": 648000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.7406e-05,
+      "loss": 8.6735,
+      "step": 648500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.7404e-05,
+      "loss": 8.7023,
+      "step": 649000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.7402000000000005e-05,
+      "loss": 8.6907,
+      "step": 649500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.74e-05,
+      "loss": 8.6788,
+      "step": 650000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.7398e-05,
+      "loss": 8.6952,
+      "step": 650500
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.7396e-05,
+      "loss": 8.6699,
+      "step": 651000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.7394e-05,
+      "loss": 8.6877,
+      "step": 651500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.7392e-05,
+      "loss": 8.6928,
+      "step": 652000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.739e-05,
+      "loss": 8.6826,
+      "step": 652500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.738800000000001e-05,
+      "loss": 8.6977,
+      "step": 653000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.7386e-05,
+      "loss": 8.6922,
+      "step": 653500
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.7384e-05,
+      "loss": 8.6922,
+      "step": 654000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.7382e-05,
+      "loss": 8.6917,
+      "step": 654500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.7380000000000004e-05,
+      "loss": 8.7062,
+      "step": 655000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.7378e-05,
+      "loss": 8.6872,
+      "step": 655500
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.7376e-05,
+      "loss": 8.6742,
+      "step": 656000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.7374000000000004e-05,
+      "loss": 8.7035,
+      "step": 656500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.7372e-05,
+      "loss": 8.6762,
+      "step": 657000
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.737e-05,
+      "loss": 8.7042,
+      "step": 657500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.7368000000000005e-05,
+      "loss": 8.687,
+      "step": 658000
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.736600000000001e-05,
+      "loss": 8.6841,
+      "step": 658500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.7364e-05,
+      "loss": 8.7086,
+      "step": 659000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.7362e-05,
+      "loss": 8.6768,
+      "step": 659500
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.736000000000001e-05,
+      "loss": 8.683,
+      "step": 660000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.7358e-05,
+      "loss": 8.6867,
+      "step": 660500
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.7356e-05,
+      "loss": 8.6706,
+      "step": 661000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.7354e-05,
+      "loss": 8.6344,
+      "step": 661500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.7352e-05,
+      "loss": 8.6803,
+      "step": 662000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.735e-05,
+      "loss": 8.691,
+      "step": 662500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.7348e-05,
+      "loss": 8.6917,
+      "step": 663000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.7346000000000004e-05,
+      "loss": 8.6831,
+      "step": 663500
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.7344000000000006e-05,
+      "loss": 8.6985,
+      "step": 664000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.7342e-05,
+      "loss": 8.6908,
+      "step": 664500
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.7340000000000004e-05,
+      "loss": 8.6704,
+      "step": 665000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.7338000000000007e-05,
+      "loss": 8.685,
+      "step": 665500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.7336e-05,
+      "loss": 8.7153,
+      "step": 666000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.7334e-05,
+      "loss": 8.7006,
+      "step": 666500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.733200000000001e-05,
+      "loss": 8.6747,
+      "step": 667000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.733e-05,
+      "loss": 8.6613,
+      "step": 667500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.7328e-05,
+      "loss": 8.7035,
+      "step": 668000
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.7326e-05,
+      "loss": 8.6896,
+      "step": 668500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.7324e-05,
+      "loss": 8.6955,
+      "step": 669000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.7322e-05,
+      "loss": 8.7121,
+      "step": 669500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.732e-05,
+      "loss": 8.6922,
+      "step": 670000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.7318e-05,
+      "loss": 8.6892,
+      "step": 670500
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.7316000000000006e-05,
+      "loss": 8.6795,
+      "step": 671000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.7314e-05,
+      "loss": 8.7076,
+      "step": 671500
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.7312000000000004e-05,
+      "loss": 8.6899,
+      "step": 672000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.7310000000000006e-05,
+      "loss": 8.7144,
+      "step": 672500
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.7308e-05,
+      "loss": 8.7106,
+      "step": 673000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.7306e-05,
+      "loss": 8.683,
+      "step": 673500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.7304000000000007e-05,
+      "loss": 8.7013,
+      "step": 674000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.7302e-05,
+      "loss": 8.6933,
+      "step": 674500
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.73e-05,
+      "loss": 8.7011,
+      "step": 675000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.7298e-05,
+      "loss": 8.6761,
+      "step": 675500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.7296e-05,
+      "loss": 8.676,
+      "step": 676000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.7294000000000005e-05,
+      "loss": 8.7049,
+      "step": 676500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.7292e-05,
+      "loss": 8.6862,
+      "step": 677000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.729e-05,
+      "loss": 8.6861,
+      "step": 677500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.7288000000000005e-05,
+      "loss": 8.6773,
+      "step": 678000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.7286e-05,
+      "loss": 8.6915,
+      "step": 678500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.7284e-05,
+      "loss": 8.6942,
+      "step": 679000
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.7282000000000006e-05,
+      "loss": 8.6837,
+      "step": 679500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.728e-05,
+      "loss": 8.6891,
+      "step": 680000
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.7278e-05,
+      "loss": 8.6752,
+      "step": 680500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.7276000000000006e-05,
+      "loss": 8.6769,
+      "step": 681000
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.7274e-05,
+      "loss": 8.6738,
+      "step": 681500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.7272e-05,
+      "loss": 8.6892,
+      "step": 682000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.7270000000000007e-05,
+      "loss": 8.7074,
+      "step": 682500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.7268e-05,
+      "loss": 8.6909,
+      "step": 683000
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.7266000000000005e-05,
+      "loss": 8.6848,
+      "step": 683500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.7264e-05,
+      "loss": 8.6926,
+      "step": 684000
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.7262e-05,
+      "loss": 8.6606,
+      "step": 684500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.7260000000000005e-05,
+      "loss": 8.696,
+      "step": 685000
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.7258e-05,
+      "loss": 8.6629,
+      "step": 685500
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.7256e-05,
+      "loss": 8.6887,
+      "step": 686000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.7254000000000005e-05,
+      "loss": 8.6852,
+      "step": 686500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.7252e-05,
+      "loss": 8.6677,
+      "step": 687000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.7249999999999997e-05,
+      "loss": 8.6951,
+      "step": 687500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.7248000000000006e-05,
+      "loss": 8.6928,
+      "step": 688000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.7246e-05,
+      "loss": 8.7003,
+      "step": 688500
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.7244000000000004e-05,
+      "loss": 8.6892,
+      "step": 689000
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.7242000000000006e-05,
+      "loss": 8.6872,
+      "step": 689500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.724e-05,
+      "loss": 8.711,
+      "step": 690000
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.7238000000000004e-05,
+      "loss": 8.7015,
+      "step": 690500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.7236e-05,
+      "loss": 8.6695,
+      "step": 691000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.7234e-05,
+      "loss": 8.7176,
+      "step": 691500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.7232000000000005e-05,
+      "loss": 8.7359,
+      "step": 692000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.723e-05,
+      "loss": 8.6934,
+      "step": 692500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.7228e-05,
+      "loss": 8.6977,
+      "step": 693000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.7226000000000005e-05,
+      "loss": 8.6904,
+      "step": 693500
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.7224e-05,
+      "loss": 8.6897,
+      "step": 694000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.7222e-05,
+      "loss": 8.6611,
+      "step": 694500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.7220000000000005e-05,
+      "loss": 8.6864,
+      "step": 695000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.7218e-05,
+      "loss": 8.6747,
+      "step": 695500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.7216e-05,
+      "loss": 8.6676,
+      "step": 696000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.7214000000000006e-05,
+      "loss": 8.7055,
+      "step": 696500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.7212e-05,
+      "loss": 8.6965,
+      "step": 697000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.7210000000000004e-05,
+      "loss": 8.6965,
+      "step": 697500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.7208e-05,
+      "loss": 8.6918,
+      "step": 698000
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.7206e-05,
+      "loss": 8.6898,
+      "step": 698500
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.7204000000000004e-05,
+      "loss": 8.7017,
+      "step": 699000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.7202e-05,
+      "loss": 8.6897,
+      "step": 699500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.72e-05,
+      "loss": 8.6989,
+      "step": 700000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.7198000000000004e-05,
+      "loss": 8.6906,
+      "step": 700500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.7196e-05,
+      "loss": 8.6889,
+      "step": 701000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.7194e-05,
+      "loss": 8.6767,
+      "step": 701500
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.7192000000000005e-05,
+      "loss": 8.6921,
+      "step": 702000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.719e-05,
+      "loss": 8.6512,
+      "step": 702500
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.7188e-05,
+      "loss": 8.6645,
+      "step": 703000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.7186000000000005e-05,
+      "loss": 8.6867,
+      "step": 703500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.7184e-05,
+      "loss": 8.6958,
+      "step": 704000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.7182e-05,
+      "loss": 8.6656,
+      "step": 704500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.718e-05,
+      "loss": 8.6697,
+      "step": 705000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.7178e-05,
+      "loss": 8.6867,
+      "step": 705500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.7176000000000004e-05,
+      "loss": 8.6707,
+      "step": 706000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.7174e-05,
+      "loss": 8.6832,
+      "step": 706500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.7172e-05,
+      "loss": 8.6698,
+      "step": 707000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.7170000000000004e-05,
+      "loss": 8.6951,
+      "step": 707500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.7168e-05,
+      "loss": 8.6829,
+      "step": 708000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.7166e-05,
+      "loss": 8.6971,
+      "step": 708500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.7164000000000004e-05,
+      "loss": 8.6797,
+      "step": 709000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.7162e-05,
+      "loss": 8.7025,
+      "step": 709500
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.716e-05,
+      "loss": 8.6876,
+      "step": 710000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.7158000000000005e-05,
+      "loss": 8.6836,
+      "step": 710500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.7156e-05,
+      "loss": 8.6867,
+      "step": 711000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.7154e-05,
+      "loss": 8.6948,
+      "step": 711500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.7152e-05,
+      "loss": 8.7156,
+      "step": 712000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.715e-05,
+      "loss": 8.6833,
+      "step": 712500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.7148e-05,
+      "loss": 8.6823,
+      "step": 713000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.7146e-05,
+      "loss": 8.6815,
+      "step": 713500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.714400000000001e-05,
+      "loss": 8.6914,
+      "step": 714000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.7142000000000004e-05,
+      "loss": 8.6879,
+      "step": 714500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.714e-05,
+      "loss": 8.6631,
+      "step": 715000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.7138e-05,
+      "loss": 8.6916,
+      "step": 715500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.7136000000000004e-05,
+      "loss": 8.6991,
+      "step": 716000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.7134e-05,
+      "loss": 8.6888,
+      "step": 716500
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.7132e-05,
+      "loss": 8.6925,
+      "step": 717000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.7130000000000004e-05,
+      "loss": 8.6895,
+      "step": 717500
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.7128e-05,
+      "loss": 8.6821,
+      "step": 718000
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.7126e-05,
+      "loss": 8.6825,
+      "step": 718500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.7124000000000005e-05,
+      "loss": 8.6447,
+      "step": 719000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.712200000000001e-05,
+      "loss": 8.6865,
+      "step": 719500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.712e-05,
+      "loss": 8.6965,
+      "step": 720000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.7118e-05,
+      "loss": 8.6994,
+      "step": 720500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.711600000000001e-05,
+      "loss": 8.7109,
+      "step": 721000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.7114e-05,
+      "loss": 8.6661,
+      "step": 721500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.7112e-05,
+      "loss": 8.6944,
+      "step": 722000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.711e-05,
+      "loss": 8.7156,
+      "step": 722500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.7108000000000004e-05,
+      "loss": 8.6916,
+      "step": 723000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.7106e-05,
+      "loss": 8.6661,
+      "step": 723500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.7104e-05,
+      "loss": 8.6902,
+      "step": 724000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.7102000000000004e-05,
+      "loss": 8.6957,
+      "step": 724500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.71e-05,
+      "loss": 8.687,
+      "step": 725000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.7098e-05,
+      "loss": 8.7003,
+      "step": 725500
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.7096000000000004e-05,
+      "loss": 8.6713,
+      "step": 726000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.709400000000001e-05,
+      "loss": 8.6699,
+      "step": 726500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.7092e-05,
+      "loss": 8.6869,
+      "step": 727000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.709e-05,
+      "loss": 8.6778,
+      "step": 727500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.708800000000001e-05,
+      "loss": 8.675,
+      "step": 728000
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.7086e-05,
+      "loss": 8.6962,
+      "step": 728500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.7084e-05,
+      "loss": 8.7078,
+      "step": 729000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.7082e-05,
+      "loss": 8.6918,
+      "step": 729500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.708e-05,
+      "loss": 8.7071,
+      "step": 730000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.7078e-05,
+      "loss": 8.6868,
+      "step": 730500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.7076e-05,
+      "loss": 8.6995,
+      "step": 731000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.7074000000000004e-05,
+      "loss": 8.6819,
+      "step": 731500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.7072000000000006e-05,
+      "loss": 8.6853,
+      "step": 732000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.707e-05,
+      "loss": 8.695,
+      "step": 732500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.7068000000000004e-05,
+      "loss": 8.6754,
+      "step": 733000
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.7066000000000006e-05,
+      "loss": 8.7053,
+      "step": 733500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.7064e-05,
+      "loss": 8.6838,
+      "step": 734000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.7062e-05,
+      "loss": 8.6893,
+      "step": 734500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.706000000000001e-05,
+      "loss": 8.7103,
+      "step": 735000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.7058e-05,
+      "loss": 8.6732,
+      "step": 735500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.7056e-05,
+      "loss": 8.7037,
+      "step": 736000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.7054e-05,
+      "loss": 8.6949,
+      "step": 736500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.7052e-05,
+      "loss": 8.7115,
+      "step": 737000
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.705e-05,
+      "loss": 8.6782,
+      "step": 737500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.7048e-05,
+      "loss": 8.6852,
+      "step": 738000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.7046e-05,
+      "loss": 8.6756,
+      "step": 738500
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.7044000000000006e-05,
+      "loss": 8.6945,
+      "step": 739000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.7042e-05,
+      "loss": 8.6926,
+      "step": 739500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.7040000000000004e-05,
+      "loss": 8.6959,
+      "step": 740000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.7038000000000006e-05,
+      "loss": 8.6791,
+      "step": 740500
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.7036e-05,
+      "loss": 8.6837,
+      "step": 741000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.7034e-05,
+      "loss": 8.6567,
+      "step": 741500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.7032000000000006e-05,
+      "loss": 8.6724,
+      "step": 742000
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.703e-05,
+      "loss": 8.6793,
+      "step": 742500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.7028e-05,
+      "loss": 8.7018,
+      "step": 743000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.7026e-05,
+      "loss": 8.6535,
+      "step": 743500
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.7024e-05,
+      "loss": 8.6763,
+      "step": 744000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.7022000000000005e-05,
+      "loss": 8.6904,
+      "step": 744500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.702e-05,
+      "loss": 8.7006,
+      "step": 745000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.7018e-05,
+      "loss": 8.705,
+      "step": 745500
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.7016000000000005e-05,
+      "loss": 8.685,
+      "step": 746000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.7014e-05,
+      "loss": 8.6814,
+      "step": 746500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.7012e-05,
+      "loss": 8.6848,
+      "step": 747000
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.7010000000000006e-05,
+      "loss": 8.6957,
+      "step": 747500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.7008e-05,
+      "loss": 8.6929,
+      "step": 748000
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.7006e-05,
+      "loss": 8.6925,
+      "step": 748500
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.7004000000000006e-05,
+      "loss": 8.7093,
+      "step": 749000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.7002e-05,
+      "loss": 8.6777,
+      "step": 749500
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.7e-05,
+      "loss": 8.6913,
+      "step": 750000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.6998000000000006e-05,
+      "loss": 8.7201,
+      "step": 750500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.6996e-05,
+      "loss": 8.6985,
+      "step": 751000
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.6994000000000004e-05,
+      "loss": 8.673,
+      "step": 751500
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.6992e-05,
+      "loss": 8.6662,
+      "step": 752000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.699e-05,
+      "loss": 8.7033,
+      "step": 752500
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.6988000000000005e-05,
+      "loss": 8.7036,
+      "step": 753000
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.6986e-05,
+      "loss": 8.7168,
+      "step": 753500
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.6984e-05,
+      "loss": 8.6793,
+      "step": 754000
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.6982000000000005e-05,
+      "loss": 8.658,
+      "step": 754500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.698e-05,
+      "loss": 8.6972,
+      "step": 755000
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.6977999999999996e-05,
+      "loss": 8.681,
+      "step": 755500
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.6976000000000006e-05,
+      "loss": 8.6965,
+      "step": 756000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.6974e-05,
+      "loss": 8.6773,
+      "step": 756500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.6972000000000004e-05,
+      "loss": 8.6668,
+      "step": 757000
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.6970000000000006e-05,
+      "loss": 8.696,
+      "step": 757500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.6968e-05,
+      "loss": 8.6799,
+      "step": 758000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.6966000000000004e-05,
+      "loss": 8.6923,
+      "step": 758500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.6964e-05,
+      "loss": 8.6856,
+      "step": 759000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.6962e-05,
+      "loss": 8.6911,
+      "step": 759500
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.6960000000000004e-05,
+      "loss": 8.6814,
+      "step": 760000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.6958e-05,
+      "loss": 8.6729,
+      "step": 760500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.6956e-05,
+      "loss": 8.7033,
+      "step": 761000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.6954000000000005e-05,
+      "loss": 8.6683,
+      "step": 761500
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.6952e-05,
+      "loss": 8.6567,
+      "step": 762000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.695e-05,
+      "loss": 8.6747,
+      "step": 762500
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.6948000000000005e-05,
+      "loss": 8.6623,
+      "step": 763000
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.6946e-05,
+      "loss": 8.69,
+      "step": 763500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.6944e-05,
+      "loss": 8.6797,
+      "step": 764000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.6942000000000006e-05,
+      "loss": 8.694,
+      "step": 764500
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.694e-05,
+      "loss": 8.6899,
+      "step": 765000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.6938000000000004e-05,
+      "loss": 8.67,
+      "step": 765500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.6936e-05,
+      "loss": 8.6741,
+      "step": 766000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.6934e-05,
+      "loss": 8.7204,
+      "step": 766500
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.6932000000000004e-05,
+      "loss": 8.6861,
+      "step": 767000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.693e-05,
+      "loss": 8.7019,
+      "step": 767500
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.6928e-05,
+      "loss": 8.6781,
+      "step": 768000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.6926000000000004e-05,
+      "loss": 8.6855,
+      "step": 768500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.6924e-05,
+      "loss": 8.6801,
+      "step": 769000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.6922e-05,
+      "loss": 8.7049,
+      "step": 769500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.6920000000000005e-05,
+      "loss": 8.6789,
+      "step": 770000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.6918e-05,
+      "loss": 8.6692,
+      "step": 770500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.6916e-05,
+      "loss": 8.7287,
+      "step": 771000
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.6914000000000005e-05,
+      "loss": 8.6889,
+      "step": 771500
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.6912e-05,
+      "loss": 8.666,
+      "step": 772000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.691e-05,
+      "loss": 8.6938,
+      "step": 772500
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.6908e-05,
+      "loss": 8.6645,
+      "step": 773000
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.6906e-05,
+      "loss": 8.6783,
+      "step": 773500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.6904000000000004e-05,
+      "loss": 8.656,
+      "step": 774000
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.6902e-05,
+      "loss": 8.6943,
+      "step": 774500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.69e-05,
+      "loss": 8.7057,
+      "step": 775000
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.6898000000000004e-05,
+      "loss": 8.6791,
+      "step": 775500
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.6896e-05,
+      "loss": 8.6785,
+      "step": 776000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.6894e-05,
+      "loss": 8.6989,
+      "step": 776500
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.6892000000000004e-05,
+      "loss": 8.6844,
+      "step": 777000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.689e-05,
+      "loss": 8.6658,
+      "step": 777500
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.6888e-05,
+      "loss": 8.6335,
+      "step": 778000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.6886000000000005e-05,
+      "loss": 8.6875,
+      "step": 778500
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.6884e-05,
+      "loss": 8.6967,
+      "step": 779000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.6882e-05,
+      "loss": 8.66,
+      "step": 779500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.688e-05,
+      "loss": 8.6612,
+      "step": 780000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.6878e-05,
+      "loss": 8.6867,
+      "step": 780500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.6876e-05,
+      "loss": 8.6855,
+      "step": 781000
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.6874e-05,
+      "loss": 8.6826,
+      "step": 781500
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.687200000000001e-05,
+      "loss": 8.6773,
+      "step": 782000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.6870000000000004e-05,
+      "loss": 8.6837,
+      "step": 782500
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.6868e-05,
+      "loss": 8.6954,
+      "step": 783000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.6866e-05,
+      "loss": 8.6884,
+      "step": 783500
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.6864000000000004e-05,
+      "loss": 8.7012,
+      "step": 784000
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.6862e-05,
+      "loss": 8.686,
+      "step": 784500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.686e-05,
+      "loss": 8.6754,
+      "step": 785000
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.6858000000000004e-05,
+      "loss": 8.6742,
+      "step": 785500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.6856e-05,
+      "loss": 8.7039,
+      "step": 786000
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.6854e-05,
+      "loss": 8.6997,
+      "step": 786500
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.6852000000000005e-05,
+      "loss": 8.6888,
+      "step": 787000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.685000000000001e-05,
+      "loss": 8.6968,
+      "step": 787500
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.6848e-05,
+      "loss": 8.6881,
+      "step": 788000
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.6846e-05,
+      "loss": 8.7016,
+      "step": 788500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.684400000000001e-05,
+      "loss": 8.674,
+      "step": 789000
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.6842e-05,
+      "loss": 8.6957,
+      "step": 789500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.684e-05,
+      "loss": 8.6614,
+      "step": 790000
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.6838e-05,
+      "loss": 8.681,
+      "step": 790500
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.6836000000000004e-05,
+      "loss": 8.6727,
+      "step": 791000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.6834e-05,
+      "loss": 8.6681,
+      "step": 791500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.6832e-05,
+      "loss": 8.6964,
+      "step": 792000
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.6830000000000004e-05,
+      "loss": 8.6826,
+      "step": 792500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.6828e-05,
+      "loss": 8.7072,
+      "step": 793000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.6826e-05,
+      "loss": 8.6884,
+      "step": 793500
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.6824000000000004e-05,
+      "loss": 8.7096,
+      "step": 794000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.682200000000001e-05,
+      "loss": 8.6898,
+      "step": 794500
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.682e-05,
+      "loss": 8.6621,
+      "step": 795000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.6818e-05,
+      "loss": 8.7063,
+      "step": 795500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.681600000000001e-05,
+      "loss": 8.6877,
+      "step": 796000
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.6814e-05,
+      "loss": 8.691,
+      "step": 796500
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.6812e-05,
+      "loss": 8.6942,
+      "step": 797000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.681e-05,
+      "loss": 8.6936,
+      "step": 797500
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.6808e-05,
+      "loss": 8.688,
+      "step": 798000
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.6806e-05,
+      "loss": 8.7132,
+      "step": 798500
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.6804e-05,
+      "loss": 8.6968,
+      "step": 799000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.6802000000000004e-05,
+      "loss": 8.6962,
+      "step": 799500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 8.6979,
+      "step": 800000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.6798e-05,
+      "loss": 8.6787,
+      "step": 800500
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.6796000000000004e-05,
+      "loss": 8.6847,
+      "step": 801000
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.6794000000000006e-05,
+      "loss": 8.7164,
+      "step": 801500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.6792e-05,
+      "loss": 8.6892,
+      "step": 802000
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.679e-05,
+      "loss": 8.6569,
+      "step": 802500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.678800000000001e-05,
+      "loss": 8.7136,
+      "step": 803000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.6786e-05,
+      "loss": 8.6801,
+      "step": 803500
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.6784e-05,
+      "loss": 8.6836,
+      "step": 804000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.6782e-05,
+      "loss": 8.685,
+      "step": 804500
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.678e-05,
+      "loss": 8.7094,
+      "step": 805000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.6778e-05,
+      "loss": 8.691,
+      "step": 805500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.6776e-05,
+      "loss": 8.6758,
+      "step": 806000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.6774e-05,
+      "loss": 8.7133,
+      "step": 806500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.6772000000000006e-05,
+      "loss": 8.6901,
+      "step": 807000
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.677e-05,
+      "loss": 8.6884,
+      "step": 807500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.6768000000000004e-05,
+      "loss": 8.6815,
+      "step": 808000
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.6766000000000006e-05,
+      "loss": 8.6945,
+      "step": 808500
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.6764e-05,
+      "loss": 8.6881,
+      "step": 809000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.6762e-05,
+      "loss": 8.6891,
+      "step": 809500
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.6760000000000006e-05,
+      "loss": 8.6968,
+      "step": 810000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.6758e-05,
+      "loss": 8.6841,
+      "step": 810500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.6756e-05,
+      "loss": 8.6876,
+      "step": 811000
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.675400000000001e-05,
+      "loss": 8.6979,
+      "step": 811500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.6752e-05,
+      "loss": 8.678,
+      "step": 812000
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.6750000000000005e-05,
+      "loss": 8.6972,
+      "step": 812500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.6748e-05,
+      "loss": 8.6811,
+      "step": 813000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.6746e-05,
+      "loss": 8.6973,
+      "step": 813500
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.6744000000000005e-05,
+      "loss": 8.6943,
+      "step": 814000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.6742e-05,
+      "loss": 8.6855,
+      "step": 814500
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.674e-05,
+      "loss": 8.6963,
+      "step": 815000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.6738000000000006e-05,
+      "loss": 8.7114,
+      "step": 815500
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.6736e-05,
+      "loss": 8.6996,
+      "step": 816000
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.6734e-05,
+      "loss": 8.6951,
+      "step": 816500
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.6732000000000006e-05,
+      "loss": 8.6772,
+      "step": 817000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.673e-05,
+      "loss": 8.6724,
+      "step": 817500
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.6728e-05,
+      "loss": 8.7174,
+      "step": 818000
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.6726000000000006e-05,
+      "loss": 8.6783,
+      "step": 818500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.6724e-05,
+      "loss": 8.6842,
+      "step": 819000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.6722000000000004e-05,
+      "loss": 8.6893,
+      "step": 819500
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.672e-05,
+      "loss": 8.6598,
+      "step": 820000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.6718e-05,
+      "loss": 8.6785,
+      "step": 820500
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.6716000000000005e-05,
+      "loss": 8.6983,
+      "step": 821000
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.6714e-05,
+      "loss": 8.6691,
+      "step": 821500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.6712e-05,
+      "loss": 8.6649,
+      "step": 822000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.6710000000000005e-05,
+      "loss": 8.6774,
+      "step": 822500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.6708e-05,
+      "loss": 8.662,
+      "step": 823000
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.6706e-05,
+      "loss": 8.7077,
+      "step": 823500
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.6704000000000005e-05,
+      "loss": 8.6697,
+      "step": 824000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.6702e-05,
+      "loss": 8.6873,
+      "step": 824500
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.6700000000000003e-05,
+      "loss": 8.6606,
+      "step": 825000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.6698000000000006e-05,
+      "loss": 8.6672,
+      "step": 825500
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.6696e-05,
+      "loss": 8.7246,
+      "step": 826000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.6694000000000004e-05,
+      "loss": 8.6949,
+      "step": 826500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.6692e-05,
+      "loss": 8.6766,
+      "step": 827000
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.669e-05,
+      "loss": 8.6795,
+      "step": 827500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.6688000000000004e-05,
+      "loss": 8.6992,
+      "step": 828000
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.6686e-05,
+      "loss": 8.6833,
+      "step": 828500
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.6684e-05,
+      "loss": 8.692,
+      "step": 829000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.6682000000000005e-05,
+      "loss": 8.686,
+      "step": 829500
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.668e-05,
+      "loss": 8.695,
+      "step": 830000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.6678e-05,
+      "loss": 8.6997,
+      "step": 830500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.6676000000000005e-05,
+      "loss": 8.7094,
+      "step": 831000
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.6674e-05,
+      "loss": 8.7044,
+      "step": 831500
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.6672e-05,
+      "loss": 8.7036,
+      "step": 832000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.6670000000000005e-05,
+      "loss": 8.6943,
+      "step": 832500
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.6668e-05,
+      "loss": 8.6921,
+      "step": 833000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.6666000000000003e-05,
+      "loss": 8.6799,
+      "step": 833500
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.6664e-05,
+      "loss": 8.6947,
+      "step": 834000
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.6662e-05,
+      "loss": 8.6906,
+      "step": 834500
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.6660000000000004e-05,
+      "loss": 8.694,
+      "step": 835000
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.6658e-05,
+      "loss": 8.6796,
+      "step": 835500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.6656e-05,
+      "loss": 8.6982,
+      "step": 836000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.6654000000000004e-05,
+      "loss": 8.7206,
+      "step": 836500
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.6652e-05,
+      "loss": 8.6691,
+      "step": 837000
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.665e-05,
+      "loss": 8.6577,
+      "step": 837500
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.6648000000000005e-05,
+      "loss": 8.714,
+      "step": 838000
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.6646e-05,
+      "loss": 8.651,
+      "step": 838500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.6644e-05,
+      "loss": 8.6644,
+      "step": 839000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.6642000000000005e-05,
+      "loss": 8.684,
+      "step": 839500
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.664e-05,
+      "loss": 8.6915,
+      "step": 840000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.6638e-05,
+      "loss": 8.6944,
+      "step": 840500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.6636e-05,
+      "loss": 8.694,
+      "step": 841000
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.6634e-05,
+      "loss": 8.664,
+      "step": 841500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.6632000000000003e-05,
+      "loss": 8.6761,
+      "step": 842000
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.663e-05,
+      "loss": 8.6971,
+      "step": 842500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.6628e-05,
+      "loss": 8.6782,
+      "step": 843000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.6626000000000004e-05,
+      "loss": 8.6581,
+      "step": 843500
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.6624e-05,
+      "loss": 8.6781,
+      "step": 844000
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.6622e-05,
+      "loss": 8.6633,
+      "step": 844500
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.6620000000000004e-05,
+      "loss": 8.6702,
+      "step": 845000
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.6618e-05,
+      "loss": 8.6883,
+      "step": 845500
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.6616e-05,
+      "loss": 8.6945,
+      "step": 846000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.6614000000000005e-05,
+      "loss": 8.6573,
+      "step": 846500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.6612e-05,
+      "loss": 8.672,
+      "step": 847000
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.661e-05,
+      "loss": 8.659,
+      "step": 847500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.6608e-05,
+      "loss": 8.6699,
+      "step": 848000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.6606e-05,
+      "loss": 8.6829,
+      "step": 848500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.6604e-05,
+      "loss": 8.6908,
+      "step": 849000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.6602e-05,
+      "loss": 8.6943,
+      "step": 849500
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 8.6991,
+      "step": 850000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.6598000000000003e-05,
+      "loss": 8.6771,
+      "step": 850500
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.6596e-05,
+      "loss": 8.719,
+      "step": 851000
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.6594e-05,
+      "loss": 8.6948,
+      "step": 851500
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.6592000000000004e-05,
+      "loss": 8.6727,
+      "step": 852000
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.659e-05,
+      "loss": 8.6538,
+      "step": 852500
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.6588e-05,
+      "loss": 8.6852,
+      "step": 853000
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.6586000000000004e-05,
+      "loss": 8.7012,
+      "step": 853500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.6584e-05,
+      "loss": 8.679,
+      "step": 854000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.6582e-05,
+      "loss": 8.6946,
+      "step": 854500
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.6580000000000005e-05,
+      "loss": 8.6983,
+      "step": 855000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.657800000000001e-05,
+      "loss": 8.6911,
+      "step": 855500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.6576e-05,
+      "loss": 8.7233,
+      "step": 856000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.6574e-05,
+      "loss": 8.6793,
+      "step": 856500
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.657200000000001e-05,
+      "loss": 8.6707,
+      "step": 857000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.657e-05,
+      "loss": 8.6818,
+      "step": 857500
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.6568e-05,
+      "loss": 8.6773,
+      "step": 858000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.6566e-05,
+      "loss": 8.7184,
+      "step": 858500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.6564000000000003e-05,
+      "loss": 8.6863,
+      "step": 859000
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.6562e-05,
+      "loss": 8.6887,
+      "step": 859500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.656e-05,
+      "loss": 8.6982,
+      "step": 860000
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.6558000000000004e-05,
+      "loss": 8.7029,
+      "step": 860500
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.6556e-05,
+      "loss": 8.6883,
+      "step": 861000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.6554e-05,
+      "loss": 8.6869,
+      "step": 861500
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.6552000000000004e-05,
+      "loss": 8.6949,
+      "step": 862000
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.655000000000001e-05,
+      "loss": 8.6647,
+      "step": 862500
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.6548e-05,
+      "loss": 8.6981,
+      "step": 863000
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.6546e-05,
+      "loss": 8.706,
+      "step": 863500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.654400000000001e-05,
+      "loss": 8.6744,
+      "step": 864000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.6542e-05,
+      "loss": 8.6722,
+      "step": 864500
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.654e-05,
+      "loss": 8.6963,
+      "step": 865000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.6538e-05,
+      "loss": 8.6823,
+      "step": 865500
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.6536e-05,
+      "loss": 8.6683,
+      "step": 866000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.6534e-05,
+      "loss": 8.6902,
+      "step": 866500
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.6532e-05,
+      "loss": 8.679,
+      "step": 867000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.6530000000000003e-05,
+      "loss": 8.689,
+      "step": 867500
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.6528000000000006e-05,
+      "loss": 8.682,
+      "step": 868000
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.6526e-05,
+      "loss": 8.6504,
+      "step": 868500
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.6524000000000004e-05,
+      "loss": 8.6971,
+      "step": 869000
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.6522000000000006e-05,
+      "loss": 8.695,
+      "step": 869500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.652e-05,
+      "loss": 8.671,
+      "step": 870000
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.6518e-05,
+      "loss": 8.6795,
+      "step": 870500
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.651600000000001e-05,
+      "loss": 8.6897,
+      "step": 871000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.6514e-05,
+      "loss": 8.686,
+      "step": 871500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.6512e-05,
+      "loss": 8.6801,
+      "step": 872000
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.651e-05,
+      "loss": 8.7021,
+      "step": 872500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.6508e-05,
+      "loss": 8.68,
+      "step": 873000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.6506e-05,
+      "loss": 8.6635,
+      "step": 873500
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.6504e-05,
+      "loss": 8.6801,
+      "step": 874000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.6502e-05,
+      "loss": 8.6499,
+      "step": 874500
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 8.7053,
+      "step": 875000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.6498e-05,
+      "loss": 8.6667,
+      "step": 875500
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.6496000000000003e-05,
+      "loss": 8.6919,
+      "step": 876000
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.6494000000000006e-05,
+      "loss": 8.6901,
+      "step": 876500
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.6492e-05,
+      "loss": 8.7098,
+      "step": 877000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.649e-05,
+      "loss": 8.6854,
+      "step": 877500
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.6488000000000006e-05,
+      "loss": 8.6944,
+      "step": 878000
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.6486e-05,
+      "loss": 8.704,
+      "step": 878500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.6484e-05,
+      "loss": 8.6889,
+      "step": 879000
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.6482000000000007e-05,
+      "loss": 8.6797,
+      "step": 879500
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.648e-05,
+      "loss": 8.6706,
+      "step": 880000
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.6478000000000005e-05,
+      "loss": 8.689,
+      "step": 880500
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.6476e-05,
+      "loss": 8.7023,
+      "step": 881000
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.6474e-05,
+      "loss": 8.6985,
+      "step": 881500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.6472000000000005e-05,
+      "loss": 8.6837,
+      "step": 882000
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.647e-05,
+      "loss": 8.6838,
+      "step": 882500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.6468e-05,
+      "loss": 8.7124,
+      "step": 883000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.6466000000000005e-05,
+      "loss": 8.7076,
+      "step": 883500
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.6464e-05,
+      "loss": 8.7046,
+      "step": 884000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.6462e-05,
+      "loss": 8.6597,
+      "step": 884500
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.6460000000000006e-05,
+      "loss": 8.6988,
+      "step": 885000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.6458e-05,
+      "loss": 8.6622,
+      "step": 885500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.6456e-05,
+      "loss": 8.6962,
+      "step": 886000
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.6454000000000006e-05,
+      "loss": 8.6824,
+      "step": 886500
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.6452e-05,
+      "loss": 8.6729,
+      "step": 887000
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.6450000000000004e-05,
+      "loss": 8.7063,
+      "step": 887500
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.6448e-05,
+      "loss": 8.6514,
+      "step": 888000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.6446e-05,
+      "loss": 8.7005,
+      "step": 888500
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.6444000000000005e-05,
+      "loss": 8.6967,
+      "step": 889000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.6442e-05,
+      "loss": 8.6805,
+      "step": 889500
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.644e-05,
+      "loss": 8.6851,
+      "step": 890000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.6438000000000005e-05,
+      "loss": 8.7003,
+      "step": 890500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.6436e-05,
+      "loss": 8.6987,
+      "step": 891000
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.6434e-05,
+      "loss": 8.694,
+      "step": 891500
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.6432000000000005e-05,
+      "loss": 8.6898,
+      "step": 892000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.643e-05,
+      "loss": 8.6917,
+      "step": 892500
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.6428000000000003e-05,
+      "loss": 8.6972,
+      "step": 893000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.6426000000000006e-05,
+      "loss": 8.7035,
+      "step": 893500
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.6424e-05,
+      "loss": 8.7104,
+      "step": 894000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.6422000000000004e-05,
+      "loss": 8.7077,
+      "step": 894500
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.642e-05,
+      "loss": 8.6918,
+      "step": 895000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.6418e-05,
+      "loss": 8.6994,
+      "step": 895500
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.6416000000000004e-05,
+      "loss": 8.6486,
+      "step": 896000
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.6414e-05,
+      "loss": 8.6843,
+      "step": 896500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.6412e-05,
+      "loss": 8.6733,
+      "step": 897000
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.6410000000000005e-05,
+      "loss": 8.6797,
+      "step": 897500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.6408e-05,
+      "loss": 8.6684,
+      "step": 898000
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.6406e-05,
+      "loss": 8.6925,
+      "step": 898500
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.6404000000000005e-05,
+      "loss": 8.6618,
+      "step": 899000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.6402e-05,
+      "loss": 8.7055,
+      "step": 899500
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.64e-05,
+      "loss": 8.6637,
+      "step": 900000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.6398000000000005e-05,
+      "loss": 8.6877,
+      "step": 900500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.6396e-05,
+      "loss": 8.6848,
+      "step": 901000
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.6394e-05,
+      "loss": 8.6867,
+      "step": 901500
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.6392e-05,
+      "loss": 8.6845,
+      "step": 902000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.639e-05,
+      "loss": 8.6648,
+      "step": 902500
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.6388000000000004e-05,
+      "loss": 8.6905,
+      "step": 903000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.6386e-05,
+      "loss": 8.6813,
+      "step": 903500
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.6384e-05,
+      "loss": 8.7004,
+      "step": 904000
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.6382000000000004e-05,
+      "loss": 8.6759,
+      "step": 904500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.638e-05,
+      "loss": 8.6784,
+      "step": 905000
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.6378e-05,
+      "loss": 8.6783,
+      "step": 905500
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.6376000000000005e-05,
+      "loss": 8.6821,
+      "step": 906000
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.6374e-05,
+      "loss": 8.704,
+      "step": 906500
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.6372e-05,
+      "loss": 8.681,
+      "step": 907000
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.6370000000000005e-05,
+      "loss": 8.664,
+      "step": 907500
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.6368e-05,
+      "loss": 8.6814,
+      "step": 908000
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.6366e-05,
+      "loss": 8.6596,
+      "step": 908500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.6364e-05,
+      "loss": 8.6903,
+      "step": 909000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.6362e-05,
+      "loss": 8.696,
+      "step": 909500
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.636e-05,
+      "loss": 8.6709,
+      "step": 910000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.6358e-05,
+      "loss": 8.6807,
+      "step": 910500
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.635600000000001e-05,
+      "loss": 8.6804,
+      "step": 911000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.6354000000000004e-05,
+      "loss": 8.6734,
+      "step": 911500
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.6352e-05,
+      "loss": 8.6951,
+      "step": 912000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.635e-05,
+      "loss": 8.6676,
+      "step": 912500
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.6348000000000004e-05,
+      "loss": 8.6494,
+      "step": 913000
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.6346e-05,
+      "loss": 8.6651,
+      "step": 913500
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.6344e-05,
+      "loss": 8.6902,
+      "step": 914000
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.6342000000000005e-05,
+      "loss": 8.6748,
+      "step": 914500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.634e-05,
+      "loss": 8.7043,
+      "step": 915000
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.6338e-05,
+      "loss": 8.6911,
+      "step": 915500
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.6336000000000005e-05,
+      "loss": 8.6861,
+      "step": 916000
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.6334e-05,
+      "loss": 8.7087,
+      "step": 916500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.6332e-05,
+      "loss": 8.6657,
+      "step": 917000
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.633e-05,
+      "loss": 8.6511,
+      "step": 917500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.632800000000001e-05,
+      "loss": 8.6858,
+      "step": 918000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.6326e-05,
+      "loss": 8.6662,
+      "step": 918500
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.6324e-05,
+      "loss": 8.6781,
+      "step": 919000
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.6322e-05,
+      "loss": 8.697,
+      "step": 919500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.6320000000000004e-05,
+      "loss": 8.6628,
+      "step": 920000
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.6318e-05,
+      "loss": 8.6862,
+      "step": 920500
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.6316e-05,
+      "loss": 8.6931,
+      "step": 921000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.6314000000000004e-05,
+      "loss": 8.6492,
+      "step": 921500
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.6312e-05,
+      "loss": 8.6837,
+      "step": 922000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.631e-05,
+      "loss": 8.7073,
+      "step": 922500
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.6308000000000005e-05,
+      "loss": 8.6857,
+      "step": 923000
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.630600000000001e-05,
+      "loss": 8.6884,
+      "step": 923500
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.6304e-05,
+      "loss": 8.6743,
+      "step": 924000
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.6302e-05,
+      "loss": 8.6903,
+      "step": 924500
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.630000000000001e-05,
+      "loss": 8.6847,
+      "step": 925000
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.6298e-05,
+      "loss": 8.698,
+      "step": 925500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.6296e-05,
+      "loss": 8.6978,
+      "step": 926000
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.6294e-05,
+      "loss": 8.6721,
+      "step": 926500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.6292e-05,
+      "loss": 8.6805,
+      "step": 927000
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.629e-05,
+      "loss": 8.6839,
+      "step": 927500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.6288e-05,
+      "loss": 8.6973,
+      "step": 928000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.6286000000000004e-05,
+      "loss": 8.6982,
+      "step": 928500
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.6284e-05,
+      "loss": 8.6642,
+      "step": 929000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.6282e-05,
+      "loss": 8.6692,
+      "step": 929500
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.6280000000000004e-05,
+      "loss": 8.6817,
+      "step": 930000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.6278000000000007e-05,
+      "loss": 8.6982,
+      "step": 930500
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.6276e-05,
+      "loss": 8.7005,
+      "step": 931000
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.6274e-05,
+      "loss": 8.6706,
+      "step": 931500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.627200000000001e-05,
+      "loss": 8.6844,
+      "step": 932000
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.627e-05,
+      "loss": 8.6853,
+      "step": 932500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.6268e-05,
+      "loss": 8.7154,
+      "step": 933000
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.6266e-05,
+      "loss": 8.6622,
+      "step": 933500
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.6264e-05,
+      "loss": 8.7021,
+      "step": 934000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.6262e-05,
+      "loss": 8.6999,
+      "step": 934500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.626e-05,
+      "loss": 8.7232,
+      "step": 935000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.6258e-05,
+      "loss": 8.6903,
+      "step": 935500
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.6256000000000006e-05,
+      "loss": 8.7058,
+      "step": 936000
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.6254e-05,
+      "loss": 8.678,
+      "step": 936500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.6252000000000004e-05,
+      "loss": 8.6845,
+      "step": 937000
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 8.7047,
+      "step": 937500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.6248e-05,
+      "loss": 8.7072,
+      "step": 938000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.6246e-05,
+      "loss": 8.6988,
+      "step": 938500
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.6244000000000007e-05,
+      "loss": 8.708,
+      "step": 939000
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.6242e-05,
+      "loss": 8.6982,
+      "step": 939500
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.624e-05,
+      "loss": 8.6931,
+      "step": 940000
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.623800000000001e-05,
+      "loss": 8.6934,
+      "step": 940500
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.6236e-05,
+      "loss": 8.7143,
+      "step": 941000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.6234e-05,
+      "loss": 8.6808,
+      "step": 941500
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6232e-05,
+      "loss": 8.6717,
+      "step": 942000
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.623e-05,
+      "loss": 8.6726,
+      "step": 942500
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6228000000000005e-05,
+      "loss": 8.6821,
+      "step": 943000
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.6226e-05,
+      "loss": 8.6614,
+      "step": 943500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.6224e-05,
+      "loss": 8.6905,
+      "step": 944000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.6222000000000006e-05,
+      "loss": 8.6846,
+      "step": 944500
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.622e-05,
+      "loss": 8.676,
+      "step": 945000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.6218e-05,
+      "loss": 8.6599,
+      "step": 945500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.6216000000000006e-05,
+      "loss": 8.7162,
+      "step": 946000
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.6214e-05,
+      "loss": 8.6802,
+      "step": 946500
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.6212e-05,
+      "loss": 8.6806,
+      "step": 947000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.6210000000000006e-05,
+      "loss": 8.6999,
+      "step": 947500
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.6208e-05,
+      "loss": 8.7034,
+      "step": 948000
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.6206000000000005e-05,
+      "loss": 8.687,
+      "step": 948500
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.6204e-05,
+      "loss": 8.6834,
+      "step": 949000
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.6202e-05,
+      "loss": 8.6872,
+      "step": 949500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 8.6548,
+      "step": 950000
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.6198e-05,
+      "loss": 8.6762,
+      "step": 950500
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.6196e-05,
+      "loss": 8.6646,
+      "step": 951000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.6194000000000005e-05,
+      "loss": 8.6795,
+      "step": 951500
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.6192e-05,
+      "loss": 8.6805,
+      "step": 952000
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.619e-05,
+      "loss": 8.6733,
+      "step": 952500
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.6188000000000006e-05,
+      "loss": 8.6975,
+      "step": 953000
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.6186e-05,
+      "loss": 8.6797,
+      "step": 953500
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.6184e-05,
+      "loss": 8.6787,
+      "step": 954000
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.6182000000000006e-05,
+      "loss": 8.6848,
+      "step": 954500
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.618e-05,
+      "loss": 8.6965,
+      "step": 955000
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.6178000000000004e-05,
+      "loss": 8.6782,
+      "step": 955500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.6176e-05,
+      "loss": 8.6756,
+      "step": 956000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.6174e-05,
+      "loss": 8.712,
+      "step": 956500
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.6172000000000004e-05,
+      "loss": 8.6875,
+      "step": 957000
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.617e-05,
+      "loss": 8.6896,
+      "step": 957500
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.6168e-05,
+      "loss": 8.69,
+      "step": 958000
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.6166000000000005e-05,
+      "loss": 8.6729,
+      "step": 958500
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.6164e-05,
+      "loss": 8.6989,
+      "step": 959000
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.6162e-05,
+      "loss": 8.6812,
+      "step": 959500
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.6160000000000005e-05,
+      "loss": 8.7029,
+      "step": 960000
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.6158e-05,
+      "loss": 8.6698,
+      "step": 960500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.6156e-05,
+      "loss": 8.6891,
+      "step": 961000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.6154000000000006e-05,
+      "loss": 8.6966,
+      "step": 961500
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.6152e-05,
+      "loss": 8.6876,
+      "step": 962000
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.6150000000000004e-05,
+      "loss": 8.6973,
+      "step": 962500
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.6148e-05,
+      "loss": 8.6773,
+      "step": 963000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.6146e-05,
+      "loss": 8.6886,
+      "step": 963500
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.6144000000000004e-05,
+      "loss": 8.6869,
+      "step": 964000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.6142e-05,
+      "loss": 8.6692,
+      "step": 964500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.614e-05,
+      "loss": 8.6766,
+      "step": 965000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.6138000000000004e-05,
+      "loss": 8.6666,
+      "step": 965500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.6136e-05,
+      "loss": 8.687,
+      "step": 966000
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.6134e-05,
+      "loss": 8.6504,
+      "step": 966500
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.6132000000000005e-05,
+      "loss": 8.6796,
+      "step": 967000
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.613e-05,
+      "loss": 8.6913,
+      "step": 967500
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.6128e-05,
+      "loss": 8.6593,
+      "step": 968000
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.6126000000000005e-05,
+      "loss": 8.6936,
+      "step": 968500
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.6124e-05,
+      "loss": 8.6986,
+      "step": 969000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.6122e-05,
+      "loss": 8.7003,
+      "step": 969500
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.612e-05,
+      "loss": 8.7268,
+      "step": 970000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.6118e-05,
+      "loss": 8.6642,
+      "step": 970500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.6116000000000004e-05,
+      "loss": 8.7087,
+      "step": 971000
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.6114e-05,
+      "loss": 8.6864,
+      "step": 971500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.6112e-05,
+      "loss": 8.6814,
+      "step": 972000
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.6110000000000004e-05,
+      "loss": 8.6834,
+      "step": 972500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.6108e-05,
+      "loss": 8.6783,
+      "step": 973000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.6106e-05,
+      "loss": 8.6716,
+      "step": 973500
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.6104000000000004e-05,
+      "loss": 8.691,
+      "step": 974000
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.6102e-05,
+      "loss": 8.6724,
+      "step": 974500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.61e-05,
+      "loss": 8.6989,
+      "step": 975000
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.6098000000000005e-05,
+      "loss": 8.6891,
+      "step": 975500
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.6096e-05,
+      "loss": 8.7081,
+      "step": 976000
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.6094e-05,
+      "loss": 8.6749,
+      "step": 976500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.6092e-05,
+      "loss": 8.6825,
+      "step": 977000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.609e-05,
+      "loss": 8.6799,
+      "step": 977500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.6088e-05,
+      "loss": 8.6761,
+      "step": 978000
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.6086e-05,
+      "loss": 8.6895,
+      "step": 978500
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.608400000000001e-05,
+      "loss": 8.6833,
+      "step": 979000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.6082000000000004e-05,
+      "loss": 8.6868,
+      "step": 979500
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.608e-05,
+      "loss": 8.6998,
+      "step": 980000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.6078e-05,
+      "loss": 8.6742,
+      "step": 980500
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.6076000000000004e-05,
+      "loss": 8.6903,
+      "step": 981000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.6074e-05,
+      "loss": 8.7028,
+      "step": 981500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.6072e-05,
+      "loss": 8.6664,
+      "step": 982000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.6070000000000004e-05,
+      "loss": 8.7171,
+      "step": 982500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.6068e-05,
+      "loss": 8.6859,
+      "step": 983000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.6066e-05,
+      "loss": 8.6715,
+      "step": 983500
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.6064000000000005e-05,
+      "loss": 8.6685,
+      "step": 984000
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.6062e-05,
+      "loss": 8.7039,
+      "step": 984500
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.606e-05,
+      "loss": 8.6847,
+      "step": 985000
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.6058e-05,
+      "loss": 8.6586,
+      "step": 985500
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.605600000000001e-05,
+      "loss": 8.7179,
+      "step": 986000
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.6054e-05,
+      "loss": 8.6705,
+      "step": 986500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.6052e-05,
+      "loss": 8.6893,
+      "step": 987000
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.605e-05,
+      "loss": 8.6609,
+      "step": 987500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.6048000000000004e-05,
+      "loss": 8.684,
+      "step": 988000
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.6046e-05,
+      "loss": 8.6849,
+      "step": 988500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.6044e-05,
+      "loss": 8.681,
+      "step": 989000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.6042000000000004e-05,
+      "loss": 8.6703,
+      "step": 989500
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.604e-05,
+      "loss": 8.6585,
+      "step": 990000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.6038e-05,
+      "loss": 8.6652,
+      "step": 990500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.6036000000000004e-05,
+      "loss": 8.6842,
+      "step": 991000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.603400000000001e-05,
+      "loss": 8.6643,
+      "step": 991500
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.6032e-05,
+      "loss": 8.6786,
+      "step": 992000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.603e-05,
+      "loss": 8.6923,
+      "step": 992500
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.602800000000001e-05,
+      "loss": 8.687,
+      "step": 993000
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.6026e-05,
+      "loss": 8.6822,
+      "step": 993500
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.6024e-05,
+      "loss": 8.692,
+      "step": 994000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6022e-05,
+      "loss": 8.7046,
+      "step": 994500
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.602e-05,
+      "loss": 8.6693,
+      "step": 995000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6018e-05,
+      "loss": 8.6736,
+      "step": 995500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.6016e-05,
+      "loss": 8.6926,
+      "step": 996000
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.6014000000000004e-05,
+      "loss": 8.6698,
+      "step": 996500
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.6012e-05,
+      "loss": 8.6779,
+      "step": 997000
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.601e-05,
+      "loss": 8.7049,
+      "step": 997500
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.6008000000000004e-05,
+      "loss": 8.6701,
+      "step": 998000
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.6006000000000006e-05,
+      "loss": 8.6645,
+      "step": 998500
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.6004e-05,
+      "loss": 8.6496,
+      "step": 999000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.6002e-05,
+      "loss": 8.6946,
+      "step": 999500
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 8.6764,
+      "step": 1000000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.5998e-05,
+      "loss": 8.697,
+      "step": 1000500
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.5996e-05,
+      "loss": 8.7044,
+      "step": 1001000
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.5994e-05,
+      "loss": 8.7141,
+      "step": 1001500
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.5992e-05,
+      "loss": 8.6652,
+      "step": 1002000
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.599e-05,
+      "loss": 8.6987,
+      "step": 1002500
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.5988e-05,
+      "loss": 8.6945,
+      "step": 1003000
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.5986e-05,
+      "loss": 8.6813,
+      "step": 1003500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.5984000000000006e-05,
+      "loss": 8.7214,
+      "step": 1004000
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.5982e-05,
+      "loss": 8.6751,
+      "step": 1004500
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.5980000000000004e-05,
+      "loss": 8.6723,
+      "step": 1005000
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.5978000000000006e-05,
+      "loss": 8.6717,
+      "step": 1005500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.5976e-05,
+      "loss": 8.679,
+      "step": 1006000
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.5974e-05,
+      "loss": 8.658,
+      "step": 1006500
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.5972000000000006e-05,
+      "loss": 8.6902,
+      "step": 1007000
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.597e-05,
+      "loss": 8.7246,
+      "step": 1007500
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.5968e-05,
+      "loss": 8.7052,
+      "step": 1008000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.596600000000001e-05,
+      "loss": 8.6782,
+      "step": 1008500
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.5964e-05,
+      "loss": 8.6826,
+      "step": 1009000
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.5962e-05,
+      "loss": 8.6854,
+      "step": 1009500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.596e-05,
+      "loss": 8.6999,
+      "step": 1010000
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.5958e-05,
+      "loss": 8.7048,
+      "step": 1010500
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.5956000000000005e-05,
+      "loss": 8.676,
+      "step": 1011000
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.5954e-05,
+      "loss": 8.6606,
+      "step": 1011500
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.5952e-05,
+      "loss": 8.6777,
+      "step": 1012000
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.5950000000000006e-05,
+      "loss": 8.6933,
+      "step": 1012500
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.5948e-05,
+      "loss": 8.6798,
+      "step": 1013000
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.5946e-05,
+      "loss": 8.7053,
+      "step": 1013500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.5944000000000006e-05,
+      "loss": 8.6938,
+      "step": 1014000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.5942e-05,
+      "loss": 8.6784,
+      "step": 1014500
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.594e-05,
+      "loss": 8.676,
+      "step": 1015000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.5938000000000006e-05,
+      "loss": 8.6565,
+      "step": 1015500
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.5936e-05,
+      "loss": 8.6663,
+      "step": 1016000
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.5934000000000004e-05,
+      "loss": 8.6991,
+      "step": 1016500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.5932e-05,
+      "loss": 8.6868,
+      "step": 1017000
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.593e-05,
+      "loss": 8.6851,
+      "step": 1017500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.5928000000000005e-05,
+      "loss": 8.6733,
+      "step": 1018000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.5926e-05,
+      "loss": 8.6802,
+      "step": 1018500
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.5924e-05,
+      "loss": 8.6988,
+      "step": 1019000
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.5922000000000005e-05,
+      "loss": 8.6794,
+      "step": 1019500
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.592e-05,
+      "loss": 8.7025,
+      "step": 1020000
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.5918e-05,
+      "loss": 8.6893,
+      "step": 1020500
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.5916000000000006e-05,
+      "loss": 8.6904,
+      "step": 1021000
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.5914e-05,
+      "loss": 8.6713,
+      "step": 1021500
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.5912e-05,
+      "loss": 8.6845,
+      "step": 1022000
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.5910000000000006e-05,
+      "loss": 8.6934,
+      "step": 1022500
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.5908e-05,
+      "loss": 8.7045,
+      "step": 1023000
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.5906000000000004e-05,
+      "loss": 8.6809,
+      "step": 1023500
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.5904e-05,
+      "loss": 8.6934,
+      "step": 1024000
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.5902e-05,
+      "loss": 8.7136,
+      "step": 1024500
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 8.7,
+      "step": 1025000
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.5898e-05,
+      "loss": 8.6985,
+      "step": 1025500
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.5896e-05,
+      "loss": 8.6758,
+      "step": 1026000
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.5894000000000005e-05,
+      "loss": 8.679,
+      "step": 1026500
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.5892e-05,
+      "loss": 8.689,
+      "step": 1027000
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.589e-05,
+      "loss": 8.6771,
+      "step": 1027500
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.5888000000000005e-05,
+      "loss": 8.6877,
+      "step": 1028000
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.5886e-05,
+      "loss": 8.7039,
+      "step": 1028500
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.5884e-05,
+      "loss": 8.6799,
+      "step": 1029000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.5882000000000006e-05,
+      "loss": 8.6733,
+      "step": 1029500
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.588e-05,
+      "loss": 8.6921,
+      "step": 1030000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.5878000000000004e-05,
+      "loss": 8.7019,
+      "step": 1030500
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.5876e-05,
+      "loss": 8.6973,
+      "step": 1031000
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.5874e-05,
+      "loss": 8.6742,
+      "step": 1031500
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.5872000000000004e-05,
+      "loss": 8.6983,
+      "step": 1032000
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.587e-05,
+      "loss": 8.6928,
+      "step": 1032500
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.5868e-05,
+      "loss": 8.6773,
+      "step": 1033000
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.5866000000000004e-05,
+      "loss": 8.6524,
+      "step": 1033500
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.5864e-05,
+      "loss": 8.6835,
+      "step": 1034000
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.5862e-05,
+      "loss": 8.681,
+      "step": 1034500
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.5860000000000005e-05,
+      "loss": 8.6665,
+      "step": 1035000
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.5858e-05,
+      "loss": 8.6794,
+      "step": 1035500
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.5856e-05,
+      "loss": 8.6984,
+      "step": 1036000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.5854000000000005e-05,
+      "loss": 8.6684,
+      "step": 1036500
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.5852e-05,
+      "loss": 8.6814,
+      "step": 1037000
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.585e-05,
+      "loss": 8.654,
+      "step": 1037500
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.5848e-05,
+      "loss": 8.6615,
+      "step": 1038000
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.5846e-05,
+      "loss": 8.7002,
+      "step": 1038500
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.5844000000000004e-05,
+      "loss": 8.6894,
+      "step": 1039000
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.5842e-05,
+      "loss": 8.673,
+      "step": 1039500
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.584e-05,
+      "loss": 8.6806,
+      "step": 1040000
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.5838000000000004e-05,
+      "loss": 8.6709,
+      "step": 1040500
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.5836e-05,
+      "loss": 8.6795,
+      "step": 1041000
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.5834e-05,
+      "loss": 8.6674,
+      "step": 1041500
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.5832000000000004e-05,
+      "loss": 8.6833,
+      "step": 1042000
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.583e-05,
+      "loss": 8.669,
+      "step": 1042500
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.5828e-05,
+      "loss": 8.7021,
+      "step": 1043000
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.5826000000000005e-05,
+      "loss": 8.688,
+      "step": 1043500
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.5824e-05,
+      "loss": 8.6934,
+      "step": 1044000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.5822e-05,
+      "loss": 8.6851,
+      "step": 1044500
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.5820000000000005e-05,
+      "loss": 8.6744,
+      "step": 1045000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.5818e-05,
+      "loss": 8.6615,
+      "step": 1045500
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.5816e-05,
+      "loss": 8.7014,
+      "step": 1046000
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.5814e-05,
+      "loss": 8.697,
+      "step": 1046500
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.581200000000001e-05,
+      "loss": 8.6944,
+      "step": 1047000
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.5810000000000004e-05,
+      "loss": 8.6951,
+      "step": 1047500
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.5808e-05,
+      "loss": 8.6984,
+      "step": 1048000
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.5806e-05,
+      "loss": 8.6842,
+      "step": 1048500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.5804000000000004e-05,
+      "loss": 8.6664,
+      "step": 1049000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.5802e-05,
+      "loss": 8.6947,
+      "step": 1049500
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.58e-05,
+      "loss": 8.701,
+      "step": 1050000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.5798000000000004e-05,
+      "loss": 8.6766,
+      "step": 1050500
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.5796e-05,
+      "loss": 8.7013,
+      "step": 1051000
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.5794e-05,
+      "loss": 8.687,
+      "step": 1051500
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.5792000000000005e-05,
+      "loss": 8.6919,
+      "step": 1052000
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.579e-05,
+      "loss": 8.6885,
+      "step": 1052500
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.5788e-05,
+      "loss": 8.6599,
+      "step": 1053000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.5786e-05,
+      "loss": 8.7044,
+      "step": 1053500
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.578400000000001e-05,
+      "loss": 8.6804,
+      "step": 1054000
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.5782e-05,
+      "loss": 8.7054,
+      "step": 1054500
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.578e-05,
+      "loss": 8.6756,
+      "step": 1055000
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.5778e-05,
+      "loss": 8.6731,
+      "step": 1055500
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.5776000000000004e-05,
+      "loss": 8.6934,
+      "step": 1056000
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.5774e-05,
+      "loss": 8.6564,
+      "step": 1056500
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.5772e-05,
+      "loss": 8.6726,
+      "step": 1057000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.5770000000000004e-05,
+      "loss": 8.6737,
+      "step": 1057500
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.5768e-05,
+      "loss": 8.7018,
+      "step": 1058000
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.5766e-05,
+      "loss": 8.661,
+      "step": 1058500
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.5764000000000004e-05,
+      "loss": 8.6806,
+      "step": 1059000
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.576200000000001e-05,
+      "loss": 8.7031,
+      "step": 1059500
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.576e-05,
+      "loss": 8.6937,
+      "step": 1060000
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.5758e-05,
+      "loss": 8.7177,
+      "step": 1060500
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.575600000000001e-05,
+      "loss": 8.6888,
+      "step": 1061000
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.5754e-05,
+      "loss": 8.704,
+      "step": 1061500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.5752e-05,
+      "loss": 8.692,
+      "step": 1062000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.575e-05,
+      "loss": 8.6758,
+      "step": 1062500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.5748e-05,
+      "loss": 8.6767,
+      "step": 1063000
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.5746e-05,
+      "loss": 8.6824,
+      "step": 1063500
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.5744e-05,
+      "loss": 8.6717,
+      "step": 1064000
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.5742000000000004e-05,
+      "loss": 8.67,
+      "step": 1064500
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.574e-05,
+      "loss": 8.6797,
+      "step": 1065000
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.5738e-05,
+      "loss": 8.6688,
+      "step": 1065500
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.5736000000000004e-05,
+      "loss": 8.6806,
+      "step": 1066000
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.5734000000000006e-05,
+      "loss": 8.6835,
+      "step": 1066500
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.5732e-05,
+      "loss": 8.699,
+      "step": 1067000
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.573e-05,
+      "loss": 8.688,
+      "step": 1067500
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.572800000000001e-05,
+      "loss": 8.6814,
+      "step": 1068000
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.5726e-05,
+      "loss": 8.6903,
+      "step": 1068500
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.5724e-05,
+      "loss": 8.7022,
+      "step": 1069000
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.572200000000001e-05,
+      "loss": 8.7054,
+      "step": 1069500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.572e-05,
+      "loss": 8.6911,
+      "step": 1070000
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.5718e-05,
+      "loss": 8.6921,
+      "step": 1070500
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.5716e-05,
+      "loss": 8.6771,
+      "step": 1071000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.5714e-05,
+      "loss": 8.6779,
+      "step": 1071500
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.5712000000000006e-05,
+      "loss": 8.706,
+      "step": 1072000
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.571e-05,
+      "loss": 8.6893,
+      "step": 1072500
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.5708000000000004e-05,
+      "loss": 8.7018,
+      "step": 1073000
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.5706000000000006e-05,
+      "loss": 8.7055,
+      "step": 1073500
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.5704e-05,
+      "loss": 8.7174,
+      "step": 1074000
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5702e-05,
+      "loss": 8.6836,
+      "step": 1074500
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5700000000000006e-05,
+      "loss": 8.7157,
+      "step": 1075000
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5698e-05,
+      "loss": 8.6948,
+      "step": 1075500
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5696e-05,
+      "loss": 8.6763,
+      "step": 1076000
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.569400000000001e-05,
+      "loss": 8.711,
+      "step": 1076500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.5692e-05,
+      "loss": 8.6814,
+      "step": 1077000
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.569e-05,
+      "loss": 8.695,
+      "step": 1077500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.5688e-05,
+      "loss": 8.6985,
+      "step": 1078000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.5686e-05,
+      "loss": 8.6937,
+      "step": 1078500
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.5684000000000005e-05,
+      "loss": 8.674,
+      "step": 1079000
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.5682e-05,
+      "loss": 8.6655,
+      "step": 1079500
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.568e-05,
+      "loss": 8.6605,
+      "step": 1080000
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.5678000000000005e-05,
+      "loss": 8.6464,
+      "step": 1080500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.5676e-05,
+      "loss": 8.6928,
+      "step": 1081000
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.5674000000000003e-05,
+      "loss": 8.6973,
+      "step": 1081500
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.5672000000000006e-05,
+      "loss": 8.6953,
+      "step": 1082000
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.567e-05,
+      "loss": 8.7098,
+      "step": 1082500
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.5668e-05,
+      "loss": 8.6848,
+      "step": 1083000
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.5666000000000006e-05,
+      "loss": 8.6531,
+      "step": 1083500
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.5664e-05,
+      "loss": 8.6983,
+      "step": 1084000
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.5662000000000004e-05,
+      "loss": 8.6984,
+      "step": 1084500
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.566e-05,
+      "loss": 8.6849,
+      "step": 1085000
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.5658e-05,
+      "loss": 8.6652,
+      "step": 1085500
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.5656000000000005e-05,
+      "loss": 8.6564,
+      "step": 1086000
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.5654e-05,
+      "loss": 8.7027,
+      "step": 1086500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.5652e-05,
+      "loss": 8.6681,
+      "step": 1087000
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.5650000000000005e-05,
+      "loss": 8.6785,
+      "step": 1087500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.5648e-05,
+      "loss": 8.6725,
+      "step": 1088000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.5646e-05,
+      "loss": 8.6948,
+      "step": 1088500
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.5644000000000005e-05,
+      "loss": 8.672,
+      "step": 1089000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.5642e-05,
+      "loss": 8.703,
+      "step": 1089500
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.564e-05,
+      "loss": 8.6858,
+      "step": 1090000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.5638000000000006e-05,
+      "loss": 8.7108,
+      "step": 1090500
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.5636e-05,
+      "loss": 8.6863,
+      "step": 1091000
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.5634000000000004e-05,
+      "loss": 8.6943,
+      "step": 1091500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.5632e-05,
+      "loss": 8.6723,
+      "step": 1092000
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.563e-05,
+      "loss": 8.6849,
+      "step": 1092500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.5628000000000004e-05,
+      "loss": 8.6747,
+      "step": 1093000
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.5626e-05,
+      "loss": 8.683,
+      "step": 1093500
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.5624e-05,
+      "loss": 8.6828,
+      "step": 1094000
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.5622000000000005e-05,
+      "loss": 8.668,
+      "step": 1094500
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.562e-05,
+      "loss": 8.7027,
+      "step": 1095000
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.5618e-05,
+      "loss": 8.6762,
+      "step": 1095500
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.5616000000000005e-05,
+      "loss": 8.6559,
+      "step": 1096000
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.5614e-05,
+      "loss": 8.682,
+      "step": 1096500
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.5612e-05,
+      "loss": 8.6751,
+      "step": 1097000
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.5610000000000005e-05,
+      "loss": 8.6736,
+      "step": 1097500
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.5608e-05,
+      "loss": 8.6695,
+      "step": 1098000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.5606000000000003e-05,
+      "loss": 8.6935,
+      "step": 1098500
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.5604e-05,
+      "loss": 8.6871,
+      "step": 1099000
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.5602e-05,
+      "loss": 8.6717,
+      "step": 1099500
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 8.7084,
+      "step": 1100000
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.5598e-05,
+      "loss": 8.6653,
+      "step": 1100500
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.5596e-05,
+      "loss": 8.6648,
+      "step": 1101000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.5594000000000004e-05,
+      "loss": 8.695,
+      "step": 1101500
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.5592e-05,
+      "loss": 8.6835,
+      "step": 1102000
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.559e-05,
+      "loss": 8.6787,
+      "step": 1102500
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.5588000000000005e-05,
+      "loss": 8.6911,
+      "step": 1103000
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.5586e-05,
+      "loss": 8.6841,
+      "step": 1103500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.5584e-05,
+      "loss": 8.6883,
+      "step": 1104000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.5582000000000005e-05,
+      "loss": 8.6887,
+      "step": 1104500
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.558e-05,
+      "loss": 8.6936,
+      "step": 1105000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.5578e-05,
+      "loss": 8.6631,
+      "step": 1105500
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.5576e-05,
+      "loss": 8.692,
+      "step": 1106000
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.5574e-05,
+      "loss": 8.6803,
+      "step": 1106500
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.5572000000000003e-05,
+      "loss": 8.6719,
+      "step": 1107000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.557e-05,
+      "loss": 8.6884,
+      "step": 1107500
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.5568e-05,
+      "loss": 8.7145,
+      "step": 1108000
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.5566000000000004e-05,
+      "loss": 8.6818,
+      "step": 1108500
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.5564e-05,
+      "loss": 8.6981,
+      "step": 1109000
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.5562e-05,
+      "loss": 8.6619,
+      "step": 1109500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.5560000000000004e-05,
+      "loss": 8.6769,
+      "step": 1110000
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.5558e-05,
+      "loss": 8.7165,
+      "step": 1110500
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.5556e-05,
+      "loss": 8.6883,
+      "step": 1111000
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.5554000000000005e-05,
+      "loss": 8.6881,
+      "step": 1111500
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.5552e-05,
+      "loss": 8.6761,
+      "step": 1112000
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.555e-05,
+      "loss": 8.6914,
+      "step": 1112500
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.5548000000000005e-05,
+      "loss": 8.7039,
+      "step": 1113000
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.5546e-05,
+      "loss": 8.6931,
+      "step": 1113500
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.5544e-05,
+      "loss": 8.6731,
+      "step": 1114000
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.5542e-05,
+      "loss": 8.6768,
+      "step": 1114500
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.554000000000001e-05,
+      "loss": 8.6697,
+      "step": 1115000
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.5538000000000003e-05,
+      "loss": 8.679,
+      "step": 1115500
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.5536e-05,
+      "loss": 8.6865,
+      "step": 1116000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.5534e-05,
+      "loss": 8.6871,
+      "step": 1116500
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.5532000000000004e-05,
+      "loss": 8.6906,
+      "step": 1117000
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.553e-05,
+      "loss": 8.7003,
+      "step": 1117500
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.5528e-05,
+      "loss": 8.6846,
+      "step": 1118000
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.5526000000000004e-05,
+      "loss": 8.6789,
+      "step": 1118500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.5524e-05,
+      "loss": 8.667,
+      "step": 1119000
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.5522e-05,
+      "loss": 8.6753,
+      "step": 1119500
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.5520000000000005e-05,
+      "loss": 8.7123,
+      "step": 1120000
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.5518e-05,
+      "loss": 8.6812,
+      "step": 1120500
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.5516e-05,
+      "loss": 8.6849,
+      "step": 1121000
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.5514e-05,
+      "loss": 8.6712,
+      "step": 1121500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.551200000000001e-05,
+      "loss": 8.6826,
+      "step": 1122000
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.551e-05,
+      "loss": 8.6913,
+      "step": 1122500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.5508e-05,
+      "loss": 8.6773,
+      "step": 1123000
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.5506e-05,
+      "loss": 8.6498,
+      "step": 1123500
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.5504000000000003e-05,
+      "loss": 8.6851,
+      "step": 1124000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.5502e-05,
+      "loss": 8.6908,
+      "step": 1124500
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.55e-05,
+      "loss": 8.6865,
+      "step": 1125000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.5498000000000004e-05,
+      "loss": 8.69,
+      "step": 1125500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.5496e-05,
+      "loss": 8.6816,
+      "step": 1126000
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.5494e-05,
+      "loss": 8.6707,
+      "step": 1126500
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.5492000000000004e-05,
+      "loss": 8.6813,
+      "step": 1127000
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.549000000000001e-05,
+      "loss": 8.6957,
+      "step": 1127500
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.5488e-05,
+      "loss": 8.6902,
+      "step": 1128000
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.5486e-05,
+      "loss": 8.6867,
+      "step": 1128500
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.548400000000001e-05,
+      "loss": 8.7026,
+      "step": 1129000
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.5482e-05,
+      "loss": 8.7217,
+      "step": 1129500
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.548e-05,
+      "loss": 8.716,
+      "step": 1130000
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.5478e-05,
+      "loss": 8.6752,
+      "step": 1130500
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.5476e-05,
+      "loss": 8.6636,
+      "step": 1131000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.5474e-05,
+      "loss": 8.6819,
+      "step": 1131500
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.5472e-05,
+      "loss": 8.6725,
+      "step": 1132000
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.5470000000000003e-05,
+      "loss": 8.6871,
+      "step": 1132500
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.5468e-05,
+      "loss": 8.683,
+      "step": 1133000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.5466e-05,
+      "loss": 8.6964,
+      "step": 1133500
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.5464000000000004e-05,
+      "loss": 8.6957,
+      "step": 1134000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.5462000000000006e-05,
+      "loss": 8.6726,
+      "step": 1134500
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.546e-05,
+      "loss": 8.6876,
+      "step": 1135000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.5458e-05,
+      "loss": 8.6787,
+      "step": 1135500
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.5456000000000007e-05,
+      "loss": 8.6688,
+      "step": 1136000
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.5454e-05,
+      "loss": 8.6881,
+      "step": 1136500
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.5452e-05,
+      "loss": 8.6567,
+      "step": 1137000
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.545000000000001e-05,
+      "loss": 8.6703,
+      "step": 1137500
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.5448e-05,
+      "loss": 8.6733,
+      "step": 1138000
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.5446e-05,
+      "loss": 8.7065,
+      "step": 1138500
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.5444e-05,
+      "loss": 8.6827,
+      "step": 1139000
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.5442e-05,
+      "loss": 8.7059,
+      "step": 1139500
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.5440000000000005e-05,
+      "loss": 8.6822,
+      "step": 1140000
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.5438e-05,
+      "loss": 8.673,
+      "step": 1140500
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.5436000000000003e-05,
+      "loss": 8.6909,
+      "step": 1141000
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.5434000000000006e-05,
+      "loss": 8.6873,
+      "step": 1141500
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.5432e-05,
+      "loss": 8.6955,
+      "step": 1142000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.543e-05,
+      "loss": 8.7092,
+      "step": 1142500
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.5428000000000006e-05,
+      "loss": 8.7171,
+      "step": 1143000
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.5426e-05,
+      "loss": 8.6882,
+      "step": 1143500
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.5424e-05,
+      "loss": 8.7058,
+      "step": 1144000
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.5422000000000007e-05,
+      "loss": 8.6528,
+      "step": 1144500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.542e-05,
+      "loss": 8.7011,
+      "step": 1145000
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.5418e-05,
+      "loss": 8.6872,
+      "step": 1145500
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.5416e-05,
+      "loss": 8.6969,
+      "step": 1146000
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.5414e-05,
+      "loss": 8.6982,
+      "step": 1146500
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.5412000000000005e-05,
+      "loss": 8.7002,
+      "step": 1147000
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.541e-05,
+      "loss": 8.6873,
+      "step": 1147500
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.5408e-05,
+      "loss": 8.6973,
+      "step": 1148000
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.5406000000000005e-05,
+      "loss": 8.694,
+      "step": 1148500
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.5404e-05,
+      "loss": 8.6527,
+      "step": 1149000
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.5402000000000003e-05,
+      "loss": 8.6799,
+      "step": 1149500
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 8.6838,
+      "step": 1150000
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.5398e-05,
+      "loss": 8.6942,
+      "step": 1150500
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.5396e-05,
+      "loss": 8.6611,
+      "step": 1151000
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.5394000000000006e-05,
+      "loss": 8.6773,
+      "step": 1151500
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.5392e-05,
+      "loss": 8.6724,
+      "step": 1152000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.5390000000000004e-05,
+      "loss": 8.6972,
+      "step": 1152500
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.5388e-05,
+      "loss": 8.6775,
+      "step": 1153000
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.5386e-05,
+      "loss": 8.6831,
+      "step": 1153500
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.5384000000000005e-05,
+      "loss": 8.6901,
+      "step": 1154000
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.5382e-05,
+      "loss": 8.677,
+      "step": 1154500
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.538e-05,
+      "loss": 8.6739,
+      "step": 1155000
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.5378000000000005e-05,
+      "loss": 8.6889,
+      "step": 1155500
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.5376e-05,
+      "loss": 8.6539,
+      "step": 1156000
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.5374e-05,
+      "loss": 8.6665,
+      "step": 1156500
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.5372000000000005e-05,
+      "loss": 8.6792,
+      "step": 1157000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.537e-05,
+      "loss": 8.6745,
+      "step": 1157500
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.5368e-05,
+      "loss": 8.6843,
+      "step": 1158000
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.5366000000000006e-05,
+      "loss": 8.6662,
+      "step": 1158500
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.5364e-05,
+      "loss": 8.6862,
+      "step": 1159000
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.5362000000000004e-05,
+      "loss": 8.6739,
+      "step": 1159500
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.536e-05,
+      "loss": 8.6797,
+      "step": 1160000
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.5358e-05,
+      "loss": 8.6775,
+      "step": 1160500
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.5356000000000004e-05,
+      "loss": 8.6791,
+      "step": 1161000
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.5354e-05,
+      "loss": 8.6946,
+      "step": 1161500
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.5352e-05,
+      "loss": 8.6883,
+      "step": 1162000
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.5350000000000005e-05,
+      "loss": 8.6724,
+      "step": 1162500
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.5348e-05,
+      "loss": 8.6816,
+      "step": 1163000
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.5346e-05,
+      "loss": 8.6815,
+      "step": 1163500
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.5344000000000005e-05,
+      "loss": 8.6697,
+      "step": 1164000
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.5342e-05,
+      "loss": 8.7017,
+      "step": 1164500
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.534e-05,
+      "loss": 8.6955,
+      "step": 1165000
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.5338000000000005e-05,
+      "loss": 8.6704,
+      "step": 1165500
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.5336e-05,
+      "loss": 8.6589,
+      "step": 1166000
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.5334e-05,
+      "loss": 8.6468,
+      "step": 1166500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.5332e-05,
+      "loss": 8.693,
+      "step": 1167000
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.533e-05,
+      "loss": 8.6931,
+      "step": 1167500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.5328000000000004e-05,
+      "loss": 8.7162,
+      "step": 1168000
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.5326e-05,
+      "loss": 8.6884,
+      "step": 1168500
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.5324e-05,
+      "loss": 8.712,
+      "step": 1169000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.5322000000000004e-05,
+      "loss": 8.6747,
+      "step": 1169500
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.532e-05,
+      "loss": 8.6712,
+      "step": 1170000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.5318e-05,
+      "loss": 8.677,
+      "step": 1170500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.5316000000000005e-05,
+      "loss": 8.6598,
+      "step": 1171000
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.5314e-05,
+      "loss": 8.685,
+      "step": 1171500
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.5312e-05,
+      "loss": 8.6576,
+      "step": 1172000
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.5310000000000005e-05,
+      "loss": 8.6396,
+      "step": 1172500
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.5308e-05,
+      "loss": 8.6703,
+      "step": 1173000
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.5306e-05,
+      "loss": 8.6643,
+      "step": 1173500
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.5304000000000005e-05,
+      "loss": 8.6905,
+      "step": 1174000
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.5302e-05,
+      "loss": 8.7058,
+      "step": 1174500
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.53e-05,
+      "loss": 8.667,
+      "step": 1175000
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.5298e-05,
+      "loss": 8.6675,
+      "step": 1175500
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.5296e-05,
+      "loss": 8.7209,
+      "step": 1176000
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.5294000000000004e-05,
+      "loss": 8.6623,
+      "step": 1176500
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.5292e-05,
+      "loss": 8.6906,
+      "step": 1177000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.529e-05,
+      "loss": 8.6851,
+      "step": 1177500
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.5288000000000004e-05,
+      "loss": 8.6968,
+      "step": 1178000
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.5286e-05,
+      "loss": 8.6732,
+      "step": 1178500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.5284e-05,
+      "loss": 8.6564,
+      "step": 1179000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.5282000000000005e-05,
+      "loss": 8.6834,
+      "step": 1179500
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.528e-05,
+      "loss": 8.6975,
+      "step": 1180000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.5278e-05,
+      "loss": 8.704,
+      "step": 1180500
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.5276000000000005e-05,
+      "loss": 8.6882,
+      "step": 1181000
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.5274e-05,
+      "loss": 8.6667,
+      "step": 1181500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.5272e-05,
+      "loss": 8.6596,
+      "step": 1182000
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.527e-05,
+      "loss": 8.6778,
+      "step": 1182500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.526800000000001e-05,
+      "loss": 8.6934,
+      "step": 1183000
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.5266e-05,
+      "loss": 8.677,
+      "step": 1183500
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.5264e-05,
+      "loss": 8.6835,
+      "step": 1184000
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.5262e-05,
+      "loss": 8.6924,
+      "step": 1184500
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.5260000000000004e-05,
+      "loss": 8.6882,
+      "step": 1185000
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.5258e-05,
+      "loss": 8.6857,
+      "step": 1185500
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.5256e-05,
+      "loss": 8.7064,
+      "step": 1186000
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.5254000000000004e-05,
+      "loss": 8.6854,
+      "step": 1186500
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.5252e-05,
+      "loss": 8.6783,
+      "step": 1187000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.525e-05,
+      "loss": 8.698,
+      "step": 1187500
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.5248000000000005e-05,
+      "loss": 8.6608,
+      "step": 1188000
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.5246e-05,
+      "loss": 8.6811,
+      "step": 1188500
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.5244e-05,
+      "loss": 8.6703,
+      "step": 1189000
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.5242e-05,
+      "loss": 8.6833,
+      "step": 1189500
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.524000000000001e-05,
+      "loss": 8.6919,
+      "step": 1190000
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.5238e-05,
+      "loss": 8.6848,
+      "step": 1190500
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.5236e-05,
+      "loss": 8.7048,
+      "step": 1191000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.5234e-05,
+      "loss": 8.6826,
+      "step": 1191500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.5232e-05,
+      "loss": 8.6761,
+      "step": 1192000
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.523e-05,
+      "loss": 8.6748,
+      "step": 1192500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.5228e-05,
+      "loss": 8.6893,
+      "step": 1193000
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.5226000000000004e-05,
+      "loss": 8.6839,
+      "step": 1193500
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.5224e-05,
+      "loss": 8.6602,
+      "step": 1194000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.5222e-05,
+      "loss": 8.6843,
+      "step": 1194500
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.5220000000000004e-05,
+      "loss": 8.6746,
+      "step": 1195000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.5218000000000007e-05,
+      "loss": 8.6799,
+      "step": 1195500
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.5216e-05,
+      "loss": 8.6897,
+      "step": 1196000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.5214e-05,
+      "loss": 8.6752,
+      "step": 1196500
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.521200000000001e-05,
+      "loss": 8.6867,
+      "step": 1197000
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.521e-05,
+      "loss": 8.7146,
+      "step": 1197500
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.5208e-05,
+      "loss": 8.6859,
+      "step": 1198000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.5206e-05,
+      "loss": 8.6733,
+      "step": 1198500
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.5204e-05,
+      "loss": 8.6976,
+      "step": 1199000
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.5202e-05,
+      "loss": 8.7035,
+      "step": 1199500
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.52e-05,
+      "loss": 8.7009,
+      "step": 1200000
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.5198e-05,
+      "loss": 8.6647,
+      "step": 1200500
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.5196e-05,
+      "loss": 8.6873,
+      "step": 1201000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.5194e-05,
+      "loss": 8.6858,
+      "step": 1201500
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.5192000000000004e-05,
+      "loss": 8.715,
+      "step": 1202000
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.5190000000000006e-05,
+      "loss": 8.7143,
+      "step": 1202500
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.5188e-05,
+      "loss": 8.6628,
+      "step": 1203000
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.5186e-05,
+      "loss": 8.7013,
+      "step": 1203500
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.5184000000000006e-05,
+      "loss": 8.6764,
+      "step": 1204000
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.5182e-05,
+      "loss": 8.6965,
+      "step": 1204500
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.518e-05,
+      "loss": 8.6844,
+      "step": 1205000
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.517800000000001e-05,
+      "loss": 8.6834,
+      "step": 1205500
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.5176e-05,
+      "loss": 8.705,
+      "step": 1206000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.5174e-05,
+      "loss": 8.683,
+      "step": 1206500
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.5172e-05,
+      "loss": 8.6824,
+      "step": 1207000
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.517e-05,
+      "loss": 8.6833,
+      "step": 1207500
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.5168000000000005e-05,
+      "loss": 8.6748,
+      "step": 1208000
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.5166e-05,
+      "loss": 8.6791,
+      "step": 1208500
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.5164e-05,
+      "loss": 8.6444,
+      "step": 1209000
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.5162000000000006e-05,
+      "loss": 8.6819,
+      "step": 1209500
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.516e-05,
+      "loss": 8.6921,
+      "step": 1210000
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.5158000000000004e-05,
+      "loss": 8.6828,
+      "step": 1210500
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.5156000000000006e-05,
+      "loss": 8.6836,
+      "step": 1211000
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.5154e-05,
+      "loss": 8.7005,
+      "step": 1211500
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.5152e-05,
+      "loss": 8.6894,
+      "step": 1212000
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.5150000000000006e-05,
+      "loss": 8.6928,
+      "step": 1212500
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.5148e-05,
+      "loss": 8.6539,
+      "step": 1213000
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.5146e-05,
+      "loss": 8.6771,
+      "step": 1213500
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.5144e-05,
+      "loss": 8.6776,
+      "step": 1214000
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.5142e-05,
+      "loss": 8.7029,
+      "step": 1214500
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.5140000000000005e-05,
+      "loss": 8.687,
+      "step": 1215000
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.5138e-05,
+      "loss": 8.6748,
+      "step": 1215500
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.5136e-05,
+      "loss": 8.673,
+      "step": 1216000
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.5134000000000005e-05,
+      "loss": 8.6738,
+      "step": 1216500
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.5132e-05,
+      "loss": 8.6695,
+      "step": 1217000
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.513e-05,
+      "loss": 8.6698,
+      "step": 1217500
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.5128000000000006e-05,
+      "loss": 8.6836,
+      "step": 1218000
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.5126e-05,
+      "loss": 8.6994,
+      "step": 1218500
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.5124e-05,
+      "loss": 8.6496,
+      "step": 1219000
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.5122000000000006e-05,
+      "loss": 8.6957,
+      "step": 1219500
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.512e-05,
+      "loss": 8.6748,
+      "step": 1220000
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.5118000000000004e-05,
+      "loss": 8.6937,
+      "step": 1220500
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.5116e-05,
+      "loss": 8.6993,
+      "step": 1221000
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.5114e-05,
+      "loss": 8.6919,
+      "step": 1221500
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.5112000000000004e-05,
+      "loss": 8.6906,
+      "step": 1222000
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.511e-05,
+      "loss": 8.669,
+      "step": 1222500
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.5108e-05,
+      "loss": 8.678,
+      "step": 1223000
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.5106000000000005e-05,
+      "loss": 8.6985,
+      "step": 1223500
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.5104e-05,
+      "loss": 8.6903,
+      "step": 1224000
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.5102e-05,
+      "loss": 8.7035,
+      "step": 1224500
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.5100000000000005e-05,
+      "loss": 8.6777,
+      "step": 1225000
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.5098e-05,
+      "loss": 8.6877,
+      "step": 1225500
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.5096e-05,
+      "loss": 8.6793,
+      "step": 1226000
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.5094000000000006e-05,
+      "loss": 8.6831,
+      "step": 1226500
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.5092e-05,
+      "loss": 8.6731,
+      "step": 1227000
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.5090000000000004e-05,
+      "loss": 8.6718,
+      "step": 1227500
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.5088e-05,
+      "loss": 8.6769,
+      "step": 1228000
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.5086e-05,
+      "loss": 8.6699,
+      "step": 1228500
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.5084000000000004e-05,
+      "loss": 8.7083,
+      "step": 1229000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.5082e-05,
+      "loss": 8.6809,
+      "step": 1229500
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.508e-05,
+      "loss": 8.6919,
+      "step": 1230000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.5078000000000004e-05,
+      "loss": 8.6801,
+      "step": 1230500
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.5076e-05,
+      "loss": 8.6558,
+      "step": 1231000
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.5074e-05,
+      "loss": 8.6635,
+      "step": 1231500
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.5072000000000005e-05,
+      "loss": 8.6968,
+      "step": 1232000
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.507e-05,
+      "loss": 8.675,
+      "step": 1232500
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.5068e-05,
+      "loss": 8.6717,
+      "step": 1233000
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.5066000000000005e-05,
+      "loss": 8.6742,
+      "step": 1233500
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.5064e-05,
+      "loss": 8.6858,
+      "step": 1234000
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.5062e-05,
+      "loss": 8.6819,
+      "step": 1234500
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.506e-05,
+      "loss": 8.6757,
+      "step": 1235000
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.5058e-05,
+      "loss": 8.6685,
+      "step": 1235500
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.5056000000000004e-05,
+      "loss": 8.6693,
+      "step": 1236000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.5054e-05,
+      "loss": 8.6885,
+      "step": 1236500
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.5052e-05,
+      "loss": 8.6897,
+      "step": 1237000
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.5050000000000004e-05,
+      "loss": 8.6522,
+      "step": 1237500
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.5048e-05,
+      "loss": 8.6907,
+      "step": 1238000
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.5046e-05,
+      "loss": 8.6739,
+      "step": 1238500
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.5044000000000004e-05,
+      "loss": 8.7044,
+      "step": 1239000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.5042e-05,
+      "loss": 8.663,
+      "step": 1239500
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.504e-05,
+      "loss": 8.6775,
+      "step": 1240000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.5038000000000005e-05,
+      "loss": 8.6476,
+      "step": 1240500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.5036e-05,
+      "loss": 8.698,
+      "step": 1241000
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.5034e-05,
+      "loss": 8.658,
+      "step": 1241500
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.5032000000000005e-05,
+      "loss": 8.6879,
+      "step": 1242000
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.503e-05,
+      "loss": 8.6903,
+      "step": 1242500
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.5028e-05,
+      "loss": 8.6587,
+      "step": 1243000
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.5026e-05,
+      "loss": 8.6678,
+      "step": 1243500
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.5024e-05,
+      "loss": 8.6691,
+      "step": 1244000
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.5022000000000004e-05,
+      "loss": 8.6914,
+      "step": 1244500
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.502e-05,
+      "loss": 8.6846,
+      "step": 1245000
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.5018e-05,
+      "loss": 8.6758,
+      "step": 1245500
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.5016000000000004e-05,
+      "loss": 8.6839,
+      "step": 1246000
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.5014e-05,
+      "loss": 8.7049,
+      "step": 1246500
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.5012e-05,
+      "loss": 8.6841,
+      "step": 1247000
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.5010000000000004e-05,
+      "loss": 8.702,
+      "step": 1247500
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.5008e-05,
+      "loss": 8.699,
+      "step": 1248000
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.5006e-05,
+      "loss": 8.693,
+      "step": 1248500
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.5004000000000005e-05,
+      "loss": 8.6829,
+      "step": 1249000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.5002e-05,
+      "loss": 8.7048,
+      "step": 1249500
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.5e-05,
+      "loss": 8.6988,
+      "step": 1250000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.4998e-05,
+      "loss": 8.6821,
+      "step": 1250500
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.499600000000001e-05,
+      "loss": 8.6763,
+      "step": 1251000
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.4994e-05,
+      "loss": 8.7007,
+      "step": 1251500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.4992e-05,
+      "loss": 8.694,
+      "step": 1252000
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.499e-05,
+      "loss": 8.6748,
+      "step": 1252500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.4988000000000004e-05,
+      "loss": 8.6479,
+      "step": 1253000
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.4986e-05,
+      "loss": 8.6867,
+      "step": 1253500
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.4984e-05,
+      "loss": 8.6867,
+      "step": 1254000
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.4982000000000004e-05,
+      "loss": 8.6843,
+      "step": 1254500
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.498e-05,
+      "loss": 8.6788,
+      "step": 1255000
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.4978e-05,
+      "loss": 8.6696,
+      "step": 1255500
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.4976000000000004e-05,
+      "loss": 8.6893,
+      "step": 1256000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.4974e-05,
+      "loss": 8.6841,
+      "step": 1256500
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.4972e-05,
+      "loss": 8.6759,
+      "step": 1257000
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.497e-05,
+      "loss": 8.7065,
+      "step": 1257500
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.496800000000001e-05,
+      "loss": 8.6793,
+      "step": 1258000
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.4966e-05,
+      "loss": 8.6588,
+      "step": 1258500
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.4964e-05,
+      "loss": 8.6842,
+      "step": 1259000
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.4962e-05,
+      "loss": 8.704,
+      "step": 1259500
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.496e-05,
+      "loss": 8.6781,
+      "step": 1260000
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.4958e-05,
+      "loss": 8.6905,
+      "step": 1260500
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.4956e-05,
+      "loss": 8.6818,
+      "step": 1261000
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.4954000000000004e-05,
+      "loss": 8.6809,
+      "step": 1261500
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.4952e-05,
+      "loss": 8.6735,
+      "step": 1262000
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.495e-05,
+      "loss": 8.6851,
+      "step": 1262500
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.4948000000000004e-05,
+      "loss": 8.6629,
+      "step": 1263000
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.4946000000000006e-05,
+      "loss": 8.6771,
+      "step": 1263500
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.4944e-05,
+      "loss": 8.6778,
+      "step": 1264000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4942e-05,
+      "loss": 8.6912,
+      "step": 1264500
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.494000000000001e-05,
+      "loss": 8.6871,
+      "step": 1265000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4938e-05,
+      "loss": 8.6768,
+      "step": 1265500
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.4936e-05,
+      "loss": 8.6829,
+      "step": 1266000
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.493400000000001e-05,
+      "loss": 8.6759,
+      "step": 1266500
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.4932e-05,
+      "loss": 8.6955,
+      "step": 1267000
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.493e-05,
+      "loss": 8.6918,
+      "step": 1267500
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.4928e-05,
+      "loss": 8.6779,
+      "step": 1268000
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.4926e-05,
+      "loss": 8.7041,
+      "step": 1268500
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.4924e-05,
+      "loss": 8.6847,
+      "step": 1269000
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.4922e-05,
+      "loss": 8.6974,
+      "step": 1269500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.4920000000000004e-05,
+      "loss": 8.6668,
+      "step": 1270000
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.4918000000000006e-05,
+      "loss": 8.6764,
+      "step": 1270500
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.4916e-05,
+      "loss": 8.7033,
+      "step": 1271000
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.4914e-05,
+      "loss": 8.6782,
+      "step": 1271500
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.4912000000000006e-05,
+      "loss": 8.6733,
+      "step": 1272000
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.491e-05,
+      "loss": 8.6912,
+      "step": 1272500
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.4908e-05,
+      "loss": 8.6948,
+      "step": 1273000
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.490600000000001e-05,
+      "loss": 8.6829,
+      "step": 1273500
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.4904e-05,
+      "loss": 8.6759,
+      "step": 1274000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.4902e-05,
+      "loss": 8.6712,
+      "step": 1274500
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.49e-05,
+      "loss": 8.6979,
+      "step": 1275000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.4898e-05,
+      "loss": 8.677,
+      "step": 1275500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.4896000000000005e-05,
+      "loss": 8.692,
+      "step": 1276000
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.4894e-05,
+      "loss": 8.6974,
+      "step": 1276500
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.4892e-05,
+      "loss": 8.6646,
+      "step": 1277000
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.4890000000000006e-05,
+      "loss": 8.703,
+      "step": 1277500
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.4888e-05,
+      "loss": 8.6844,
+      "step": 1278000
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.4886000000000004e-05,
+      "loss": 8.6685,
+      "step": 1278500
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.4884000000000006e-05,
+      "loss": 8.6807,
+      "step": 1279000
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.4882e-05,
+      "loss": 8.6909,
+      "step": 1279500
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.488e-05,
+      "loss": 8.6838,
+      "step": 1280000
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.4878000000000006e-05,
+      "loss": 8.6648,
+      "step": 1280500
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.4876e-05,
+      "loss": 8.6703,
+      "step": 1281000
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.4874000000000004e-05,
+      "loss": 8.7103,
+      "step": 1281500
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4872e-05,
+      "loss": 8.684,
+      "step": 1282000
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.487e-05,
+      "loss": 8.645,
+      "step": 1282500
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4868000000000005e-05,
+      "loss": 8.6707,
+      "step": 1283000
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.4866e-05,
+      "loss": 8.6758,
+      "step": 1283500
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.4864e-05,
+      "loss": 8.679,
+      "step": 1284000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.4862000000000005e-05,
+      "loss": 8.672,
+      "step": 1284500
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.486e-05,
+      "loss": 8.6981,
+      "step": 1285000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.4858e-05,
+      "loss": 8.6861,
+      "step": 1285500
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.4856000000000006e-05,
+      "loss": 8.6675,
+      "step": 1286000
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.4854e-05,
+      "loss": 8.6978,
+      "step": 1286500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.4852e-05,
+      "loss": 8.7052,
+      "step": 1287000
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.4850000000000006e-05,
+      "loss": 8.6748,
+      "step": 1287500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.4848e-05,
+      "loss": 8.6747,
+      "step": 1288000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.4846000000000004e-05,
+      "loss": 8.6947,
+      "step": 1288500
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.4844e-05,
+      "loss": 8.6767,
+      "step": 1289000
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.4842e-05,
+      "loss": 8.6911,
+      "step": 1289500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.4840000000000004e-05,
+      "loss": 8.6862,
+      "step": 1290000
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.4838e-05,
+      "loss": 8.6846,
+      "step": 1290500
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.4836e-05,
+      "loss": 8.6587,
+      "step": 1291000
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.4834000000000005e-05,
+      "loss": 8.6888,
+      "step": 1291500
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.4832e-05,
+      "loss": 8.7028,
+      "step": 1292000
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.483e-05,
+      "loss": 8.669,
+      "step": 1292500
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.4828000000000005e-05,
+      "loss": 8.6689,
+      "step": 1293000
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.4826e-05,
+      "loss": 8.6661,
+      "step": 1293500
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.4824e-05,
+      "loss": 8.6998,
+      "step": 1294000
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.4822000000000006e-05,
+      "loss": 8.6838,
+      "step": 1294500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.482e-05,
+      "loss": 8.6834,
+      "step": 1295000
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.4818000000000004e-05,
+      "loss": 8.6713,
+      "step": 1295500
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.4816e-05,
+      "loss": 8.672,
+      "step": 1296000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.4814e-05,
+      "loss": 8.6941,
+      "step": 1296500
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.4812000000000004e-05,
+      "loss": 8.6948,
+      "step": 1297000
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.481e-05,
+      "loss": 8.6743,
+      "step": 1297500
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.4808e-05,
+      "loss": 8.6913,
+      "step": 1298000
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.4806000000000004e-05,
+      "loss": 8.699,
+      "step": 1298500
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.4804e-05,
+      "loss": 8.6978,
+      "step": 1299000
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.4802e-05,
+      "loss": 8.6837,
+      "step": 1299500
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 8.6797,
+      "step": 1300000
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.4798e-05,
+      "loss": 8.6873,
+      "step": 1300500
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.4796e-05,
+      "loss": 8.6678,
+      "step": 1301000
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.4794000000000005e-05,
+      "loss": 8.6768,
+      "step": 1301500
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.4792e-05,
+      "loss": 8.6882,
+      "step": 1302000
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.479e-05,
+      "loss": 8.6682,
+      "step": 1302500
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.4788000000000006e-05,
+      "loss": 8.6659,
+      "step": 1303000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.4786e-05,
+      "loss": 8.676,
+      "step": 1303500
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.4784000000000004e-05,
+      "loss": 8.6675,
+      "step": 1304000
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.4782e-05,
+      "loss": 8.6916,
+      "step": 1304500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.478e-05,
+      "loss": 8.674,
+      "step": 1305000
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.4778000000000004e-05,
+      "loss": 8.7042,
+      "step": 1305500
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.4776e-05,
+      "loss": 8.6648,
+      "step": 1306000
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.4774e-05,
+      "loss": 8.6879,
+      "step": 1306500
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.4772000000000004e-05,
+      "loss": 8.6917,
+      "step": 1307000
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.477e-05,
+      "loss": 8.6734,
+      "step": 1307500
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.4768e-05,
+      "loss": 8.6804,
+      "step": 1308000
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.4766000000000005e-05,
+      "loss": 8.6807,
+      "step": 1308500
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.4764e-05,
+      "loss": 8.6974,
+      "step": 1309000
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.4762e-05,
+      "loss": 8.6956,
+      "step": 1309500
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.4760000000000005e-05,
+      "loss": 8.6606,
+      "step": 1310000
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.4758e-05,
+      "loss": 8.6837,
+      "step": 1310500
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.4756e-05,
+      "loss": 8.6747,
+      "step": 1311000
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.4754e-05,
+      "loss": 8.6753,
+      "step": 1311500
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.4752e-05,
+      "loss": 8.6538,
+      "step": 1312000
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.4750000000000004e-05,
+      "loss": 8.6761,
+      "step": 1312500
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.4748e-05,
+      "loss": 8.68,
+      "step": 1313000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.4746e-05,
+      "loss": 8.689,
+      "step": 1313500
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.4744000000000004e-05,
+      "loss": 8.6889,
+      "step": 1314000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.4742e-05,
+      "loss": 8.6552,
+      "step": 1314500
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.474e-05,
+      "loss": 8.6928,
+      "step": 1315000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.4738000000000004e-05,
+      "loss": 8.6594,
+      "step": 1315500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.4736e-05,
+      "loss": 8.6797,
+      "step": 1316000
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.4734e-05,
+      "loss": 8.6711,
+      "step": 1316500
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.4732000000000005e-05,
+      "loss": 8.6532,
+      "step": 1317000
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.473e-05,
+      "loss": 8.6831,
+      "step": 1317500
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.4728e-05,
+      "loss": 8.6986,
+      "step": 1318000
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.4726e-05,
+      "loss": 8.6883,
+      "step": 1318500
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.472400000000001e-05,
+      "loss": 8.7141,
+      "step": 1319000
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.4722e-05,
+      "loss": 8.6755,
+      "step": 1319500
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.472e-05,
+      "loss": 8.6628,
+      "step": 1320000
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.4718e-05,
+      "loss": 8.6673,
+      "step": 1320500
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.4716000000000004e-05,
+      "loss": 8.6782,
+      "step": 1321000
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.4714e-05,
+      "loss": 8.6657,
+      "step": 1321500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.4712e-05,
+      "loss": 8.6986,
+      "step": 1322000
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.4710000000000004e-05,
+      "loss": 8.673,
+      "step": 1322500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.4708e-05,
+      "loss": 8.6761,
+      "step": 1323000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.4706e-05,
+      "loss": 8.6888,
+      "step": 1323500
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.4704000000000004e-05,
+      "loss": 8.6774,
+      "step": 1324000
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.4702e-05,
+      "loss": 8.6629,
+      "step": 1324500
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.47e-05,
+      "loss": 8.6626,
+      "step": 1325000
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.4698e-05,
+      "loss": 8.6831,
+      "step": 1325500
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.469600000000001e-05,
+      "loss": 8.6885,
+      "step": 1326000
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.4694e-05,
+      "loss": 8.7342,
+      "step": 1326500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.4692e-05,
+      "loss": 8.7031,
+      "step": 1327000
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.469e-05,
+      "loss": 8.7079,
+      "step": 1327500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.4688e-05,
+      "loss": 8.6874,
+      "step": 1328000
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.4686e-05,
+      "loss": 8.7168,
+      "step": 1328500
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.4684e-05,
+      "loss": 8.683,
+      "step": 1329000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.4682000000000004e-05,
+      "loss": 8.676,
+      "step": 1329500
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.468e-05,
+      "loss": 8.6367,
+      "step": 1330000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.4678e-05,
+      "loss": 8.6868,
+      "step": 1330500
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.4676000000000004e-05,
+      "loss": 8.7022,
+      "step": 1331000
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.4674000000000006e-05,
+      "loss": 8.7035,
+      "step": 1331500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.4672e-05,
+      "loss": 8.6841,
+      "step": 1332000
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.467e-05,
+      "loss": 8.6874,
+      "step": 1332500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.466800000000001e-05,
+      "loss": 8.7151,
+      "step": 1333000
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.4666e-05,
+      "loss": 8.6965,
+      "step": 1333500
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.4664e-05,
+      "loss": 8.6873,
+      "step": 1334000
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.466200000000001e-05,
+      "loss": 8.6843,
+      "step": 1334500
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.466e-05,
+      "loss": 8.6629,
+      "step": 1335000
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.4658e-05,
+      "loss": 8.6947,
+      "step": 1335500
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.4656e-05,
+      "loss": 8.6782,
+      "step": 1336000
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.4654e-05,
+      "loss": 8.6953,
+      "step": 1336500
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.4652e-05,
+      "loss": 8.6806,
+      "step": 1337000
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.465e-05,
+      "loss": 8.6673,
+      "step": 1337500
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.4648000000000003e-05,
+      "loss": 8.6953,
+      "step": 1338000
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.4646000000000006e-05,
+      "loss": 8.6766,
+      "step": 1338500
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.4644e-05,
+      "loss": 8.6922,
+      "step": 1339000
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.4642e-05,
+      "loss": 8.6896,
+      "step": 1339500
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.4640000000000006e-05,
+      "loss": 8.6947,
+      "step": 1340000
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.4638e-05,
+      "loss": 8.6774,
+      "step": 1340500
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.4636e-05,
+      "loss": 8.6984,
+      "step": 1341000
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.463400000000001e-05,
+      "loss": 8.6892,
+      "step": 1341500
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.4632e-05,
+      "loss": 8.6877,
+      "step": 1342000
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.463e-05,
+      "loss": 8.6854,
+      "step": 1342500
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.4628e-05,
+      "loss": 8.7147,
+      "step": 1343000
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.4626e-05,
+      "loss": 8.6754,
+      "step": 1343500
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.4624000000000005e-05,
+      "loss": 8.6926,
+      "step": 1344000
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.4622e-05,
+      "loss": 8.6954,
+      "step": 1344500
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.462e-05,
+      "loss": 8.6694,
+      "step": 1345000
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.4618000000000005e-05,
+      "loss": 8.6997,
+      "step": 1345500
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.4616e-05,
+      "loss": 8.6677,
+      "step": 1346000
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.4614000000000003e-05,
+      "loss": 8.6774,
+      "step": 1346500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.4612000000000006e-05,
+      "loss": 8.6694,
+      "step": 1347000
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.461e-05,
+      "loss": 8.6681,
+      "step": 1347500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.4608e-05,
+      "loss": 8.6813,
+      "step": 1348000
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.4606000000000006e-05,
+      "loss": 8.6797,
+      "step": 1348500
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.4604e-05,
+      "loss": 8.6917,
+      "step": 1349000
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.4602000000000004e-05,
+      "loss": 8.6854,
+      "step": 1349500
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.46e-05,
+      "loss": 8.6866,
+      "step": 1350000
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.4598e-05,
+      "loss": 8.6949,
+      "step": 1350500
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.4596000000000005e-05,
+      "loss": 8.6847,
+      "step": 1351000
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.4594e-05,
+      "loss": 8.6616,
+      "step": 1351500
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.4592e-05,
+      "loss": 8.6827,
+      "step": 1352000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.4590000000000005e-05,
+      "loss": 8.6497,
+      "step": 1352500
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.4588e-05,
+      "loss": 8.685,
+      "step": 1353000
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.4586e-05,
+      "loss": 8.6852,
+      "step": 1353500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.4584000000000005e-05,
+      "loss": 8.669,
+      "step": 1354000
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.4582e-05,
+      "loss": 8.6911,
+      "step": 1354500
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.458e-05,
+      "loss": 8.6871,
+      "step": 1355000
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.4578000000000006e-05,
+      "loss": 8.6838,
+      "step": 1355500
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.4576e-05,
+      "loss": 8.6989,
+      "step": 1356000
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.4574000000000004e-05,
+      "loss": 8.6843,
+      "step": 1356500
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.4572e-05,
+      "loss": 8.6736,
+      "step": 1357000
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.457e-05,
+      "loss": 8.6755,
+      "step": 1357500
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.4568000000000004e-05,
+      "loss": 8.6829,
+      "step": 1358000
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.4566e-05,
+      "loss": 8.6925,
+      "step": 1358500
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.4564e-05,
+      "loss": 8.6795,
+      "step": 1359000
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.4562000000000005e-05,
+      "loss": 8.6818,
+      "step": 1359500
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.456e-05,
+      "loss": 8.6958,
+      "step": 1360000
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.4558e-05,
+      "loss": 8.6859,
+      "step": 1360500
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.4556000000000005e-05,
+      "loss": 8.6639,
+      "step": 1361000
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.4554e-05,
+      "loss": 8.6765,
+      "step": 1361500
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.4552e-05,
+      "loss": 8.6821,
+      "step": 1362000
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.4550000000000005e-05,
+      "loss": 8.6866,
+      "step": 1362500
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.4548e-05,
+      "loss": 8.6553,
+      "step": 1363000
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.4546000000000003e-05,
+      "loss": 8.6824,
+      "step": 1363500
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.4544e-05,
+      "loss": 8.6674,
+      "step": 1364000
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.4542e-05,
+      "loss": 8.6805,
+      "step": 1364500
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.4540000000000004e-05,
+      "loss": 8.6714,
+      "step": 1365000
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.4538e-05,
+      "loss": 8.7068,
+      "step": 1365500
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.4536e-05,
+      "loss": 8.6629,
+      "step": 1366000
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.4534000000000004e-05,
+      "loss": 8.6774,
+      "step": 1366500
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.4532e-05,
+      "loss": 8.6886,
+      "step": 1367000
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.453e-05,
+      "loss": 8.6808,
+      "step": 1367500
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.4528000000000005e-05,
+      "loss": 8.6658,
+      "step": 1368000
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.4526e-05,
+      "loss": 8.6964,
+      "step": 1368500
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.4524e-05,
+      "loss": 8.6908,
+      "step": 1369000
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.4522000000000005e-05,
+      "loss": 8.6808,
+      "step": 1369500
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.452e-05,
+      "loss": 8.6823,
+      "step": 1370000
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.4518e-05,
+      "loss": 8.6761,
+      "step": 1370500
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.4516000000000005e-05,
+      "loss": 8.685,
+      "step": 1371000
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.4514e-05,
+      "loss": 8.6699,
+      "step": 1371500
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.4512000000000003e-05,
+      "loss": 8.6561,
+      "step": 1372000
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.451e-05,
+      "loss": 8.6768,
+      "step": 1372500
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.4508e-05,
+      "loss": 8.6836,
+      "step": 1373000
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.4506000000000004e-05,
+      "loss": 8.6828,
+      "step": 1373500
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.4504e-05,
+      "loss": 8.6874,
+      "step": 1374000
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.4502e-05,
+      "loss": 8.6723,
+      "step": 1374500
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 8.6568,
+      "step": 1375000
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.4498e-05,
+      "loss": 8.6732,
+      "step": 1375500
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.4496e-05,
+      "loss": 8.6856,
+      "step": 1376000
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.4494000000000005e-05,
+      "loss": 8.6748,
+      "step": 1376500
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.4492e-05,
+      "loss": 8.6798,
+      "step": 1377000
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.449e-05,
+      "loss": 8.6773,
+      "step": 1377500
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.4488000000000005e-05,
+      "loss": 8.686,
+      "step": 1378000
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.4486e-05,
+      "loss": 8.6831,
+      "step": 1378500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.4484e-05,
+      "loss": 8.6704,
+      "step": 1379000
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.4482e-05,
+      "loss": 8.6722,
+      "step": 1379500
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.448e-05,
+      "loss": 8.6715,
+      "step": 1380000
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.4478000000000003e-05,
+      "loss": 8.6763,
+      "step": 1380500
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.4476e-05,
+      "loss": 8.6949,
+      "step": 1381000
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.4474e-05,
+      "loss": 8.6873,
+      "step": 1381500
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.4472000000000004e-05,
+      "loss": 8.6794,
+      "step": 1382000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.447e-05,
+      "loss": 8.6865,
+      "step": 1382500
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.4468e-05,
+      "loss": 8.6937,
+      "step": 1383000
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.4466000000000004e-05,
+      "loss": 8.6774,
+      "step": 1383500
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.4464e-05,
+      "loss": 8.671,
+      "step": 1384000
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.4462e-05,
+      "loss": 8.6739,
+      "step": 1384500
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.4460000000000005e-05,
+      "loss": 8.6876,
+      "step": 1385000
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.4458e-05,
+      "loss": 8.6725,
+      "step": 1385500
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.4456e-05,
+      "loss": 8.696,
+      "step": 1386000
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.4454e-05,
+      "loss": 8.6979,
+      "step": 1386500
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.445200000000001e-05,
+      "loss": 8.6636,
+      "step": 1387000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.445e-05,
+      "loss": 8.6806,
+      "step": 1387500
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.4448e-05,
+      "loss": 8.6741,
+      "step": 1388000
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.4446e-05,
+      "loss": 8.6797,
+      "step": 1388500
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.4444000000000003e-05,
+      "loss": 8.675,
+      "step": 1389000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.4442e-05,
+      "loss": 8.6788,
+      "step": 1389500
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.444e-05,
+      "loss": 8.6812,
+      "step": 1390000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.4438000000000004e-05,
+      "loss": 8.7191,
+      "step": 1390500
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.4436e-05,
+      "loss": 8.6671,
+      "step": 1391000
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.4434e-05,
+      "loss": 8.7002,
+      "step": 1391500
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.4432000000000004e-05,
+      "loss": 8.6935,
+      "step": 1392000
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.443e-05,
+      "loss": 8.6774,
+      "step": 1392500
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.4428e-05,
+      "loss": 8.6507,
+      "step": 1393000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.4426e-05,
+      "loss": 8.7016,
+      "step": 1393500
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.442400000000001e-05,
+      "loss": 8.7073,
+      "step": 1394000
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.4422e-05,
+      "loss": 8.6642,
+      "step": 1394500
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.442e-05,
+      "loss": 8.7026,
+      "step": 1395000
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.441800000000001e-05,
+      "loss": 8.6713,
+      "step": 1395500
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.4416e-05,
+      "loss": 8.6974,
+      "step": 1396000
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.4414e-05,
+      "loss": 8.6899,
+      "step": 1396500
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.4412e-05,
+      "loss": 8.6838,
+      "step": 1397000
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.4410000000000003e-05,
+      "loss": 8.68,
+      "step": 1397500
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.4408e-05,
+      "loss": 8.7208,
+      "step": 1398000
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.4406e-05,
+      "loss": 8.6746,
+      "step": 1398500
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.4404000000000004e-05,
+      "loss": 8.666,
+      "step": 1399000
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.4402000000000006e-05,
+      "loss": 8.6831,
+      "step": 1399500
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.44e-05,
+      "loss": 8.6936,
+      "step": 1400000
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.4398e-05,
+      "loss": 8.6704,
+      "step": 1400500
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.4396000000000007e-05,
+      "loss": 8.6812,
+      "step": 1401000
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.4394e-05,
+      "loss": 8.6934,
+      "step": 1401500
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.4392e-05,
+      "loss": 8.6618,
+      "step": 1402000
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.439000000000001e-05,
+      "loss": 8.689,
+      "step": 1402500
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.4388e-05,
+      "loss": 8.6973,
+      "step": 1403000
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.4386e-05,
+      "loss": 8.677,
+      "step": 1403500
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.4384e-05,
+      "loss": 8.6716,
+      "step": 1404000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.4382e-05,
+      "loss": 8.6798,
+      "step": 1404500
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.438e-05,
+      "loss": 8.6936,
+      "step": 1405000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.4378e-05,
+      "loss": 8.7097,
+      "step": 1405500
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.4376e-05,
+      "loss": 8.6674,
+      "step": 1406000
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.4374000000000006e-05,
+      "loss": 8.6778,
+      "step": 1406500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.4372e-05,
+      "loss": 8.673,
+      "step": 1407000
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.4370000000000004e-05,
+      "loss": 8.6796,
+      "step": 1407500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.4368000000000006e-05,
+      "loss": 8.6807,
+      "step": 1408000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.4366e-05,
+      "loss": 8.684,
+      "step": 1408500
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.4364e-05,
+      "loss": 8.6913,
+      "step": 1409000
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.4362000000000007e-05,
+      "loss": 8.717,
+      "step": 1409500
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.436e-05,
+      "loss": 8.6598,
+      "step": 1410000
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.4358e-05,
+      "loss": 8.681,
+      "step": 1410500
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.4356e-05,
+      "loss": 8.6636,
+      "step": 1411000
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.4354e-05,
+      "loss": 8.6663,
+      "step": 1411500
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.4352000000000005e-05,
+      "loss": 8.6876,
+      "step": 1412000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.435e-05,
+      "loss": 8.6941,
+      "step": 1412500
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.4348e-05,
+      "loss": 8.6862,
+      "step": 1413000
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.4346000000000005e-05,
+      "loss": 8.6784,
+      "step": 1413500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.4344e-05,
+      "loss": 8.6941,
+      "step": 1414000
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.4342e-05,
+      "loss": 8.6919,
+      "step": 1414500
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.4340000000000006e-05,
+      "loss": 8.6629,
+      "step": 1415000
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.4338e-05,
+      "loss": 8.6665,
+      "step": 1415500
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.4336e-05,
+      "loss": 8.6617,
+      "step": 1416000
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.4334000000000006e-05,
+      "loss": 8.688,
+      "step": 1416500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.4332e-05,
+      "loss": 8.6783,
+      "step": 1417000
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.4330000000000004e-05,
+      "loss": 8.6857,
+      "step": 1417500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.4328e-05,
+      "loss": 8.6568,
+      "step": 1418000
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.4326e-05,
+      "loss": 8.7021,
+      "step": 1418500
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.4324000000000005e-05,
+      "loss": 8.6954,
+      "step": 1419000
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.4322e-05,
+      "loss": 8.6757,
+      "step": 1419500
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.432e-05,
+      "loss": 8.6948,
+      "step": 1420000
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.4318000000000005e-05,
+      "loss": 8.6815,
+      "step": 1420500
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.4316e-05,
+      "loss": 8.6836,
+      "step": 1421000
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.4314e-05,
+      "loss": 8.6762,
+      "step": 1421500
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.4312000000000005e-05,
+      "loss": 8.6873,
+      "step": 1422000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.431e-05,
+      "loss": 8.6883,
+      "step": 1422500
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.4307999999999997e-05,
+      "loss": 8.6727,
+      "step": 1423000
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.4306000000000006e-05,
+      "loss": 8.6866,
+      "step": 1423500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.4304e-05,
+      "loss": 8.6688,
+      "step": 1424000
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.4302000000000004e-05,
+      "loss": 8.691,
+      "step": 1424500
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.43e-05,
+      "loss": 8.6905,
+      "step": 1425000
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.4298e-05,
+      "loss": 8.6575,
+      "step": 1425500
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.4296000000000004e-05,
+      "loss": 8.6747,
+      "step": 1426000
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.4294e-05,
+      "loss": 8.6863,
+      "step": 1426500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.4292e-05,
+      "loss": 8.6581,
+      "step": 1427000
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.4290000000000005e-05,
+      "loss": 8.6637,
+      "step": 1427500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.4288e-05,
+      "loss": 8.6897,
+      "step": 1428000
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.4286e-05,
+      "loss": 8.7029,
+      "step": 1428500
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.4284000000000005e-05,
+      "loss": 8.694,
+      "step": 1429000
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.4282e-05,
+      "loss": 8.6907,
+      "step": 1429500
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.428e-05,
+      "loss": 8.6931,
+      "step": 1430000
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.4278000000000005e-05,
+      "loss": 8.6977,
+      "step": 1430500
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.4276e-05,
+      "loss": 8.68,
+      "step": 1431000
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.4274e-05,
+      "loss": 8.6818,
+      "step": 1431500
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.4272000000000006e-05,
+      "loss": 8.6891,
+      "step": 1432000
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.427e-05,
+      "loss": 8.6873,
+      "step": 1432500
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.4268000000000004e-05,
+      "loss": 8.7087,
+      "step": 1433000
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.4266e-05,
+      "loss": 8.6785,
+      "step": 1433500
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.4264e-05,
+      "loss": 8.6784,
+      "step": 1434000
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.4262000000000004e-05,
+      "loss": 8.6885,
+      "step": 1434500
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.426e-05,
+      "loss": 8.7019,
+      "step": 1435000
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.4258e-05,
+      "loss": 8.698,
+      "step": 1435500
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.4256000000000005e-05,
+      "loss": 8.7079,
+      "step": 1436000
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.4254e-05,
+      "loss": 8.661,
+      "step": 1436500
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.4252e-05,
+      "loss": 8.6553,
+      "step": 1437000
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.4250000000000005e-05,
+      "loss": 8.6629,
+      "step": 1437500
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.4248e-05,
+      "loss": 8.6927,
+      "step": 1438000
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.4246e-05,
+      "loss": 8.6908,
+      "step": 1438500
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.4244000000000005e-05,
+      "loss": 8.6646,
+      "step": 1439000
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.4242e-05,
+      "loss": 8.6639,
+      "step": 1439500
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.424e-05,
+      "loss": 8.6975,
+      "step": 1440000
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.4238e-05,
+      "loss": 8.67,
+      "step": 1440500
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4236e-05,
+      "loss": 8.6734,
+      "step": 1441000
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4234000000000004e-05,
+      "loss": 8.6764,
+      "step": 1441500
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.4232e-05,
+      "loss": 8.676,
+      "step": 1442000
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.423e-05,
+      "loss": 8.646,
+      "step": 1442500
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.4228000000000004e-05,
+      "loss": 8.7027,
+      "step": 1443000
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.4226e-05,
+      "loss": 8.6741,
+      "step": 1443500
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.4224e-05,
+      "loss": 8.6565,
+      "step": 1444000
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.4222000000000005e-05,
+      "loss": 8.6644,
+      "step": 1444500
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.422e-05,
+      "loss": 8.6796,
+      "step": 1445000
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.4218e-05,
+      "loss": 8.6761,
+      "step": 1445500
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.4216000000000005e-05,
+      "loss": 8.6797,
+      "step": 1446000
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.4214e-05,
+      "loss": 8.6821,
+      "step": 1446500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.4212e-05,
+      "loss": 8.6804,
+      "step": 1447000
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.421e-05,
+      "loss": 8.6837,
+      "step": 1447500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.4208e-05,
+      "loss": 8.6764,
+      "step": 1448000
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.4206e-05,
+      "loss": 8.7004,
+      "step": 1448500
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.4204e-05,
+      "loss": 8.6578,
+      "step": 1449000
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.4202e-05,
+      "loss": 8.657,
+      "step": 1449500
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 8.6699,
+      "step": 1450000
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.4198e-05,
+      "loss": 8.6896,
+      "step": 1450500
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.4196e-05,
+      "loss": 8.6403,
+      "step": 1451000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.4194000000000004e-05,
+      "loss": 8.6853,
+      "step": 1451500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.4192e-05,
+      "loss": 8.6746,
+      "step": 1452000
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.419e-05,
+      "loss": 8.6948,
+      "step": 1452500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.4188000000000005e-05,
+      "loss": 8.6854,
+      "step": 1453000
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.4186e-05,
+      "loss": 8.6629,
+      "step": 1453500
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.4184e-05,
+      "loss": 8.6872,
+      "step": 1454000
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.4182e-05,
+      "loss": 8.673,
+      "step": 1454500
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.418000000000001e-05,
+      "loss": 8.6516,
+      "step": 1455000
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.4178e-05,
+      "loss": 8.669,
+      "step": 1455500
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.4176e-05,
+      "loss": 8.6677,
+      "step": 1456000
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.4174e-05,
+      "loss": 8.6699,
+      "step": 1456500
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.4172e-05,
+      "loss": 8.665,
+      "step": 1457000
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.417e-05,
+      "loss": 8.7142,
+      "step": 1457500
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.4168e-05,
+      "loss": 8.6866,
+      "step": 1458000
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4166000000000004e-05,
+      "loss": 8.6632,
+      "step": 1458500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4164e-05,
+      "loss": 8.6833,
+      "step": 1459000
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.4162e-05,
+      "loss": 8.6737,
+      "step": 1459500
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.4160000000000004e-05,
+      "loss": 8.7098,
+      "step": 1460000
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.4158e-05,
+      "loss": 8.6475,
+      "step": 1460500
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.4156e-05,
+      "loss": 8.6624,
+      "step": 1461000
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.4154e-05,
+      "loss": 8.6969,
+      "step": 1461500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.415200000000001e-05,
+      "loss": 8.6833,
+      "step": 1462000
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.415e-05,
+      "loss": 8.689,
+      "step": 1462500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.4148e-05,
+      "loss": 8.6595,
+      "step": 1463000
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.414600000000001e-05,
+      "loss": 8.6789,
+      "step": 1463500
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.4144e-05,
+      "loss": 8.6726,
+      "step": 1464000
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.4142e-05,
+      "loss": 8.6825,
+      "step": 1464500
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.414e-05,
+      "loss": 8.6773,
+      "step": 1465000
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.4138e-05,
+      "loss": 8.6724,
+      "step": 1465500
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.4136e-05,
+      "loss": 8.6929,
+      "step": 1466000
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.4134e-05,
+      "loss": 8.6711,
+      "step": 1466500
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.4132000000000004e-05,
+      "loss": 8.6716,
+      "step": 1467000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.4130000000000006e-05,
+      "loss": 8.71,
+      "step": 1467500
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.4128e-05,
+      "loss": 8.6614,
+      "step": 1468000
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.4126e-05,
+      "loss": 8.6917,
+      "step": 1468500
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.4124000000000006e-05,
+      "loss": 8.6831,
+      "step": 1469000
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.4122e-05,
+      "loss": 8.668,
+      "step": 1469500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.412e-05,
+      "loss": 8.667,
+      "step": 1470000
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.411800000000001e-05,
+      "loss": 8.6614,
+      "step": 1470500
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.4116e-05,
+      "loss": 8.6821,
+      "step": 1471000
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.4114e-05,
+      "loss": 8.6653,
+      "step": 1471500
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.4112e-05,
+      "loss": 8.6863,
+      "step": 1472000
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.411e-05,
+      "loss": 8.6759,
+      "step": 1472500
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.4108000000000005e-05,
+      "loss": 8.6805,
+      "step": 1473000
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.4106e-05,
+      "loss": 8.7015,
+      "step": 1473500
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.4104e-05,
+      "loss": 8.6562,
+      "step": 1474000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.4102000000000006e-05,
+      "loss": 8.6747,
+      "step": 1474500
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.41e-05,
+      "loss": 8.6789,
+      "step": 1475000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.4098000000000004e-05,
+      "loss": 8.6524,
+      "step": 1475500
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.4096000000000006e-05,
+      "loss": 8.6618,
+      "step": 1476000
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.4094e-05,
+      "loss": 8.6817,
+      "step": 1476500
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.4092e-05,
+      "loss": 8.6948,
+      "step": 1477000
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.4090000000000006e-05,
+      "loss": 8.6682,
+      "step": 1477500
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.4088e-05,
+      "loss": 8.6452,
+      "step": 1478000
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.4086e-05,
+      "loss": 8.6851,
+      "step": 1478500
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.4084e-05,
+      "loss": 8.6777,
+      "step": 1479000
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.4082e-05,
+      "loss": 8.6732,
+      "step": 1479500
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.4080000000000005e-05,
+      "loss": 8.6927,
+      "step": 1480000
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.4078e-05,
+      "loss": 8.6913,
+      "step": 1480500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.4076e-05,
+      "loss": 8.6612,
+      "step": 1481000
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.4074000000000005e-05,
+      "loss": 8.7139,
+      "step": 1481500
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.4072e-05,
+      "loss": 8.6856,
+      "step": 1482000
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.407e-05,
+      "loss": 8.6723,
+      "step": 1482500
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.4068000000000006e-05,
+      "loss": 8.6858,
+      "step": 1483000
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.4066e-05,
+      "loss": 8.6761,
+      "step": 1483500
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.4064e-05,
+      "loss": 8.6919,
+      "step": 1484000
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.4062000000000006e-05,
+      "loss": 8.6761,
+      "step": 1484500
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.406e-05,
+      "loss": 8.689,
+      "step": 1485000
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.4058000000000004e-05,
+      "loss": 8.6675,
+      "step": 1485500
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.4056e-05,
+      "loss": 8.6809,
+      "step": 1486000
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.4054e-05,
+      "loss": 8.7032,
+      "step": 1486500
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.4052000000000004e-05,
+      "loss": 8.7098,
+      "step": 1487000
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.405e-05,
+      "loss": 8.6612,
+      "step": 1487500
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.4048e-05,
+      "loss": 8.6746,
+      "step": 1488000
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.4046000000000005e-05,
+      "loss": 8.6695,
+      "step": 1488500
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.4044e-05,
+      "loss": 8.6714,
+      "step": 1489000
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.4042e-05,
+      "loss": 8.6852,
+      "step": 1489500
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.4040000000000005e-05,
+      "loss": 8.6772,
+      "step": 1490000
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.4038e-05,
+      "loss": 8.6697,
+      "step": 1490500
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.4035999999999996e-05,
+      "loss": 8.6456,
+      "step": 1491000
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.4034000000000006e-05,
+      "loss": 8.6959,
+      "step": 1491500
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.4032e-05,
+      "loss": 8.6616,
+      "step": 1492000
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.4030000000000004e-05,
+      "loss": 8.6814,
+      "step": 1492500
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.4028e-05,
+      "loss": 8.6567,
+      "step": 1493000
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.4026e-05,
+      "loss": 8.6518,
+      "step": 1493500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.4024000000000004e-05,
+      "loss": 8.6799,
+      "step": 1494000
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.4022e-05,
+      "loss": 8.6895,
+      "step": 1494500
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.402e-05,
+      "loss": 8.677,
+      "step": 1495000
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.4018000000000004e-05,
+      "loss": 8.6905,
+      "step": 1495500
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.4016e-05,
+      "loss": 8.6613,
+      "step": 1496000
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.4014e-05,
+      "loss": 8.6763,
+      "step": 1496500
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.4012000000000005e-05,
+      "loss": 8.6546,
+      "step": 1497000
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.401e-05,
+      "loss": 8.6658,
+      "step": 1497500
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.4008e-05,
+      "loss": 8.6746,
+      "step": 1498000
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.4006000000000005e-05,
+      "loss": 8.666,
+      "step": 1498500
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.4004e-05,
+      "loss": 8.6715,
+      "step": 1499000
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.4002e-05,
+      "loss": 8.6713,
+      "step": 1499500
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 8.6739,
+      "step": 1500000
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.3998e-05,
+      "loss": 8.6672,
+      "step": 1500500
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.3996000000000004e-05,
+      "loss": 8.6654,
+      "step": 1501000
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.3994e-05,
+      "loss": 8.71,
+      "step": 1501500
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.3992e-05,
+      "loss": 8.6906,
+      "step": 1502000
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.3990000000000004e-05,
+      "loss": 8.7011,
+      "step": 1502500
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.3988e-05,
+      "loss": 8.6885,
+      "step": 1503000
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.3986e-05,
+      "loss": 8.6526,
+      "step": 1503500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.3984000000000004e-05,
+      "loss": 8.6879,
+      "step": 1504000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.3982e-05,
+      "loss": 8.6624,
+      "step": 1504500
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.398e-05,
+      "loss": 8.7053,
+      "step": 1505000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.3978000000000005e-05,
+      "loss": 8.6805,
+      "step": 1505500
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.3976e-05,
+      "loss": 8.6868,
+      "step": 1506000
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.3974e-05,
+      "loss": 8.7021,
+      "step": 1506500
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.3972000000000005e-05,
+      "loss": 8.6909,
+      "step": 1507000
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.397e-05,
+      "loss": 8.6705,
+      "step": 1507500
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.3968e-05,
+      "loss": 8.6963,
+      "step": 1508000
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.3966e-05,
+      "loss": 8.6846,
+      "step": 1508500
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.3964e-05,
+      "loss": 8.6831,
+      "step": 1509000
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.3962000000000004e-05,
+      "loss": 8.6882,
+      "step": 1509500
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.396e-05,
+      "loss": 8.6903,
+      "step": 1510000
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.3958e-05,
+      "loss": 8.6875,
+      "step": 1510500
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.3956000000000004e-05,
+      "loss": 8.6809,
+      "step": 1511000
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.3954e-05,
+      "loss": 8.6633,
+      "step": 1511500
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.3952e-05,
+      "loss": 8.6722,
+      "step": 1512000
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.3950000000000004e-05,
+      "loss": 8.6903,
+      "step": 1512500
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.3948e-05,
+      "loss": 8.6771,
+      "step": 1513000
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.3946e-05,
+      "loss": 8.6747,
+      "step": 1513500
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.3944000000000005e-05,
+      "loss": 8.691,
+      "step": 1514000
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.3942e-05,
+      "loss": 8.6689,
+      "step": 1514500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.394e-05,
+      "loss": 8.6619,
+      "step": 1515000
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.3938e-05,
+      "loss": 8.6476,
+      "step": 1515500
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.3936e-05,
+      "loss": 8.6996,
+      "step": 1516000
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.3934e-05,
+      "loss": 8.6747,
+      "step": 1516500
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.3932e-05,
+      "loss": 8.6774,
+      "step": 1517000
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.393e-05,
+      "loss": 8.6869,
+      "step": 1517500
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.3928000000000004e-05,
+      "loss": 8.6653,
+      "step": 1518000
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.3926e-05,
+      "loss": 8.6809,
+      "step": 1518500
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.3924e-05,
+      "loss": 8.6973,
+      "step": 1519000
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.3922000000000004e-05,
+      "loss": 8.6989,
+      "step": 1519500
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.392e-05,
+      "loss": 8.6899,
+      "step": 1520000
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.3918e-05,
+      "loss": 8.6943,
+      "step": 1520500
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.3916000000000004e-05,
+      "loss": 8.6592,
+      "step": 1521000
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.3914e-05,
+      "loss": 8.6819,
+      "step": 1521500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.3912e-05,
+      "loss": 8.6925,
+      "step": 1522000
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.391e-05,
+      "loss": 8.6697,
+      "step": 1522500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.390800000000001e-05,
+      "loss": 8.6586,
+      "step": 1523000
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.3906e-05,
+      "loss": 8.6722,
+      "step": 1523500
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.3904e-05,
+      "loss": 8.6586,
+      "step": 1524000
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.390200000000001e-05,
+      "loss": 8.6938,
+      "step": 1524500
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.39e-05,
+      "loss": 8.6725,
+      "step": 1525000
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.3898e-05,
+      "loss": 8.6467,
+      "step": 1525500
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.3896e-05,
+      "loss": 8.7079,
+      "step": 1526000
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.3894000000000004e-05,
+      "loss": 8.6854,
+      "step": 1526500
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.3892e-05,
+      "loss": 8.7007,
+      "step": 1527000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.389e-05,
+      "loss": 8.6889,
+      "step": 1527500
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.3888000000000004e-05,
+      "loss": 8.7038,
+      "step": 1528000
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.3886e-05,
+      "loss": 8.6536,
+      "step": 1528500
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.3884e-05,
+      "loss": 8.6831,
+      "step": 1529000
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.3882e-05,
+      "loss": 8.6792,
+      "step": 1529500
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.388000000000001e-05,
+      "loss": 8.6932,
+      "step": 1530000
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.3878e-05,
+      "loss": 8.6891,
+      "step": 1530500
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.3876e-05,
+      "loss": 8.6866,
+      "step": 1531000
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.387400000000001e-05,
+      "loss": 8.672,
+      "step": 1531500
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.3872e-05,
+      "loss": 8.672,
+      "step": 1532000
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.387e-05,
+      "loss": 8.7002,
+      "step": 1532500
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.3868e-05,
+      "loss": 8.6728,
+      "step": 1533000
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.3866e-05,
+      "loss": 8.6812,
+      "step": 1533500
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.3864e-05,
+      "loss": 8.644,
+      "step": 1534000
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.3862e-05,
+      "loss": 8.6899,
+      "step": 1534500
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.3860000000000004e-05,
+      "loss": 8.688,
+      "step": 1535000
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.3858000000000006e-05,
+      "loss": 8.6622,
+      "step": 1535500
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.3856e-05,
+      "loss": 8.6991,
+      "step": 1536000
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.3854000000000004e-05,
+      "loss": 8.699,
+      "step": 1536500
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.3852000000000006e-05,
+      "loss": 8.6681,
+      "step": 1537000
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.385e-05,
+      "loss": 8.6747,
+      "step": 1537500
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.3848e-05,
+      "loss": 8.6442,
+      "step": 1538000
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.384600000000001e-05,
+      "loss": 8.6469,
+      "step": 1538500
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.3844e-05,
+      "loss": 8.6613,
+      "step": 1539000
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.3842e-05,
+      "loss": 8.6914,
+      "step": 1539500
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.384e-05,
+      "loss": 8.6721,
+      "step": 1540000
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.3838e-05,
+      "loss": 8.6913,
+      "step": 1540500
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.3836000000000005e-05,
+      "loss": 8.6717,
+      "step": 1541000
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.3834e-05,
+      "loss": 8.6817,
+      "step": 1541500
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.3832e-05,
+      "loss": 8.6997,
+      "step": 1542000
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.3830000000000006e-05,
+      "loss": 8.7035,
+      "step": 1542500
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.3828e-05,
+      "loss": 8.6876,
+      "step": 1543000
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.3826000000000004e-05,
+      "loss": 8.6981,
+      "step": 1543500
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.3824000000000006e-05,
+      "loss": 8.7151,
+      "step": 1544000
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.3822e-05,
+      "loss": 8.7211,
+      "step": 1544500
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.382e-05,
+      "loss": 8.699,
+      "step": 1545000
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.3818000000000006e-05,
+      "loss": 8.6745,
+      "step": 1545500
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.3816e-05,
+      "loss": 8.6977,
+      "step": 1546000
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.3814e-05,
+      "loss": 8.6718,
+      "step": 1546500
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.3812e-05,
+      "loss": 8.7128,
+      "step": 1547000
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.381e-05,
+      "loss": 8.6882,
+      "step": 1547500
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.3808000000000005e-05,
+      "loss": 8.6773,
+      "step": 1548000
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.3806e-05,
+      "loss": 8.6646,
+      "step": 1548500
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.3804e-05,
+      "loss": 8.6737,
+      "step": 1549000
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.3802000000000005e-05,
+      "loss": 8.65,
+      "step": 1549500
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.38e-05,
+      "loss": 8.6774,
+      "step": 1550000
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.3798e-05,
+      "loss": 8.6685,
+      "step": 1550500
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.3796000000000006e-05,
+      "loss": 8.6778,
+      "step": 1551000
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.3794e-05,
+      "loss": 8.6895,
+      "step": 1551500
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3792e-05,
+      "loss": 8.6687,
+      "step": 1552000
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3790000000000006e-05,
+      "loss": 8.6951,
+      "step": 1552500
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3788e-05,
+      "loss": 8.6959,
+      "step": 1553000
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3786000000000004e-05,
+      "loss": 8.67,
+      "step": 1553500
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3784e-05,
+      "loss": 8.6531,
+      "step": 1554000
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3782e-05,
+      "loss": 8.7009,
+      "step": 1554500
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3780000000000004e-05,
+      "loss": 8.6707,
+      "step": 1555000
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3778e-05,
+      "loss": 8.676,
+      "step": 1555500
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3776e-05,
+      "loss": 8.6667,
+      "step": 1556000
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3774000000000005e-05,
+      "loss": 8.7089,
+      "step": 1556500
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3772e-05,
+      "loss": 8.6605,
+      "step": 1557000
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.377e-05,
+      "loss": 8.7017,
+      "step": 1557500
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3768000000000005e-05,
+      "loss": 8.6731,
+      "step": 1558000
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3766e-05,
+      "loss": 8.6783,
+      "step": 1558500
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3763999999999996e-05,
+      "loss": 8.715,
+      "step": 1559000
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3762000000000006e-05,
+      "loss": 8.6533,
+      "step": 1559500
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.376e-05,
+      "loss": 8.6847,
+      "step": 1560000
+    }
+  ],
+  "max_steps": 12500000,
+  "num_train_epochs": 100,
+  "total_flos": 4.871898683434107e+17,
+  "trial_name": null,
+  "trial_params": null
+}