diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.582010582010582,
+  "global_step": 500000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 10.2046,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 9.366,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 8.9116,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 8.5519,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 8.1293,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 7.6255,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 7.1213,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 6.7092,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9e-06,
+      "loss": 6.4643,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 6.311,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 6.2174,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2e-05,
+      "loss": 6.1559,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 6.1035,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 6.0585,
+      "step": 1400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5e-05,
+      "loss": 6.0161,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 5.9838,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 5.9601,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8e-05,
+      "loss": 5.9298,
+      "step": 1800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9e-05,
+      "loss": 5.9038,
+      "step": 1900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 5.8812,
+      "step": 2000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1e-05,
+      "loss": 5.8652,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 5.8491,
+      "step": 2200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 5.8159,
+      "step": 2300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4e-05,
+      "loss": 5.8046,
+      "step": 2400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5e-05,
+      "loss": 5.7949,
+      "step": 2500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 5.7781,
+      "step": 2600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 5.7541,
+      "step": 2700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 5.7444,
+      "step": 2800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9e-05,
+      "loss": 5.7199,
+      "step": 2900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3e-05,
+      "loss": 5.7058,
+      "step": 3000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1e-05,
+      "loss": 5.6956,
+      "step": 3100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 5.6666,
+      "step": 3200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3e-05,
+      "loss": 5.6611,
+      "step": 3300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 5.6509,
+      "step": 3400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5e-05,
+      "loss": 5.6268,
+      "step": 3500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.6e-05,
+      "loss": 5.6069,
+      "step": 3600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.7e-05,
+      "loss": 5.595,
+      "step": 3700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8e-05,
+      "loss": 5.5835,
+      "step": 3800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 5.5517,
+      "step": 3900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4e-05,
+      "loss": 5.538,
+      "step": 4000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.1e-05,
+      "loss": 5.508,
+      "step": 4100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.2e-05,
+      "loss": 5.4989,
+      "step": 4200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.3e-05,
+      "loss": 5.4678,
+      "step": 4300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 5.4355,
+      "step": 4400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5e-05,
+      "loss": 5.3605,
+      "step": 4500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 5.2785,
+      "step": 4600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7e-05,
+      "loss": 5.2045,
+      "step": 4700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8e-05,
+      "loss": 5.1103,
+      "step": 4800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9e-05,
+      "loss": 5.0325,
+      "step": 4900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5e-05,
+      "loss": 4.9641,
+      "step": 5000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 4.8674,
+      "step": 5100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 4.7503,
+      "step": 5200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 4.6245,
+      "step": 5300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 4.5214,
+      "step": 5400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 4.4142,
+      "step": 5500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 4.3115,
+      "step": 5600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 4.2359,
+      "step": 5700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.8e-05,
+      "loss": 4.1521,
+      "step": 5800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9e-05,
+      "loss": 4.0841,
+      "step": 5900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6e-05,
+      "loss": 4.0192,
+      "step": 6000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.1e-05,
+      "loss": 3.9532,
+      "step": 6100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.2e-05,
+      "loss": 3.8956,
+      "step": 6200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.3e-05,
+      "loss": 3.8454,
+      "step": 6300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 3.7992,
+      "step": 6400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 3.753,
+      "step": 6500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.6e-05,
+      "loss": 3.7035,
+      "step": 6600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.7e-05,
+      "loss": 3.6755,
+      "step": 6700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 3.6067,
+      "step": 6800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.9e-05,
+      "loss": 3.5867,
+      "step": 6900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7e-05,
+      "loss": 3.5409,
+      "step": 7000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.1e-05,
+      "loss": 3.4989,
+      "step": 7100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.2e-05,
+      "loss": 3.463,
+      "step": 7200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.3e-05,
+      "loss": 3.4175,
+      "step": 7300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.4e-05,
+      "loss": 3.3896,
+      "step": 7400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 3.3591,
+      "step": 7500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.6e-05,
+      "loss": 3.3243,
+      "step": 7600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.7e-05,
+      "loss": 3.284,
+      "step": 7700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 3.2571,
+      "step": 7800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 3.2228,
+      "step": 7900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8e-05,
+      "loss": 3.1848,
+      "step": 8000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.1e-05,
+      "loss": 3.1748,
+      "step": 8100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.2e-05,
+      "loss": 3.1406,
+      "step": 8200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.3e-05,
+      "loss": 3.123,
+      "step": 8300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.4e-05,
+      "loss": 3.1,
+      "step": 8400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.5e-05,
+      "loss": 3.0738,
+      "step": 8500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.6e-05,
+      "loss": 3.05,
+      "step": 8600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.7e-05,
+      "loss": 3.0246,
+      "step": 8700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 2.9948,
+      "step": 8800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 2.9928,
+      "step": 8900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9e-05,
+      "loss": 2.9668,
+      "step": 9000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.1e-05,
+      "loss": 2.9411,
+      "step": 9100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 2.9251,
+      "step": 9200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 2.9019,
+      "step": 9300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4e-05,
+      "loss": 2.8918,
+      "step": 9400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.5e-05,
+      "loss": 2.8718,
+      "step": 9500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.6e-05,
+      "loss": 2.8649,
+      "step": 9600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.7e-05,
+      "loss": 2.8558,
+      "step": 9700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.8e-05,
+      "loss": 2.8363,
+      "step": 9800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 2.8145,
+      "step": 9900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001,
+      "loss": 2.8047,
+      "step": 10000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.99795918367347e-05,
+      "loss": 2.785,
+      "step": 10100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.995918367346939e-05,
+      "loss": 2.7711,
+      "step": 10200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.993877551020409e-05,
+      "loss": 2.7626,
+      "step": 10300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.991836734693878e-05,
+      "loss": 2.7413,
+      "step": 10400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.989795918367347e-05,
+      "loss": 2.7286,
+      "step": 10500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.987755102040817e-05,
+      "loss": 2.7203,
+      "step": 10600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.985714285714287e-05,
+      "loss": 2.6978,
+      "step": 10700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.983673469387755e-05,
+      "loss": 2.6935,
+      "step": 10800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.981632653061225e-05,
+      "loss": 2.6829,
+      "step": 10900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.979591836734695e-05,
+      "loss": 2.6686,
+      "step": 11000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.977551020408163e-05,
+      "loss": 2.6699,
+      "step": 11100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.975510204081633e-05,
+      "loss": 2.6482,
+      "step": 11200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.973469387755102e-05,
+      "loss": 2.637,
+      "step": 11300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.971428571428571e-05,
+      "loss": 2.6153,
+      "step": 11400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.969387755102041e-05,
+      "loss": 2.612,
+      "step": 11500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.96734693877551e-05,
+      "loss": 2.6096,
+      "step": 11600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.96530612244898e-05,
+      "loss": 2.5922,
+      "step": 11700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.96326530612245e-05,
+      "loss": 2.5857,
+      "step": 11800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.961224489795918e-05,
+      "loss": 2.5749,
+      "step": 11900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.959183673469388e-05,
+      "loss": 2.5663,
+      "step": 12000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.957142857142858e-05,
+      "loss": 2.5588,
+      "step": 12100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.955102040816326e-05,
+      "loss": 2.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.953061224489797e-05,
+      "loss": 2.5398,
+      "step": 12300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.951020408163266e-05,
+      "loss": 2.5352,
+      "step": 12400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.948979591836736e-05,
+      "loss": 2.5278,
+      "step": 12500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.946938775510205e-05,
+      "loss": 2.5242,
+      "step": 12600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.944897959183674e-05,
+      "loss": 2.5046,
+      "step": 12700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.942857142857144e-05,
+      "loss": 2.4943,
+      "step": 12800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.940816326530614e-05,
+      "loss": 2.4989,
+      "step": 12900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.938775510204082e-05,
+      "loss": 2.4866,
+      "step": 13000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.936734693877552e-05,
+      "loss": 2.4764,
+      "step": 13100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.934693877551022e-05,
+      "loss": 2.4728,
+      "step": 13200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.93265306122449e-05,
+      "loss": 2.4532,
+      "step": 13300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.93061224489796e-05,
+      "loss": 2.4532,
+      "step": 13400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.92857142857143e-05,
+      "loss": 2.464,
+      "step": 13500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.926530612244898e-05,
+      "loss": 2.4405,
+      "step": 13600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.924489795918368e-05,
+      "loss": 2.4397,
+      "step": 13700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.922448979591838e-05,
+      "loss": 2.4286,
+      "step": 13800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.920408163265306e-05,
+      "loss": 2.4184,
+      "step": 13900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.91838775510204e-05,
+      "loss": 2.4207,
+      "step": 14000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.91634693877551e-05,
+      "loss": 2.4132,
+      "step": 14100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.91430612244898e-05,
+      "loss": 2.4126,
+      "step": 14200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.912265306122449e-05,
+      "loss": 2.4032,
+      "step": 14300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.910224489795919e-05,
+      "loss": 2.3902,
+      "step": 14400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.908183673469388e-05,
+      "loss": 2.3934,
+      "step": 14500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.906142857142857e-05,
+      "loss": 2.3873,
+      "step": 14600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.904102040816327e-05,
+      "loss": 2.3686,
+      "step": 14700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.902061224489797e-05,
+      "loss": 2.3741,
+      "step": 14800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.900020408163265e-05,
+      "loss": 2.3662,
+      "step": 14900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.897979591836735e-05,
+      "loss": 2.3625,
+      "step": 15000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.895938775510205e-05,
+      "loss": 2.3601,
+      "step": 15100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.893897959183675e-05,
+      "loss": 2.36,
+      "step": 15200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.891857142857144e-05,
+      "loss": 2.3391,
+      "step": 15300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.889816326530613e-05,
+      "loss": 2.3479,
+      "step": 15400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.887775510204083e-05,
+      "loss": 2.3399,
+      "step": 15500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.885734693877553e-05,
+      "loss": 2.3281,
+      "step": 15600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.883693877551021e-05,
+      "loss": 2.3202,
+      "step": 15700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.881653061224491e-05,
+      "loss": 2.3204,
+      "step": 15800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.87961224489796e-05,
+      "loss": 2.3178,
+      "step": 15900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.877571428571429e-05,
+      "loss": 2.3155,
+      "step": 16000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.875530612244899e-05,
+      "loss": 2.3132,
+      "step": 16100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.873489795918367e-05,
+      "loss": 2.3096,
+      "step": 16200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.871448979591837e-05,
+      "loss": 2.298,
+      "step": 16300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.869408163265307e-05,
+      "loss": 2.3049,
+      "step": 16400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.867367346938776e-05,
+      "loss": 2.2992,
+      "step": 16500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.865326530612245e-05,
+      "loss": 2.2926,
+      "step": 16600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.863285714285715e-05,
+      "loss": 2.2843,
+      "step": 16700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.861244897959184e-05,
+      "loss": 2.2873,
+      "step": 16800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.859204081632654e-05,
+      "loss": 2.2797,
+      "step": 16900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.857163265306123e-05,
+      "loss": 2.2736,
+      "step": 17000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.855122448979592e-05,
+      "loss": 2.2744,
+      "step": 17100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.853081632653062e-05,
+      "loss": 2.2719,
+      "step": 17200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.851040816326532e-05,
+      "loss": 2.2618,
+      "step": 17300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.849e-05,
+      "loss": 2.2613,
+      "step": 17400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.84695918367347e-05,
+      "loss": 2.2648,
+      "step": 17500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.844938775510204e-05,
+      "loss": 2.2479,
+      "step": 17600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.842897959183674e-05,
+      "loss": 2.2456,
+      "step": 17700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.840857142857142e-05,
+      "loss": 2.2376,
+      "step": 17800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.838816326530613e-05,
+      "loss": 2.2329,
+      "step": 17900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.836775510204082e-05,
+      "loss": 2.2304,
+      "step": 18000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.834734693877552e-05,
+      "loss": 2.2331,
+      "step": 18100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.832693877551022e-05,
+      "loss": 2.2272,
+      "step": 18200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.83065306122449e-05,
+      "loss": 2.2291,
+      "step": 18300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.82861224489796e-05,
+      "loss": 2.2169,
+      "step": 18400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.82657142857143e-05,
+      "loss": 2.2227,
+      "step": 18500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.824530612244898e-05,
+      "loss": 2.2115,
+      "step": 18600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.822489795918368e-05,
+      "loss": 2.2094,
+      "step": 18700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.820448979591838e-05,
+      "loss": 2.2061,
+      "step": 18800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.818408163265306e-05,
+      "loss": 2.1997,
+      "step": 18900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.816367346938776e-05,
+      "loss": 2.1995,
+      "step": 19000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.814326530612246e-05,
+      "loss": 2.1977,
+      "step": 19100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.812285714285715e-05,
+      "loss": 2.1991,
+      "step": 19200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.810244897959184e-05,
+      "loss": 2.1947,
+      "step": 19300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.808204081632654e-05,
+      "loss": 2.1963,
+      "step": 19400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.806163265306123e-05,
+      "loss": 2.1837,
+      "step": 19500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.804122448979593e-05,
+      "loss": 2.1804,
+      "step": 19600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.802081632653062e-05,
+      "loss": 2.1775,
+      "step": 19700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.800040816326531e-05,
+      "loss": 2.1752,
+      "step": 19800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.798000000000001e-05,
+      "loss": 2.1774,
+      "step": 19900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.795959183673469e-05,
+      "loss": 2.1665,
+      "step": 20000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.793918367346939e-05,
+      "loss": 2.1627,
+      "step": 20100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.791877551020409e-05,
+      "loss": 2.1643,
+      "step": 20200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.789857142857143e-05,
+      "loss": 2.1728,
+      "step": 20300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.787816326530613e-05,
+      "loss": 2.1629,
+      "step": 20400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.785775510204081e-05,
+      "loss": 2.1569,
+      "step": 20500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.783734693877552e-05,
+      "loss": 2.1567,
+      "step": 20600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.781693877551021e-05,
+      "loss": 2.1566,
+      "step": 20700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.779673469387755e-05,
+      "loss": 2.1454,
+      "step": 20800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.777632653061225e-05,
+      "loss": 2.157,
+      "step": 20900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.775591836734695e-05,
+      "loss": 2.1419,
+      "step": 21000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.773551020408163e-05,
+      "loss": 2.1389,
+      "step": 21100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.771510204081633e-05,
+      "loss": 2.1391,
+      "step": 21200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.769469387755103e-05,
+      "loss": 2.1343,
+      "step": 21300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.767428571428571e-05,
+      "loss": 2.1381,
+      "step": 21400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.765387755102041e-05,
+      "loss": 2.1352,
+      "step": 21500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.763346938775511e-05,
+      "loss": 2.1352,
+      "step": 21600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.76130612244898e-05,
+      "loss": 2.1262,
+      "step": 21700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.759265306122449e-05,
+      "loss": 2.133,
+      "step": 21800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.757224489795919e-05,
+      "loss": 2.1228,
+      "step": 21900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.755183673469388e-05,
+      "loss": 2.1245,
+      "step": 22000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.753142857142857e-05,
+      "loss": 2.1176,
+      "step": 22100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.751102040816327e-05,
+      "loss": 2.1164,
+      "step": 22200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.749061224489796e-05,
+      "loss": 2.1119,
+      "step": 22300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.747020408163266e-05,
+      "loss": 2.1172,
+      "step": 22400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.744979591836735e-05,
+      "loss": 2.1059,
+      "step": 22500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.742938775510204e-05,
+      "loss": 2.1087,
+      "step": 22600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.740897959183674e-05,
+      "loss": 2.1014,
+      "step": 22700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.738857142857142e-05,
+      "loss": 2.11,
+      "step": 22800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.736816326530612e-05,
+      "loss": 2.0921,
+      "step": 22900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.734775510204082e-05,
+      "loss": 2.0993,
+      "step": 23000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.73273469387755e-05,
+      "loss": 2.1137,
+      "step": 23100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.73069387755102e-05,
+      "loss": 2.0831,
+      "step": 23200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.72865306122449e-05,
+      "loss": 2.0913,
+      "step": 23300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.72661224489796e-05,
+      "loss": 2.0882,
+      "step": 23400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.72457142857143e-05,
+      "loss": 2.0866,
+      "step": 23500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.7225306122449e-05,
+      "loss": 2.0787,
+      "step": 23600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.720489795918368e-05,
+      "loss": 2.0838,
+      "step": 23700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.718448979591838e-05,
+      "loss": 2.0832,
+      "step": 23800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.716408163265306e-05,
+      "loss": 2.0752,
+      "step": 23900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.714387755102042e-05,
+      "loss": 2.0825,
+      "step": 24000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.71234693877551e-05,
+      "loss": 2.0727,
+      "step": 24100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.71030612244898e-05,
+      "loss": 2.0677,
+      "step": 24200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.70826530612245e-05,
+      "loss": 2.0616,
+      "step": 24300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706224489795918e-05,
+      "loss": 2.0599,
+      "step": 24400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704183673469388e-05,
+      "loss": 2.0761,
+      "step": 24500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702142857142857e-05,
+      "loss": 2.0647,
+      "step": 24600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700102040816327e-05,
+      "loss": 2.064,
+      "step": 24700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698061224489796e-05,
+      "loss": 2.0585,
+      "step": 24800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.696020408163265e-05,
+      "loss": 2.058,
+      "step": 24900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.693979591836735e-05,
+      "loss": 2.0575,
+      "step": 25000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.691938775510205e-05,
+      "loss": 2.0467,
+      "step": 25100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.689897959183673e-05,
+      "loss": 2.0557,
+      "step": 25200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.687857142857143e-05,
+      "loss": 2.0391,
+      "step": 25300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.685816326530613e-05,
+      "loss": 2.0377,
+      "step": 25400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.683775510204081e-05,
+      "loss": 2.0564,
+      "step": 25500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.681734693877551e-05,
+      "loss": 2.0432,
+      "step": 25600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.679693877551021e-05,
+      "loss": 2.0504,
+      "step": 25700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.67765306122449e-05,
+      "loss": 2.0465,
+      "step": 25800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.675612244897959e-05,
+      "loss": 2.0407,
+      "step": 25900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.673571428571429e-05,
+      "loss": 2.0432,
+      "step": 26000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.671530612244898e-05,
+      "loss": 2.0386,
+      "step": 26100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.669489795918369e-05,
+      "loss": 2.023,
+      "step": 26200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.667469387755103e-05,
+      "loss": 2.0236,
+      "step": 26300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.665428571428571e-05,
+      "loss": 2.023,
+      "step": 26400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.663387755102041e-05,
+      "loss": 2.032,
+      "step": 26500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.661367346938775e-05,
+      "loss": 2.0356,
+      "step": 26600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.659326530612245e-05,
+      "loss": 2.0287,
+      "step": 26700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.657285714285715e-05,
+      "loss": 2.0249,
+      "step": 26800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.655244897959183e-05,
+      "loss": 2.0241,
+      "step": 26900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.653204081632653e-05,
+      "loss": 2.0162,
+      "step": 27000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.651163265306123e-05,
+      "loss": 2.0054,
+      "step": 27100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.649122448979593e-05,
+      "loss": 2.0278,
+      "step": 27200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.647081632653063e-05,
+      "loss": 2.0101,
+      "step": 27300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.645040816326531e-05,
+      "loss": 2.0098,
+      "step": 27400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.643000000000001e-05,
+      "loss": 2.0176,
+      "step": 27500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.640959183673471e-05,
+      "loss": 2.0118,
+      "step": 27600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.63891836734694e-05,
+      "loss": 1.999,
+      "step": 27700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.636877551020409e-05,
+      "loss": 2.0061,
+      "step": 27800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.634836734693879e-05,
+      "loss": 1.9981,
+      "step": 27900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.632795918367347e-05,
+      "loss": 2.0041,
+      "step": 28000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.630755102040817e-05,
+      "loss": 1.9992,
+      "step": 28100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.628714285714286e-05,
+      "loss": 2.0004,
+      "step": 28200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.626673469387756e-05,
+      "loss": 2.0035,
+      "step": 28300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.624632653061225e-05,
+      "loss": 1.99,
+      "step": 28400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.622591836734694e-05,
+      "loss": 2.0057,
+      "step": 28500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.620551020408164e-05,
+      "loss": 1.9969,
+      "step": 28600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.618510204081634e-05,
+      "loss": 2.0034,
+      "step": 28700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.616469387755102e-05,
+      "loss": 1.9936,
+      "step": 28800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.614428571428572e-05,
+      "loss": 1.9894,
+      "step": 28900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.612387755102042e-05,
+      "loss": 1.9868,
+      "step": 29000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.61034693877551e-05,
+      "loss": 1.986,
+      "step": 29100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.60830612244898e-05,
+      "loss": 1.9815,
+      "step": 29200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.60626530612245e-05,
+      "loss": 1.9786,
+      "step": 29300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.604224489795918e-05,
+      "loss": 1.9814,
+      "step": 29400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.602183673469388e-05,
+      "loss": 1.9706,
+      "step": 29500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.600142857142858e-05,
+      "loss": 1.9836,
+      "step": 29600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.598102040816327e-05,
+      "loss": 1.9775,
+      "step": 29700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.596061224489796e-05,
+      "loss": 1.9759,
+      "step": 29800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.594020408163266e-05,
+      "loss": 1.9742,
+      "step": 29900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.591979591836735e-05,
+      "loss": 1.9741,
+      "step": 30000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.589938775510205e-05,
+      "loss": 1.9729,
+      "step": 30100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.587897959183674e-05,
+      "loss": 1.9715,
+      "step": 30200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.585857142857143e-05,
+      "loss": 1.9599,
+      "step": 30300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.583816326530613e-05,
+      "loss": 1.969,
+      "step": 30400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.581775510204081e-05,
+      "loss": 1.9579,
+      "step": 30500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.579734693877551e-05,
+      "loss": 1.9747,
+      "step": 30600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.577693877551021e-05,
+      "loss": 1.9598,
+      "step": 30700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.57565306122449e-05,
+      "loss": 1.9604,
+      "step": 30800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.573612244897959e-05,
+      "loss": 1.9619,
+      "step": 30900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.571591836734695e-05,
+      "loss": 1.9631,
+      "step": 31000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.569551020408164e-05,
+      "loss": 1.9627,
+      "step": 31100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.567510204081633e-05,
+      "loss": 1.9624,
+      "step": 31200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.565469387755103e-05,
+      "loss": 1.957,
+      "step": 31300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.563428571428573e-05,
+      "loss": 1.9582,
+      "step": 31400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.561387755102041e-05,
+      "loss": 1.951,
+      "step": 31500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.559346938775511e-05,
+      "loss": 1.9566,
+      "step": 31600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.557306122448981e-05,
+      "loss": 1.9542,
+      "step": 31700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.555265306122449e-05,
+      "loss": 1.9523,
+      "step": 31800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.553224489795919e-05,
+      "loss": 1.95,
+      "step": 31900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.551183673469389e-05,
+      "loss": 1.9511,
+      "step": 32000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.549142857142857e-05,
+      "loss": 1.9521,
+      "step": 32100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.547102040816327e-05,
+      "loss": 1.9352,
+      "step": 32200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.545061224489796e-05,
+      "loss": 1.9526,
+      "step": 32300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.543020408163266e-05,
+      "loss": 1.949,
+      "step": 32400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.540979591836735e-05,
+      "loss": 1.9462,
+      "step": 32500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.538938775510204e-05,
+      "loss": 1.9307,
+      "step": 32600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.536897959183674e-05,
+      "loss": 1.9331,
+      "step": 32700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.534877551020409e-05,
+      "loss": 1.9348,
+      "step": 32800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.532836734693879e-05,
+      "loss": 1.9337,
+      "step": 32900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.530795918367347e-05,
+      "loss": 1.9364,
+      "step": 33000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.528755102040817e-05,
+      "loss": 1.9354,
+      "step": 33100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.526714285714287e-05,
+      "loss": 1.9338,
+      "step": 33200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.524673469387756e-05,
+      "loss": 1.9324,
+      "step": 33300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.522632653061225e-05,
+      "loss": 1.9245,
+      "step": 33400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.520591836734695e-05,
+      "loss": 1.9257,
+      "step": 33500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.518551020408164e-05,
+      "loss": 1.9342,
+      "step": 33600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.516510204081634e-05,
+      "loss": 1.9234,
+      "step": 33700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.514469387755103e-05,
+      "loss": 1.9315,
+      "step": 33800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.512428571428572e-05,
+      "loss": 1.9247,
+      "step": 33900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.510387755102042e-05,
+      "loss": 1.9185,
+      "step": 34000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.50834693877551e-05,
+      "loss": 1.9257,
+      "step": 34100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.50630612244898e-05,
+      "loss": 1.9159,
+      "step": 34200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.50426530612245e-05,
+      "loss": 1.9157,
+      "step": 34300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.502224489795918e-05,
+      "loss": 1.9157,
+      "step": 34400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.500183673469388e-05,
+      "loss": 1.9231,
+      "step": 34500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.498142857142858e-05,
+      "loss": 1.9168,
+      "step": 34600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.496102040816327e-05,
+      "loss": 1.9025,
+      "step": 34700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.494061224489796e-05,
+      "loss": 1.915,
+      "step": 34800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.492020408163266e-05,
+      "loss": 1.9145,
+      "step": 34900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.489979591836735e-05,
+      "loss": 1.9116,
+      "step": 35000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.487938775510205e-05,
+      "loss": 1.9127,
+      "step": 35100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.485897959183674e-05,
+      "loss": 1.9149,
+      "step": 35200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.483877551020408e-05,
+      "loss": 1.8986,
+      "step": 35300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.481836734693877e-05,
+      "loss": 1.9075,
+      "step": 35400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.479795918367348e-05,
+      "loss": 1.9033,
+      "step": 35500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.477755102040818e-05,
+      "loss": 1.907,
+      "step": 35600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.475714285714286e-05,
+      "loss": 1.8974,
+      "step": 35700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.473673469387756e-05,
+      "loss": 1.9046,
+      "step": 35800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.471632653061225e-05,
+      "loss": 1.902,
+      "step": 35900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.469591836734695e-05,
+      "loss": 1.9043,
+      "step": 36000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.467551020408164e-05,
+      "loss": 1.9067,
+      "step": 36100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.465510204081633e-05,
+      "loss": 1.8967,
+      "step": 36200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.463469387755103e-05,
+      "loss": 1.8988,
+      "step": 36300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.461428571428573e-05,
+      "loss": 1.8956,
+      "step": 36400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.459387755102041e-05,
+      "loss": 1.9023,
+      "step": 36500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.457346938775511e-05,
+      "loss": 1.8922,
+      "step": 36600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.455306122448981e-05,
+      "loss": 1.8957,
+      "step": 36700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.453265306122449e-05,
+      "loss": 1.8954,
+      "step": 36800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.451224489795919e-05,
+      "loss": 1.8881,
+      "step": 36900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.449183673469389e-05,
+      "loss": 1.8904,
+      "step": 37000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.447142857142857e-05,
+      "loss": 1.8925,
+      "step": 37100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.445102040816327e-05,
+      "loss": 1.8846,
+      "step": 37200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.443061224489797e-05,
+      "loss": 1.8902,
+      "step": 37300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.441020408163266e-05,
+      "loss": 1.891,
+      "step": 37400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.438979591836735e-05,
+      "loss": 1.8818,
+      "step": 37500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.436938775510205e-05,
+      "loss": 1.8877,
+      "step": 37600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.434897959183674e-05,
+      "loss": 1.8846,
+      "step": 37700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.432857142857143e-05,
+      "loss": 1.8811,
+      "step": 37800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.430816326530612e-05,
+      "loss": 1.8765,
+      "step": 37900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.428775510204082e-05,
+      "loss": 1.8753,
+      "step": 38000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.426734693877552e-05,
+      "loss": 1.8803,
+      "step": 38100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.424714285714287e-05,
+      "loss": 1.869,
+      "step": 38200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.422673469387756e-05,
+      "loss": 1.879,
+      "step": 38300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.420632653061225e-05,
+      "loss": 1.8771,
+      "step": 38400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.418591836734695e-05,
+      "loss": 1.874,
+      "step": 38500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.416551020408164e-05,
+      "loss": 1.8774,
+      "step": 38600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.414510204081634e-05,
+      "loss": 1.8701,
+      "step": 38700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.412469387755103e-05,
+      "loss": 1.8816,
+      "step": 38800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.410428571428572e-05,
+      "loss": 1.8751,
+      "step": 38900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.408387755102042e-05,
+      "loss": 1.8697,
+      "step": 39000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.406346938775512e-05,
+      "loss": 1.8759,
+      "step": 39100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.40430612244898e-05,
+      "loss": 1.868,
+      "step": 39200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.40226530612245e-05,
+      "loss": 1.877,
+      "step": 39300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.40022448979592e-05,
+      "loss": 1.8579,
+      "step": 39400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.398183673469388e-05,
+      "loss": 1.864,
+      "step": 39500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.396142857142858e-05,
+      "loss": 1.8668,
+      "step": 39600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.394102040816328e-05,
+      "loss": 1.8649,
+      "step": 39700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.392061224489796e-05,
+      "loss": 1.8577,
+      "step": 39800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.390020408163266e-05,
+      "loss": 1.8644,
+      "step": 39900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.387979591836735e-05,
+      "loss": 1.8623,
+      "step": 40000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.385938775510204e-05,
+      "loss": 1.8702,
+      "step": 40100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.383918367346939e-05,
+      "loss": 1.8701,
+      "step": 40200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.381877551020408e-05,
+      "loss": 1.8628,
+      "step": 40300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.379836734693878e-05,
+      "loss": 1.8651,
+      "step": 40400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.377795918367347e-05,
+      "loss": 1.8549,
+      "step": 40500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.375755102040817e-05,
+      "loss": 1.8583,
+      "step": 40600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.373714285714285e-05,
+      "loss": 1.8585,
+      "step": 40700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.371673469387755e-05,
+      "loss": 1.8534,
+      "step": 40800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.369632653061225e-05,
+      "loss": 1.8506,
+      "step": 40900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.367591836734695e-05,
+      "loss": 1.8517,
+      "step": 41000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.365551020408164e-05,
+      "loss": 1.8433,
+      "step": 41100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.363510204081634e-05,
+      "loss": 1.8522,
+      "step": 41200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.361469387755103e-05,
+      "loss": 1.8476,
+      "step": 41300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.359428571428573e-05,
+      "loss": 1.8474,
+      "step": 41400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.357387755102042e-05,
+      "loss": 1.8538,
+      "step": 41500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.355346938775511e-05,
+      "loss": 1.8529,
+      "step": 41600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.35330612244898e-05,
+      "loss": 1.8456,
+      "step": 41700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.351265306122449e-05,
+      "loss": 1.8524,
+      "step": 41800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.349224489795919e-05,
+      "loss": 1.8396,
+      "step": 41900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.347183673469389e-05,
+      "loss": 1.8365,
+      "step": 42000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.345142857142857e-05,
+      "loss": 1.8428,
+      "step": 42100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.343102040816327e-05,
+      "loss": 1.8381,
+      "step": 42200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.341061224489797e-05,
+      "loss": 1.8452,
+      "step": 42300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.339020408163265e-05,
+      "loss": 1.8318,
+      "step": 42400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.336979591836735e-05,
+      "loss": 1.8449,
+      "step": 42500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.334938775510205e-05,
+      "loss": 1.8372,
+      "step": 42600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.332897959183674e-05,
+      "loss": 1.8401,
+      "step": 42700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.330857142857143e-05,
+      "loss": 1.8383,
+      "step": 42800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.328816326530613e-05,
+      "loss": 1.8455,
+      "step": 42900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.326775510204082e-05,
+      "loss": 1.8365,
+      "step": 43000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.324734693877552e-05,
+      "loss": 1.8363,
+      "step": 43100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.322693877551021e-05,
+      "loss": 1.8314,
+      "step": 43200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.32065306122449e-05,
+      "loss": 1.8402,
+      "step": 43300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.31861224489796e-05,
+      "loss": 1.8266,
+      "step": 43400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.31657142857143e-05,
+      "loss": 1.8325,
+      "step": 43500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.314530612244898e-05,
+      "loss": 1.8309,
+      "step": 43600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.312489795918368e-05,
+      "loss": 1.8327,
+      "step": 43700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.310448979591836e-05,
+      "loss": 1.831,
+      "step": 43800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.308408163265306e-05,
+      "loss": 1.8328,
+      "step": 43900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.306367346938776e-05,
+      "loss": 1.8282,
+      "step": 44000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.304326530612245e-05,
+      "loss": 1.8296,
+      "step": 44100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.302285714285714e-05,
+      "loss": 1.8363,
+      "step": 44200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.30026530612245e-05,
+      "loss": 1.8332,
+      "step": 44300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.29822448979592e-05,
+      "loss": 1.8283,
+      "step": 44400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.296183673469388e-05,
+      "loss": 1.8268,
+      "step": 44500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.294142857142858e-05,
+      "loss": 1.8202,
+      "step": 44600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.292102040816328e-05,
+      "loss": 1.8201,
+      "step": 44700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.290061224489796e-05,
+      "loss": 1.8212,
+      "step": 44800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.28804081632653e-05,
+      "loss": 1.8187,
+      "step": 44900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.286e-05,
+      "loss": 1.8206,
+      "step": 45000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.28395918367347e-05,
+      "loss": 1.8195,
+      "step": 45100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.281918367346939e-05,
+      "loss": 1.8204,
+      "step": 45200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.279877551020408e-05,
+      "loss": 1.8205,
+      "step": 45300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.277836734693878e-05,
+      "loss": 1.8046,
+      "step": 45400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.275795918367347e-05,
+      "loss": 1.8107,
+      "step": 45500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.273755102040817e-05,
+      "loss": 1.8048,
+      "step": 45600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.271714285714286e-05,
+      "loss": 1.8062,
+      "step": 45700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.269673469387755e-05,
+      "loss": 1.8215,
+      "step": 45800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.267632653061225e-05,
+      "loss": 1.8174,
+      "step": 45900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.265591836734694e-05,
+      "loss": 1.8117,
+      "step": 46000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.263551020408163e-05,
+      "loss": 1.814,
+      "step": 46100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.261510204081633e-05,
+      "loss": 1.8195,
+      "step": 46200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.259469387755103e-05,
+      "loss": 1.8141,
+      "step": 46300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.257428571428571e-05,
+      "loss": 1.8168,
+      "step": 46400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.255387755102042e-05,
+      "loss": 1.8141,
+      "step": 46500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.253346938775511e-05,
+      "loss": 1.8012,
+      "step": 46600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.25130612244898e-05,
+      "loss": 1.8045,
+      "step": 46700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.24926530612245e-05,
+      "loss": 1.802,
+      "step": 46800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.247224489795919e-05,
+      "loss": 1.7992,
+      "step": 46900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.245183673469389e-05,
+      "loss": 1.802,
+      "step": 47000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.243142857142859e-05,
+      "loss": 1.8078,
+      "step": 47100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.241122448979593e-05,
+      "loss": 1.8033,
+      "step": 47200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.239081632653061e-05,
+      "loss": 1.8092,
+      "step": 47300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.237040816326531e-05,
+      "loss": 1.8075,
+      "step": 47400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.235000000000001e-05,
+      "loss": 1.8011,
+      "step": 47500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.23295918367347e-05,
+      "loss": 1.7966,
+      "step": 47600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.230918367346939e-05,
+      "loss": 1.7932,
+      "step": 47700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.228877551020409e-05,
+      "loss": 1.7997,
+      "step": 47800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.226836734693877e-05,
+      "loss": 1.7804,
+      "step": 47900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.224795918367347e-05,
+      "loss": 1.7995,
+      "step": 48000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.222755102040817e-05,
+      "loss": 1.7975,
+      "step": 48100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.220714285714286e-05,
+      "loss": 1.7941,
+      "step": 48200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.218673469387755e-05,
+      "loss": 1.7945,
+      "step": 48300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.216632653061224e-05,
+      "loss": 1.8069,
+      "step": 48400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.214591836734694e-05,
+      "loss": 1.7861,
+      "step": 48500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.212551020408164e-05,
+      "loss": 1.7893,
+      "step": 48600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.210510204081632e-05,
+      "loss": 1.7968,
+      "step": 48700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.208469387755102e-05,
+      "loss": 1.7837,
+      "step": 48800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.206428571428572e-05,
+      "loss": 1.7927,
+      "step": 48900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.20438775510204e-05,
+      "loss": 1.7835,
+      "step": 49000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.20234693877551e-05,
+      "loss": 1.786,
+      "step": 49100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.20030612244898e-05,
+      "loss": 1.7881,
+      "step": 49200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.198285714285715e-05,
+      "loss": 1.7879,
+      "step": 49300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.196244897959184e-05,
+      "loss": 1.7805,
+      "step": 49400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.194204081632654e-05,
+      "loss": 1.7913,
+      "step": 49500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.192163265306124e-05,
+      "loss": 1.7811,
+      "step": 49600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.190122448979592e-05,
+      "loss": 1.7742,
+      "step": 49700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.188081632653062e-05,
+      "loss": 1.7848,
+      "step": 49800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.186040816326532e-05,
+      "loss": 1.7778,
+      "step": 49900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.184020408163266e-05,
+      "loss": 1.7794,
+      "step": 50000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.181979591836734e-05,
+      "loss": 1.781,
+      "step": 50100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.179938775510204e-05,
+      "loss": 1.7792,
+      "step": 50200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.177897959183674e-05,
+      "loss": 1.7769,
+      "step": 50300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.175857142857144e-05,
+      "loss": 1.7735,
+      "step": 50400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.173816326530614e-05,
+      "loss": 1.7859,
+      "step": 50500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.171775510204082e-05,
+      "loss": 1.7767,
+      "step": 50600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.169734693877552e-05,
+      "loss": 1.7792,
+      "step": 50700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.167693877551022e-05,
+      "loss": 1.7791,
+      "step": 50800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.16565306122449e-05,
+      "loss": 1.7757,
+      "step": 50900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.16361224489796e-05,
+      "loss": 1.7723,
+      "step": 51000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.16157142857143e-05,
+      "loss": 1.7866,
+      "step": 51100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.159530612244898e-05,
+      "loss": 1.775,
+      "step": 51200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.157489795918368e-05,
+      "loss": 1.7649,
+      "step": 51300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.155448979591838e-05,
+      "loss": 1.7735,
+      "step": 51400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.153408163265307e-05,
+      "loss": 1.7694,
+      "step": 51500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.151367346938776e-05,
+      "loss": 1.7719,
+      "step": 51600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.149326530612246e-05,
+      "loss": 1.7653,
+      "step": 51700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.147285714285715e-05,
+      "loss": 1.7661,
+      "step": 51800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.145244897959185e-05,
+      "loss": 1.7754,
+      "step": 51900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.143204081632653e-05,
+      "loss": 1.7763,
+      "step": 52000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.141163265306123e-05,
+      "loss": 1.7632,
+      "step": 52100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.139122448979593e-05,
+      "loss": 1.7677,
+      "step": 52200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.137081632653061e-05,
+      "loss": 1.7678,
+      "step": 52300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.135040816326531e-05,
+      "loss": 1.7653,
+      "step": 52400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.133000000000001e-05,
+      "loss": 1.7606,
+      "step": 52500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.130959183673469e-05,
+      "loss": 1.763,
+      "step": 52600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.128918367346939e-05,
+      "loss": 1.7578,
+      "step": 52700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.126877551020409e-05,
+      "loss": 1.7688,
+      "step": 52800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.124836734693877e-05,
+      "loss": 1.7617,
+      "step": 52900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.122795918367347e-05,
+      "loss": 1.7671,
+      "step": 53000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.120755102040817e-05,
+      "loss": 1.7587,
+      "step": 53100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.118714285714286e-05,
+      "loss": 1.7603,
+      "step": 53200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.116673469387755e-05,
+      "loss": 1.7619,
+      "step": 53300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.114632653061225e-05,
+      "loss": 1.7661,
+      "step": 53400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.112591836734694e-05,
+      "loss": 1.7594,
+      "step": 53500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.110551020408164e-05,
+      "loss": 1.7596,
+      "step": 53600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.108510204081633e-05,
+      "loss": 1.7566,
+      "step": 53700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.106469387755102e-05,
+      "loss": 1.7634,
+      "step": 53800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.104428571428572e-05,
+      "loss": 1.7567,
+      "step": 53900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.10238775510204e-05,
+      "loss": 1.7582,
+      "step": 54000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.10034693877551e-05,
+      "loss": 1.7571,
+      "step": 54100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.09830612244898e-05,
+      "loss": 1.7602,
+      "step": 54200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.096265306122448e-05,
+      "loss": 1.7547,
+      "step": 54300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.094224489795918e-05,
+      "loss": 1.756,
+      "step": 54400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.092204081632654e-05,
+      "loss": 1.7572,
+      "step": 54500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.090163265306123e-05,
+      "loss": 1.7537,
+      "step": 54600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.088122448979592e-05,
+      "loss": 1.7488,
+      "step": 54700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.086081632653062e-05,
+      "loss": 1.749,
+      "step": 54800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.084040816326532e-05,
+      "loss": 1.7595,
+      "step": 54900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.082e-05,
+      "loss": 1.748,
+      "step": 55000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.07995918367347e-05,
+      "loss": 1.7583,
+      "step": 55100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.07791836734694e-05,
+      "loss": 1.7517,
+      "step": 55200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.075877551020408e-05,
+      "loss": 1.7597,
+      "step": 55300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.073836734693878e-05,
+      "loss": 1.7466,
+      "step": 55400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.071795918367348e-05,
+      "loss": 1.7591,
+      "step": 55500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.069755102040816e-05,
+      "loss": 1.7553,
+      "step": 55600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.067714285714286e-05,
+      "loss": 1.7522,
+      "step": 55700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.065673469387755e-05,
+      "loss": 1.7545,
+      "step": 55800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.063632653061225e-05,
+      "loss": 1.7465,
+      "step": 55900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.061591836734694e-05,
+      "loss": 1.7485,
+      "step": 56000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.059551020408163e-05,
+      "loss": 1.7433,
+      "step": 56100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.057510204081633e-05,
+      "loss": 1.7459,
+      "step": 56200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.055469387755103e-05,
+      "loss": 1.7517,
+      "step": 56300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.053428571428571e-05,
+      "loss": 1.7461,
+      "step": 56400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.051387755102041e-05,
+      "loss": 1.7453,
+      "step": 56500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.049346938775511e-05,
+      "loss": 1.7475,
+      "step": 56600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.047326530612246e-05,
+      "loss": 1.7362,
+      "step": 56700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.045285714285715e-05,
+      "loss": 1.7442,
+      "step": 56800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.043244897959184e-05,
+      "loss": 1.7404,
+      "step": 56900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.041204081632654e-05,
+      "loss": 1.7508,
+      "step": 57000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.039163265306123e-05,
+      "loss": 1.74,
+      "step": 57100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.037122448979593e-05,
+      "loss": 1.7411,
+      "step": 57200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.035081632653062e-05,
+      "loss": 1.7488,
+      "step": 57300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.033040816326531e-05,
+      "loss": 1.7416,
+      "step": 57400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.031000000000001e-05,
+      "loss": 1.7335,
+      "step": 57500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.028959183673469e-05,
+      "loss": 1.7294,
+      "step": 57600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.026918367346939e-05,
+      "loss": 1.7356,
+      "step": 57700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.024877551020409e-05,
+      "loss": 1.737,
+      "step": 57800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.022836734693877e-05,
+      "loss": 1.7334,
+      "step": 57900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.020795918367347e-05,
+      "loss": 1.739,
+      "step": 58000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.018755102040817e-05,
+      "loss": 1.7376,
+      "step": 58100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.016714285714286e-05,
+      "loss": 1.7282,
+      "step": 58200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.014673469387755e-05,
+      "loss": 1.7373,
+      "step": 58300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.012632653061225e-05,
+      "loss": 1.7267,
+      "step": 58400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.010591836734694e-05,
+      "loss": 1.7359,
+      "step": 58500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.008551020408164e-05,
+      "loss": 1.7355,
+      "step": 58600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.006510204081633e-05,
+      "loss": 1.7232,
+      "step": 58700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.004469387755102e-05,
+      "loss": 1.7437,
+      "step": 58800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.002428571428572e-05,
+      "loss": 1.7298,
+      "step": 58900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.000387755102042e-05,
+      "loss": 1.741,
+      "step": 59000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.998367346938777e-05,
+      "loss": 1.7277,
+      "step": 59100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.996326530612245e-05,
+      "loss": 1.7175,
+      "step": 59200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.994285714285715e-05,
+      "loss": 1.7265,
+      "step": 59300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.992244897959185e-05,
+      "loss": 1.7356,
+      "step": 59400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.990204081632654e-05,
+      "loss": 1.7379,
+      "step": 59500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.988163265306123e-05,
+      "loss": 1.7247,
+      "step": 59600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.986122448979592e-05,
+      "loss": 1.7357,
+      "step": 59700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.984081632653062e-05,
+      "loss": 1.7294,
+      "step": 59800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.982040816326532e-05,
+      "loss": 1.7194,
+      "step": 59900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.98e-05,
+      "loss": 1.732,
+      "step": 60000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.97795918367347e-05,
+      "loss": 1.7268,
+      "step": 60100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.97591836734694e-05,
+      "loss": 1.7266,
+      "step": 60200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.973877551020408e-05,
+      "loss": 1.7285,
+      "step": 60300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.971836734693878e-05,
+      "loss": 1.7191,
+      "step": 60400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.969795918367348e-05,
+      "loss": 1.7208,
+      "step": 60500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.967755102040816e-05,
+      "loss": 1.7213,
+      "step": 60600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.965714285714286e-05,
+      "loss": 1.7199,
+      "step": 60700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.963673469387756e-05,
+      "loss": 1.7276,
+      "step": 60800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.961632653061225e-05,
+      "loss": 1.725,
+      "step": 60900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.959591836734694e-05,
+      "loss": 1.7227,
+      "step": 61000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.957551020408164e-05,
+      "loss": 1.7272,
+      "step": 61100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.955510204081633e-05,
+      "loss": 1.715,
+      "step": 61200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.953489795918367e-05,
+      "loss": 1.7319,
+      "step": 61300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.951448979591838e-05,
+      "loss": 1.7271,
+      "step": 61400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.949408163265306e-05,
+      "loss": 1.7232,
+      "step": 61500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.947367346938776e-05,
+      "loss": 1.7178,
+      "step": 61600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.945326530612246e-05,
+      "loss": 1.7232,
+      "step": 61700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.943285714285715e-05,
+      "loss": 1.7197,
+      "step": 61800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.941244897959184e-05,
+      "loss": 1.7137,
+      "step": 61900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.939204081632654e-05,
+      "loss": 1.7153,
+      "step": 62000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.937183673469388e-05,
+      "loss": 1.716,
+      "step": 62100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.935142857142857e-05,
+      "loss": 1.7187,
+      "step": 62200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.933102040816327e-05,
+      "loss": 1.7137,
+      "step": 62300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.931061224489797e-05,
+      "loss": 1.7138,
+      "step": 62400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.929020408163265e-05,
+      "loss": 1.7162,
+      "step": 62500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.926979591836735e-05,
+      "loss": 1.7143,
+      "step": 62600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.924938775510205e-05,
+      "loss": 1.7056,
+      "step": 62700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.922897959183673e-05,
+      "loss": 1.7168,
+      "step": 62800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.920857142857143e-05,
+      "loss": 1.7115,
+      "step": 62900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.918816326530613e-05,
+      "loss": 1.7186,
+      "step": 63000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.916775510204081e-05,
+      "loss": 1.7073,
+      "step": 63100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.914734693877551e-05,
+      "loss": 1.706,
+      "step": 63200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.912693877551021e-05,
+      "loss": 1.7135,
+      "step": 63300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.91065306122449e-05,
+      "loss": 1.7051,
+      "step": 63400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.908612244897959e-05,
+      "loss": 1.702,
+      "step": 63500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.906571428571429e-05,
+      "loss": 1.7151,
+      "step": 63600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.904530612244898e-05,
+      "loss": 1.706,
+      "step": 63700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.902489795918367e-05,
+      "loss": 1.7042,
+      "step": 63800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.900448979591837e-05,
+      "loss": 1.7106,
+      "step": 63900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.898408163265306e-05,
+      "loss": 1.7157,
+      "step": 64000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.896367346938776e-05,
+      "loss": 1.707,
+      "step": 64100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.894326530612245e-05,
+      "loss": 1.7065,
+      "step": 64200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.892285714285715e-05,
+      "loss": 1.7028,
+      "step": 64300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.890244897959185e-05,
+      "loss": 1.7071,
+      "step": 64400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.888204081632654e-05,
+      "loss": 1.7122,
+      "step": 64500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.886163265306123e-05,
+      "loss": 1.7008,
+      "step": 64600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.884122448979593e-05,
+      "loss": 1.6919,
+      "step": 64700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.882081632653062e-05,
+      "loss": 1.6971,
+      "step": 64800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.880061224489796e-05,
+      "loss": 1.6916,
+      "step": 64900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.878020408163266e-05,
+      "loss": 1.7018,
+      "step": 65000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.875979591836735e-05,
+      "loss": 1.6992,
+      "step": 65100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.873938775510204e-05,
+      "loss": 1.7048,
+      "step": 65200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.871897959183674e-05,
+      "loss": 1.698,
+      "step": 65300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.869857142857144e-05,
+      "loss": 1.7036,
+      "step": 65400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.867816326530612e-05,
+      "loss": 1.7019,
+      "step": 65500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.865775510204082e-05,
+      "loss": 1.7049,
+      "step": 65600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.863734693877552e-05,
+      "loss": 1.6903,
+      "step": 65700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.86169387755102e-05,
+      "loss": 1.7013,
+      "step": 65800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.85965306122449e-05,
+      "loss": 1.6985,
+      "step": 65900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.85761224489796e-05,
+      "loss": 1.6899,
+      "step": 66000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.855571428571428e-05,
+      "loss": 1.7036,
+      "step": 66100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.853530612244898e-05,
+      "loss": 1.6986,
+      "step": 66200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.851489795918367e-05,
+      "loss": 1.6894,
+      "step": 66300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.849448979591837e-05,
+      "loss": 1.6953,
+      "step": 66400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.847408163265306e-05,
+      "loss": 1.6989,
+      "step": 66500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.845367346938775e-05,
+      "loss": 1.6929,
+      "step": 66600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.843326530612245e-05,
+      "loss": 1.6943,
+      "step": 66700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.841285714285715e-05,
+      "loss": 1.6957,
+      "step": 66800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.839244897959184e-05,
+      "loss": 1.6895,
+      "step": 66900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.837204081632654e-05,
+      "loss": 1.6964,
+      "step": 67000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.835163265306123e-05,
+      "loss": 1.6901,
+      "step": 67100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.833122448979593e-05,
+      "loss": 1.7021,
+      "step": 67200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.831081632653062e-05,
+      "loss": 1.6999,
+      "step": 67300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.829040816326531e-05,
+      "loss": 1.695,
+      "step": 67400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.827000000000001e-05,
+      "loss": 1.6981,
+      "step": 67500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.82495918367347e-05,
+      "loss": 1.6905,
+      "step": 67600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.822918367346939e-05,
+      "loss": 1.6954,
+      "step": 67700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.820877551020409e-05,
+      "loss": 1.6976,
+      "step": 67800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.818836734693879e-05,
+      "loss": 1.69,
+      "step": 67900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.816795918367347e-05,
+      "loss": 1.6851,
+      "step": 68000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.814755102040817e-05,
+      "loss": 1.6923,
+      "step": 68100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.812714285714287e-05,
+      "loss": 1.6876,
+      "step": 68200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.810673469387755e-05,
+      "loss": 1.6924,
+      "step": 68300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.808632653061225e-05,
+      "loss": 1.6842,
+      "step": 68400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.806591836734695e-05,
+      "loss": 1.6818,
+      "step": 68500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.804551020408163e-05,
+      "loss": 1.6891,
+      "step": 68600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.802510204081633e-05,
+      "loss": 1.6815,
+      "step": 68700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.800469387755103e-05,
+      "loss": 1.6905,
+      "step": 68800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.798428571428572e-05,
+      "loss": 1.6864,
+      "step": 68900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.796408163265306e-05,
+      "loss": 1.6832,
+      "step": 69000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.794367346938776e-05,
+      "loss": 1.6885,
+      "step": 69100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.792326530612245e-05,
+      "loss": 1.6906,
+      "step": 69200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.790285714285714e-05,
+      "loss": 1.6818,
+      "step": 69300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.788265306122449e-05,
+      "loss": 1.6789,
+      "step": 69400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.786224489795919e-05,
+      "loss": 1.6823,
+      "step": 69500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.784183673469389e-05,
+      "loss": 1.683,
+      "step": 69600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.782142857142857e-05,
+      "loss": 1.6849,
+      "step": 69700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.780102040816327e-05,
+      "loss": 1.6844,
+      "step": 69800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.778061224489796e-05,
+      "loss": 1.6798,
+      "step": 69900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.776040816326531e-05,
+      "loss": 1.6715,
+      "step": 70000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.774e-05,
+      "loss": 1.6838,
+      "step": 70100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.77195918367347e-05,
+      "loss": 1.6847,
+      "step": 70200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.76991836734694e-05,
+      "loss": 1.6865,
+      "step": 70300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.767877551020408e-05,
+      "loss": 1.6788,
+      "step": 70400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.765836734693878e-05,
+      "loss": 1.6836,
+      "step": 70500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.763795918367346e-05,
+      "loss": 1.6776,
+      "step": 70600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.761755102040817e-05,
+      "loss": 1.6734,
+      "step": 70700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.759714285714287e-05,
+      "loss": 1.6777,
+      "step": 70800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.757673469387756e-05,
+      "loss": 1.6768,
+      "step": 70900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.755632653061225e-05,
+      "loss": 1.6786,
+      "step": 71000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.753591836734695e-05,
+      "loss": 1.6835,
+      "step": 71100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.751551020408164e-05,
+      "loss": 1.6744,
+      "step": 71200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.749510204081634e-05,
+      "loss": 1.6714,
+      "step": 71300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.747469387755103e-05,
+      "loss": 1.6782,
+      "step": 71400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.745428571428572e-05,
+      "loss": 1.6745,
+      "step": 71500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.743387755102042e-05,
+      "loss": 1.6676,
+      "step": 71600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.74134693877551e-05,
+      "loss": 1.6725,
+      "step": 71700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.73930612244898e-05,
+      "loss": 1.6674,
+      "step": 71800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.73726530612245e-05,
+      "loss": 1.676,
+      "step": 71900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.735224489795918e-05,
+      "loss": 1.6717,
+      "step": 72000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.733183673469388e-05,
+      "loss": 1.6677,
+      "step": 72100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.731142857142858e-05,
+      "loss": 1.6684,
+      "step": 72200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.729102040816327e-05,
+      "loss": 1.6703,
+      "step": 72300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.727061224489796e-05,
+      "loss": 1.6717,
+      "step": 72400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.725020408163266e-05,
+      "loss": 1.672,
+      "step": 72500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.722979591836735e-05,
+      "loss": 1.6617,
+      "step": 72600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.720938775510205e-05,
+      "loss": 1.673,
+      "step": 72700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.718897959183674e-05,
+      "loss": 1.6649,
+      "step": 72800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.716857142857143e-05,
+      "loss": 1.6754,
+      "step": 72900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.714816326530613e-05,
+      "loss": 1.6693,
+      "step": 73000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.712775510204083e-05,
+      "loss": 1.6612,
+      "step": 73100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.710734693877551e-05,
+      "loss": 1.6712,
+      "step": 73200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.708693877551021e-05,
+      "loss": 1.6641,
+      "step": 73300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.706653061224491e-05,
+      "loss": 1.664,
+      "step": 73400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.704612244897959e-05,
+      "loss": 1.6667,
+      "step": 73500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.702571428571429e-05,
+      "loss": 1.6604,
+      "step": 73600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.700530612244898e-05,
+      "loss": 1.6719,
+      "step": 73700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.698489795918367e-05,
+      "loss": 1.6651,
+      "step": 73800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.696448979591837e-05,
+      "loss": 1.6675,
+      "step": 73900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.694408163265306e-05,
+      "loss": 1.6611,
+      "step": 74000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.692367346938776e-05,
+      "loss": 1.6659,
+      "step": 74100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.690326530612245e-05,
+      "loss": 1.6709,
+      "step": 74200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.688285714285714e-05,
+      "loss": 1.6553,
+      "step": 74300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.686244897959184e-05,
+      "loss": 1.6654,
+      "step": 74400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.684204081632653e-05,
+      "loss": 1.6549,
+      "step": 74500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.682183673469389e-05,
+      "loss": 1.6685,
+      "step": 74600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.680142857142857e-05,
+      "loss": 1.6552,
+      "step": 74700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.678102040816327e-05,
+      "loss": 1.6568,
+      "step": 74800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.676061224489797e-05,
+      "loss": 1.6621,
+      "step": 74900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.674020408163266e-05,
+      "loss": 1.665,
+      "step": 75000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.671979591836735e-05,
+      "loss": 1.6602,
+      "step": 75100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.669938775510205e-05,
+      "loss": 1.6588,
+      "step": 75200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.667897959183674e-05,
+      "loss": 1.659,
+      "step": 75300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.665857142857144e-05,
+      "loss": 1.6661,
+      "step": 75400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.663816326530612e-05,
+      "loss": 1.6632,
+      "step": 75500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.661775510204082e-05,
+      "loss": 1.6646,
+      "step": 75600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.659734693877552e-05,
+      "loss": 1.6618,
+      "step": 75700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.65769387755102e-05,
+      "loss": 1.6599,
+      "step": 75800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.65565306122449e-05,
+      "loss": 1.6611,
+      "step": 75900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.65361224489796e-05,
+      "loss": 1.6468,
+      "step": 76000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.651571428571428e-05,
+      "loss": 1.6629,
+      "step": 76100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.649530612244898e-05,
+      "loss": 1.6565,
+      "step": 76200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.647489795918368e-05,
+      "loss": 1.6618,
+      "step": 76300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.645448979591836e-05,
+      "loss": 1.6577,
+      "step": 76400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.643408163265306e-05,
+      "loss": 1.6609,
+      "step": 76500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.641367346938776e-05,
+      "loss": 1.656,
+      "step": 76600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.639326530612245e-05,
+      "loss": 1.6584,
+      "step": 76700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.637285714285714e-05,
+      "loss": 1.6525,
+      "step": 76800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.63526530612245e-05,
+      "loss": 1.6497,
+      "step": 76900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.63322448979592e-05,
+      "loss": 1.6528,
+      "step": 77000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.631183673469388e-05,
+      "loss": 1.6534,
+      "step": 77100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.629142857142858e-05,
+      "loss": 1.6549,
+      "step": 77200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.627102040816328e-05,
+      "loss": 1.6509,
+      "step": 77300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.625061224489796e-05,
+      "loss": 1.6506,
+      "step": 77400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.623020408163266e-05,
+      "loss": 1.6498,
+      "step": 77500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.620979591836735e-05,
+      "loss": 1.6564,
+      "step": 77600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.618938775510205e-05,
+      "loss": 1.6501,
+      "step": 77700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.616897959183674e-05,
+      "loss": 1.6487,
+      "step": 77800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.614857142857143e-05,
+      "loss": 1.6444,
+      "step": 77900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.612816326530613e-05,
+      "loss": 1.6475,
+      "step": 78000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.610775510204083e-05,
+      "loss": 1.6496,
+      "step": 78100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.608734693877551e-05,
+      "loss": 1.6542,
+      "step": 78200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.606693877551021e-05,
+      "loss": 1.6604,
+      "step": 78300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.60465306122449e-05,
+      "loss": 1.6454,
+      "step": 78400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.602612244897959e-05,
+      "loss": 1.6475,
+      "step": 78500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.600571428571429e-05,
+      "loss": 1.6516,
+      "step": 78600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.598530612244899e-05,
+      "loss": 1.6451,
+      "step": 78700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.596489795918367e-05,
+      "loss": 1.6511,
+      "step": 78800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.594448979591837e-05,
+      "loss": 1.6444,
+      "step": 78900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.592408163265307e-05,
+      "loss": 1.6485,
+      "step": 79000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.590367346938775e-05,
+      "loss": 1.6492,
+      "step": 79100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.588326530612245e-05,
+      "loss": 1.6386,
+      "step": 79200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.586285714285715e-05,
+      "loss": 1.6482,
+      "step": 79300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.584244897959184e-05,
+      "loss": 1.6493,
+      "step": 79400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.582204081632653e-05,
+      "loss": 1.6512,
+      "step": 79500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580163265306122e-05,
+      "loss": 1.6423,
+      "step": 79600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.578122448979592e-05,
+      "loss": 1.637,
+      "step": 79700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.576081632653062e-05,
+      "loss": 1.6464,
+      "step": 79800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.57404081632653e-05,
+      "loss": 1.6383,
+      "step": 79900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.572e-05,
+      "loss": 1.6486,
+      "step": 80000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.56995918367347e-05,
+      "loss": 1.6384,
+      "step": 80100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.567918367346938e-05,
+      "loss": 1.6405,
+      "step": 80200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.565897959183674e-05,
+      "loss": 1.6375,
+      "step": 80300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.563857142857144e-05,
+      "loss": 1.6406,
+      "step": 80400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.561816326530613e-05,
+      "loss": 1.6478,
+      "step": 80500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.559775510204082e-05,
+      "loss": 1.6418,
+      "step": 80600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.557734693877552e-05,
+      "loss": 1.6416,
+      "step": 80700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.555693877551021e-05,
+      "loss": 1.6395,
+      "step": 80800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.55365306122449e-05,
+      "loss": 1.6435,
+      "step": 80900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.55161224489796e-05,
+      "loss": 1.6446,
+      "step": 81000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.54957142857143e-05,
+      "loss": 1.6399,
+      "step": 81100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.547530612244898e-05,
+      "loss": 1.6338,
+      "step": 81200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.545489795918368e-05,
+      "loss": 1.6412,
+      "step": 81300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.543448979591836e-05,
+      "loss": 1.6326,
+      "step": 81400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.541408163265306e-05,
+      "loss": 1.6488,
+      "step": 81500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.539367346938776e-05,
+      "loss": 1.6392,
+      "step": 81600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.537326530612245e-05,
+      "loss": 1.6355,
+      "step": 81700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.535285714285714e-05,
+      "loss": 1.6371,
+      "step": 81800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.533244897959184e-05,
+      "loss": 1.6325,
+      "step": 81900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.531204081632653e-05,
+      "loss": 1.634,
+      "step": 82000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.529163265306123e-05,
+      "loss": 1.6442,
+      "step": 82100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.527122448979592e-05,
+      "loss": 1.6302,
+      "step": 82200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.525081632653061e-05,
+      "loss": 1.6329,
+      "step": 82300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.523040816326531e-05,
+      "loss": 1.6356,
+      "step": 82400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.521e-05,
+      "loss": 1.6328,
+      "step": 82500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.518959183673469e-05,
+      "loss": 1.6346,
+      "step": 82600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.516918367346939e-05,
+      "loss": 1.6301,
+      "step": 82700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.514877551020409e-05,
+      "loss": 1.6325,
+      "step": 82800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.512836734693877e-05,
+      "loss": 1.6276,
+      "step": 82900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.510795918367347e-05,
+      "loss": 1.638,
+      "step": 83000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.508755102040817e-05,
+      "loss": 1.6347,
+      "step": 83100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.506734693877551e-05,
+      "loss": 1.6288,
+      "step": 83200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.504693877551021e-05,
+      "loss": 1.6267,
+      "step": 83300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.50265306122449e-05,
+      "loss": 1.6386,
+      "step": 83400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.500612244897959e-05,
+      "loss": 1.634,
+      "step": 83500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.498591836734695e-05,
+      "loss": 1.6339,
+      "step": 83600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.496551020408163e-05,
+      "loss": 1.6281,
+      "step": 83700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.494510204081633e-05,
+      "loss": 1.6272,
+      "step": 83800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.492469387755103e-05,
+      "loss": 1.6306,
+      "step": 83900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.490428571428571e-05,
+      "loss": 1.6375,
+      "step": 84000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.488387755102041e-05,
+      "loss": 1.6298,
+      "step": 84100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.48634693877551e-05,
+      "loss": 1.6336,
+      "step": 84200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.48430612244898e-05,
+      "loss": 1.6233,
+      "step": 84300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.482265306122449e-05,
+      "loss": 1.6354,
+      "step": 84400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.480224489795919e-05,
+      "loss": 1.6274,
+      "step": 84500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.478183673469389e-05,
+      "loss": 1.6248,
+      "step": 84600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.476142857142859e-05,
+      "loss": 1.6284,
+      "step": 84700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.474102040816327e-05,
+      "loss": 1.6327,
+      "step": 84800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.472061224489797e-05,
+      "loss": 1.6178,
+      "step": 84900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.470020408163265e-05,
+      "loss": 1.6245,
+      "step": 85000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.467979591836735e-05,
+      "loss": 1.6274,
+      "step": 85100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.465938775510205e-05,
+      "loss": 1.6248,
+      "step": 85200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.463897959183674e-05,
+      "loss": 1.6219,
+      "step": 85300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.461857142857143e-05,
+      "loss": 1.623,
+      "step": 85400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.459816326530613e-05,
+      "loss": 1.6258,
+      "step": 85500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.457775510204082e-05,
+      "loss": 1.6235,
+      "step": 85600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.455734693877552e-05,
+      "loss": 1.6183,
+      "step": 85700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.453693877551021e-05,
+      "loss": 1.6229,
+      "step": 85800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.45165306122449e-05,
+      "loss": 1.6289,
+      "step": 85900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.44961224489796e-05,
+      "loss": 1.6321,
+      "step": 86000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.44757142857143e-05,
+      "loss": 1.625,
+      "step": 86100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.445530612244898e-05,
+      "loss": 1.6288,
+      "step": 86200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.443489795918368e-05,
+      "loss": 1.6262,
+      "step": 86300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.441448979591838e-05,
+      "loss": 1.6222,
+      "step": 86400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.439408163265306e-05,
+      "loss": 1.6253,
+      "step": 86500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.437367346938776e-05,
+      "loss": 1.6219,
+      "step": 86600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.435326530612246e-05,
+      "loss": 1.6215,
+      "step": 86700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.433285714285714e-05,
+      "loss": 1.6131,
+      "step": 86800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.431244897959184e-05,
+      "loss": 1.6179,
+      "step": 86900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.429204081632654e-05,
+      "loss": 1.6131,
+      "step": 87000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.427163265306123e-05,
+      "loss": 1.6247,
+      "step": 87100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.425122448979592e-05,
+      "loss": 1.6259,
+      "step": 87200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.423081632653061e-05,
+      "loss": 1.6216,
+      "step": 87300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.421040816326531e-05,
+      "loss": 1.6218,
+      "step": 87400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.419e-05,
+      "loss": 1.61,
+      "step": 87500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.416959183673469e-05,
+      "loss": 1.6201,
+      "step": 87600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.414918367346939e-05,
+      "loss": 1.6161,
+      "step": 87700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.412877551020409e-05,
+      "loss": 1.617,
+      "step": 87800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.410836734693877e-05,
+      "loss": 1.622,
+      "step": 87900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.408795918367347e-05,
+      "loss": 1.6149,
+      "step": 88000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.406755102040817e-05,
+      "loss": 1.6155,
+      "step": 88100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.404714285714285e-05,
+      "loss": 1.6139,
+      "step": 88200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.402673469387755e-05,
+      "loss": 1.6086,
+      "step": 88300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.400632653061225e-05,
+      "loss": 1.6116,
+      "step": 88400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.39861224489796e-05,
+      "loss": 1.621,
+      "step": 88500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.396571428571429e-05,
+      "loss": 1.6108,
+      "step": 88600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.394530612244899e-05,
+      "loss": 1.6223,
+      "step": 88700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.392489795918369e-05,
+      "loss": 1.6185,
+      "step": 88800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.390448979591837e-05,
+      "loss": 1.6239,
+      "step": 88900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.388408163265307e-05,
+      "loss": 1.605,
+      "step": 89000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.386367346938775e-05,
+      "loss": 1.6105,
+      "step": 89100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.384326530612245e-05,
+      "loss": 1.6151,
+      "step": 89200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.382285714285715e-05,
+      "loss": 1.6145,
+      "step": 89300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.380244897959184e-05,
+      "loss": 1.623,
+      "step": 89400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.378204081632653e-05,
+      "loss": 1.6176,
+      "step": 89500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.376163265306123e-05,
+      "loss": 1.6083,
+      "step": 89600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.374122448979592e-05,
+      "loss": 1.6112,
+      "step": 89700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.372081632653062e-05,
+      "loss": 1.613,
+      "step": 89800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.370040816326531e-05,
+      "loss": 1.6176,
+      "step": 89900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.368e-05,
+      "loss": 1.6185,
+      "step": 90000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.36595918367347e-05,
+      "loss": 1.6205,
+      "step": 90100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.36391836734694e-05,
+      "loss": 1.617,
+      "step": 90200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.361877551020408e-05,
+      "loss": 1.6221,
+      "step": 90300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.359836734693878e-05,
+      "loss": 1.6177,
+      "step": 90400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.357816326530613e-05,
+      "loss": 1.607,
+      "step": 90500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.355775510204083e-05,
+      "loss": 1.6132,
+      "step": 90600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.353734693877552e-05,
+      "loss": 1.6007,
+      "step": 90700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.351693877551021e-05,
+      "loss": 1.6057,
+      "step": 90800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.34965306122449e-05,
+      "loss": 1.6116,
+      "step": 90900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.34761224489796e-05,
+      "loss": 1.6121,
+      "step": 91000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.34557142857143e-05,
+      "loss": 1.6077,
+      "step": 91100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.343530612244898e-05,
+      "loss": 1.616,
+      "step": 91200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.341489795918368e-05,
+      "loss": 1.6066,
+      "step": 91300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.339448979591838e-05,
+      "loss": 1.6093,
+      "step": 91400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.337408163265306e-05,
+      "loss": 1.6054,
+      "step": 91500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.335367346938776e-05,
+      "loss": 1.6099,
+      "step": 91600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.333326530612246e-05,
+      "loss": 1.6069,
+      "step": 91700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.331285714285714e-05,
+      "loss": 1.6152,
+      "step": 91800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.329244897959184e-05,
+      "loss": 1.6062,
+      "step": 91900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.327204081632654e-05,
+      "loss": 1.6017,
+      "step": 92000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.325163265306123e-05,
+      "loss": 1.6047,
+      "step": 92100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.323122448979592e-05,
+      "loss": 1.6063,
+      "step": 92200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.321081632653062e-05,
+      "loss": 1.6036,
+      "step": 92300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.31904081632653e-05,
+      "loss": 1.6064,
+      "step": 92400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.317e-05,
+      "loss": 1.6035,
+      "step": 92500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.31495918367347e-05,
+      "loss": 1.6031,
+      "step": 92600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.312938775510204e-05,
+      "loss": 1.6052,
+      "step": 92700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.310897959183674e-05,
+      "loss": 1.6037,
+      "step": 92800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.308857142857144e-05,
+      "loss": 1.6044,
+      "step": 92900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.306816326530613e-05,
+      "loss": 1.6047,
+      "step": 93000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.304775510204082e-05,
+      "loss": 1.601,
+      "step": 93100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.302734693877552e-05,
+      "loss": 1.5984,
+      "step": 93200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.300693877551021e-05,
+      "loss": 1.6022,
+      "step": 93300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.29865306122449e-05,
+      "loss": 1.5999,
+      "step": 93400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.29661224489796e-05,
+      "loss": 1.6109,
+      "step": 93500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.294571428571429e-05,
+      "loss": 1.604,
+      "step": 93600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.292551020408163e-05,
+      "loss": 1.6048,
+      "step": 93700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.290510204081633e-05,
+      "loss": 1.6008,
+      "step": 93800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.288469387755103e-05,
+      "loss": 1.6005,
+      "step": 93900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.286428571428571e-05,
+      "loss": 1.5987,
+      "step": 94000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.284387755102041e-05,
+      "loss": 1.6,
+      "step": 94100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.282346938775511e-05,
+      "loss": 1.602,
+      "step": 94200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.280306122448979e-05,
+      "loss": 1.6111,
+      "step": 94300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.278265306122449e-05,
+      "loss": 1.599,
+      "step": 94400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.276224489795919e-05,
+      "loss": 1.606,
+      "step": 94500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.274183673469387e-05,
+      "loss": 1.5968,
+      "step": 94600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.272142857142857e-05,
+      "loss": 1.5868,
+      "step": 94700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.270102040816327e-05,
+      "loss": 1.6044,
+      "step": 94800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.268061224489796e-05,
+      "loss": 1.594,
+      "step": 94900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.266020408163265e-05,
+      "loss": 1.5909,
+      "step": 95000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.263979591836735e-05,
+      "loss": 1.5955,
+      "step": 95100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.261938775510204e-05,
+      "loss": 1.588,
+      "step": 95200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.259897959183674e-05,
+      "loss": 1.5873,
+      "step": 95300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.257857142857143e-05,
+      "loss": 1.5995,
+      "step": 95400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.255816326530612e-05,
+      "loss": 1.5928,
+      "step": 95500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.253775510204083e-05,
+      "loss": 1.5925,
+      "step": 95600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.251734693877552e-05,
+      "loss": 1.5995,
+      "step": 95700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.249693877551021e-05,
+      "loss": 1.5906,
+      "step": 95800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.247653061224491e-05,
+      "loss": 1.5933,
+      "step": 95900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.24561224489796e-05,
+      "loss": 1.5945,
+      "step": 96000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.24357142857143e-05,
+      "loss": 1.5872,
+      "step": 96100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.2415306122449e-05,
+      "loss": 1.5889,
+      "step": 96200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.239489795918368e-05,
+      "loss": 1.5898,
+      "step": 96300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.237448979591838e-05,
+      "loss": 1.585,
+      "step": 96400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.235408163265306e-05,
+      "loss": 1.5857,
+      "step": 96500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.233367346938776e-05,
+      "loss": 1.5885,
+      "step": 96600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.231326530612246e-05,
+      "loss": 1.5938,
+      "step": 96700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.229285714285714e-05,
+      "loss": 1.5889,
+      "step": 96800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.227244897959184e-05,
+      "loss": 1.5863,
+      "step": 96900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.225204081632654e-05,
+      "loss": 1.5901,
+      "step": 97000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.223163265306122e-05,
+      "loss": 1.587,
+      "step": 97100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.221122448979592e-05,
+      "loss": 1.5944,
+      "step": 97200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.219081632653062e-05,
+      "loss": 1.5867,
+      "step": 97300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.21704081632653e-05,
+      "loss": 1.5812,
+      "step": 97400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.215e-05,
+      "loss": 1.5864,
+      "step": 97500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.21295918367347e-05,
+      "loss": 1.5928,
+      "step": 97600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.210938775510204e-05,
+      "loss": 1.5907,
+      "step": 97700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.208897959183673e-05,
+      "loss": 1.5866,
+      "step": 97800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.206857142857143e-05,
+      "loss": 1.6052,
+      "step": 97900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.204816326530613e-05,
+      "loss": 1.5815,
+      "step": 98000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.202775510204081e-05,
+      "loss": 1.5801,
+      "step": 98100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.200734693877551e-05,
+      "loss": 1.5841,
+      "step": 98200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.198693877551021e-05,
+      "loss": 1.5855,
+      "step": 98300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.19665306122449e-05,
+      "loss": 1.584,
+      "step": 98400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.19461224489796e-05,
+      "loss": 1.5811,
+      "step": 98500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.192571428571429e-05,
+      "loss": 1.5792,
+      "step": 98600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.190530612244899e-05,
+      "loss": 1.5816,
+      "step": 98700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.188489795918369e-05,
+      "loss": 1.5791,
+      "step": 98800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.186448979591837e-05,
+      "loss": 1.5806,
+      "step": 98900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.184408163265307e-05,
+      "loss": 1.5861,
+      "step": 99000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.182367346938777e-05,
+      "loss": 1.5813,
+      "step": 99100
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.180326530612245e-05,
+      "loss": 1.5888,
+      "step": 99200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.178285714285715e-05,
+      "loss": 1.5832,
+      "step": 99300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.176244897959185e-05,
+      "loss": 1.5824,
+      "step": 99400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.174204081632653e-05,
+      "loss": 1.5756,
+      "step": 99500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.172163265306123e-05,
+      "loss": 1.5854,
+      "step": 99600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.170122448979593e-05,
+      "loss": 1.5929,
+      "step": 99700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.168081632653061e-05,
+      "loss": 1.574,
+      "step": 99800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.166040816326531e-05,
+      "loss": 1.5755,
+      "step": 99900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.164000000000001e-05,
+      "loss": 1.5811,
+      "step": 100000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.16195918367347e-05,
+      "loss": 1.5787,
+      "step": 100100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.159938775510204e-05,
+      "loss": 1.5844,
+      "step": 100200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.157897959183674e-05,
+      "loss": 1.5802,
+      "step": 100300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.155857142857143e-05,
+      "loss": 1.5799,
+      "step": 100400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.153816326530612e-05,
+      "loss": 1.5914,
+      "step": 100500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.151775510204082e-05,
+      "loss": 1.5856,
+      "step": 100600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.149734693877552e-05,
+      "loss": 1.5786,
+      "step": 100700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.14769387755102e-05,
+      "loss": 1.5799,
+      "step": 100800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.14565306122449e-05,
+      "loss": 1.5756,
+      "step": 100900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.14361224489796e-05,
+      "loss": 1.5801,
+      "step": 101000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.14157142857143e-05,
+      "loss": 1.5814,
+      "step": 101100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.1395306122449e-05,
+      "loss": 1.5797,
+      "step": 101200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.137489795918368e-05,
+      "loss": 1.58,
+      "step": 101300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.135448979591838e-05,
+      "loss": 1.5733,
+      "step": 101400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.133408163265307e-05,
+      "loss": 1.5771,
+      "step": 101500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.131367346938776e-05,
+      "loss": 1.5803,
+      "step": 101600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.129326530612246e-05,
+      "loss": 1.5738,
+      "step": 101700
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.127285714285716e-05,
+      "loss": 1.5729,
+      "step": 101800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.125244897959184e-05,
+      "loss": 1.5772,
+      "step": 101900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.123204081632654e-05,
+      "loss": 1.5771,
+      "step": 102000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.121163265306124e-05,
+      "loss": 1.5776,
+      "step": 102100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.119122448979592e-05,
+      "loss": 1.583,
+      "step": 102200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.117081632653062e-05,
+      "loss": 1.5745,
+      "step": 102300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.11504081632653e-05,
+      "loss": 1.5749,
+      "step": 102400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.113e-05,
+      "loss": 1.5867,
+      "step": 102500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.11095918367347e-05,
+      "loss": 1.5795,
+      "step": 102600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.108938775510204e-05,
+      "loss": 1.5763,
+      "step": 102700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.106897959183674e-05,
+      "loss": 1.5683,
+      "step": 102800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.104857142857143e-05,
+      "loss": 1.5766,
+      "step": 102900
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.102816326530612e-05,
+      "loss": 1.5716,
+      "step": 103000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.100775510204082e-05,
+      "loss": 1.581,
+      "step": 103100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.098734693877551e-05,
+      "loss": 1.578,
+      "step": 103200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.09669387755102e-05,
+      "loss": 1.5702,
+      "step": 103300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.094653061224489e-05,
+      "loss": 1.572,
+      "step": 103400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.092612244897959e-05,
+      "loss": 1.5754,
+      "step": 103500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.090571428571429e-05,
+      "loss": 1.5761,
+      "step": 103600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.088530612244897e-05,
+      "loss": 1.5736,
+      "step": 103700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.086489795918367e-05,
+      "loss": 1.5668,
+      "step": 103800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.084448979591838e-05,
+      "loss": 1.5761,
+      "step": 103900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.082408163265307e-05,
+      "loss": 1.5777,
+      "step": 104000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.080367346938777e-05,
+      "loss": 1.5655,
+      "step": 104100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.078326530612245e-05,
+      "loss": 1.579,
+      "step": 104200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.076285714285715e-05,
+      "loss": 1.5827,
+      "step": 104300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.074244897959185e-05,
+      "loss": 1.5745,
+      "step": 104400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.072204081632653e-05,
+      "loss": 1.5625,
+      "step": 104500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.070163265306123e-05,
+      "loss": 1.5732,
+      "step": 104600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.068122448979593e-05,
+      "loss": 1.5714,
+      "step": 104700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.066081632653061e-05,
+      "loss": 1.5755,
+      "step": 104800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.064040816326531e-05,
+      "loss": 1.5785,
+      "step": 104900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.062000000000001e-05,
+      "loss": 1.5689,
+      "step": 105000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.05995918367347e-05,
+      "loss": 1.563,
+      "step": 105100
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.05791836734694e-05,
+      "loss": 1.5673,
+      "step": 105200
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.055877551020409e-05,
+      "loss": 1.5768,
+      "step": 105300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.053836734693878e-05,
+      "loss": 1.57,
+      "step": 105400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.051795918367348e-05,
+      "loss": 1.568,
+      "step": 105500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.049755102040817e-05,
+      "loss": 1.567,
+      "step": 105600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.047714285714286e-05,
+      "loss": 1.5693,
+      "step": 105700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.045673469387756e-05,
+      "loss": 1.5817,
+      "step": 105800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.043632653061226e-05,
+      "loss": 1.5695,
+      "step": 105900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.041591836734694e-05,
+      "loss": 1.5707,
+      "step": 106000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.039551020408164e-05,
+      "loss": 1.5695,
+      "step": 106100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.037510204081634e-05,
+      "loss": 1.5738,
+      "step": 106200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.035469387755102e-05,
+      "loss": 1.5671,
+      "step": 106300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.033428571428572e-05,
+      "loss": 1.568,
+      "step": 106400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.03138775510204e-05,
+      "loss": 1.5706,
+      "step": 106500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.02934693877551e-05,
+      "loss": 1.5664,
+      "step": 106600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.02730612244898e-05,
+      "loss": 1.5629,
+      "step": 106700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.025265306122449e-05,
+      "loss": 1.5689,
+      "step": 106800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.023244897959184e-05,
+      "loss": 1.564,
+      "step": 106900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.021204081632654e-05,
+      "loss": 1.5708,
+      "step": 107000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.019163265306124e-05,
+      "loss": 1.5595,
+      "step": 107100
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.017122448979592e-05,
+      "loss": 1.5546,
+      "step": 107200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.015081632653062e-05,
+      "loss": 1.5598,
+      "step": 107300
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.013040816326532e-05,
+      "loss": 1.5639,
+      "step": 107400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.011e-05,
+      "loss": 1.558,
+      "step": 107500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.00895918367347e-05,
+      "loss": 1.5644,
+      "step": 107600
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.00691836734694e-05,
+      "loss": 1.5677,
+      "step": 107700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.004877551020409e-05,
+      "loss": 1.5615,
+      "step": 107800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.002836734693878e-05,
+      "loss": 1.5615,
+      "step": 107900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.000795918367348e-05,
+      "loss": 1.5653,
+      "step": 108000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.998755102040817e-05,
+      "loss": 1.5624,
+      "step": 108100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.996714285714287e-05,
+      "loss": 1.5565,
+      "step": 108200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.994673469387755e-05,
+      "loss": 1.5667,
+      "step": 108300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.99265306122449e-05,
+      "loss": 1.5662,
+      "step": 108400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.990612244897959e-05,
+      "loss": 1.5653,
+      "step": 108500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.988571428571429e-05,
+      "loss": 1.5635,
+      "step": 108600
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.986530612244899e-05,
+      "loss": 1.5638,
+      "step": 108700
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.984489795918367e-05,
+      "loss": 1.5647,
+      "step": 108800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.982448979591837e-05,
+      "loss": 1.5653,
+      "step": 108900
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.980408163265305e-05,
+      "loss": 1.5694,
+      "step": 109000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.978367346938775e-05,
+      "loss": 1.5644,
+      "step": 109100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.976326530612245e-05,
+      "loss": 1.5638,
+      "step": 109200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.974285714285714e-05,
+      "loss": 1.5624,
+      "step": 109300
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.972244897959183e-05,
+      "loss": 1.5654,
+      "step": 109400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.970204081632655e-05,
+      "loss": 1.5575,
+      "step": 109500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.968163265306123e-05,
+      "loss": 1.5626,
+      "step": 109600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.966122448979593e-05,
+      "loss": 1.5577,
+      "step": 109700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.964081632653063e-05,
+      "loss": 1.5663,
+      "step": 109800
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.962040816326531e-05,
+      "loss": 1.5587,
+      "step": 109900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 1.5641,
+      "step": 110000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.95795918367347e-05,
+      "loss": 1.5583,
+      "step": 110100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.95591836734694e-05,
+      "loss": 1.5579,
+      "step": 110200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.953877551020409e-05,
+      "loss": 1.5577,
+      "step": 110300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.951836734693878e-05,
+      "loss": 1.5526,
+      "step": 110400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.949795918367348e-05,
+      "loss": 1.5622,
+      "step": 110500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.947755102040817e-05,
+      "loss": 1.5611,
+      "step": 110600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.945734693877551e-05,
+      "loss": 1.5596,
+      "step": 110700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.94369387755102e-05,
+      "loss": 1.5589,
+      "step": 110800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.94165306122449e-05,
+      "loss": 1.5554,
+      "step": 110900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.93961224489796e-05,
+      "loss": 1.559,
+      "step": 111000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.937571428571428e-05,
+      "loss": 1.5522,
+      "step": 111100
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.935530612244898e-05,
+      "loss": 1.559,
+      "step": 111200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.933489795918368e-05,
+      "loss": 1.5542,
+      "step": 111300
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.931448979591836e-05,
+      "loss": 1.5536,
+      "step": 111400
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.929408163265306e-05,
+      "loss": 1.5498,
+      "step": 111500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.927367346938776e-05,
+      "loss": 1.5556,
+      "step": 111600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.925326530612244e-05,
+      "loss": 1.5617,
+      "step": 111700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.923285714285714e-05,
+      "loss": 1.5517,
+      "step": 111800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.921244897959184e-05,
+      "loss": 1.5526,
+      "step": 111900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.919204081632653e-05,
+      "loss": 1.5576,
+      "step": 112000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.917163265306122e-05,
+      "loss": 1.5517,
+      "step": 112100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.915122448979592e-05,
+      "loss": 1.55,
+      "step": 112200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.913081632653062e-05,
+      "loss": 1.551,
+      "step": 112300
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.911040816326532e-05,
+      "loss": 1.5561,
+      "step": 112400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.909e-05,
+      "loss": 1.5479,
+      "step": 112500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.90695918367347e-05,
+      "loss": 1.5498,
+      "step": 112600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.90491836734694e-05,
+      "loss": 1.5467,
+      "step": 112700
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.902877551020408e-05,
+      "loss": 1.5532,
+      "step": 112800
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.900836734693878e-05,
+      "loss": 1.5613,
+      "step": 112900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.898795918367348e-05,
+      "loss": 1.5587,
+      "step": 113000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.896755102040817e-05,
+      "loss": 1.5509,
+      "step": 113100
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.894714285714286e-05,
+      "loss": 1.551,
+      "step": 113200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.89269387755102e-05,
+      "loss": 1.5522,
+      "step": 113300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.89065306122449e-05,
+      "loss": 1.5469,
+      "step": 113400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.888612244897959e-05,
+      "loss": 1.552,
+      "step": 113500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.886571428571429e-05,
+      "loss": 1.5568,
+      "step": 113600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.884530612244899e-05,
+      "loss": 1.552,
+      "step": 113700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.882489795918367e-05,
+      "loss": 1.543,
+      "step": 113800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.880448979591837e-05,
+      "loss": 1.5518,
+      "step": 113900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.878408163265307e-05,
+      "loss": 1.5574,
+      "step": 114000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.876367346938775e-05,
+      "loss": 1.5493,
+      "step": 114100
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.874326530612245e-05,
+      "loss": 1.5509,
+      "step": 114200
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.872285714285715e-05,
+      "loss": 1.5438,
+      "step": 114300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.870244897959183e-05,
+      "loss": 1.5488,
+      "step": 114400
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.868204081632653e-05,
+      "loss": 1.5539,
+      "step": 114500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.866163265306123e-05,
+      "loss": 1.5434,
+      "step": 114600
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.864122448979591e-05,
+      "loss": 1.5465,
+      "step": 114700
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.862081632653061e-05,
+      "loss": 1.5467,
+      "step": 114800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.86004081632653e-05,
+      "loss": 1.5436,
+      "step": 114900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.858000000000001e-05,
+      "loss": 1.5447,
+      "step": 115000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.855959183673471e-05,
+      "loss": 1.5457,
+      "step": 115100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.853918367346939e-05,
+      "loss": 1.547,
+      "step": 115200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.851877551020409e-05,
+      "loss": 1.5493,
+      "step": 115300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.849857142857143e-05,
+      "loss": 1.5352,
+      "step": 115400
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.847816326530613e-05,
+      "loss": 1.5444,
+      "step": 115500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.845775510204082e-05,
+      "loss": 1.5563,
+      "step": 115600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.843734693877551e-05,
+      "loss": 1.5493,
+      "step": 115700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.841693877551021e-05,
+      "loss": 1.5445,
+      "step": 115800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.83965306122449e-05,
+      "loss": 1.548,
+      "step": 115900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.83761224489796e-05,
+      "loss": 1.5448,
+      "step": 116000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.83557142857143e-05,
+      "loss": 1.5489,
+      "step": 116100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.833530612244898e-05,
+      "loss": 1.5479,
+      "step": 116200
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.831489795918368e-05,
+      "loss": 1.5432,
+      "step": 116300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.829448979591838e-05,
+      "loss": 1.5413,
+      "step": 116400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.827408163265306e-05,
+      "loss": 1.5402,
+      "step": 116500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.825367346938776e-05,
+      "loss": 1.5461,
+      "step": 116600
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.823326530612244e-05,
+      "loss": 1.5456,
+      "step": 116700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.821285714285714e-05,
+      "loss": 1.5503,
+      "step": 116800
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.819244897959184e-05,
+      "loss": 1.5402,
+      "step": 116900
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.817204081632652e-05,
+      "loss": 1.5477,
+      "step": 117000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.815163265306122e-05,
+      "loss": 1.5453,
+      "step": 117100
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.813122448979592e-05,
+      "loss": 1.5508,
+      "step": 117200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.81108163265306e-05,
+      "loss": 1.5441,
+      "step": 117300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.80904081632653e-05,
+      "loss": 1.5413,
+      "step": 117400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.807020408163266e-05,
+      "loss": 1.547,
+      "step": 117500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.804979591836736e-05,
+      "loss": 1.5493,
+      "step": 117600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.802938775510204e-05,
+      "loss": 1.5385,
+      "step": 117700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.800897959183674e-05,
+      "loss": 1.5419,
+      "step": 117800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.798857142857144e-05,
+      "loss": 1.5409,
+      "step": 117900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.796816326530612e-05,
+      "loss": 1.5382,
+      "step": 118000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.794775510204082e-05,
+      "loss": 1.5337,
+      "step": 118100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.792734693877552e-05,
+      "loss": 1.5345,
+      "step": 118200
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.79069387755102e-05,
+      "loss": 1.5412,
+      "step": 118300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.78865306122449e-05,
+      "loss": 1.5362,
+      "step": 118400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.786612244897959e-05,
+      "loss": 1.5429,
+      "step": 118500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.784571428571429e-05,
+      "loss": 1.5394,
+      "step": 118600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.782530612244899e-05,
+      "loss": 1.5422,
+      "step": 118700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.780489795918367e-05,
+      "loss": 1.5367,
+      "step": 118800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.778448979591837e-05,
+      "loss": 1.54,
+      "step": 118900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.776408163265307e-05,
+      "loss": 1.539,
+      "step": 119000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.774367346938775e-05,
+      "loss": 1.5392,
+      "step": 119100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.772326530612245e-05,
+      "loss": 1.5401,
+      "step": 119200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.770285714285715e-05,
+      "loss": 1.5364,
+      "step": 119300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.768244897959183e-05,
+      "loss": 1.5404,
+      "step": 119400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.766204081632653e-05,
+      "loss": 1.5374,
+      "step": 119500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.764163265306123e-05,
+      "loss": 1.5386,
+      "step": 119600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.762122448979591e-05,
+      "loss": 1.5436,
+      "step": 119700
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.760081632653061e-05,
+      "loss": 1.5483,
+      "step": 119800
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.758040816326531e-05,
+      "loss": 1.5424,
+      "step": 119900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.756e-05,
+      "loss": 1.5456,
+      "step": 120000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.75395918367347e-05,
+      "loss": 1.5483,
+      "step": 120100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.751918367346939e-05,
+      "loss": 1.5346,
+      "step": 120200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.749877551020408e-05,
+      "loss": 1.5387,
+      "step": 120300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.747836734693878e-05,
+      "loss": 1.5327,
+      "step": 120400
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.745795918367347e-05,
+      "loss": 1.5299,
+      "step": 120500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.743755102040817e-05,
+      "loss": 1.5365,
+      "step": 120600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.741714285714287e-05,
+      "loss": 1.5332,
+      "step": 120700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.739673469387756e-05,
+      "loss": 1.5309,
+      "step": 120800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.737632653061225e-05,
+      "loss": 1.5325,
+      "step": 120900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.735591836734695e-05,
+      "loss": 1.5408,
+      "step": 121000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.733551020408164e-05,
+      "loss": 1.5385,
+      "step": 121100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.731510204081634e-05,
+      "loss": 1.5373,
+      "step": 121200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.729489795918368e-05,
+      "loss": 1.5339,
+      "step": 121300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.727448979591837e-05,
+      "loss": 1.5329,
+      "step": 121400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.725408163265306e-05,
+      "loss": 1.5338,
+      "step": 121500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.723367346938776e-05,
+      "loss": 1.5435,
+      "step": 121600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.721326530612246e-05,
+      "loss": 1.5353,
+      "step": 121700
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.719285714285714e-05,
+      "loss": 1.5441,
+      "step": 121800
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.717244897959184e-05,
+      "loss": 1.5382,
+      "step": 121900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.715204081632654e-05,
+      "loss": 1.5281,
+      "step": 122000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.713163265306122e-05,
+      "loss": 1.5395,
+      "step": 122100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.711122448979592e-05,
+      "loss": 1.5414,
+      "step": 122200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.709081632653062e-05,
+      "loss": 1.5403,
+      "step": 122300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.70704081632653e-05,
+      "loss": 1.5377,
+      "step": 122400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.705e-05,
+      "loss": 1.5357,
+      "step": 122500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.702959183673469e-05,
+      "loss": 1.5267,
+      "step": 122600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.700918367346939e-05,
+      "loss": 1.5313,
+      "step": 122700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.698877551020408e-05,
+      "loss": 1.5313,
+      "step": 122800
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.696836734693877e-05,
+      "loss": 1.5416,
+      "step": 122900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.694795918367347e-05,
+      "loss": 1.5324,
+      "step": 123000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.692755102040817e-05,
+      "loss": 1.5337,
+      "step": 123100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.690714285714285e-05,
+      "loss": 1.5251,
+      "step": 123200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.688673469387756e-05,
+      "loss": 1.5319,
+      "step": 123300
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.686632653061225e-05,
+      "loss": 1.5345,
+      "step": 123400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.684591836734695e-05,
+      "loss": 1.5258,
+      "step": 123500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.682551020408164e-05,
+      "loss": 1.5312,
+      "step": 123600
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.680530612244898e-05,
+      "loss": 1.5261,
+      "step": 123700
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.678489795918368e-05,
+      "loss": 1.5297,
+      "step": 123800
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.676469387755102e-05,
+      "loss": 1.5295,
+      "step": 123900
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.674428571428571e-05,
+      "loss": 1.5279,
+      "step": 124000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.672387755102041e-05,
+      "loss": 1.531,
+      "step": 124100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.67034693877551e-05,
+      "loss": 1.5365,
+      "step": 124200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.66830612244898e-05,
+      "loss": 1.5293,
+      "step": 124300
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.66626530612245e-05,
+      "loss": 1.533,
+      "step": 124400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.664224489795919e-05,
+      "loss": 1.5199,
+      "step": 124500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.662183673469389e-05,
+      "loss": 1.5206,
+      "step": 124600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.660142857142858e-05,
+      "loss": 1.5226,
+      "step": 124700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.658102040816327e-05,
+      "loss": 1.5292,
+      "step": 124800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.656061224489797e-05,
+      "loss": 1.5311,
+      "step": 124900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.654020408163267e-05,
+      "loss": 1.5315,
+      "step": 125000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.651979591836735e-05,
+      "loss": 1.5295,
+      "step": 125100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.649938775510205e-05,
+      "loss": 1.5281,
+      "step": 125200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.647897959183675e-05,
+      "loss": 1.5298,
+      "step": 125300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.645857142857143e-05,
+      "loss": 1.5273,
+      "step": 125400
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.643816326530613e-05,
+      "loss": 1.5355,
+      "step": 125500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.641775510204083e-05,
+      "loss": 1.5248,
+      "step": 125600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.639734693877551e-05,
+      "loss": 1.5235,
+      "step": 125700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.637693877551021e-05,
+      "loss": 1.5206,
+      "step": 125800
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.635653061224491e-05,
+      "loss": 1.5353,
+      "step": 125900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.63361224489796e-05,
+      "loss": 1.5237,
+      "step": 126000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.631571428571429e-05,
+      "loss": 1.5284,
+      "step": 126100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.629530612244898e-05,
+      "loss": 1.5201,
+      "step": 126200
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.627489795918368e-05,
+      "loss": 1.5184,
+      "step": 126300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.625448979591837e-05,
+      "loss": 1.5288,
+      "step": 126400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.623408163265306e-05,
+      "loss": 1.5185,
+      "step": 126500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.621367346938776e-05,
+      "loss": 1.5197,
+      "step": 126600
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.619326530612246e-05,
+      "loss": 1.5183,
+      "step": 126700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.617285714285714e-05,
+      "loss": 1.526,
+      "step": 126800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.615244897959184e-05,
+      "loss": 1.5263,
+      "step": 126900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.613204081632654e-05,
+      "loss": 1.5355,
+      "step": 127000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.611163265306122e-05,
+      "loss": 1.5153,
+      "step": 127100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.609122448979592e-05,
+      "loss": 1.5238,
+      "step": 127200
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.607081632653062e-05,
+      "loss": 1.5249,
+      "step": 127300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.60504081632653e-05,
+      "loss": 1.5221,
+      "step": 127400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.603e-05,
+      "loss": 1.5308,
+      "step": 127500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.60095918367347e-05,
+      "loss": 1.5278,
+      "step": 127600
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.598918367346939e-05,
+      "loss": 1.5243,
+      "step": 127700
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.596877551020408e-05,
+      "loss": 1.5192,
+      "step": 127800
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.594836734693878e-05,
+      "loss": 1.5234,
+      "step": 127900
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.592795918367347e-05,
+      "loss": 1.5179,
+      "step": 128000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.590755102040817e-05,
+      "loss": 1.5199,
+      "step": 128100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.588714285714285e-05,
+      "loss": 1.5272,
+      "step": 128200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.586673469387755e-05,
+      "loss": 1.5181,
+      "step": 128300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.584632653061225e-05,
+      "loss": 1.5181,
+      "step": 128400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.582591836734693e-05,
+      "loss": 1.5264,
+      "step": 128500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.580591836734694e-05,
+      "loss": 1.5234,
+      "step": 128600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.578551020408163e-05,
+      "loss": 1.5209,
+      "step": 128700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.576510204081633e-05,
+      "loss": 1.5243,
+      "step": 128800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.574469387755102e-05,
+      "loss": 1.5177,
+      "step": 128900
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.572428571428571e-05,
+      "loss": 1.5149,
+      "step": 129000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.57038775510204e-05,
+      "loss": 1.5152,
+      "step": 129100
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.56834693877551e-05,
+      "loss": 1.5168,
+      "step": 129200
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.566306122448979e-05,
+      "loss": 1.5249,
+      "step": 129300
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.564265306122449e-05,
+      "loss": 1.5187,
+      "step": 129400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.562224489795919e-05,
+      "loss": 1.5214,
+      "step": 129500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.560183673469387e-05,
+      "loss": 1.5098,
+      "step": 129600
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.558142857142857e-05,
+      "loss": 1.5203,
+      "step": 129700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.556102040816327e-05,
+      "loss": 1.5166,
+      "step": 129800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.554061224489797e-05,
+      "loss": 1.5162,
+      "step": 129900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.552020408163266e-05,
+      "loss": 1.5211,
+      "step": 130000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.549979591836735e-05,
+      "loss": 1.516,
+      "step": 130100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.547938775510205e-05,
+      "loss": 1.5189,
+      "step": 130200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.545897959183675e-05,
+      "loss": 1.5171,
+      "step": 130300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.543857142857143e-05,
+      "loss": 1.5168,
+      "step": 130400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.541816326530613e-05,
+      "loss": 1.5272,
+      "step": 130500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.539775510204083e-05,
+      "loss": 1.5202,
+      "step": 130600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.537734693877551e-05,
+      "loss": 1.5196,
+      "step": 130700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.535693877551021e-05,
+      "loss": 1.5122,
+      "step": 130800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.533653061224491e-05,
+      "loss": 1.5158,
+      "step": 130900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.53161224489796e-05,
+      "loss": 1.5127,
+      "step": 131000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.529571428571429e-05,
+      "loss": 1.5193,
+      "step": 131100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.527530612244899e-05,
+      "loss": 1.5264,
+      "step": 131200
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.525489795918368e-05,
+      "loss": 1.5092,
+      "step": 131300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.523448979591837e-05,
+      "loss": 1.5148,
+      "step": 131400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.521408163265307e-05,
+      "loss": 1.5172,
+      "step": 131500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.519367346938776e-05,
+      "loss": 1.5146,
+      "step": 131600
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.517326530612246e-05,
+      "loss": 1.5272,
+      "step": 131700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.515285714285714e-05,
+      "loss": 1.5156,
+      "step": 131800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.513244897959184e-05,
+      "loss": 1.5165,
+      "step": 131900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.511204081632654e-05,
+      "loss": 1.5151,
+      "step": 132000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.509163265306122e-05,
+      "loss": 1.5146,
+      "step": 132100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.507122448979592e-05,
+      "loss": 1.5261,
+      "step": 132200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.505081632653062e-05,
+      "loss": 1.5172,
+      "step": 132300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.50304081632653e-05,
+      "loss": 1.5118,
+      "step": 132400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.501e-05,
+      "loss": 1.5136,
+      "step": 132500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.49895918367347e-05,
+      "loss": 1.5107,
+      "step": 132600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.496918367346938e-05,
+      "loss": 1.5172,
+      "step": 132700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.494877551020408e-05,
+      "loss": 1.52,
+      "step": 132800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.492836734693878e-05,
+      "loss": 1.5162,
+      "step": 132900
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.490795918367347e-05,
+      "loss": 1.5069,
+      "step": 133000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.488755102040816e-05,
+      "loss": 1.5107,
+      "step": 133100
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.486714285714286e-05,
+      "loss": 1.5024,
+      "step": 133200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.484673469387755e-05,
+      "loss": 1.5077,
+      "step": 133300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.482632653061225e-05,
+      "loss": 1.5104,
+      "step": 133400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.480591836734694e-05,
+      "loss": 1.5059,
+      "step": 133500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.478551020408163e-05,
+      "loss": 1.5116,
+      "step": 133600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.476510204081633e-05,
+      "loss": 1.5048,
+      "step": 133700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.474469387755103e-05,
+      "loss": 1.51,
+      "step": 133800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.472428571428571e-05,
+      "loss": 1.5127,
+      "step": 133900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.470387755102041e-05,
+      "loss": 1.5077,
+      "step": 134000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.46834693877551e-05,
+      "loss": 1.51,
+      "step": 134100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.466326530612245e-05,
+      "loss": 1.5082,
+      "step": 134200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.464285714285715e-05,
+      "loss": 1.5151,
+      "step": 134300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.462244897959185e-05,
+      "loss": 1.5116,
+      "step": 134400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.460204081632653e-05,
+      "loss": 1.507,
+      "step": 134500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.458163265306123e-05,
+      "loss": 1.5088,
+      "step": 134600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.456122448979593e-05,
+      "loss": 1.5147,
+      "step": 134700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.454081632653061e-05,
+      "loss": 1.5039,
+      "step": 134800
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.452040816326531e-05,
+      "loss": 1.5101,
+      "step": 134900
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.450000000000001e-05,
+      "loss": 1.5042,
+      "step": 135000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.447959183673469e-05,
+      "loss": 1.5117,
+      "step": 135100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.445918367346939e-05,
+      "loss": 1.5149,
+      "step": 135200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.443877551020409e-05,
+      "loss": 1.506,
+      "step": 135300
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.441836734693877e-05,
+      "loss": 1.5112,
+      "step": 135400
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.439795918367347e-05,
+      "loss": 1.5203,
+      "step": 135500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.437755102040817e-05,
+      "loss": 1.5137,
+      "step": 135600
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.435714285714286e-05,
+      "loss": 1.5105,
+      "step": 135700
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.433673469387755e-05,
+      "loss": 1.5049,
+      "step": 135800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.431632653061224e-05,
+      "loss": 1.5072,
+      "step": 135900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.429591836734694e-05,
+      "loss": 1.5087,
+      "step": 136000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.427551020408164e-05,
+      "loss": 1.5089,
+      "step": 136100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.425530612244899e-05,
+      "loss": 1.506,
+      "step": 136200
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.423489795918368e-05,
+      "loss": 1.5045,
+      "step": 136300
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.421448979591837e-05,
+      "loss": 1.5111,
+      "step": 136400
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.419408163265307e-05,
+      "loss": 1.509,
+      "step": 136500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.417367346938776e-05,
+      "loss": 1.5069,
+      "step": 136600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.415326530612246e-05,
+      "loss": 1.513,
+      "step": 136700
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.413285714285715e-05,
+      "loss": 1.5051,
+      "step": 136800
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.411244897959184e-05,
+      "loss": 1.4956,
+      "step": 136900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.409204081632654e-05,
+      "loss": 1.5124,
+      "step": 137000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.407163265306124e-05,
+      "loss": 1.5031,
+      "step": 137100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.405122448979592e-05,
+      "loss": 1.506,
+      "step": 137200
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.403081632653062e-05,
+      "loss": 1.5137,
+      "step": 137300
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.401040816326532e-05,
+      "loss": 1.5057,
+      "step": 137400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.399e-05,
+      "loss": 1.5096,
+      "step": 137500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.39695918367347e-05,
+      "loss": 1.5013,
+      "step": 137600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.394918367346938e-05,
+      "loss": 1.4997,
+      "step": 137700
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.392877551020408e-05,
+      "loss": 1.5044,
+      "step": 137800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.390836734693878e-05,
+      "loss": 1.4998,
+      "step": 137900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.388795918367347e-05,
+      "loss": 1.4997,
+      "step": 138000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.386755102040816e-05,
+      "loss": 1.5011,
+      "step": 138100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.384714285714286e-05,
+      "loss": 1.5059,
+      "step": 138200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.382673469387755e-05,
+      "loss": 1.4967,
+      "step": 138300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.380632653061225e-05,
+      "loss": 1.5058,
+      "step": 138400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.378591836734694e-05,
+      "loss": 1.507,
+      "step": 138500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.376551020408163e-05,
+      "loss": 1.5055,
+      "step": 138600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.374530612244898e-05,
+      "loss": 1.4997,
+      "step": 138700
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.372489795918368e-05,
+      "loss": 1.5018,
+      "step": 138800
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.370448979591838e-05,
+      "loss": 1.5025,
+      "step": 138900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.368408163265307e-05,
+      "loss": 1.4973,
+      "step": 139000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.366367346938776e-05,
+      "loss": 1.5026,
+      "step": 139100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.364326530612246e-05,
+      "loss": 1.4982,
+      "step": 139200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.362285714285715e-05,
+      "loss": 1.5012,
+      "step": 139300
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.360244897959184e-05,
+      "loss": 1.496,
+      "step": 139400
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.358204081632653e-05,
+      "loss": 1.4993,
+      "step": 139500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.356163265306123e-05,
+      "loss": 1.5052,
+      "step": 139600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.354122448979593e-05,
+      "loss": 1.4959,
+      "step": 139700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.352081632653061e-05,
+      "loss": 1.5008,
+      "step": 139800
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.350040816326531e-05,
+      "loss": 1.5012,
+      "step": 139900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.348000000000001e-05,
+      "loss": 1.4916,
+      "step": 140000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.345959183673469e-05,
+      "loss": 1.5001,
+      "step": 140100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.343918367346939e-05,
+      "loss": 1.4972,
+      "step": 140200
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.341877551020409e-05,
+      "loss": 1.508,
+      "step": 140300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.339836734693877e-05,
+      "loss": 1.5028,
+      "step": 140400
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.337795918367347e-05,
+      "loss": 1.4981,
+      "step": 140500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.335755102040817e-05,
+      "loss": 1.5012,
+      "step": 140600
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.333714285714286e-05,
+      "loss": 1.5039,
+      "step": 140700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.331673469387755e-05,
+      "loss": 1.4998,
+      "step": 140800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.329632653061225e-05,
+      "loss": 1.5055,
+      "step": 140900
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.327591836734694e-05,
+      "loss": 1.5009,
+      "step": 141000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.325551020408164e-05,
+      "loss": 1.4966,
+      "step": 141100
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.323510204081633e-05,
+      "loss": 1.4976,
+      "step": 141200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.321469387755102e-05,
+      "loss": 1.5013,
+      "step": 141300
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.319428571428572e-05,
+      "loss": 1.5025,
+      "step": 141400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.317387755102042e-05,
+      "loss": 1.4963,
+      "step": 141500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.31534693877551e-05,
+      "loss": 1.5062,
+      "step": 141600
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.31330612244898e-05,
+      "loss": 1.496,
+      "step": 141700
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.311265306122448e-05,
+      "loss": 1.5063,
+      "step": 141800
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.309224489795918e-05,
+      "loss": 1.493,
+      "step": 141900
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.307183673469388e-05,
+      "loss": 1.4966,
+      "step": 142000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.305142857142857e-05,
+      "loss": 1.4851,
+      "step": 142100
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.303102040816326e-05,
+      "loss": 1.4939,
+      "step": 142200
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.301061224489796e-05,
+      "loss": 1.4906,
+      "step": 142300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.299020408163265e-05,
+      "loss": 1.483,
+      "step": 142400
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.296979591836735e-05,
+      "loss": 1.4873,
+      "step": 142500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.294938775510204e-05,
+      "loss": 1.4995,
+      "step": 142600
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.29291836734694e-05,
+      "loss": 1.4982,
+      "step": 142700
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.290877551020408e-05,
+      "loss": 1.4879,
+      "step": 142800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.288836734693878e-05,
+      "loss": 1.4877,
+      "step": 142900
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.286816326530612e-05,
+      "loss": 1.4944,
+      "step": 143000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.284775510204082e-05,
+      "loss": 1.4964,
+      "step": 143100
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.28273469387755e-05,
+      "loss": 1.4956,
+      "step": 143200
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.28069387755102e-05,
+      "loss": 1.4902,
+      "step": 143300
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.27865306122449e-05,
+      "loss": 1.5054,
+      "step": 143400
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.276612244897959e-05,
+      "loss": 1.4937,
+      "step": 143500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.274571428571428e-05,
+      "loss": 1.4873,
+      "step": 143600
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.272530612244898e-05,
+      "loss": 1.4904,
+      "step": 143700
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.270489795918368e-05,
+      "loss": 1.4901,
+      "step": 143800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.268448979591838e-05,
+      "loss": 1.4897,
+      "step": 143900
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.266408163265306e-05,
+      "loss": 1.4886,
+      "step": 144000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.264367346938776e-05,
+      "loss": 1.4874,
+      "step": 144100
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.262326530612246e-05,
+      "loss": 1.4873,
+      "step": 144200
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.260285714285715e-05,
+      "loss": 1.4831,
+      "step": 144300
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.258244897959184e-05,
+      "loss": 1.4915,
+      "step": 144400
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.256204081632654e-05,
+      "loss": 1.4896,
+      "step": 144500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.254163265306123e-05,
+      "loss": 1.494,
+      "step": 144600
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.252122448979593e-05,
+      "loss": 1.4907,
+      "step": 144700
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.250081632653062e-05,
+      "loss": 1.4886,
+      "step": 144800
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.248040816326531e-05,
+      "loss": 1.4901,
+      "step": 144900
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.246000000000001e-05,
+      "loss": 1.4895,
+      "step": 145000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.24395918367347e-05,
+      "loss": 1.4902,
+      "step": 145100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.241918367346939e-05,
+      "loss": 1.4891,
+      "step": 145200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.239877551020409e-05,
+      "loss": 1.4909,
+      "step": 145300
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.237836734693877e-05,
+      "loss": 1.4858,
+      "step": 145400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.235795918367347e-05,
+      "loss": 1.4814,
+      "step": 145500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.233755102040817e-05,
+      "loss": 1.4896,
+      "step": 145600
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.231714285714286e-05,
+      "loss": 1.4909,
+      "step": 145700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.229673469387755e-05,
+      "loss": 1.4817,
+      "step": 145800
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.227632653061225e-05,
+      "loss": 1.4996,
+      "step": 145900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.225612244897959e-05,
+      "loss": 1.4846,
+      "step": 146000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.223571428571428e-05,
+      "loss": 1.4848,
+      "step": 146100
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.221530612244898e-05,
+      "loss": 1.4927,
+      "step": 146200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.219489795918367e-05,
+      "loss": 1.4827,
+      "step": 146300
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.217448979591837e-05,
+      "loss": 1.4828,
+      "step": 146400
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.215408163265307e-05,
+      "loss": 1.4836,
+      "step": 146500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.213367346938777e-05,
+      "loss": 1.492,
+      "step": 146600
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.211326530612245e-05,
+      "loss": 1.486,
+      "step": 146700
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.209285714285715e-05,
+      "loss": 1.4876,
+      "step": 146800
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.207244897959185e-05,
+      "loss": 1.4951,
+      "step": 146900
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.205204081632654e-05,
+      "loss": 1.4902,
+      "step": 147000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.203163265306123e-05,
+      "loss": 1.4879,
+      "step": 147100
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.201122448979592e-05,
+      "loss": 1.4919,
+      "step": 147200
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.199081632653062e-05,
+      "loss": 1.4878,
+      "step": 147300
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.197040816326532e-05,
+      "loss": 1.4864,
+      "step": 147400
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.195e-05,
+      "loss": 1.491,
+      "step": 147500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.19295918367347e-05,
+      "loss": 1.487,
+      "step": 147600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.19091836734694e-05,
+      "loss": 1.4847,
+      "step": 147700
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.188877551020408e-05,
+      "loss": 1.4858,
+      "step": 147800
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.186836734693878e-05,
+      "loss": 1.4811,
+      "step": 147900
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.184795918367348e-05,
+      "loss": 1.4852,
+      "step": 148000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.182755102040816e-05,
+      "loss": 1.4834,
+      "step": 148100
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.180714285714286e-05,
+      "loss": 1.4882,
+      "step": 148200
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.178673469387756e-05,
+      "loss": 1.4876,
+      "step": 148300
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.176632653061225e-05,
+      "loss": 1.4891,
+      "step": 148400
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.174612244897959e-05,
+      "loss": 1.4888,
+      "step": 148500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.172571428571428e-05,
+      "loss": 1.4906,
+      "step": 148600
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.170530612244898e-05,
+      "loss": 1.4777,
+      "step": 148700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.168489795918367e-05,
+      "loss": 1.4833,
+      "step": 148800
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.166448979591837e-05,
+      "loss": 1.4839,
+      "step": 148900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.164408163265306e-05,
+      "loss": 1.497,
+      "step": 149000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.162367346938775e-05,
+      "loss": 1.485,
+      "step": 149100
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.160326530612246e-05,
+      "loss": 1.4841,
+      "step": 149200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.158285714285715e-05,
+      "loss": 1.4745,
+      "step": 149300
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.156244897959184e-05,
+      "loss": 1.4864,
+      "step": 149400
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.154204081632654e-05,
+      "loss": 1.4819,
+      "step": 149500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.152163265306123e-05,
+      "loss": 1.4867,
+      "step": 149600
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.150122448979593e-05,
+      "loss": 1.4801,
+      "step": 149700
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.148081632653062e-05,
+      "loss": 1.4701,
+      "step": 149800
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.146040816326531e-05,
+      "loss": 1.4851,
+      "step": 149900
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.144000000000001e-05,
+      "loss": 1.4769,
+      "step": 150000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.14195918367347e-05,
+      "loss": 1.4796,
+      "step": 150100
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.139918367346939e-05,
+      "loss": 1.4736,
+      "step": 150200
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.137877551020409e-05,
+      "loss": 1.4836,
+      "step": 150300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.135836734693879e-05,
+      "loss": 1.4787,
+      "step": 150400
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.133795918367347e-05,
+      "loss": 1.4834,
+      "step": 150500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.131755102040817e-05,
+      "loss": 1.4826,
+      "step": 150600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.129714285714287e-05,
+      "loss": 1.4824,
+      "step": 150700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.127673469387755e-05,
+      "loss": 1.4822,
+      "step": 150800
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.125632653061225e-05,
+      "loss": 1.4784,
+      "step": 150900
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.123591836734694e-05,
+      "loss": 1.4739,
+      "step": 151000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.121551020408163e-05,
+      "loss": 1.4851,
+      "step": 151100
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.119510204081633e-05,
+      "loss": 1.475,
+      "step": 151200
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.117469387755102e-05,
+      "loss": 1.4827,
+      "step": 151300
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.115428571428572e-05,
+      "loss": 1.4764,
+      "step": 151400
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.113387755102041e-05,
+      "loss": 1.4717,
+      "step": 151500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.11134693877551e-05,
+      "loss": 1.4854,
+      "step": 151600
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.109326530612245e-05,
+      "loss": 1.4696,
+      "step": 151700
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.107285714285714e-05,
+      "loss": 1.4815,
+      "step": 151800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.105244897959184e-05,
+      "loss": 1.4826,
+      "step": 151900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.103204081632654e-05,
+      "loss": 1.4785,
+      "step": 152000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.101163265306123e-05,
+      "loss": 1.4766,
+      "step": 152100
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.099122448979593e-05,
+      "loss": 1.4779,
+      "step": 152200
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.097081632653062e-05,
+      "loss": 1.4775,
+      "step": 152300
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.095040816326532e-05,
+      "loss": 1.4779,
+      "step": 152400
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.093000000000001e-05,
+      "loss": 1.4756,
+      "step": 152500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.09095918367347e-05,
+      "loss": 1.482,
+      "step": 152600
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.08891836734694e-05,
+      "loss": 1.4791,
+      "step": 152700
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.086877551020408e-05,
+      "loss": 1.4696,
+      "step": 152800
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.084836734693878e-05,
+      "loss": 1.4822,
+      "step": 152900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.082795918367348e-05,
+      "loss": 1.4795,
+      "step": 153000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.080755102040816e-05,
+      "loss": 1.4753,
+      "step": 153100
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.078714285714286e-05,
+      "loss": 1.4844,
+      "step": 153200
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.076673469387756e-05,
+      "loss": 1.4719,
+      "step": 153300
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.074632653061224e-05,
+      "loss": 1.4787,
+      "step": 153400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.072591836734694e-05,
+      "loss": 1.4747,
+      "step": 153500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.070551020408164e-05,
+      "loss": 1.477,
+      "step": 153600
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.068510204081633e-05,
+      "loss": 1.4793,
+      "step": 153700
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 7.066469387755102e-05,
+      "loss": 1.4716,
+      "step": 153800
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 7.064448979591837e-05,
+      "loss": 1.4776,
+      "step": 153900
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 7.062408163265306e-05,
+      "loss": 1.4762,
+      "step": 154000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 7.060367346938775e-05,
+      "loss": 1.4777,
+      "step": 154100
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 7.058326530612245e-05,
+      "loss": 1.4844,
+      "step": 154200
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 7.056285714285715e-05,
+      "loss": 1.4796,
+      "step": 154300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 7.054244897959183e-05,
+      "loss": 1.4695,
+      "step": 154400
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 7.052204081632653e-05,
+      "loss": 1.4783,
+      "step": 154500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 7.050163265306123e-05,
+      "loss": 1.4821,
+      "step": 154600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 7.048122448979593e-05,
+      "loss": 1.4741,
+      "step": 154700
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.046081632653062e-05,
+      "loss": 1.4705,
+      "step": 154800
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.044040816326531e-05,
+      "loss": 1.4731,
+      "step": 154900
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.042000000000001e-05,
+      "loss": 1.4795,
+      "step": 155000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.03995918367347e-05,
+      "loss": 1.4709,
+      "step": 155100
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 7.037918367346939e-05,
+      "loss": 1.4758,
+      "step": 155200
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.035877551020409e-05,
+      "loss": 1.4721,
+      "step": 155300
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.033836734693879e-05,
+      "loss": 1.4812,
+      "step": 155400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.031795918367347e-05,
+      "loss": 1.4775,
+      "step": 155500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.029755102040817e-05,
+      "loss": 1.4835,
+      "step": 155600
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.027714285714287e-05,
+      "loss": 1.4796,
+      "step": 155700
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.025673469387755e-05,
+      "loss": 1.4798,
+      "step": 155800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.023632653061225e-05,
+      "loss": 1.4718,
+      "step": 155900
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.021591836734695e-05,
+      "loss": 1.4766,
+      "step": 156000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.019571428571429e-05,
+      "loss": 1.4761,
+      "step": 156100
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.017530612244898e-05,
+      "loss": 1.4791,
+      "step": 156200
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.015489795918367e-05,
+      "loss": 1.463,
+      "step": 156300
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.013448979591837e-05,
+      "loss": 1.4671,
+      "step": 156400
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.011408163265306e-05,
+      "loss": 1.4708,
+      "step": 156500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.009367346938776e-05,
+      "loss": 1.4685,
+      "step": 156600
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.007326530612245e-05,
+      "loss": 1.4649,
+      "step": 156700
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.005285714285714e-05,
+      "loss": 1.4684,
+      "step": 156800
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.003244897959184e-05,
+      "loss": 1.4714,
+      "step": 156900
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.001204081632654e-05,
+      "loss": 1.4737,
+      "step": 157000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.999163265306122e-05,
+      "loss": 1.4659,
+      "step": 157100
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.997122448979592e-05,
+      "loss": 1.4762,
+      "step": 157200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.995081632653062e-05,
+      "loss": 1.4687,
+      "step": 157300
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.99304081632653e-05,
+      "loss": 1.473,
+      "step": 157400
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.991000000000001e-05,
+      "loss": 1.4657,
+      "step": 157500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.98895918367347e-05,
+      "loss": 1.4718,
+      "step": 157600
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.98691836734694e-05,
+      "loss": 1.4785,
+      "step": 157700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.98487755102041e-05,
+      "loss": 1.4707,
+      "step": 157800
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.982836734693878e-05,
+      "loss": 1.4636,
+      "step": 157900
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.980795918367348e-05,
+      "loss": 1.4707,
+      "step": 158000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.978755102040818e-05,
+      "loss": 1.4648,
+      "step": 158100
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.976714285714286e-05,
+      "loss": 1.4699,
+      "step": 158200
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.974673469387756e-05,
+      "loss": 1.4668,
+      "step": 158300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.972632653061226e-05,
+      "loss": 1.4699,
+      "step": 158400
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.970591836734694e-05,
+      "loss": 1.4691,
+      "step": 158500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.968571428571428e-05,
+      "loss": 1.4771,
+      "step": 158600
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.966530612244898e-05,
+      "loss": 1.4666,
+      "step": 158700
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.964489795918368e-05,
+      "loss": 1.4655,
+      "step": 158800
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.962448979591837e-05,
+      "loss": 1.4689,
+      "step": 158900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.960408163265306e-05,
+      "loss": 1.4633,
+      "step": 159000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.958367346938776e-05,
+      "loss": 1.4655,
+      "step": 159100
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.956326530612245e-05,
+      "loss": 1.4622,
+      "step": 159200
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.954285714285714e-05,
+      "loss": 1.4644,
+      "step": 159300
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.952244897959184e-05,
+      "loss": 1.4699,
+      "step": 159400
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.950204081632653e-05,
+      "loss": 1.4622,
+      "step": 159500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.948163265306123e-05,
+      "loss": 1.4687,
+      "step": 159600
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.946122448979591e-05,
+      "loss": 1.4754,
+      "step": 159700
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.944081632653061e-05,
+      "loss": 1.4585,
+      "step": 159800
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.942040816326531e-05,
+      "loss": 1.4764,
+      "step": 159900
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 1.477,
+      "step": 160000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.937959183673469e-05,
+      "loss": 1.4675,
+      "step": 160100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.935918367346939e-05,
+      "loss": 1.4614,
+      "step": 160200
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.933877551020409e-05,
+      "loss": 1.4692,
+      "step": 160300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.931836734693879e-05,
+      "loss": 1.4738,
+      "step": 160400
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.929795918367347e-05,
+      "loss": 1.4686,
+      "step": 160500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.927755102040817e-05,
+      "loss": 1.4662,
+      "step": 160600
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.925714285714287e-05,
+      "loss": 1.4678,
+      "step": 160700
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.923673469387755e-05,
+      "loss": 1.4637,
+      "step": 160800
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.921632653061225e-05,
+      "loss": 1.4693,
+      "step": 160900
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.919591836734695e-05,
+      "loss": 1.4652,
+      "step": 161000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.917551020408163e-05,
+      "loss": 1.4715,
+      "step": 161100
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.915510204081633e-05,
+      "loss": 1.4652,
+      "step": 161200
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.913469387755103e-05,
+      "loss": 1.4633,
+      "step": 161300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.911428571428572e-05,
+      "loss": 1.4604,
+      "step": 161400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.909387755102041e-05,
+      "loss": 1.4643,
+      "step": 161500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.907346938775511e-05,
+      "loss": 1.4675,
+      "step": 161600
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.90530612244898e-05,
+      "loss": 1.4678,
+      "step": 161700
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.90326530612245e-05,
+      "loss": 1.4709,
+      "step": 161800
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.90122448979592e-05,
+      "loss": 1.4652,
+      "step": 161900
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.899183673469388e-05,
+      "loss": 1.4665,
+      "step": 162000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.897142857142858e-05,
+      "loss": 1.4581,
+      "step": 162100
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.895102040816328e-05,
+      "loss": 1.4595,
+      "step": 162200
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.893061224489796e-05,
+      "loss": 1.4611,
+      "step": 162300
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.891020408163266e-05,
+      "loss": 1.4668,
+      "step": 162400
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.888979591836736e-05,
+      "loss": 1.4709,
+      "step": 162500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.886938775510204e-05,
+      "loss": 1.4499,
+      "step": 162600
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.884918367346938e-05,
+      "loss": 1.4748,
+      "step": 162700
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.882877551020408e-05,
+      "loss": 1.4707,
+      "step": 162800
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.880836734693878e-05,
+      "loss": 1.47,
+      "step": 162900
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.878795918367346e-05,
+      "loss": 1.4619,
+      "step": 163000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.876755102040818e-05,
+      "loss": 1.4681,
+      "step": 163100
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.874714285714286e-05,
+      "loss": 1.474,
+      "step": 163200
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.872673469387756e-05,
+      "loss": 1.4628,
+      "step": 163300
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.870632653061226e-05,
+      "loss": 1.4585,
+      "step": 163400
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.868591836734694e-05,
+      "loss": 1.4648,
+      "step": 163500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.866551020408164e-05,
+      "loss": 1.4648,
+      "step": 163600
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.864530612244898e-05,
+      "loss": 1.4679,
+      "step": 163700
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.862489795918368e-05,
+      "loss": 1.4626,
+      "step": 163800
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.860448979591836e-05,
+      "loss": 1.4632,
+      "step": 163900
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.858408163265306e-05,
+      "loss": 1.4665,
+      "step": 164000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.856367346938776e-05,
+      "loss": 1.4637,
+      "step": 164100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.854326530612245e-05,
+      "loss": 1.4647,
+      "step": 164200
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.852285714285714e-05,
+      "loss": 1.4545,
+      "step": 164300
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.850244897959184e-05,
+      "loss": 1.468,
+      "step": 164400
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.848204081632653e-05,
+      "loss": 1.4599,
+      "step": 164500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.846163265306123e-05,
+      "loss": 1.4655,
+      "step": 164600
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.844122448979592e-05,
+      "loss": 1.46,
+      "step": 164700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.842081632653061e-05,
+      "loss": 1.4621,
+      "step": 164800
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.840040816326531e-05,
+      "loss": 1.4609,
+      "step": 164900
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.838e-05,
+      "loss": 1.4595,
+      "step": 165000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.835959183673469e-05,
+      "loss": 1.4637,
+      "step": 165100
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.833918367346939e-05,
+      "loss": 1.4643,
+      "step": 165200
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.831877551020407e-05,
+      "loss": 1.4607,
+      "step": 165300
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.829836734693877e-05,
+      "loss": 1.4653,
+      "step": 165400
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.827795918367347e-05,
+      "loss": 1.4641,
+      "step": 165500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.825755102040816e-05,
+      "loss": 1.4606,
+      "step": 165600
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.823714285714285e-05,
+      "loss": 1.4565,
+      "step": 165700
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.821673469387757e-05,
+      "loss": 1.4664,
+      "step": 165800
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.819632653061225e-05,
+      "loss": 1.4644,
+      "step": 165900
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.817591836734695e-05,
+      "loss": 1.464,
+      "step": 166000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.815551020408165e-05,
+      "loss": 1.4656,
+      "step": 166100
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.813510204081633e-05,
+      "loss": 1.4568,
+      "step": 166200
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.811469387755103e-05,
+      "loss": 1.4682,
+      "step": 166300
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.809428571428572e-05,
+      "loss": 1.462,
+      "step": 166400
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.807387755102041e-05,
+      "loss": 1.466,
+      "step": 166500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.805346938775511e-05,
+      "loss": 1.4559,
+      "step": 166600
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.80330612244898e-05,
+      "loss": 1.4558,
+      "step": 166700
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.80126530612245e-05,
+      "loss": 1.4545,
+      "step": 166800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.79922448979592e-05,
+      "loss": 1.4587,
+      "step": 166900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.797183673469388e-05,
+      "loss": 1.4619,
+      "step": 167000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.795142857142858e-05,
+      "loss": 1.4557,
+      "step": 167100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.793102040816328e-05,
+      "loss": 1.4665,
+      "step": 167200
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.791061224489796e-05,
+      "loss": 1.466,
+      "step": 167300
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.789020408163266e-05,
+      "loss": 1.4621,
+      "step": 167400
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.786979591836736e-05,
+      "loss": 1.4557,
+      "step": 167500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.784938775510204e-05,
+      "loss": 1.4589,
+      "step": 167600
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.782897959183674e-05,
+      "loss": 1.4491,
+      "step": 167700
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.780857142857144e-05,
+      "loss": 1.4534,
+      "step": 167800
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.778836734693878e-05,
+      "loss": 1.4545,
+      "step": 167900
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.776795918367346e-05,
+      "loss": 1.4624,
+      "step": 168000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.774755102040816e-05,
+      "loss": 1.4559,
+      "step": 168100
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.772714285714286e-05,
+      "loss": 1.459,
+      "step": 168200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.770673469387755e-05,
+      "loss": 1.4538,
+      "step": 168300
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.768632653061224e-05,
+      "loss": 1.4577,
+      "step": 168400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.766591836734694e-05,
+      "loss": 1.4527,
+      "step": 168500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.764551020408164e-05,
+      "loss": 1.4545,
+      "step": 168600
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.762510204081634e-05,
+      "loss": 1.4587,
+      "step": 168700
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.760469387755102e-05,
+      "loss": 1.4558,
+      "step": 168800
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.758428571428572e-05,
+      "loss": 1.4493,
+      "step": 168900
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.756408163265306e-05,
+      "loss": 1.4554,
+      "step": 169000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.754367346938776e-05,
+      "loss": 1.4521,
+      "step": 169100
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.752326530612245e-05,
+      "loss": 1.4568,
+      "step": 169200
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.750285714285714e-05,
+      "loss": 1.4459,
+      "step": 169300
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.748244897959184e-05,
+      "loss": 1.4529,
+      "step": 169400
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.746204081632653e-05,
+      "loss": 1.4647,
+      "step": 169500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.744163265306123e-05,
+      "loss": 1.4514,
+      "step": 169600
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.742122448979592e-05,
+      "loss": 1.4546,
+      "step": 169700
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.740081632653061e-05,
+      "loss": 1.4522,
+      "step": 169800
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.738040816326531e-05,
+      "loss": 1.4545,
+      "step": 169900
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.736e-05,
+      "loss": 1.4591,
+      "step": 170000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.733959183673469e-05,
+      "loss": 1.4543,
+      "step": 170100
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.731918367346939e-05,
+      "loss": 1.451,
+      "step": 170200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.729877551020409e-05,
+      "loss": 1.4572,
+      "step": 170300
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.727836734693877e-05,
+      "loss": 1.4536,
+      "step": 170400
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.725795918367347e-05,
+      "loss": 1.4485,
+      "step": 170500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.723755102040817e-05,
+      "loss": 1.4536,
+      "step": 170600
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.721714285714285e-05,
+      "loss": 1.4581,
+      "step": 170700
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.719673469387755e-05,
+      "loss": 1.4488,
+      "step": 170800
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.717632653061225e-05,
+      "loss": 1.4555,
+      "step": 170900
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.715612244897959e-05,
+      "loss": 1.4499,
+      "step": 171000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.713571428571429e-05,
+      "loss": 1.4531,
+      "step": 171100
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.711530612244899e-05,
+      "loss": 1.4526,
+      "step": 171200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.709489795918367e-05,
+      "loss": 1.4559,
+      "step": 171300
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.707448979591837e-05,
+      "loss": 1.4425,
+      "step": 171400
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.705408163265307e-05,
+      "loss": 1.4522,
+      "step": 171500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.703367346938775e-05,
+      "loss": 1.4468,
+      "step": 171600
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.701326530612245e-05,
+      "loss": 1.4513,
+      "step": 171700
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.699285714285715e-05,
+      "loss": 1.4513,
+      "step": 171800
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.697244897959184e-05,
+      "loss": 1.4582,
+      "step": 171900
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.695204081632653e-05,
+      "loss": 1.4562,
+      "step": 172000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.693163265306123e-05,
+      "loss": 1.4456,
+      "step": 172100
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.691122448979592e-05,
+      "loss": 1.4475,
+      "step": 172200
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.689081632653062e-05,
+      "loss": 1.4458,
+      "step": 172300
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.687040816326531e-05,
+      "loss": 1.444,
+      "step": 172400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.685e-05,
+      "loss": 1.4507,
+      "step": 172500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.68295918367347e-05,
+      "loss": 1.451,
+      "step": 172600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.68091836734694e-05,
+      "loss": 1.4528,
+      "step": 172700
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.678877551020408e-05,
+      "loss": 1.4532,
+      "step": 172800
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.676836734693878e-05,
+      "loss": 1.4425,
+      "step": 172900
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.674795918367346e-05,
+      "loss": 1.4472,
+      "step": 173000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.672755102040816e-05,
+      "loss": 1.4475,
+      "step": 173100
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.670714285714286e-05,
+      "loss": 1.4531,
+      "step": 173200
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.668673469387754e-05,
+      "loss": 1.4475,
+      "step": 173300
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.666632653061224e-05,
+      "loss": 1.4532,
+      "step": 173400
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.66461224489796e-05,
+      "loss": 1.461,
+      "step": 173500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.66257142857143e-05,
+      "loss": 1.4396,
+      "step": 173600
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.660530612244898e-05,
+      "loss": 1.4489,
+      "step": 173700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.658489795918368e-05,
+      "loss": 1.4457,
+      "step": 173800
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.656448979591838e-05,
+      "loss": 1.4496,
+      "step": 173900
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.654408163265306e-05,
+      "loss": 1.4456,
+      "step": 174000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.652367346938776e-05,
+      "loss": 1.4436,
+      "step": 174100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.650326530612246e-05,
+      "loss": 1.4414,
+      "step": 174200
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.648285714285714e-05,
+      "loss": 1.451,
+      "step": 174300
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.646244897959184e-05,
+      "loss": 1.4498,
+      "step": 174400
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.644204081632654e-05,
+      "loss": 1.4499,
+      "step": 174500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.642163265306123e-05,
+      "loss": 1.4449,
+      "step": 174600
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.640122448979592e-05,
+      "loss": 1.4533,
+      "step": 174700
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.638081632653061e-05,
+      "loss": 1.4513,
+      "step": 174800
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.636040816326531e-05,
+      "loss": 1.4592,
+      "step": 174900
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.634e-05,
+      "loss": 1.4425,
+      "step": 175000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.631959183673469e-05,
+      "loss": 1.4557,
+      "step": 175100
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.629918367346939e-05,
+      "loss": 1.4598,
+      "step": 175200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.627877551020409e-05,
+      "loss": 1.4538,
+      "step": 175300
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.625836734693877e-05,
+      "loss": 1.4389,
+      "step": 175400
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.623795918367347e-05,
+      "loss": 1.4462,
+      "step": 175500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.621755102040817e-05,
+      "loss": 1.446,
+      "step": 175600
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.619714285714285e-05,
+      "loss": 1.4461,
+      "step": 175700
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.617673469387755e-05,
+      "loss": 1.4489,
+      "step": 175800
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.615632653061225e-05,
+      "loss": 1.4474,
+      "step": 175900
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.613591836734693e-05,
+      "loss": 1.444,
+      "step": 176000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.611551020408163e-05,
+      "loss": 1.4533,
+      "step": 176100
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.609530612244899e-05,
+      "loss": 1.4454,
+      "step": 176200
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.607489795918369e-05,
+      "loss": 1.4361,
+      "step": 176300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.605448979591837e-05,
+      "loss": 1.4436,
+      "step": 176400
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.603428571428571e-05,
+      "loss": 1.4406,
+      "step": 176500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.601387755102041e-05,
+      "loss": 1.4432,
+      "step": 176600
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.599346938775511e-05,
+      "loss": 1.4469,
+      "step": 176700
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.597306122448979e-05,
+      "loss": 1.4446,
+      "step": 176800
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.595265306122449e-05,
+      "loss": 1.442,
+      "step": 176900
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.593224489795919e-05,
+      "loss": 1.4429,
+      "step": 177000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.591183673469387e-05,
+      "loss": 1.4458,
+      "step": 177100
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.589142857142857e-05,
+      "loss": 1.4481,
+      "step": 177200
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.587102040816326e-05,
+      "loss": 1.4408,
+      "step": 177300
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.585061224489796e-05,
+      "loss": 1.4464,
+      "step": 177400
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.583020408163265e-05,
+      "loss": 1.4403,
+      "step": 177500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.580979591836734e-05,
+      "loss": 1.4466,
+      "step": 177600
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.578938775510204e-05,
+      "loss": 1.4415,
+      "step": 177700
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.576897959183674e-05,
+      "loss": 1.4466,
+      "step": 177800
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.574857142857143e-05,
+      "loss": 1.4457,
+      "step": 177900
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.572816326530613e-05,
+      "loss": 1.4421,
+      "step": 178000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.570775510204083e-05,
+      "loss": 1.4445,
+      "step": 178100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.568734693877552e-05,
+      "loss": 1.4483,
+      "step": 178200
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.566693877551021e-05,
+      "loss": 1.4496,
+      "step": 178300
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.56465306122449e-05,
+      "loss": 1.4471,
+      "step": 178400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.56261224489796e-05,
+      "loss": 1.4483,
+      "step": 178500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.56057142857143e-05,
+      "loss": 1.4459,
+      "step": 178600
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.558530612244898e-05,
+      "loss": 1.4456,
+      "step": 178700
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.556489795918368e-05,
+      "loss": 1.4442,
+      "step": 178800
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.554448979591838e-05,
+      "loss": 1.4451,
+      "step": 178900
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.552408163265306e-05,
+      "loss": 1.4377,
+      "step": 179000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.550367346938776e-05,
+      "loss": 1.4498,
+      "step": 179100
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.548326530612246e-05,
+      "loss": 1.439,
+      "step": 179200
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.546285714285714e-05,
+      "loss": 1.4383,
+      "step": 179300
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.544244897959184e-05,
+      "loss": 1.4344,
+      "step": 179400
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.542204081632654e-05,
+      "loss": 1.4415,
+      "step": 179500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.540163265306122e-05,
+      "loss": 1.443,
+      "step": 179600
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.538122448979592e-05,
+      "loss": 1.4432,
+      "step": 179700
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.536081632653062e-05,
+      "loss": 1.4444,
+      "step": 179800
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.53404081632653e-05,
+      "loss": 1.4433,
+      "step": 179900
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.532e-05,
+      "loss": 1.4456,
+      "step": 180000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.52995918367347e-05,
+      "loss": 1.4504,
+      "step": 180100
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.527918367346939e-05,
+      "loss": 1.4386,
+      "step": 180200
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.525877551020409e-05,
+      "loss": 1.4451,
+      "step": 180300
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.523836734693878e-05,
+      "loss": 1.4381,
+      "step": 180400
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.521795918367347e-05,
+      "loss": 1.4386,
+      "step": 180500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.519755102040817e-05,
+      "loss": 1.4445,
+      "step": 180600
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.517734693877552e-05,
+      "loss": 1.4447,
+      "step": 180700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.515693877551021e-05,
+      "loss": 1.4458,
+      "step": 180800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.51365306122449e-05,
+      "loss": 1.4426,
+      "step": 180900
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.51161224489796e-05,
+      "loss": 1.4437,
+      "step": 181000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.509571428571429e-05,
+      "loss": 1.436,
+      "step": 181100
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.507530612244899e-05,
+      "loss": 1.4431,
+      "step": 181200
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.505489795918369e-05,
+      "loss": 1.4411,
+      "step": 181300
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.503448979591837e-05,
+      "loss": 1.4425,
+      "step": 181400
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.501408163265307e-05,
+      "loss": 1.442,
+      "step": 181500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.499367346938777e-05,
+      "loss": 1.4388,
+      "step": 181600
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.497326530612245e-05,
+      "loss": 1.4404,
+      "step": 181700
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.495285714285715e-05,
+      "loss": 1.4486,
+      "step": 181800
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.493244897959185e-05,
+      "loss": 1.4376,
+      "step": 181900
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.491204081632653e-05,
+      "loss": 1.4356,
+      "step": 182000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.489163265306123e-05,
+      "loss": 1.4375,
+      "step": 182100
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.487122448979593e-05,
+      "loss": 1.4396,
+      "step": 182200
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.485081632653061e-05,
+      "loss": 1.4366,
+      "step": 182300
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.483040816326531e-05,
+      "loss": 1.44,
+      "step": 182400
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.481e-05,
+      "loss": 1.4448,
+      "step": 182500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.47895918367347e-05,
+      "loss": 1.4372,
+      "step": 182600
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.47691836734694e-05,
+      "loss": 1.4427,
+      "step": 182700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.474877551020408e-05,
+      "loss": 1.4434,
+      "step": 182800
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.472836734693878e-05,
+      "loss": 1.4332,
+      "step": 182900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.470795918367348e-05,
+      "loss": 1.4402,
+      "step": 183000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.468755102040816e-05,
+      "loss": 1.4364,
+      "step": 183100
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.466714285714286e-05,
+      "loss": 1.4406,
+      "step": 183200
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.464673469387756e-05,
+      "loss": 1.4308,
+      "step": 183300
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.462632653061224e-05,
+      "loss": 1.4375,
+      "step": 183400
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.460591836734694e-05,
+      "loss": 1.4368,
+      "step": 183500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.458551020408164e-05,
+      "loss": 1.4438,
+      "step": 183600
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.456510204081632e-05,
+      "loss": 1.4479,
+      "step": 183700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.454469387755102e-05,
+      "loss": 1.4405,
+      "step": 183800
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.452428571428572e-05,
+      "loss": 1.4366,
+      "step": 183900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.45038775510204e-05,
+      "loss": 1.4372,
+      "step": 184000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.44834693877551e-05,
+      "loss": 1.4367,
+      "step": 184100
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.446326530612246e-05,
+      "loss": 1.4404,
+      "step": 184200
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.444285714285714e-05,
+      "loss": 1.4326,
+      "step": 184300
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.442244897959184e-05,
+      "loss": 1.4479,
+      "step": 184400
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.440204081632654e-05,
+      "loss": 1.4353,
+      "step": 184500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.438163265306122e-05,
+      "loss": 1.4333,
+      "step": 184600
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.436122448979592e-05,
+      "loss": 1.4371,
+      "step": 184700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.434081632653062e-05,
+      "loss": 1.4392,
+      "step": 184800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.43204081632653e-05,
+      "loss": 1.4349,
+      "step": 184900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.43e-05,
+      "loss": 1.4387,
+      "step": 185000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.42795918367347e-05,
+      "loss": 1.4401,
+      "step": 185100
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.425918367346939e-05,
+      "loss": 1.4369,
+      "step": 185200
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.423877551020409e-05,
+      "loss": 1.4377,
+      "step": 185300
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.421836734693878e-05,
+      "loss": 1.4337,
+      "step": 185400
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.419795918367347e-05,
+      "loss": 1.4405,
+      "step": 185500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.417755102040817e-05,
+      "loss": 1.4305,
+      "step": 185600
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.415714285714287e-05,
+      "loss": 1.4364,
+      "step": 185700
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.413673469387755e-05,
+      "loss": 1.4351,
+      "step": 185800
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.411632653061225e-05,
+      "loss": 1.4293,
+      "step": 185900
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.409591836734695e-05,
+      "loss": 1.4342,
+      "step": 186000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.407551020408163e-05,
+      "loss": 1.4455,
+      "step": 186100
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.405530612244899e-05,
+      "loss": 1.4377,
+      "step": 186200
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.403489795918368e-05,
+      "loss": 1.4333,
+      "step": 186300
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.401448979591837e-05,
+      "loss": 1.4367,
+      "step": 186400
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.399408163265307e-05,
+      "loss": 1.434,
+      "step": 186500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.397367346938777e-05,
+      "loss": 1.4362,
+      "step": 186600
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.395326530612245e-05,
+      "loss": 1.4346,
+      "step": 186700
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.393285714285715e-05,
+      "loss": 1.4307,
+      "step": 186800
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.391244897959185e-05,
+      "loss": 1.4335,
+      "step": 186900
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.389204081632653e-05,
+      "loss": 1.4326,
+      "step": 187000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.387163265306123e-05,
+      "loss": 1.4343,
+      "step": 187100
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.385122448979593e-05,
+      "loss": 1.4377,
+      "step": 187200
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.383081632653061e-05,
+      "loss": 1.4354,
+      "step": 187300
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.381040816326531e-05,
+      "loss": 1.4357,
+      "step": 187400
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.379000000000001e-05,
+      "loss": 1.4269,
+      "step": 187500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.37695918367347e-05,
+      "loss": 1.4449,
+      "step": 187600
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.37491836734694e-05,
+      "loss": 1.4258,
+      "step": 187700
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.372877551020409e-05,
+      "loss": 1.4349,
+      "step": 187800
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.370836734693878e-05,
+      "loss": 1.4298,
+      "step": 187900
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.368795918367348e-05,
+      "loss": 1.4251,
+      "step": 188000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.366755102040816e-05,
+      "loss": 1.4376,
+      "step": 188100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.364714285714286e-05,
+      "loss": 1.4268,
+      "step": 188200
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.362673469387756e-05,
+      "loss": 1.4285,
+      "step": 188300
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.360632653061224e-05,
+      "loss": 1.4305,
+      "step": 188400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.358591836734694e-05,
+      "loss": 1.4288,
+      "step": 188500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.356551020408164e-05,
+      "loss": 1.4387,
+      "step": 188600
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.354510204081632e-05,
+      "loss": 1.4235,
+      "step": 188700
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.352469387755102e-05,
+      "loss": 1.4333,
+      "step": 188800
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.350448979591836e-05,
+      "loss": 1.4287,
+      "step": 188900
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.348408163265307e-05,
+      "loss": 1.4286,
+      "step": 189000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.346367346938776e-05,
+      "loss": 1.4227,
+      "step": 189100
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.344326530612246e-05,
+      "loss": 1.4155,
+      "step": 189200
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.342285714285716e-05,
+      "loss": 1.4221,
+      "step": 189300
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.340244897959184e-05,
+      "loss": 1.4314,
+      "step": 189400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.338204081632654e-05,
+      "loss": 1.4283,
+      "step": 189500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.336163265306124e-05,
+      "loss": 1.4354,
+      "step": 189600
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.334122448979592e-05,
+      "loss": 1.4281,
+      "step": 189700
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.332081632653062e-05,
+      "loss": 1.4304,
+      "step": 189800
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.33004081632653e-05,
+      "loss": 1.4311,
+      "step": 189900
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.328e-05,
+      "loss": 1.4265,
+      "step": 190000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.32595918367347e-05,
+      "loss": 1.4252,
+      "step": 190100
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.323918367346939e-05,
+      "loss": 1.4302,
+      "step": 190200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.321877551020409e-05,
+      "loss": 1.4273,
+      "step": 190300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.319836734693878e-05,
+      "loss": 1.431,
+      "step": 190400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.317795918367347e-05,
+      "loss": 1.4249,
+      "step": 190500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.315755102040817e-05,
+      "loss": 1.4218,
+      "step": 190600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.313714285714287e-05,
+      "loss": 1.423,
+      "step": 190700
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.311673469387755e-05,
+      "loss": 1.4276,
+      "step": 190800
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.309632653061225e-05,
+      "loss": 1.4213,
+      "step": 190900
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.307591836734695e-05,
+      "loss": 1.4147,
+      "step": 191000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.305551020408163e-05,
+      "loss": 1.4253,
+      "step": 191100
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.303510204081633e-05,
+      "loss": 1.425,
+      "step": 191200
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.301469387755103e-05,
+      "loss": 1.4224,
+      "step": 191300
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.299428571428571e-05,
+      "loss": 1.4329,
+      "step": 191400
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.297387755102041e-05,
+      "loss": 1.4255,
+      "step": 191500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.295346938775511e-05,
+      "loss": 1.4216,
+      "step": 191600
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.29330612244898e-05,
+      "loss": 1.4159,
+      "step": 191700
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.291265306122449e-05,
+      "loss": 1.4251,
+      "step": 191800
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.289244897959185e-05,
+      "loss": 1.4245,
+      "step": 191900
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.287204081632653e-05,
+      "loss": 1.4248,
+      "step": 192000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.285163265306123e-05,
+      "loss": 1.4188,
+      "step": 192100
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.283122448979593e-05,
+      "loss": 1.4197,
+      "step": 192200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.281081632653061e-05,
+      "loss": 1.4295,
+      "step": 192300
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.279040816326531e-05,
+      "loss": 1.426,
+      "step": 192400
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.277000000000001e-05,
+      "loss": 1.4273,
+      "step": 192500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.27495918367347e-05,
+      "loss": 1.4299,
+      "step": 192600
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.27291836734694e-05,
+      "loss": 1.4154,
+      "step": 192700
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.270877551020409e-05,
+      "loss": 1.4249,
+      "step": 192800
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.268836734693878e-05,
+      "loss": 1.4183,
+      "step": 192900
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.266795918367348e-05,
+      "loss": 1.4252,
+      "step": 193000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.264755102040817e-05,
+      "loss": 1.4217,
+      "step": 193100
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.262734693877551e-05,
+      "loss": 1.4244,
+      "step": 193200
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.26069387755102e-05,
+      "loss": 1.4244,
+      "step": 193300
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.25865306122449e-05,
+      "loss": 1.4235,
+      "step": 193400
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.25661224489796e-05,
+      "loss": 1.4225,
+      "step": 193500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.254571428571428e-05,
+      "loss": 1.4244,
+      "step": 193600
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.252530612244898e-05,
+      "loss": 1.4259,
+      "step": 193700
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.250489795918368e-05,
+      "loss": 1.4301,
+      "step": 193800
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.248448979591836e-05,
+      "loss": 1.4201,
+      "step": 193900
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.246408163265306e-05,
+      "loss": 1.4182,
+      "step": 194000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.244367346938776e-05,
+      "loss": 1.4292,
+      "step": 194100
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.242326530612244e-05,
+      "loss": 1.4224,
+      "step": 194200
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.240285714285714e-05,
+      "loss": 1.4178,
+      "step": 194300
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.238244897959184e-05,
+      "loss": 1.4167,
+      "step": 194400
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.236204081632654e-05,
+      "loss": 1.4292,
+      "step": 194500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.234163265306124e-05,
+      "loss": 1.4121,
+      "step": 194600
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.232122448979592e-05,
+      "loss": 1.4211,
+      "step": 194700
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.230081632653062e-05,
+      "loss": 1.4181,
+      "step": 194800
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.228040816326532e-05,
+      "loss": 1.422,
+      "step": 194900
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.226e-05,
+      "loss": 1.4266,
+      "step": 195000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.22395918367347e-05,
+      "loss": 1.4206,
+      "step": 195100
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.22191836734694e-05,
+      "loss": 1.4197,
+      "step": 195200
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.219877551020409e-05,
+      "loss": 1.4274,
+      "step": 195300
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.217836734693878e-05,
+      "loss": 1.4238,
+      "step": 195400
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.215795918367348e-05,
+      "loss": 1.4234,
+      "step": 195500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.213755102040817e-05,
+      "loss": 1.4275,
+      "step": 195600
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.211714285714287e-05,
+      "loss": 1.4313,
+      "step": 195700
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.209673469387755e-05,
+      "loss": 1.4172,
+      "step": 195800
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.207632653061225e-05,
+      "loss": 1.4196,
+      "step": 195900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.205591836734695e-05,
+      "loss": 1.4224,
+      "step": 196000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.203551020408163e-05,
+      "loss": 1.424,
+      "step": 196100
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.201510204081633e-05,
+      "loss": 1.4144,
+      "step": 196200
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.199469387755103e-05,
+      "loss": 1.4276,
+      "step": 196300
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.197428571428571e-05,
+      "loss": 1.427,
+      "step": 196400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.195387755102041e-05,
+      "loss": 1.4167,
+      "step": 196500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.193346938775511e-05,
+      "loss": 1.4221,
+      "step": 196600
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.19130612244898e-05,
+      "loss": 1.4281,
+      "step": 196700
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.189265306122449e-05,
+      "loss": 1.4188,
+      "step": 196800
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.187224489795919e-05,
+      "loss": 1.4206,
+      "step": 196900
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.185183673469388e-05,
+      "loss": 1.4125,
+      "step": 197000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.183142857142857e-05,
+      "loss": 1.4118,
+      "step": 197100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.181122448979592e-05,
+      "loss": 1.4289,
+      "step": 197200
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.179081632653063e-05,
+      "loss": 1.4164,
+      "step": 197300
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.177040816326531e-05,
+      "loss": 1.4257,
+      "step": 197400
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.175000000000001e-05,
+      "loss": 1.4117,
+      "step": 197500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.17295918367347e-05,
+      "loss": 1.4185,
+      "step": 197600
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.17091836734694e-05,
+      "loss": 1.4199,
+      "step": 197700
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.168877551020409e-05,
+      "loss": 1.4126,
+      "step": 197800
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.166836734693878e-05,
+      "loss": 1.4199,
+      "step": 197900
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.164795918367347e-05,
+      "loss": 1.4156,
+      "step": 198000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.162755102040817e-05,
+      "loss": 1.4157,
+      "step": 198100
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.160714285714286e-05,
+      "loss": 1.421,
+      "step": 198200
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.158673469387756e-05,
+      "loss": 1.4257,
+      "step": 198300
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.156632653061225e-05,
+      "loss": 1.4168,
+      "step": 198400
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.154591836734694e-05,
+      "loss": 1.4145,
+      "step": 198500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.152551020408164e-05,
+      "loss": 1.419,
+      "step": 198600
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.150510204081634e-05,
+      "loss": 1.4159,
+      "step": 198700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.148469387755102e-05,
+      "loss": 1.4189,
+      "step": 198800
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.146428571428572e-05,
+      "loss": 1.4137,
+      "step": 198900
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.144387755102042e-05,
+      "loss": 1.4187,
+      "step": 199000
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.14234693877551e-05,
+      "loss": 1.4223,
+      "step": 199100
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.14030612244898e-05,
+      "loss": 1.4292,
+      "step": 199200
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.13826530612245e-05,
+      "loss": 1.4133,
+      "step": 199300
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.136224489795918e-05,
+      "loss": 1.4122,
+      "step": 199400
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.134183673469388e-05,
+      "loss": 1.4174,
+      "step": 199500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.132142857142858e-05,
+      "loss": 1.4114,
+      "step": 199600
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.130102040816327e-05,
+      "loss": 1.4132,
+      "step": 199700
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.128061224489796e-05,
+      "loss": 1.4167,
+      "step": 199800
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.126020408163265e-05,
+      "loss": 1.4112,
+      "step": 199900
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.123979591836735e-05,
+      "loss": 1.4204,
+      "step": 200000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.121938775510205e-05,
+      "loss": 1.4199,
+      "step": 200100
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.119897959183673e-05,
+      "loss": 1.4143,
+      "step": 200200
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.117857142857143e-05,
+      "loss": 1.4089,
+      "step": 200300
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.115816326530613e-05,
+      "loss": 1.4135,
+      "step": 200400
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.113775510204081e-05,
+      "loss": 1.4137,
+      "step": 200500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.111734693877551e-05,
+      "loss": 1.4073,
+      "step": 200600
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.109693877551021e-05,
+      "loss": 1.4108,
+      "step": 200700
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.10765306122449e-05,
+      "loss": 1.4244,
+      "step": 200800
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.105612244897959e-05,
+      "loss": 1.4208,
+      "step": 200900
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.103571428571429e-05,
+      "loss": 1.4136,
+      "step": 201000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.101530612244898e-05,
+      "loss": 1.4216,
+      "step": 201100
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.099489795918367e-05,
+      "loss": 1.4109,
+      "step": 201200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.0974489795918365e-05,
+      "loss": 1.4162,
+      "step": 201300
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.0954081632653056e-05,
+      "loss": 1.4099,
+      "step": 201400
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.0933673469387755e-05,
+      "loss": 1.4188,
+      "step": 201500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.0913265306122446e-05,
+      "loss": 1.4116,
+      "step": 201600
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.089285714285714e-05,
+      "loss": 1.4163,
+      "step": 201700
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.087244897959184e-05,
+      "loss": 1.4141,
+      "step": 201800
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.085204081632654e-05,
+      "loss": 1.4135,
+      "step": 201900
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.083163265306123e-05,
+      "loss": 1.4166,
+      "step": 202000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.0811224489795925e-05,
+      "loss": 1.4144,
+      "step": 202100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.079081632653062e-05,
+      "loss": 1.4151,
+      "step": 202200
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.0770408163265315e-05,
+      "loss": 1.4212,
+      "step": 202300
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.0750000000000006e-05,
+      "loss": 1.408,
+      "step": 202400
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.072979591836735e-05,
+      "loss": 1.4142,
+      "step": 202500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.0709387755102045e-05,
+      "loss": 1.4018,
+      "step": 202600
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.068897959183674e-05,
+      "loss": 1.4106,
+      "step": 202700
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.066857142857143e-05,
+      "loss": 1.4162,
+      "step": 202800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.064816326530613e-05,
+      "loss": 1.4122,
+      "step": 202900
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.062775510204082e-05,
+      "loss": 1.4079,
+      "step": 203000
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.060734693877551e-05,
+      "loss": 1.4124,
+      "step": 203100
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.05869387755102e-05,
+      "loss": 1.4136,
+      "step": 203200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.05665306122449e-05,
+      "loss": 1.4119,
+      "step": 203300
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.0546326530612254e-05,
+      "loss": 1.415,
+      "step": 203400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.0525918367346946e-05,
+      "loss": 1.4218,
+      "step": 203500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.050551020408164e-05,
+      "loss": 1.4143,
+      "step": 203600
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.0485102040816336e-05,
+      "loss": 1.4201,
+      "step": 203700
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.046469387755103e-05,
+      "loss": 1.4185,
+      "step": 203800
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.044428571428572e-05,
+      "loss": 1.4195,
+      "step": 203900
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.042387755102041e-05,
+      "loss": 1.4096,
+      "step": 204000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.040346938775511e-05,
+      "loss": 1.4074,
+      "step": 204100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.03830612244898e-05,
+      "loss": 1.4086,
+      "step": 204200
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.036265306122449e-05,
+      "loss": 1.4111,
+      "step": 204300
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.034224489795919e-05,
+      "loss": 1.405,
+      "step": 204400
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.032183673469388e-05,
+      "loss": 1.4101,
+      "step": 204500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.0301428571428574e-05,
+      "loss": 1.4148,
+      "step": 204600
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.028102040816327e-05,
+      "loss": 1.4137,
+      "step": 204700
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.0260612244897964e-05,
+      "loss": 1.4115,
+      "step": 204800
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.0240204081632655e-05,
+      "loss": 1.4087,
+      "step": 204900
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.021979591836735e-05,
+      "loss": 1.4085,
+      "step": 205000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.0199387755102045e-05,
+      "loss": 1.4082,
+      "step": 205100
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.017897959183674e-05,
+      "loss": 1.4124,
+      "step": 205200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.015857142857143e-05,
+      "loss": 1.4086,
+      "step": 205300
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.013816326530613e-05,
+      "loss": 1.4181,
+      "step": 205400
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.011775510204082e-05,
+      "loss": 1.4173,
+      "step": 205500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.009734693877551e-05,
+      "loss": 1.4258,
+      "step": 205600
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.007693877551021e-05,
+      "loss": 1.4115,
+      "step": 205700
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.00565306122449e-05,
+      "loss": 1.4147,
+      "step": 205800
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.003612244897959e-05,
+      "loss": 1.4112,
+      "step": 205900
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.001571428571428e-05,
+      "loss": 1.4077,
+      "step": 206000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.999530612244898e-05,
+      "loss": 1.4079,
+      "step": 206100
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.997489795918367e-05,
+      "loss": 1.4088,
+      "step": 206200
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.9954489795918364e-05,
+      "loss": 1.4112,
+      "step": 206300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.993408163265306e-05,
+      "loss": 1.4124,
+      "step": 206400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.9913673469387754e-05,
+      "loss": 1.4074,
+      "step": 206500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.9893265306122446e-05,
+      "loss": 1.413,
+      "step": 206600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.9872857142857144e-05,
+      "loss": 1.4145,
+      "step": 206700
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.9852448979591836e-05,
+      "loss": 1.4045,
+      "step": 206800
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.983204081632653e-05,
+      "loss": 1.4172,
+      "step": 206900
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.9811632653061226e-05,
+      "loss": 1.4078,
+      "step": 207000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.979122448979592e-05,
+      "loss": 1.4117,
+      "step": 207100
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.977081632653061e-05,
+      "loss": 1.4072,
+      "step": 207200
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.97504081632653e-05,
+      "loss": 1.4051,
+      "step": 207300
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.9730000000000006e-05,
+      "loss": 1.4093,
+      "step": 207400
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.9709591836734704e-05,
+      "loss": 1.4099,
+      "step": 207500
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.9689183673469396e-05,
+      "loss": 1.4056,
+      "step": 207600
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.966877551020409e-05,
+      "loss": 1.408,
+      "step": 207700
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.9648367346938786e-05,
+      "loss": 1.4046,
+      "step": 207800
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.962795918367348e-05,
+      "loss": 1.4112,
+      "step": 207900
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.960755102040817e-05,
+      "loss": 1.4077,
+      "step": 208000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.958714285714286e-05,
+      "loss": 1.412,
+      "step": 208100
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.956673469387756e-05,
+      "loss": 1.408,
+      "step": 208200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.954632653061225e-05,
+      "loss": 1.4075,
+      "step": 208300
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.952591836734694e-05,
+      "loss": 1.404,
+      "step": 208400
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.950551020408164e-05,
+      "loss": 1.4147,
+      "step": 208500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.948530612244898e-05,
+      "loss": 1.4089,
+      "step": 208600
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.946489795918367e-05,
+      "loss": 1.417,
+      "step": 208700
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.944448979591837e-05,
+      "loss": 1.4023,
+      "step": 208800
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.942408163265306e-05,
+      "loss": 1.4093,
+      "step": 208900
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.9403673469387754e-05,
+      "loss": 1.4139,
+      "step": 209000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.9383265306122446e-05,
+      "loss": 1.4026,
+      "step": 209100
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.9362857142857144e-05,
+      "loss": 1.4187,
+      "step": 209200
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.9342448979591836e-05,
+      "loss": 1.4034,
+      "step": 209300
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.932204081632653e-05,
+      "loss": 1.4013,
+      "step": 209400
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.930183673469388e-05,
+      "loss": 1.3987,
+      "step": 209500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.928142857142857e-05,
+      "loss": 1.4042,
+      "step": 209600
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.926102040816327e-05,
+      "loss": 1.4077,
+      "step": 209700
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.924061224489796e-05,
+      "loss": 1.4009,
+      "step": 209800
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.9220204081632655e-05,
+      "loss": 1.4147,
+      "step": 209900
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.919979591836735e-05,
+      "loss": 1.4082,
+      "step": 210000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.9179387755102045e-05,
+      "loss": 1.4001,
+      "step": 210100
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.9158979591836736e-05,
+      "loss": 1.4195,
+      "step": 210200
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.9138571428571435e-05,
+      "loss": 1.4051,
+      "step": 210300
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.9118163265306126e-05,
+      "loss": 1.409,
+      "step": 210400
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.909775510204082e-05,
+      "loss": 1.4046,
+      "step": 210500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.9077346938775516e-05,
+      "loss": 1.4012,
+      "step": 210600
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.905693877551021e-05,
+      "loss": 1.4025,
+      "step": 210700
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.90365306122449e-05,
+      "loss": 1.408,
+      "step": 210800
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.901612244897959e-05,
+      "loss": 1.4045,
+      "step": 210900
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.899571428571429e-05,
+      "loss": 1.4053,
+      "step": 211000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.897530612244898e-05,
+      "loss": 1.4055,
+      "step": 211100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.895489795918367e-05,
+      "loss": 1.3999,
+      "step": 211200
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.893448979591837e-05,
+      "loss": 1.4076,
+      "step": 211300
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.891408163265306e-05,
+      "loss": 1.3992,
+      "step": 211400
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.8893673469387754e-05,
+      "loss": 1.4015,
+      "step": 211500
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.887326530612245e-05,
+      "loss": 1.404,
+      "step": 211600
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.8852857142857144e-05,
+      "loss": 1.4057,
+      "step": 211700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.8832448979591836e-05,
+      "loss": 1.4067,
+      "step": 211800
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.881204081632653e-05,
+      "loss": 1.4045,
+      "step": 211900
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.8791632653061226e-05,
+      "loss": 1.4099,
+      "step": 212000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.877122448979592e-05,
+      "loss": 1.405,
+      "step": 212100
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.875081632653061e-05,
+      "loss": 1.4077,
+      "step": 212200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.873040816326531e-05,
+      "loss": 1.4061,
+      "step": 212300
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.871e-05,
+      "loss": 1.4098,
+      "step": 212400
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.868959183673469e-05,
+      "loss": 1.4043,
+      "step": 212500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.866918367346939e-05,
+      "loss": 1.3997,
+      "step": 212600
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.864877551020408e-05,
+      "loss": 1.4013,
+      "step": 212700
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.862836734693877e-05,
+      "loss": 1.401,
+      "step": 212800
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.860795918367348e-05,
+      "loss": 1.4035,
+      "step": 212900
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.858755102040817e-05,
+      "loss": 1.4077,
+      "step": 213000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.856714285714287e-05,
+      "loss": 1.411,
+      "step": 213100
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.854673469387756e-05,
+      "loss": 1.3997,
+      "step": 213200
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.852632653061225e-05,
+      "loss": 1.4058,
+      "step": 213300
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.850591836734695e-05,
+      "loss": 1.4071,
+      "step": 213400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.848551020408164e-05,
+      "loss": 1.4006,
+      "step": 213500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.846510204081633e-05,
+      "loss": 1.4025,
+      "step": 213600
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.844469387755103e-05,
+      "loss": 1.4034,
+      "step": 213700
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.842428571428572e-05,
+      "loss": 1.3961,
+      "step": 213800
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.840408163265306e-05,
+      "loss": 1.3938,
+      "step": 213900
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.8383673469387754e-05,
+      "loss": 1.404,
+      "step": 214000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.836326530612245e-05,
+      "loss": 1.4045,
+      "step": 214100
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.8342857142857144e-05,
+      "loss": 1.4049,
+      "step": 214200
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.8322448979591835e-05,
+      "loss": 1.4032,
+      "step": 214300
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.8302040816326534e-05,
+      "loss": 1.4043,
+      "step": 214400
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.8281632653061225e-05,
+      "loss": 1.4072,
+      "step": 214500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.826122448979592e-05,
+      "loss": 1.3978,
+      "step": 214600
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.824081632653061e-05,
+      "loss": 1.4076,
+      "step": 214700
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.822040816326531e-05,
+      "loss": 1.392,
+      "step": 214800
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.82e-05,
+      "loss": 1.3962,
+      "step": 214900
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.817959183673469e-05,
+      "loss": 1.3993,
+      "step": 215000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.815918367346939e-05,
+      "loss": 1.3982,
+      "step": 215100
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.813877551020408e-05,
+      "loss": 1.3922,
+      "step": 215200
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.811836734693877e-05,
+      "loss": 1.397,
+      "step": 215300
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.809795918367347e-05,
+      "loss": 1.4086,
+      "step": 215400
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.807755102040816e-05,
+      "loss": 1.4039,
+      "step": 215500
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.805714285714285e-05,
+      "loss": 1.3971,
+      "step": 215600
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.803673469387756e-05,
+      "loss": 1.3937,
+      "step": 215700
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.801632653061225e-05,
+      "loss": 1.3922,
+      "step": 215800
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.799591836734695e-05,
+      "loss": 1.3989,
+      "step": 215900
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.797551020408164e-05,
+      "loss": 1.4012,
+      "step": 216000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.795510204081633e-05,
+      "loss": 1.4098,
+      "step": 216100
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.793469387755103e-05,
+      "loss": 1.4027,
+      "step": 216200
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.791428571428572e-05,
+      "loss": 1.3988,
+      "step": 216300
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.789387755102041e-05,
+      "loss": 1.4035,
+      "step": 216400
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.787346938775511e-05,
+      "loss": 1.3945,
+      "step": 216500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.78530612244898e-05,
+      "loss": 1.3958,
+      "step": 216600
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.7832653061224494e-05,
+      "loss": 1.4012,
+      "step": 216700
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.7812244897959186e-05,
+      "loss": 1.3956,
+      "step": 216800
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.7791836734693884e-05,
+      "loss": 1.397,
+      "step": 216900
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.7771428571428576e-05,
+      "loss": 1.4022,
+      "step": 217000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.775102040816327e-05,
+      "loss": 1.3978,
+      "step": 217100
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.7730612244897966e-05,
+      "loss": 1.3999,
+      "step": 217200
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.771020408163266e-05,
+      "loss": 1.4066,
+      "step": 217300
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.769e-05,
+      "loss": 1.3896,
+      "step": 217400
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.7669591836734697e-05,
+      "loss": 1.4034,
+      "step": 217500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.764918367346939e-05,
+      "loss": 1.3997,
+      "step": 217600
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.762877551020408e-05,
+      "loss": 1.3975,
+      "step": 217700
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.760836734693877e-05,
+      "loss": 1.3996,
+      "step": 217800
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.758795918367347e-05,
+      "loss": 1.3945,
+      "step": 217900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.756755102040816e-05,
+      "loss": 1.3963,
+      "step": 218000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.754714285714285e-05,
+      "loss": 1.3992,
+      "step": 218100
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.752673469387755e-05,
+      "loss": 1.4032,
+      "step": 218200
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.750632653061224e-05,
+      "loss": 1.3951,
+      "step": 218300
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.748591836734695e-05,
+      "loss": 1.4036,
+      "step": 218400
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.746551020408164e-05,
+      "loss": 1.3971,
+      "step": 218500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.744510204081633e-05,
+      "loss": 1.4007,
+      "step": 218600
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.742469387755103e-05,
+      "loss": 1.3972,
+      "step": 218700
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.740428571428572e-05,
+      "loss": 1.4015,
+      "step": 218800
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.738387755102041e-05,
+      "loss": 1.4013,
+      "step": 218900
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.736346938775511e-05,
+      "loss": 1.3963,
+      "step": 219000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.73430612244898e-05,
+      "loss": 1.3968,
+      "step": 219100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.7322653061224494e-05,
+      "loss": 1.39,
+      "step": 219200
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.730244897959184e-05,
+      "loss": 1.3921,
+      "step": 219300
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.728204081632653e-05,
+      "loss": 1.3942,
+      "step": 219400
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.7261632653061225e-05,
+      "loss": 1.3987,
+      "step": 219500
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.7241224489795916e-05,
+      "loss": 1.3988,
+      "step": 219600
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.7220816326530615e-05,
+      "loss": 1.3955,
+      "step": 219700
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.7200408163265306e-05,
+      "loss": 1.3946,
+      "step": 219800
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.718e-05,
+      "loss": 1.4045,
+      "step": 219900
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.7159591836734696e-05,
+      "loss": 1.3956,
+      "step": 220000
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.713918367346939e-05,
+      "loss": 1.3978,
+      "step": 220100
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.711877551020408e-05,
+      "loss": 1.3982,
+      "step": 220200
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.709836734693878e-05,
+      "loss": 1.3949,
+      "step": 220300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.707795918367347e-05,
+      "loss": 1.3947,
+      "step": 220400
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.705755102040816e-05,
+      "loss": 1.4004,
+      "step": 220500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.703714285714285e-05,
+      "loss": 1.4058,
+      "step": 220600
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.701673469387755e-05,
+      "loss": 1.4008,
+      "step": 220700
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.699632653061224e-05,
+      "loss": 1.3931,
+      "step": 220800
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.6975918367346934e-05,
+      "loss": 1.3989,
+      "step": 220900
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.695551020408163e-05,
+      "loss": 1.3918,
+      "step": 221000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.6935102040816324e-05,
+      "loss": 1.3937,
+      "step": 221100
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.691469387755103e-05,
+      "loss": 1.3974,
+      "step": 221200
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.689428571428572e-05,
+      "loss": 1.3933,
+      "step": 221300
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.687387755102041e-05,
+      "loss": 1.3934,
+      "step": 221400
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.685367346938776e-05,
+      "loss": 1.3922,
+      "step": 221500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.683326530612245e-05,
+      "loss": 1.3911,
+      "step": 221600
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.681285714285714e-05,
+      "loss": 1.3908,
+      "step": 221700
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.679244897959184e-05,
+      "loss": 1.3965,
+      "step": 221800
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.677204081632653e-05,
+      "loss": 1.4021,
+      "step": 221900
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.6751632653061225e-05,
+      "loss": 1.3955,
+      "step": 222000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.673122448979592e-05,
+      "loss": 1.3934,
+      "step": 222100
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.6710816326530615e-05,
+      "loss": 1.3968,
+      "step": 222200
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.6690408163265306e-05,
+      "loss": 1.3932,
+      "step": 222300
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.667e-05,
+      "loss": 1.3976,
+      "step": 222400
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.6649591836734696e-05,
+      "loss": 1.3959,
+      "step": 222500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.662918367346939e-05,
+      "loss": 1.392,
+      "step": 222600
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.660877551020408e-05,
+      "loss": 1.3955,
+      "step": 222700
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.658836734693878e-05,
+      "loss": 1.3874,
+      "step": 222800
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.656795918367347e-05,
+      "loss": 1.3968,
+      "step": 222900
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.654755102040816e-05,
+      "loss": 1.3904,
+      "step": 223000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.652714285714286e-05,
+      "loss": 1.3912,
+      "step": 223100
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.650673469387755e-05,
+      "loss": 1.3996,
+      "step": 223200
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.648632653061224e-05,
+      "loss": 1.3897,
+      "step": 223300
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.6465918367346934e-05,
+      "loss": 1.3922,
+      "step": 223400
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.644551020408163e-05,
+      "loss": 1.4015,
+      "step": 223500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.6425102040816324e-05,
+      "loss": 1.3882,
+      "step": 223600
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.6404693877551016e-05,
+      "loss": 1.3984,
+      "step": 223700
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.6384285714285714e-05,
+      "loss": 1.3913,
+      "step": 223800
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.6363877551020405e-05,
+      "loss": 1.3874,
+      "step": 223900
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.634346938775511e-05,
+      "loss": 1.393,
+      "step": 224000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.63230612244898e-05,
+      "loss": 1.3916,
+      "step": 224100
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.63026530612245e-05,
+      "loss": 1.391,
+      "step": 224200
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.628224489795919e-05,
+      "loss": 1.3869,
+      "step": 224300
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.6261836734693884e-05,
+      "loss": 1.3942,
+      "step": 224400
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.6241428571428575e-05,
+      "loss": 1.3863,
+      "step": 224500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.6221020408163274e-05,
+      "loss": 1.3916,
+      "step": 224600
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.6200612244897965e-05,
+      "loss": 1.3891,
+      "step": 224700
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.618020408163266e-05,
+      "loss": 1.3901,
+      "step": 224800
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.6159795918367355e-05,
+      "loss": 1.3934,
+      "step": 224900
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.613938775510205e-05,
+      "loss": 1.3956,
+      "step": 225000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.611897959183674e-05,
+      "loss": 1.3877,
+      "step": 225100
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.609857142857144e-05,
+      "loss": 1.3958,
+      "step": 225200
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.607816326530613e-05,
+      "loss": 1.3907,
+      "step": 225300
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.605775510204082e-05,
+      "loss": 1.3949,
+      "step": 225400
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.603734693877551e-05,
+      "loss": 1.3919,
+      "step": 225500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.601693877551021e-05,
+      "loss": 1.3983,
+      "step": 225600
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.59965306122449e-05,
+      "loss": 1.3869,
+      "step": 225700
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.597612244897959e-05,
+      "loss": 1.3941,
+      "step": 225800
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.595571428571429e-05,
+      "loss": 1.3953,
+      "step": 225900
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.593551020408163e-05,
+      "loss": 1.3968,
+      "step": 226000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.5915102040816324e-05,
+      "loss": 1.3905,
+      "step": 226100
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.589469387755102e-05,
+      "loss": 1.3922,
+      "step": 226200
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.5874285714285714e-05,
+      "loss": 1.389,
+      "step": 226300
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.5853877551020405e-05,
+      "loss": 1.3823,
+      "step": 226400
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.58334693877551e-05,
+      "loss": 1.3936,
+      "step": 226500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.5813061224489795e-05,
+      "loss": 1.3956,
+      "step": 226600
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.57926530612245e-05,
+      "loss": 1.3901,
+      "step": 226700
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.577224489795919e-05,
+      "loss": 1.3857,
+      "step": 226800
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.5751836734693884e-05,
+      "loss": 1.3895,
+      "step": 226900
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.573142857142858e-05,
+      "loss": 1.3905,
+      "step": 227000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.5711020408163273e-05,
+      "loss": 1.3901,
+      "step": 227100
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.5690612244897965e-05,
+      "loss": 1.3913,
+      "step": 227200
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.567020408163266e-05,
+      "loss": 1.3875,
+      "step": 227300
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.5649795918367355e-05,
+      "loss": 1.382,
+      "step": 227400
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.5629387755102047e-05,
+      "loss": 1.3959,
+      "step": 227500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.560897959183674e-05,
+      "loss": 1.386,
+      "step": 227600
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.5588571428571437e-05,
+      "loss": 1.3884,
+      "step": 227700
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.556816326530613e-05,
+      "loss": 1.3885,
+      "step": 227800
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.554775510204082e-05,
+      "loss": 1.3945,
+      "step": 227900
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.552734693877552e-05,
+      "loss": 1.3819,
+      "step": 228000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.550693877551021e-05,
+      "loss": 1.3916,
+      "step": 228100
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.54865306122449e-05,
+      "loss": 1.3913,
+      "step": 228200
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.546612244897959e-05,
+      "loss": 1.3866,
+      "step": 228300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.544591836734694e-05,
+      "loss": 1.3878,
+      "step": 228400
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.542551020408163e-05,
+      "loss": 1.3955,
+      "step": 228500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5405102040816324e-05,
+      "loss": 1.3854,
+      "step": 228600
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.538469387755102e-05,
+      "loss": 1.3889,
+      "step": 228700
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5364285714285713e-05,
+      "loss": 1.3907,
+      "step": 228800
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5343877551020405e-05,
+      "loss": 1.3943,
+      "step": 228900
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.5323469387755103e-05,
+      "loss": 1.3867,
+      "step": 229000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.5303061224489795e-05,
+      "loss": 1.3864,
+      "step": 229100
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.5282653061224487e-05,
+      "loss": 1.3888,
+      "step": 229200
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.526224489795918e-05,
+      "loss": 1.3842,
+      "step": 229300
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.5241836734693877e-05,
+      "loss": 1.3886,
+      "step": 229400
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.522142857142858e-05,
+      "loss": 1.3955,
+      "step": 229500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.520122448979592e-05,
+      "loss": 1.3783,
+      "step": 229600
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.5180816326530614e-05,
+      "loss": 1.3809,
+      "step": 229700
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.516040816326531e-05,
+      "loss": 1.386,
+      "step": 229800
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.5140000000000004e-05,
+      "loss": 1.3852,
+      "step": 229900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.5119591836734696e-05,
+      "loss": 1.3826,
+      "step": 230000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.509918367346939e-05,
+      "loss": 1.3838,
+      "step": 230100
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.5078775510204086e-05,
+      "loss": 1.3899,
+      "step": 230200
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.505836734693878e-05,
+      "loss": 1.3862,
+      "step": 230300
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.503795918367347e-05,
+      "loss": 1.388,
+      "step": 230400
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.501755102040817e-05,
+      "loss": 1.3864,
+      "step": 230500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.499714285714286e-05,
+      "loss": 1.3791,
+      "step": 230600
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.497673469387755e-05,
+      "loss": 1.3915,
+      "step": 230700
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.495632653061225e-05,
+      "loss": 1.3864,
+      "step": 230800
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.493591836734694e-05,
+      "loss": 1.3748,
+      "step": 230900
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.491551020408163e-05,
+      "loss": 1.3906,
+      "step": 231000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.489510204081632e-05,
+      "loss": 1.3836,
+      "step": 231100
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.487469387755102e-05,
+      "loss": 1.3837,
+      "step": 231200
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.485428571428571e-05,
+      "loss": 1.3869,
+      "step": 231300
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.4833877551020405e-05,
+      "loss": 1.3804,
+      "step": 231400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.48134693877551e-05,
+      "loss": 1.3866,
+      "step": 231500
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.4793061224489795e-05,
+      "loss": 1.3783,
+      "step": 231600
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.4772653061224486e-05,
+      "loss": 1.3925,
+      "step": 231700
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.4752244897959185e-05,
+      "loss": 1.3832,
+      "step": 231800
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.4731836734693876e-05,
+      "loss": 1.387,
+      "step": 231900
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.471163265306123e-05,
+      "loss": 1.3928,
+      "step": 232000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.469122448979592e-05,
+      "loss": 1.3731,
+      "step": 232100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.4670816326530614e-05,
+      "loss": 1.3833,
+      "step": 232200
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.465040816326531e-05,
+      "loss": 1.3912,
+      "step": 232300
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.4630000000000004e-05,
+      "loss": 1.3806,
+      "step": 232400
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.4609591836734695e-05,
+      "loss": 1.3874,
+      "step": 232500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.4589183673469394e-05,
+      "loss": 1.3715,
+      "step": 232600
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.4568775510204085e-05,
+      "loss": 1.394,
+      "step": 232700
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.454836734693878e-05,
+      "loss": 1.3899,
+      "step": 232800
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.452795918367347e-05,
+      "loss": 1.388,
+      "step": 232900
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.450755102040817e-05,
+      "loss": 1.3854,
+      "step": 233000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.448714285714286e-05,
+      "loss": 1.3906,
+      "step": 233100
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.446673469387755e-05,
+      "loss": 1.3852,
+      "step": 233200
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.444632653061225e-05,
+      "loss": 1.3783,
+      "step": 233300
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.442591836734694e-05,
+      "loss": 1.3822,
+      "step": 233400
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.440551020408163e-05,
+      "loss": 1.3801,
+      "step": 233500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.438510204081633e-05,
+      "loss": 1.3828,
+      "step": 233600
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.436469387755102e-05,
+      "loss": 1.3793,
+      "step": 233700
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.434428571428571e-05,
+      "loss": 1.3888,
+      "step": 233800
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.4323877551020405e-05,
+      "loss": 1.3893,
+      "step": 233900
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.43034693877551e-05,
+      "loss": 1.3863,
+      "step": 234000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.4283061224489795e-05,
+      "loss": 1.3819,
+      "step": 234100
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.4262653061224486e-05,
+      "loss": 1.3797,
+      "step": 234200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.4242244897959185e-05,
+      "loss": 1.3859,
+      "step": 234300
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.4221836734693876e-05,
+      "loss": 1.3799,
+      "step": 234400
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.420142857142857e-05,
+      "loss": 1.3792,
+      "step": 234500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.4181020408163266e-05,
+      "loss": 1.3811,
+      "step": 234600
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.416061224489796e-05,
+      "loss": 1.3802,
+      "step": 234700
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.414020408163265e-05,
+      "loss": 1.3867,
+      "step": 234800
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.411979591836734e-05,
+      "loss": 1.3771,
+      "step": 234900
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.4099387755102046e-05,
+      "loss": 1.3859,
+      "step": 235000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.4078979591836744e-05,
+      "loss": 1.3873,
+      "step": 235100
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.4058571428571436e-05,
+      "loss": 1.3838,
+      "step": 235200
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.403816326530613e-05,
+      "loss": 1.3731,
+      "step": 235300
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.4017755102040826e-05,
+      "loss": 1.3794,
+      "step": 235400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.399734693877552e-05,
+      "loss": 1.3826,
+      "step": 235500
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.397693877551021e-05,
+      "loss": 1.3785,
+      "step": 235600
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.395653061224491e-05,
+      "loss": 1.3761,
+      "step": 235700
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.39361224489796e-05,
+      "loss": 1.3798,
+      "step": 235800
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.391571428571429e-05,
+      "loss": 1.3788,
+      "step": 235900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.389530612244898e-05,
+      "loss": 1.3782,
+      "step": 236000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.387489795918368e-05,
+      "loss": 1.3784,
+      "step": 236100
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.385448979591837e-05,
+      "loss": 1.379,
+      "step": 236200
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.3834081632653064e-05,
+      "loss": 1.3728,
+      "step": 236300
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.381367346938776e-05,
+      "loss": 1.3765,
+      "step": 236400
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.3793265306122454e-05,
+      "loss": 1.3776,
+      "step": 236500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.3773061224489794e-05,
+      "loss": 1.3735,
+      "step": 236600
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.375265306122449e-05,
+      "loss": 1.3798,
+      "step": 236700
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.3732244897959184e-05,
+      "loss": 1.3693,
+      "step": 236800
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.3711836734693876e-05,
+      "loss": 1.3738,
+      "step": 236900
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.369142857142857e-05,
+      "loss": 1.3739,
+      "step": 237000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.3671020408163266e-05,
+      "loss": 1.3664,
+      "step": 237100
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.365061224489796e-05,
+      "loss": 1.375,
+      "step": 237200
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.363020408163265e-05,
+      "loss": 1.3725,
+      "step": 237300
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.360979591836735e-05,
+      "loss": 1.3726,
+      "step": 237400
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.358938775510204e-05,
+      "loss": 1.3735,
+      "step": 237500
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.356897959183673e-05,
+      "loss": 1.3666,
+      "step": 237600
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.354857142857143e-05,
+      "loss": 1.3782,
+      "step": 237700
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.352816326530613e-05,
+      "loss": 1.3752,
+      "step": 237800
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.3507755102040826e-05,
+      "loss": 1.3752,
+      "step": 237900
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 5.348734693877552e-05,
+      "loss": 1.3802,
+      "step": 238000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 5.346693877551021e-05,
+      "loss": 1.3747,
+      "step": 238100
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 5.344653061224491e-05,
+      "loss": 1.3736,
+      "step": 238200
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 5.34261224489796e-05,
+      "loss": 1.3809,
+      "step": 238300
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 5.340571428571429e-05,
+      "loss": 1.3719,
+      "step": 238400
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 5.338530612244899e-05,
+      "loss": 1.384,
+      "step": 238500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 5.336489795918368e-05,
+      "loss": 1.3688,
+      "step": 238600
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 5.334448979591837e-05,
+      "loss": 1.3824,
+      "step": 238700
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 5.3324081632653063e-05,
+      "loss": 1.3654,
+      "step": 238800
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 5.330367346938776e-05,
+      "loss": 1.3766,
+      "step": 238900
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 5.3283265306122453e-05,
+      "loss": 1.3762,
+      "step": 239000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 5.3263061224489794e-05,
+      "loss": 1.3794,
+      "step": 239100
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 5.324265306122449e-05,
+      "loss": 1.3763,
+      "step": 239200
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 5.3222244897959184e-05,
+      "loss": 1.3725,
+      "step": 239300
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 5.3201836734693876e-05,
+      "loss": 1.3763,
+      "step": 239400
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 5.3181428571428574e-05,
+      "loss": 1.3667,
+      "step": 239500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 5.3161020408163266e-05,
+      "loss": 1.3746,
+      "step": 239600
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 5.314061224489796e-05,
+      "loss": 1.3727,
+      "step": 239700
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 5.312020408163265e-05,
+      "loss": 1.3799,
+      "step": 239800
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 5.309979591836735e-05,
+      "loss": 1.3714,
+      "step": 239900
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 5.307938775510204e-05,
+      "loss": 1.3683,
+      "step": 240000
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 5.305897959183673e-05,
+      "loss": 1.3774,
+      "step": 240100
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 5.303857142857143e-05,
+      "loss": 1.369,
+      "step": 240200
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 5.301816326530612e-05,
+      "loss": 1.3806,
+      "step": 240300
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 5.299775510204081e-05,
+      "loss": 1.3725,
+      "step": 240400
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 5.297734693877551e-05,
+      "loss": 1.3733,
+      "step": 240500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 5.295693877551021e-05,
+      "loss": 1.383,
+      "step": 240600
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 5.293653061224491e-05,
+      "loss": 1.3655,
+      "step": 240700
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 5.29161224489796e-05,
+      "loss": 1.3752,
+      "step": 240800
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 5.289571428571429e-05,
+      "loss": 1.3796,
+      "step": 240900
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 5.287530612244899e-05,
+      "loss": 1.3732,
+      "step": 241000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 5.285489795918368e-05,
+      "loss": 1.3668,
+      "step": 241100
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 5.283448979591837e-05,
+      "loss": 1.3679,
+      "step": 241200
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 5.281408163265307e-05,
+      "loss": 1.376,
+      "step": 241300
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 5.279367346938776e-05,
+      "loss": 1.3758,
+      "step": 241400
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 5.277326530612245e-05,
+      "loss": 1.3716,
+      "step": 241500
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 5.275285714285715e-05,
+      "loss": 1.3811,
+      "step": 241600
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 5.273244897959184e-05,
+      "loss": 1.3784,
+      "step": 241700
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 5.2712040816326535e-05,
+      "loss": 1.3682,
+      "step": 241800
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 5.2691632653061226e-05,
+      "loss": 1.3745,
+      "step": 241900
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 5.2671224489795925e-05,
+      "loss": 1.37,
+      "step": 242000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 5.2650816326530616e-05,
+      "loss": 1.3733,
+      "step": 242100
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 5.263040816326531e-05,
+      "loss": 1.3629,
+      "step": 242200
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 5.2610000000000006e-05,
+      "loss": 1.3735,
+      "step": 242300
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 5.258979591836735e-05,
+      "loss": 1.368,
+      "step": 242400
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 5.256938775510204e-05,
+      "loss": 1.3758,
+      "step": 242500
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 5.254897959183673e-05,
+      "loss": 1.3735,
+      "step": 242600
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 5.252857142857143e-05,
+      "loss": 1.3688,
+      "step": 242700
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 5.250816326530612e-05,
+      "loss": 1.3777,
+      "step": 242800
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 5.248775510204081e-05,
+      "loss": 1.3649,
+      "step": 242900
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 5.246734693877551e-05,
+      "loss": 1.3765,
+      "step": 243000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 5.24469387755102e-05,
+      "loss": 1.3756,
+      "step": 243100
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 5.242653061224489e-05,
+      "loss": 1.3668,
+      "step": 243200
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 5.24061224489796e-05,
+      "loss": 1.37,
+      "step": 243300
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 5.23857142857143e-05,
+      "loss": 1.3694,
+      "step": 243400
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 5.236530612244899e-05,
+      "loss": 1.3737,
+      "step": 243500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 5.234489795918368e-05,
+      "loss": 1.3688,
+      "step": 243600
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 5.232448979591837e-05,
+      "loss": 1.3714,
+      "step": 243700
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 5.230408163265307e-05,
+      "loss": 1.3711,
+      "step": 243800
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 5.228367346938776e-05,
+      "loss": 1.3686,
+      "step": 243900
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 5.226326530612245e-05,
+      "loss": 1.3589,
+      "step": 244000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 5.224285714285715e-05,
+      "loss": 1.3818,
+      "step": 244100
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 5.222244897959184e-05,
+      "loss": 1.3743,
+      "step": 244200
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 5.2202040816326535e-05,
+      "loss": 1.3707,
+      "step": 244300
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 5.2181836734693875e-05,
+      "loss": 1.3706,
+      "step": 244400
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 5.2161428571428574e-05,
+      "loss": 1.3684,
+      "step": 244500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 5.2141020408163265e-05,
+      "loss": 1.3695,
+      "step": 244600
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 5.212061224489796e-05,
+      "loss": 1.3676,
+      "step": 244700
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 5.2100204081632655e-05,
+      "loss": 1.3762,
+      "step": 244800
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 5.207979591836735e-05,
+      "loss": 1.3768,
+      "step": 244900
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 5.205938775510204e-05,
+      "loss": 1.3673,
+      "step": 245000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 5.203897959183674e-05,
+      "loss": 1.3667,
+      "step": 245100
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 5.201857142857143e-05,
+      "loss": 1.3749,
+      "step": 245200
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 5.199816326530612e-05,
+      "loss": 1.3715,
+      "step": 245300
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 5.197775510204081e-05,
+      "loss": 1.3654,
+      "step": 245400
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.195734693877551e-05,
+      "loss": 1.372,
+      "step": 245500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.19369387755102e-05,
+      "loss": 1.3669,
+      "step": 245600
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.191653061224489e-05,
+      "loss": 1.3771,
+      "step": 245700
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.189612244897959e-05,
+      "loss": 1.3663,
+      "step": 245800
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.187571428571428e-05,
+      "loss": 1.371,
+      "step": 245900
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 5.1855306122448975e-05,
+      "loss": 1.3738,
+      "step": 246000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 5.183489795918368e-05,
+      "loss": 1.3744,
+      "step": 246100
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 5.181448979591838e-05,
+      "loss": 1.3747,
+      "step": 246200
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 5.179428571428572e-05,
+      "loss": 1.3614,
+      "step": 246300
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 5.177387755102041e-05,
+      "loss": 1.3677,
+      "step": 246400
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 5.17534693877551e-05,
+      "loss": 1.3644,
+      "step": 246500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 5.17330612244898e-05,
+      "loss": 1.3696,
+      "step": 246600
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 5.171265306122449e-05,
+      "loss": 1.3718,
+      "step": 246700
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 5.1692244897959184e-05,
+      "loss": 1.3666,
+      "step": 246800
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 5.167183673469388e-05,
+      "loss": 1.3634,
+      "step": 246900
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 5.1651428571428574e-05,
+      "loss": 1.3669,
+      "step": 247000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 5.1631020408163265e-05,
+      "loss": 1.3611,
+      "step": 247100
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 5.1610612244897963e-05,
+      "loss": 1.3688,
+      "step": 247200
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 5.1590204081632655e-05,
+      "loss": 1.3583,
+      "step": 247300
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 5.156979591836735e-05,
+      "loss": 1.3703,
+      "step": 247400
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 5.154938775510204e-05,
+      "loss": 1.374,
+      "step": 247500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 5.1528979591836737e-05,
+      "loss": 1.3702,
+      "step": 247600
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 5.150857142857143e-05,
+      "loss": 1.3673,
+      "step": 247700
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 5.148816326530612e-05,
+      "loss": 1.375,
+      "step": 247800
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 5.146775510204082e-05,
+      "loss": 1.3668,
+      "step": 247900
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 5.144734693877551e-05,
+      "loss": 1.3645,
+      "step": 248000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 5.14269387755102e-05,
+      "loss": 1.37,
+      "step": 248100
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 5.14065306122449e-05,
+      "loss": 1.3684,
+      "step": 248200
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 5.138612244897959e-05,
+      "loss": 1.3699,
+      "step": 248300
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 5.136571428571428e-05,
+      "loss": 1.3612,
+      "step": 248400
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 5.1345306122448974e-05,
+      "loss": 1.3729,
+      "step": 248500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 5.132489795918367e-05,
+      "loss": 1.3631,
+      "step": 248600
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 5.1304489795918364e-05,
+      "loss": 1.3682,
+      "step": 248700
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 5.1284081632653056e-05,
+      "loss": 1.3639,
+      "step": 248800
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 5.126367346938776e-05,
+      "loss": 1.3699,
+      "step": 248900
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 5.124326530612246e-05,
+      "loss": 1.3675,
+      "step": 249000
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 5.122285714285715e-05,
+      "loss": 1.3771,
+      "step": 249100
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 5.120244897959184e-05,
+      "loss": 1.3675,
+      "step": 249200
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.1182040816326534e-05,
+      "loss": 1.3714,
+      "step": 249300
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.116163265306123e-05,
+      "loss": 1.3658,
+      "step": 249400
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.1141224489795924e-05,
+      "loss": 1.3634,
+      "step": 249500
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.1120816326530616e-05,
+      "loss": 1.3612,
+      "step": 249600
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.1100408163265314e-05,
+      "loss": 1.3638,
+      "step": 249700
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.1080000000000006e-05,
+      "loss": 1.369,
+      "step": 249800
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.10595918367347e-05,
+      "loss": 1.3629,
+      "step": 249900
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.1039183673469396e-05,
+      "loss": 1.3725,
+      "step": 250000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.101877551020409e-05,
+      "loss": 1.3663,
+      "step": 250100
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.099836734693878e-05,
+      "loss": 1.3668,
+      "step": 250200
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.097795918367348e-05,
+      "loss": 1.371,
+      "step": 250300
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.095755102040817e-05,
+      "loss": 1.3677,
+      "step": 250400
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.093714285714286e-05,
+      "loss": 1.3704,
+      "step": 250500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.091673469387755e-05,
+      "loss": 1.3606,
+      "step": 250600
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.089632653061225e-05,
+      "loss": 1.3665,
+      "step": 250700
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.087591836734694e-05,
+      "loss": 1.3671,
+      "step": 250800
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.085551020408163e-05,
+      "loss": 1.3599,
+      "step": 250900
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.083510204081633e-05,
+      "loss": 1.3704,
+      "step": 251000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.081469387755102e-05,
+      "loss": 1.3655,
+      "step": 251100
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.0794285714285715e-05,
+      "loss": 1.3656,
+      "step": 251200
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.077387755102041e-05,
+      "loss": 1.3651,
+      "step": 251300
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.0753469387755105e-05,
+      "loss": 1.3607,
+      "step": 251400
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.0733265306122446e-05,
+      "loss": 1.3639,
+      "step": 251500
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.071285714285715e-05,
+      "loss": 1.3607,
+      "step": 251600
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.069244897959184e-05,
+      "loss": 1.3622,
+      "step": 251700
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.067224489795919e-05,
+      "loss": 1.3674,
+      "step": 251800
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.065183673469388e-05,
+      "loss": 1.3547,
+      "step": 251900
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.063142857142857e-05,
+      "loss": 1.3675,
+      "step": 252000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.0611020408163265e-05,
+      "loss": 1.3667,
+      "step": 252100
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.059061224489796e-05,
+      "loss": 1.366,
+      "step": 252200
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.0570204081632655e-05,
+      "loss": 1.3609,
+      "step": 252300
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.0549795918367346e-05,
+      "loss": 1.3662,
+      "step": 252400
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.0529387755102045e-05,
+      "loss": 1.3656,
+      "step": 252500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.0508979591836736e-05,
+      "loss": 1.3602,
+      "step": 252600
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.048857142857143e-05,
+      "loss": 1.3609,
+      "step": 252700
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.0468163265306126e-05,
+      "loss": 1.37,
+      "step": 252800
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.044775510204082e-05,
+      "loss": 1.3692,
+      "step": 252900
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.042734693877551e-05,
+      "loss": 1.3616,
+      "step": 253000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.04069387755102e-05,
+      "loss": 1.3564,
+      "step": 253100
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.03865306122449e-05,
+      "loss": 1.3671,
+      "step": 253200
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.036612244897959e-05,
+      "loss": 1.3626,
+      "step": 253300
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.034571428571428e-05,
+      "loss": 1.3637,
+      "step": 253400
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.032530612244898e-05,
+      "loss": 1.3721,
+      "step": 253500
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.030489795918367e-05,
+      "loss": 1.3671,
+      "step": 253600
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.0284489795918364e-05,
+      "loss": 1.3607,
+      "step": 253700
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.026408163265306e-05,
+      "loss": 1.3707,
+      "step": 253800
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.0243673469387754e-05,
+      "loss": 1.3642,
+      "step": 253900
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.0223265306122445e-05,
+      "loss": 1.3632,
+      "step": 254000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.020285714285714e-05,
+      "loss": 1.357,
+      "step": 254100
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.0182448979591835e-05,
+      "loss": 1.3646,
+      "step": 254200
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.016204081632653e-05,
+      "loss": 1.3602,
+      "step": 254300
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.014163265306123e-05,
+      "loss": 1.3586,
+      "step": 254400
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.0121224489795924e-05,
+      "loss": 1.3614,
+      "step": 254500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.010081632653062e-05,
+      "loss": 1.3678,
+      "step": 254600
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.0080408163265314e-05,
+      "loss": 1.354,
+      "step": 254700
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.0060000000000005e-05,
+      "loss": 1.3695,
+      "step": 254800
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.0039591836734704e-05,
+      "loss": 1.3717,
+      "step": 254900
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.0019183673469395e-05,
+      "loss": 1.3631,
+      "step": 255000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.999877551020408e-05,
+      "loss": 1.354,
+      "step": 255100
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.997836734693878e-05,
+      "loss": 1.3627,
+      "step": 255200
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.995795918367347e-05,
+      "loss": 1.3661,
+      "step": 255300
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.993755102040816e-05,
+      "loss": 1.3643,
+      "step": 255400
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.991714285714286e-05,
+      "loss": 1.3659,
+      "step": 255500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.989673469387756e-05,
+      "loss": 1.3579,
+      "step": 255600
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.987632653061225e-05,
+      "loss": 1.3626,
+      "step": 255700
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.985591836734694e-05,
+      "loss": 1.3663,
+      "step": 255800
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.983551020408164e-05,
+      "loss": 1.3669,
+      "step": 255900
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.981510204081633e-05,
+      "loss": 1.3658,
+      "step": 256000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.979469387755102e-05,
+      "loss": 1.3646,
+      "step": 256100
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.9774285714285714e-05,
+      "loss": 1.3545,
+      "step": 256200
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.975387755102041e-05,
+      "loss": 1.3558,
+      "step": 256300
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.9733469387755104e-05,
+      "loss": 1.3715,
+      "step": 256400
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.9713061224489796e-05,
+      "loss": 1.3543,
+      "step": 256500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.9692653061224494e-05,
+      "loss": 1.3574,
+      "step": 256600
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.9672244897959186e-05,
+      "loss": 1.366,
+      "step": 256700
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.965183673469388e-05,
+      "loss": 1.3644,
+      "step": 256800
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.9631428571428576e-05,
+      "loss": 1.3538,
+      "step": 256900
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.961102040816327e-05,
+      "loss": 1.3584,
+      "step": 257000
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.959061224489796e-05,
+      "loss": 1.364,
+      "step": 257100
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.957020408163266e-05,
+      "loss": 1.3648,
+      "step": 257200
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.954979591836735e-05,
+      "loss": 1.3677,
+      "step": 257300
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.952938775510204e-05,
+      "loss": 1.3627,
+      "step": 257400
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.950897959183673e-05,
+      "loss": 1.3604,
+      "step": 257500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.948857142857143e-05,
+      "loss": 1.3644,
+      "step": 257600
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.946816326530612e-05,
+      "loss": 1.3597,
+      "step": 257700
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.944775510204082e-05,
+      "loss": 1.3584,
+      "step": 257800
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.942734693877551e-05,
+      "loss": 1.3579,
+      "step": 257900
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.940693877551021e-05,
+      "loss": 1.3517,
+      "step": 258000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.93865306122449e-05,
+      "loss": 1.3607,
+      "step": 258100
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.9366122448979594e-05,
+      "loss": 1.361,
+      "step": 258200
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.934591836734694e-05,
+      "loss": 1.3581,
+      "step": 258300
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.932571428571429e-05,
+      "loss": 1.364,
+      "step": 258400
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.930530612244898e-05,
+      "loss": 1.3659,
+      "step": 258500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.928489795918367e-05,
+      "loss": 1.3594,
+      "step": 258600
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.926448979591837e-05,
+      "loss": 1.3557,
+      "step": 258700
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.924408163265307e-05,
+      "loss": 1.3619,
+      "step": 258800
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.922367346938776e-05,
+      "loss": 1.3588,
+      "step": 258900
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.920326530612245e-05,
+      "loss": 1.3564,
+      "step": 259000
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.918285714285715e-05,
+      "loss": 1.3562,
+      "step": 259100
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.916244897959184e-05,
+      "loss": 1.3595,
+      "step": 259200
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.914204081632653e-05,
+      "loss": 1.3622,
+      "step": 259300
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.9121632653061225e-05,
+      "loss": 1.3632,
+      "step": 259400
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.910142857142858e-05,
+      "loss": 1.3518,
+      "step": 259500
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.908102040816327e-05,
+      "loss": 1.3604,
+      "step": 259600
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.906061224489796e-05,
+      "loss": 1.3563,
+      "step": 259700
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.9040204081632654e-05,
+      "loss": 1.3605,
+      "step": 259800
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.901979591836735e-05,
+      "loss": 1.3585,
+      "step": 259900
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.8999387755102044e-05,
+      "loss": 1.3552,
+      "step": 260000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.8978979591836736e-05,
+      "loss": 1.3663,
+      "step": 260100
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.8958571428571434e-05,
+      "loss": 1.3552,
+      "step": 260200
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.8938163265306126e-05,
+      "loss": 1.3525,
+      "step": 260300
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.891775510204082e-05,
+      "loss": 1.3558,
+      "step": 260400
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.8897346938775515e-05,
+      "loss": 1.3521,
+      "step": 260500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.887693877551021e-05,
+      "loss": 1.3548,
+      "step": 260600
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.88565306122449e-05,
+      "loss": 1.3492,
+      "step": 260700
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.883612244897959e-05,
+      "loss": 1.3522,
+      "step": 260800
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.881571428571429e-05,
+      "loss": 1.3587,
+      "step": 260900
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.879530612244898e-05,
+      "loss": 1.356,
+      "step": 261000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.877489795918367e-05,
+      "loss": 1.3501,
+      "step": 261100
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.875448979591837e-05,
+      "loss": 1.361,
+      "step": 261200
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.873408163265306e-05,
+      "loss": 1.3567,
+      "step": 261300
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.871367346938775e-05,
+      "loss": 1.3509,
+      "step": 261400
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.869326530612245e-05,
+      "loss": 1.356,
+      "step": 261500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.867285714285715e-05,
+      "loss": 1.3551,
+      "step": 261600
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.865244897959184e-05,
+      "loss": 1.3602,
+      "step": 261700
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.863204081632653e-05,
+      "loss": 1.3582,
+      "step": 261800
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.861163265306123e-05,
+      "loss": 1.3511,
+      "step": 261900
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.859122448979592e-05,
+      "loss": 1.3636,
+      "step": 262000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.857102040816327e-05,
+      "loss": 1.3549,
+      "step": 262100
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.855061224489796e-05,
+      "loss": 1.3544,
+      "step": 262200
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.853020408163266e-05,
+      "loss": 1.3574,
+      "step": 262300
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.850979591836735e-05,
+      "loss": 1.3552,
+      "step": 262400
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.8489387755102044e-05,
+      "loss": 1.358,
+      "step": 262500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.8468979591836735e-05,
+      "loss": 1.3516,
+      "step": 262600
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.8448571428571434e-05,
+      "loss": 1.3533,
+      "step": 262700
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.8428163265306125e-05,
+      "loss": 1.3445,
+      "step": 262800
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.840775510204082e-05,
+      "loss": 1.357,
+      "step": 262900
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.8387346938775515e-05,
+      "loss": 1.3589,
+      "step": 263000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.836693877551021e-05,
+      "loss": 1.3528,
+      "step": 263100
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.83465306122449e-05,
+      "loss": 1.3559,
+      "step": 263200
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.83261224489796e-05,
+      "loss": 1.3541,
+      "step": 263300
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.830571428571429e-05,
+      "loss": 1.3598,
+      "step": 263400
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.828530612244898e-05,
+      "loss": 1.3551,
+      "step": 263500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.826489795918367e-05,
+      "loss": 1.3555,
+      "step": 263600
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.824448979591837e-05,
+      "loss": 1.356,
+      "step": 263700
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.822408163265306e-05,
+      "loss": 1.3579,
+      "step": 263800
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.820367346938775e-05,
+      "loss": 1.349,
+      "step": 263900
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.818326530612245e-05,
+      "loss": 1.3577,
+      "step": 264000
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.816285714285714e-05,
+      "loss": 1.3547,
+      "step": 264100
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.8142448979591835e-05,
+      "loss": 1.3527,
+      "step": 264200
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.812204081632653e-05,
+      "loss": 1.3547,
+      "step": 264300
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.810163265306123e-05,
+      "loss": 1.355,
+      "step": 264400
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.808122448979592e-05,
+      "loss": 1.3486,
+      "step": 264500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.8060816326530614e-05,
+      "loss": 1.3456,
+      "step": 264600
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.804040816326531e-05,
+      "loss": 1.3508,
+      "step": 264700
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.8020000000000004e-05,
+      "loss": 1.357,
+      "step": 264800
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.7999591836734696e-05,
+      "loss": 1.3491,
+      "step": 264900
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.7979183673469394e-05,
+      "loss": 1.3544,
+      "step": 265000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.7958775510204086e-05,
+      "loss": 1.3528,
+      "step": 265100
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.793836734693878e-05,
+      "loss": 1.3547,
+      "step": 265200
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.791795918367347e-05,
+      "loss": 1.3523,
+      "step": 265300
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.789755102040817e-05,
+      "loss": 1.3545,
+      "step": 265400
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.787714285714286e-05,
+      "loss": 1.3518,
+      "step": 265500
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.785673469387755e-05,
+      "loss": 1.3518,
+      "step": 265600
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.783632653061225e-05,
+      "loss": 1.3509,
+      "step": 265700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.781591836734694e-05,
+      "loss": 1.3615,
+      "step": 265800
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.779551020408163e-05,
+      "loss": 1.3546,
+      "step": 265900
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.777510204081633e-05,
+      "loss": 1.3509,
+      "step": 266000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.775469387755102e-05,
+      "loss": 1.3513,
+      "step": 266100
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.7734285714285714e-05,
+      "loss": 1.3494,
+      "step": 266200
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.7713877551020405e-05,
+      "loss": 1.349,
+      "step": 266300
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.769367346938776e-05,
+      "loss": 1.3497,
+      "step": 266400
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.767326530612245e-05,
+      "loss": 1.3567,
+      "step": 266500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.765285714285714e-05,
+      "loss": 1.3459,
+      "step": 266600
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.7632448979591834e-05,
+      "loss": 1.3518,
+      "step": 266700
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.761204081632653e-05,
+      "loss": 1.3449,
+      "step": 266800
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.7591632653061224e-05,
+      "loss": 1.3547,
+      "step": 266900
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.7571224489795916e-05,
+      "loss": 1.3518,
+      "step": 267000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.7550816326530614e-05,
+      "loss": 1.355,
+      "step": 267100
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.753040816326531e-05,
+      "loss": 1.3594,
+      "step": 267200
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.7510000000000004e-05,
+      "loss": 1.3578,
+      "step": 267300
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.7489795918367345e-05,
+      "loss": 1.3415,
+      "step": 267400
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.746938775510204e-05,
+      "loss": 1.3566,
+      "step": 267500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.744897959183674e-05,
+      "loss": 1.3475,
+      "step": 267600
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.742857142857143e-05,
+      "loss": 1.3459,
+      "step": 267700
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.7408163265306125e-05,
+      "loss": 1.3535,
+      "step": 267800
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.738775510204082e-05,
+      "loss": 1.3524,
+      "step": 267900
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.7367346938775515e-05,
+      "loss": 1.3577,
+      "step": 268000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.7346938775510206e-05,
+      "loss": 1.3492,
+      "step": 268100
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.7326530612244905e-05,
+      "loss": 1.3512,
+      "step": 268200
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.7306122448979596e-05,
+      "loss": 1.3496,
+      "step": 268300
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.728571428571429e-05,
+      "loss": 1.3584,
+      "step": 268400
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.726530612244898e-05,
+      "loss": 1.3619,
+      "step": 268500
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.724489795918368e-05,
+      "loss": 1.3485,
+      "step": 268600
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.722448979591837e-05,
+      "loss": 1.3514,
+      "step": 268700
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.720408163265306e-05,
+      "loss": 1.3496,
+      "step": 268800
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.718367346938776e-05,
+      "loss": 1.3482,
+      "step": 268900
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.716326530612245e-05,
+      "loss": 1.3519,
+      "step": 269000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.714285714285714e-05,
+      "loss": 1.3434,
+      "step": 269100
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.712244897959184e-05,
+      "loss": 1.3528,
+      "step": 269200
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.710204081632653e-05,
+      "loss": 1.3501,
+      "step": 269300
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.7081632653061224e-05,
+      "loss": 1.3531,
+      "step": 269400
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.7061224489795916e-05,
+      "loss": 1.3506,
+      "step": 269500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.7040816326530614e-05,
+      "loss": 1.3509,
+      "step": 269600
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.7020408163265306e-05,
+      "loss": 1.3494,
+      "step": 269700
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.7e-05,
+      "loss": 1.3502,
+      "step": 269800
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.6979591836734696e-05,
+      "loss": 1.3478,
+      "step": 269900
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.6959183673469394e-05,
+      "loss": 1.3474,
+      "step": 270000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.6938775510204086e-05,
+      "loss": 1.3509,
+      "step": 270100
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.691836734693878e-05,
+      "loss": 1.3512,
+      "step": 270200
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.6897959183673475e-05,
+      "loss": 1.3464,
+      "step": 270300
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.687755102040817e-05,
+      "loss": 1.3499,
+      "step": 270400
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.6857346938775515e-05,
+      "loss": 1.3504,
+      "step": 270500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.6836938775510206e-05,
+      "loss": 1.3491,
+      "step": 270600
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.6816530612244905e-05,
+      "loss": 1.3474,
+      "step": 270700
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.6796122448979596e-05,
+      "loss": 1.3493,
+      "step": 270800
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.677571428571429e-05,
+      "loss": 1.3453,
+      "step": 270900
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.6755306122448986e-05,
+      "loss": 1.3475,
+      "step": 271000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.673489795918368e-05,
+      "loss": 1.3389,
+      "step": 271100
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.671448979591837e-05,
+      "loss": 1.3562,
+      "step": 271200
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.669408163265306e-05,
+      "loss": 1.347,
+      "step": 271300
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.667367346938776e-05,
+      "loss": 1.3447,
+      "step": 271400
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.665326530612245e-05,
+      "loss": 1.3423,
+      "step": 271500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.663285714285714e-05,
+      "loss": 1.3465,
+      "step": 271600
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.661244897959184e-05,
+      "loss": 1.3532,
+      "step": 271700
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.659204081632653e-05,
+      "loss": 1.3523,
+      "step": 271800
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.6571632653061224e-05,
+      "loss": 1.3499,
+      "step": 271900
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.655122448979592e-05,
+      "loss": 1.3379,
+      "step": 272000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.6530816326530614e-05,
+      "loss": 1.3507,
+      "step": 272100
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.6510408163265305e-05,
+      "loss": 1.3504,
+      "step": 272200
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.649e-05,
+      "loss": 1.3476,
+      "step": 272300
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.6469591836734695e-05,
+      "loss": 1.3441,
+      "step": 272400
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.644918367346939e-05,
+      "loss": 1.348,
+      "step": 272500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.642877551020408e-05,
+      "loss": 1.3441,
+      "step": 272600
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.640836734693878e-05,
+      "loss": 1.3603,
+      "step": 272700
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.6387959183673475e-05,
+      "loss": 1.3454,
+      "step": 272800
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.636755102040817e-05,
+      "loss": 1.3501,
+      "step": 272900
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.634714285714286e-05,
+      "loss": 1.3571,
+      "step": 273000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.632673469387756e-05,
+      "loss": 1.3456,
+      "step": 273100
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.630632653061225e-05,
+      "loss": 1.3444,
+      "step": 273200
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.628591836734694e-05,
+      "loss": 1.3509,
+      "step": 273300
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.626551020408164e-05,
+      "loss": 1.353,
+      "step": 273400
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.624510204081633e-05,
+      "loss": 1.3496,
+      "step": 273500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.622469387755102e-05,
+      "loss": 1.3433,
+      "step": 273600
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.620428571428572e-05,
+      "loss": 1.3516,
+      "step": 273700
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.618387755102041e-05,
+      "loss": 1.3459,
+      "step": 273800
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.61634693877551e-05,
+      "loss": 1.3496,
+      "step": 273900
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.6143061224489795e-05,
+      "loss": 1.3548,
+      "step": 274000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.612265306122449e-05,
+      "loss": 1.3448,
+      "step": 274100
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.6102244897959185e-05,
+      "loss": 1.3405,
+      "step": 274200
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.6081836734693876e-05,
+      "loss": 1.3427,
+      "step": 274300
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.6061428571428575e-05,
+      "loss": 1.3477,
+      "step": 274400
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.6041020408163266e-05,
+      "loss": 1.3471,
+      "step": 274500
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.6020816326530614e-05,
+      "loss": 1.3434,
+      "step": 274600
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.6000408163265305e-05,
+      "loss": 1.3432,
+      "step": 274700
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.5980000000000004e-05,
+      "loss": 1.3433,
+      "step": 274800
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.5959591836734695e-05,
+      "loss": 1.347,
+      "step": 274900
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.593918367346939e-05,
+      "loss": 1.3484,
+      "step": 275000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.591877551020408e-05,
+      "loss": 1.3401,
+      "step": 275100
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.589836734693878e-05,
+      "loss": 1.3389,
+      "step": 275200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.587795918367347e-05,
+      "loss": 1.3432,
+      "step": 275300
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.585755102040817e-05,
+      "loss": 1.3445,
+      "step": 275400
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.5837142857142865e-05,
+      "loss": 1.353,
+      "step": 275500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.581673469387756e-05,
+      "loss": 1.3429,
+      "step": 275600
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.579632653061225e-05,
+      "loss": 1.3466,
+      "step": 275700
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.577591836734694e-05,
+      "loss": 1.3392,
+      "step": 275800
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.575551020408164e-05,
+      "loss": 1.3471,
+      "step": 275900
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.573510204081633e-05,
+      "loss": 1.3481,
+      "step": 276000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.571469387755102e-05,
+      "loss": 1.3497,
+      "step": 276100
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.569428571428572e-05,
+      "loss": 1.3433,
+      "step": 276200
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.567387755102041e-05,
+      "loss": 1.3518,
+      "step": 276300
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.56534693877551e-05,
+      "loss": 1.3487,
+      "step": 276400
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.56330612244898e-05,
+      "loss": 1.3424,
+      "step": 276500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.561265306122449e-05,
+      "loss": 1.348,
+      "step": 276600
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.5592244897959184e-05,
+      "loss": 1.3455,
+      "step": 276700
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.5571836734693876e-05,
+      "loss": 1.35,
+      "step": 276800
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.5551428571428574e-05,
+      "loss": 1.344,
+      "step": 276900
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.5531020408163266e-05,
+      "loss": 1.3352,
+      "step": 277000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.5510816326530613e-05,
+      "loss": 1.3472,
+      "step": 277100
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.5490408163265305e-05,
+      "loss": 1.3417,
+      "step": 277200
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.5470000000000003e-05,
+      "loss": 1.343,
+      "step": 277300
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.5449591836734695e-05,
+      "loss": 1.3394,
+      "step": 277400
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.5429183673469387e-05,
+      "loss": 1.3352,
+      "step": 277500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.5408775510204085e-05,
+      "loss": 1.3409,
+      "step": 277600
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.5388367346938777e-05,
+      "loss": 1.342,
+      "step": 277700
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.536795918367347e-05,
+      "loss": 1.3487,
+      "step": 277800
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.5347551020408166e-05,
+      "loss": 1.3347,
+      "step": 277900
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.532714285714286e-05,
+      "loss": 1.3452,
+      "step": 278000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.530673469387755e-05,
+      "loss": 1.3467,
+      "step": 278100
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.528632653061225e-05,
+      "loss": 1.3372,
+      "step": 278200
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.5265918367346946e-05,
+      "loss": 1.345,
+      "step": 278300
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.524551020408164e-05,
+      "loss": 1.3479,
+      "step": 278400
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.522510204081633e-05,
+      "loss": 1.3459,
+      "step": 278500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.520469387755102e-05,
+      "loss": 1.3469,
+      "step": 278600
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.518428571428572e-05,
+      "loss": 1.3439,
+      "step": 278700
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.516387755102041e-05,
+      "loss": 1.3485,
+      "step": 278800
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.51434693877551e-05,
+      "loss": 1.3465,
+      "step": 278900
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.51230612244898e-05,
+      "loss": 1.3453,
+      "step": 279000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.510265306122449e-05,
+      "loss": 1.3403,
+      "step": 279100
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.5082244897959184e-05,
+      "loss": 1.3401,
+      "step": 279200
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.506183673469388e-05,
+      "loss": 1.342,
+      "step": 279300
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.5041428571428574e-05,
+      "loss": 1.35,
+      "step": 279400
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.5021020408163266e-05,
+      "loss": 1.3474,
+      "step": 279500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.500061224489796e-05,
+      "loss": 1.3487,
+      "step": 279600
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.4980204081632656e-05,
+      "loss": 1.3438,
+      "step": 279700
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.495979591836735e-05,
+      "loss": 1.3435,
+      "step": 279800
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.493938775510204e-05,
+      "loss": 1.3462,
+      "step": 279900
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.491897959183674e-05,
+      "loss": 1.3392,
+      "step": 280000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.489857142857143e-05,
+      "loss": 1.3455,
+      "step": 280100
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.487816326530612e-05,
+      "loss": 1.3355,
+      "step": 280200
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.485775510204082e-05,
+      "loss": 1.3454,
+      "step": 280300
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.483734693877551e-05,
+      "loss": 1.3421,
+      "step": 280400
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.481693877551021e-05,
+      "loss": 1.3418,
+      "step": 280500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.47965306122449e-05,
+      "loss": 1.3387,
+      "step": 280600
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.477632653061225e-05,
+      "loss": 1.3361,
+      "step": 280700
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.475591836734694e-05,
+      "loss": 1.3422,
+      "step": 280800
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.473551020408163e-05,
+      "loss": 1.337,
+      "step": 280900
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.471510204081633e-05,
+      "loss": 1.3355,
+      "step": 281000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.469469387755103e-05,
+      "loss": 1.3407,
+      "step": 281100
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.467428571428572e-05,
+      "loss": 1.3399,
+      "step": 281200
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.465387755102041e-05,
+      "loss": 1.3435,
+      "step": 281300
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.46334693877551e-05,
+      "loss": 1.3507,
+      "step": 281400
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.46130612244898e-05,
+      "loss": 1.344,
+      "step": 281500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.459265306122449e-05,
+      "loss": 1.3464,
+      "step": 281600
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.4572244897959184e-05,
+      "loss": 1.345,
+      "step": 281700
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.455183673469388e-05,
+      "loss": 1.3423,
+      "step": 281800
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.4531428571428574e-05,
+      "loss": 1.3356,
+      "step": 281900
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.4511020408163265e-05,
+      "loss": 1.3349,
+      "step": 282000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.4490612244897964e-05,
+      "loss": 1.3366,
+      "step": 282100
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.4470204081632655e-05,
+      "loss": 1.3421,
+      "step": 282200
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.444979591836735e-05,
+      "loss": 1.332,
+      "step": 282300
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.4429387755102045e-05,
+      "loss": 1.3393,
+      "step": 282400
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.440897959183674e-05,
+      "loss": 1.3377,
+      "step": 282500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.438857142857143e-05,
+      "loss": 1.3301,
+      "step": 282600
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.436816326530612e-05,
+      "loss": 1.342,
+      "step": 282700
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.434775510204082e-05,
+      "loss": 1.3413,
+      "step": 282800
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.432734693877551e-05,
+      "loss": 1.3391,
+      "step": 282900
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.43069387755102e-05,
+      "loss": 1.3439,
+      "step": 283000
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.42865306122449e-05,
+      "loss": 1.3376,
+      "step": 283100
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.42661224489796e-05,
+      "loss": 1.3525,
+      "step": 283200
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.424571428571429e-05,
+      "loss": 1.3381,
+      "step": 283300
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.422530612244898e-05,
+      "loss": 1.34,
+      "step": 283400
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.420489795918368e-05,
+      "loss": 1.3415,
+      "step": 283500
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.418448979591837e-05,
+      "loss": 1.3354,
+      "step": 283600
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.416408163265306e-05,
+      "loss": 1.3298,
+      "step": 283700
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.414367346938776e-05,
+      "loss": 1.339,
+      "step": 283800
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.412326530612245e-05,
+      "loss": 1.3375,
+      "step": 283900
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.4102857142857145e-05,
+      "loss": 1.3392,
+      "step": 284000
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.4082448979591836e-05,
+      "loss": 1.3396,
+      "step": 284100
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.4062040816326535e-05,
+      "loss": 1.3316,
+      "step": 284200
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.404183673469388e-05,
+      "loss": 1.3385,
+      "step": 284300
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.4021428571428574e-05,
+      "loss": 1.343,
+      "step": 284400
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.4001020408163265e-05,
+      "loss": 1.3366,
+      "step": 284500
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.3980612244897964e-05,
+      "loss": 1.3309,
+      "step": 284600
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.3960204081632655e-05,
+      "loss": 1.3334,
+      "step": 284700
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.393979591836735e-05,
+      "loss": 1.3319,
+      "step": 284800
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.3919387755102045e-05,
+      "loss": 1.3326,
+      "step": 284900
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.389897959183674e-05,
+      "loss": 1.3369,
+      "step": 285000
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.387857142857143e-05,
+      "loss": 1.3434,
+      "step": 285100
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.385816326530613e-05,
+      "loss": 1.3365,
+      "step": 285200
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.383775510204082e-05,
+      "loss": 1.3283,
+      "step": 285300
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.381734693877551e-05,
+      "loss": 1.3421,
+      "step": 285400
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.37969387755102e-05,
+      "loss": 1.3341,
+      "step": 285500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.37765306122449e-05,
+      "loss": 1.3389,
+      "step": 285600
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.375612244897959e-05,
+      "loss": 1.3407,
+      "step": 285700
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.373571428571428e-05,
+      "loss": 1.3303,
+      "step": 285800
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.371530612244898e-05,
+      "loss": 1.3381,
+      "step": 285900
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.369489795918368e-05,
+      "loss": 1.3354,
+      "step": 286000
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.367448979591837e-05,
+      "loss": 1.3276,
+      "step": 286100
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.365408163265306e-05,
+      "loss": 1.3286,
+      "step": 286200
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.363367346938776e-05,
+      "loss": 1.3329,
+      "step": 286300
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.361326530612245e-05,
+      "loss": 1.3352,
+      "step": 286400
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.3592857142857144e-05,
+      "loss": 1.342,
+      "step": 286500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.357244897959184e-05,
+      "loss": 1.3327,
+      "step": 286600
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.355224489795919e-05,
+      "loss": 1.3269,
+      "step": 286700
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.353183673469388e-05,
+      "loss": 1.333,
+      "step": 286800
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.3511428571428574e-05,
+      "loss": 1.3328,
+      "step": 286900
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.349102040816327e-05,
+      "loss": 1.3344,
+      "step": 287000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.3470612244897963e-05,
+      "loss": 1.3306,
+      "step": 287100
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.3450204081632655e-05,
+      "loss": 1.3337,
+      "step": 287200
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.342979591836735e-05,
+      "loss": 1.3351,
+      "step": 287300
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.3409387755102045e-05,
+      "loss": 1.3411,
+      "step": 287400
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.3388979591836737e-05,
+      "loss": 1.3384,
+      "step": 287500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.336857142857143e-05,
+      "loss": 1.342,
+      "step": 287600
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.3348163265306127e-05,
+      "loss": 1.3367,
+      "step": 287700
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.332775510204082e-05,
+      "loss": 1.3284,
+      "step": 287800
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.330734693877551e-05,
+      "loss": 1.341,
+      "step": 287900
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.328693877551021e-05,
+      "loss": 1.3353,
+      "step": 288000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.32665306122449e-05,
+      "loss": 1.3286,
+      "step": 288100
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.324612244897959e-05,
+      "loss": 1.3329,
+      "step": 288200
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.322571428571428e-05,
+      "loss": 1.3288,
+      "step": 288300
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.320530612244898e-05,
+      "loss": 1.3316,
+      "step": 288400
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.318489795918367e-05,
+      "loss": 1.3349,
+      "step": 288500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.3164489795918364e-05,
+      "loss": 1.3362,
+      "step": 288600
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.314408163265306e-05,
+      "loss": 1.3388,
+      "step": 288700
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.312367346938776e-05,
+      "loss": 1.3298,
+      "step": 288800
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.310326530612245e-05,
+      "loss": 1.3323,
+      "step": 288900
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.3082857142857144e-05,
+      "loss": 1.3256,
+      "step": 289000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.306244897959184e-05,
+      "loss": 1.3281,
+      "step": 289100
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.3042040816326534e-05,
+      "loss": 1.3363,
+      "step": 289200
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.3021632653061226e-05,
+      "loss": 1.3302,
+      "step": 289300
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.3001224489795924e-05,
+      "loss": 1.3341,
+      "step": 289400
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.2980816326530616e-05,
+      "loss": 1.329,
+      "step": 289500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.296040816326531e-05,
+      "loss": 1.3314,
+      "step": 289600
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.2940000000000006e-05,
+      "loss": 1.3385,
+      "step": 289700
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.29195918367347e-05,
+      "loss": 1.3387,
+      "step": 289800
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.289918367346939e-05,
+      "loss": 1.3306,
+      "step": 289900
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.287877551020408e-05,
+      "loss": 1.3336,
+      "step": 290000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.285836734693878e-05,
+      "loss": 1.336,
+      "step": 290100
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.283795918367347e-05,
+      "loss": 1.3374,
+      "step": 290200
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.281755102040816e-05,
+      "loss": 1.3331,
+      "step": 290300
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.279714285714286e-05,
+      "loss": 1.3254,
+      "step": 290400
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.277673469387755e-05,
+      "loss": 1.3313,
+      "step": 290500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.2756326530612243e-05,
+      "loss": 1.3255,
+      "step": 290600
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.273591836734694e-05,
+      "loss": 1.3337,
+      "step": 290700
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.271571428571429e-05,
+      "loss": 1.3215,
+      "step": 290800
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.269530612244898e-05,
+      "loss": 1.3266,
+      "step": 290900
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.267489795918367e-05,
+      "loss": 1.3328,
+      "step": 291000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.265448979591837e-05,
+      "loss": 1.3328,
+      "step": 291100
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.263408163265306e-05,
+      "loss": 1.3267,
+      "step": 291200
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.2613673469387754e-05,
+      "loss": 1.3329,
+      "step": 291300
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.2593265306122446e-05,
+      "loss": 1.3281,
+      "step": 291400
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.257285714285715e-05,
+      "loss": 1.3345,
+      "step": 291500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.255244897959184e-05,
+      "loss": 1.3294,
+      "step": 291600
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.253224489795918e-05,
+      "loss": 1.3314,
+      "step": 291700
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.2511836734693875e-05,
+      "loss": 1.3254,
+      "step": 291800
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.249142857142857e-05,
+      "loss": 1.3333,
+      "step": 291900
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.247102040816327e-05,
+      "loss": 1.3368,
+      "step": 292000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.245061224489796e-05,
+      "loss": 1.3384,
+      "step": 292100
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.2430204081632655e-05,
+      "loss": 1.3312,
+      "step": 292200
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.240979591836735e-05,
+      "loss": 1.336,
+      "step": 292300
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.2389387755102045e-05,
+      "loss": 1.3343,
+      "step": 292400
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.2368979591836736e-05,
+      "loss": 1.3281,
+      "step": 292500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.2348571428571435e-05,
+      "loss": 1.3312,
+      "step": 292600
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.2328163265306126e-05,
+      "loss": 1.3395,
+      "step": 292700
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.230775510204082e-05,
+      "loss": 1.3291,
+      "step": 292800
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.2287346938775516e-05,
+      "loss": 1.3359,
+      "step": 292900
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.226693877551021e-05,
+      "loss": 1.3228,
+      "step": 293000
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.22465306122449e-05,
+      "loss": 1.3324,
+      "step": 293100
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.222612244897959e-05,
+      "loss": 1.3299,
+      "step": 293200
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.220571428571429e-05,
+      "loss": 1.3326,
+      "step": 293300
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.218530612244898e-05,
+      "loss": 1.3344,
+      "step": 293400
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.216489795918367e-05,
+      "loss": 1.324,
+      "step": 293500
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.214448979591837e-05,
+      "loss": 1.324,
+      "step": 293600
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.212408163265306e-05,
+      "loss": 1.3296,
+      "step": 293700
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.2103673469387754e-05,
+      "loss": 1.3252,
+      "step": 293800
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.208326530612245e-05,
+      "loss": 1.3327,
+      "step": 293900
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.2062857142857144e-05,
+      "loss": 1.3246,
+      "step": 294000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.204265306122449e-05,
+      "loss": 1.3382,
+      "step": 294100
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.202224489795918e-05,
+      "loss": 1.3306,
+      "step": 294200
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.2001836734693875e-05,
+      "loss": 1.3293,
+      "step": 294300
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.198163265306123e-05,
+      "loss": 1.3291,
+      "step": 294400
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.196122448979592e-05,
+      "loss": 1.3355,
+      "step": 294500
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.194081632653061e-05,
+      "loss": 1.3385,
+      "step": 294600
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.1920408163265304e-05,
+      "loss": 1.3289,
+      "step": 294700
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.19e-05,
+      "loss": 1.3301,
+      "step": 294800
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.1879591836734694e-05,
+      "loss": 1.3195,
+      "step": 294900
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.1859183673469385e-05,
+      "loss": 1.3366,
+      "step": 295000
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.1838775510204084e-05,
+      "loss": 1.3306,
+      "step": 295100
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.1818367346938775e-05,
+      "loss": 1.3331,
+      "step": 295200
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.1797959183673473e-05,
+      "loss": 1.3288,
+      "step": 295300
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.1777551020408165e-05,
+      "loss": 1.3346,
+      "step": 295400
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.1757142857142863e-05,
+      "loss": 1.3198,
+      "step": 295500
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.1736734693877555e-05,
+      "loss": 1.324,
+      "step": 295600
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.1716326530612247e-05,
+      "loss": 1.328,
+      "step": 295700
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.1695918367346945e-05,
+      "loss": 1.3288,
+      "step": 295800
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.1675510204081637e-05,
+      "loss": 1.3293,
+      "step": 295900
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.165510204081633e-05,
+      "loss": 1.3365,
+      "step": 296000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.1634693877551026e-05,
+      "loss": 1.3325,
+      "step": 296100
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.161428571428572e-05,
+      "loss": 1.3273,
+      "step": 296200
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.159387755102041e-05,
+      "loss": 1.3268,
+      "step": 296300
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.15734693877551e-05,
+      "loss": 1.3318,
+      "step": 296400
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.15530612244898e-05,
+      "loss": 1.3242,
+      "step": 296500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.153265306122449e-05,
+      "loss": 1.3341,
+      "step": 296600
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.151224489795918e-05,
+      "loss": 1.3291,
+      "step": 296700
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.149204081632653e-05,
+      "loss": 1.3285,
+      "step": 296800
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.147163265306123e-05,
+      "loss": 1.3295,
+      "step": 296900
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.145122448979592e-05,
+      "loss": 1.3319,
+      "step": 297000
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.143081632653061e-05,
+      "loss": 1.332,
+      "step": 297100
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.141040816326531e-05,
+      "loss": 1.3232,
+      "step": 297200
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.139e-05,
+      "loss": 1.326,
+      "step": 297300
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.1369591836734693e-05,
+      "loss": 1.3215,
+      "step": 297400
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.1349183673469385e-05,
+      "loss": 1.3292,
+      "step": 297500
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.132877551020408e-05,
+      "loss": 1.3246,
+      "step": 297600
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.1308367346938775e-05,
+      "loss": 1.3297,
+      "step": 297700
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.1287959183673467e-05,
+      "loss": 1.3299,
+      "step": 297800
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.1267551020408165e-05,
+      "loss": 1.3294,
+      "step": 297900
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.1247142857142856e-05,
+      "loss": 1.324,
+      "step": 298000
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.1226734693877555e-05,
+      "loss": 1.3301,
+      "step": 298100
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.1206326530612246e-05,
+      "loss": 1.335,
+      "step": 298200
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.1185918367346945e-05,
+      "loss": 1.3306,
+      "step": 298300
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.1165510204081636e-05,
+      "loss": 1.3263,
+      "step": 298400
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.114510204081633e-05,
+      "loss": 1.3282,
+      "step": 298500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.1124693877551026e-05,
+      "loss": 1.3322,
+      "step": 298600
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.110428571428572e-05,
+      "loss": 1.3278,
+      "step": 298700
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.108387755102041e-05,
+      "loss": 1.3255,
+      "step": 298800
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.106346938775511e-05,
+      "loss": 1.3229,
+      "step": 298900
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.10430612244898e-05,
+      "loss": 1.3337,
+      "step": 299000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.102265306122449e-05,
+      "loss": 1.3301,
+      "step": 299100
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.100224489795918e-05,
+      "loss": 1.3248,
+      "step": 299200
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.098183673469388e-05,
+      "loss": 1.3266,
+      "step": 299300
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.096142857142857e-05,
+      "loss": 1.3258,
+      "step": 299400
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.0941020408163264e-05,
+      "loss": 1.3233,
+      "step": 299500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.092061224489796e-05,
+      "loss": 1.3349,
+      "step": 299600
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.0900204081632654e-05,
+      "loss": 1.3269,
+      "step": 299700
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.0879795918367346e-05,
+      "loss": 1.3183,
+      "step": 299800
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.0859387755102044e-05,
+      "loss": 1.3233,
+      "step": 299900
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.0838979591836736e-05,
+      "loss": 1.3278,
+      "step": 300000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.081857142857143e-05,
+      "loss": 1.3298,
+      "step": 300100
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.079816326530612e-05,
+      "loss": 1.3276,
+      "step": 300200
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.0777755102040824e-05,
+      "loss": 1.3239,
+      "step": 300300
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.0757346938775515e-05,
+      "loss": 1.3323,
+      "step": 300400
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.073693877551021e-05,
+      "loss": 1.329,
+      "step": 300500
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.07165306122449e-05,
+      "loss": 1.3304,
+      "step": 300600
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.06961224489796e-05,
+      "loss": 1.3177,
+      "step": 300700
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.067571428571429e-05,
+      "loss": 1.3218,
+      "step": 300800
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.065530612244898e-05,
+      "loss": 1.3271,
+      "step": 300900
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.063489795918368e-05,
+      "loss": 1.3224,
+      "step": 301000
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.0614693877551026e-05,
+      "loss": 1.3159,
+      "step": 301100
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.059428571428572e-05,
+      "loss": 1.3284,
+      "step": 301200
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.057387755102041e-05,
+      "loss": 1.3276,
+      "step": 301300
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.055346938775511e-05,
+      "loss": 1.3288,
+      "step": 301400
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.05330612244898e-05,
+      "loss": 1.3234,
+      "step": 301500
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.051265306122449e-05,
+      "loss": 1.3255,
+      "step": 301600
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.049224489795919e-05,
+      "loss": 1.3282,
+      "step": 301700
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.047183673469388e-05,
+      "loss": 1.3234,
+      "step": 301800
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.045142857142857e-05,
+      "loss": 1.3254,
+      "step": 301900
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.0431020408163264e-05,
+      "loss": 1.3226,
+      "step": 302000
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.041061224489796e-05,
+      "loss": 1.3284,
+      "step": 302100
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.0390204081632654e-05,
+      "loss": 1.3249,
+      "step": 302200
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.0369795918367345e-05,
+      "loss": 1.3257,
+      "step": 302300
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.0349387755102044e-05,
+      "loss": 1.3319,
+      "step": 302400
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.0328979591836735e-05,
+      "loss": 1.3203,
+      "step": 302500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.030857142857143e-05,
+      "loss": 1.3167,
+      "step": 302600
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.0288163265306125e-05,
+      "loss": 1.3221,
+      "step": 302700
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.026775510204082e-05,
+      "loss": 1.3205,
+      "step": 302800
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.024734693877551e-05,
+      "loss": 1.328,
+      "step": 302900
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.02269387755102e-05,
+      "loss": 1.3292,
+      "step": 303000
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.0206530612244905e-05,
+      "loss": 1.3229,
+      "step": 303100
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.0186326530612246e-05,
+      "loss": 1.3306,
+      "step": 303200
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.016591836734694e-05,
+      "loss": 1.326,
+      "step": 303300
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.014551020408163e-05,
+      "loss": 1.3293,
+      "step": 303400
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.012510204081633e-05,
+      "loss": 1.3209,
+      "step": 303500
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.0104693877551026e-05,
+      "loss": 1.3244,
+      "step": 303600
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.008428571428572e-05,
+      "loss": 1.3213,
+      "step": 303700
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.006387755102041e-05,
+      "loss": 1.3211,
+      "step": 303800
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.004346938775511e-05,
+      "loss": 1.3206,
+      "step": 303900
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.00230612244898e-05,
+      "loss": 1.3316,
+      "step": 304000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.000265306122449e-05,
+      "loss": 1.3205,
+      "step": 304100
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.998224489795919e-05,
+      "loss": 1.3256,
+      "step": 304200
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.996183673469388e-05,
+      "loss": 1.3185,
+      "step": 304300
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.994142857142857e-05,
+      "loss": 1.3267,
+      "step": 304400
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.992102040816327e-05,
+      "loss": 1.3304,
+      "step": 304500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.990061224489796e-05,
+      "loss": 1.3165,
+      "step": 304600
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9880204081632654e-05,
+      "loss": 1.326,
+      "step": 304700
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9859795918367345e-05,
+      "loss": 1.3202,
+      "step": 304800
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9839387755102044e-05,
+      "loss": 1.3171,
+      "step": 304900
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.9818979591836735e-05,
+      "loss": 1.323,
+      "step": 305000
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.979857142857143e-05,
+      "loss": 1.321,
+      "step": 305100
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.9778163265306125e-05,
+      "loss": 1.3249,
+      "step": 305200
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.975775510204082e-05,
+      "loss": 1.3184,
+      "step": 305300
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.973734693877551e-05,
+      "loss": 1.3143,
+      "step": 305400
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.971693877551021e-05,
+      "loss": 1.3207,
+      "step": 305500
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.96965306122449e-05,
+      "loss": 1.3271,
+      "step": 305600
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.967612244897959e-05,
+      "loss": 1.3215,
+      "step": 305700
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.965571428571429e-05,
+      "loss": 1.3137,
+      "step": 305800
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.9635306122448987e-05,
+      "loss": 1.3218,
+      "step": 305900
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.961489795918368e-05,
+      "loss": 1.3197,
+      "step": 306000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.959448979591837e-05,
+      "loss": 1.3151,
+      "step": 306100
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.957428571428571e-05,
+      "loss": 1.3235,
+      "step": 306200
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.955387755102041e-05,
+      "loss": 1.3356,
+      "step": 306300
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.953346938775511e-05,
+      "loss": 1.3146,
+      "step": 306400
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.95130612244898e-05,
+      "loss": 1.3164,
+      "step": 306500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.94926530612245e-05,
+      "loss": 1.3195,
+      "step": 306600
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.947224489795919e-05,
+      "loss": 1.3221,
+      "step": 306700
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.945183673469388e-05,
+      "loss": 1.3194,
+      "step": 306800
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.943142857142857e-05,
+      "loss": 1.3209,
+      "step": 306900
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.941102040816327e-05,
+      "loss": 1.3189,
+      "step": 307000
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.939061224489796e-05,
+      "loss": 1.3166,
+      "step": 307100
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.9370204081632653e-05,
+      "loss": 1.3185,
+      "step": 307200
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.934979591836735e-05,
+      "loss": 1.3212,
+      "step": 307300
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.932938775510204e-05,
+      "loss": 1.3183,
+      "step": 307400
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9308979591836735e-05,
+      "loss": 1.3144,
+      "step": 307500
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.928857142857143e-05,
+      "loss": 1.3148,
+      "step": 307600
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9268163265306125e-05,
+      "loss": 1.3184,
+      "step": 307700
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9247755102040816e-05,
+      "loss": 1.3162,
+      "step": 307800
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.922734693877551e-05,
+      "loss": 1.3134,
+      "step": 307900
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.9206938775510206e-05,
+      "loss": 1.3265,
+      "step": 308000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.91865306122449e-05,
+      "loss": 1.3164,
+      "step": 308100
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.916612244897959e-05,
+      "loss": 1.3278,
+      "step": 308200
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.914571428571429e-05,
+      "loss": 1.3117,
+      "step": 308300
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.912530612244898e-05,
+      "loss": 1.3206,
+      "step": 308400
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.910489795918367e-05,
+      "loss": 1.312,
+      "step": 308500
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.908448979591837e-05,
+      "loss": 1.3182,
+      "step": 308600
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.906408163265307e-05,
+      "loss": 1.3242,
+      "step": 308700
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.904367346938776e-05,
+      "loss": 1.3133,
+      "step": 308800
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.902326530612245e-05,
+      "loss": 1.3246,
+      "step": 308900
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.900285714285715e-05,
+      "loss": 1.319,
+      "step": 309000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.898244897959184e-05,
+      "loss": 1.3172,
+      "step": 309100
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.896204081632653e-05,
+      "loss": 1.3197,
+      "step": 309200
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.8941632653061224e-05,
+      "loss": 1.3124,
+      "step": 309300
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.892122448979592e-05,
+      "loss": 1.3159,
+      "step": 309400
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.8900816326530614e-05,
+      "loss": 1.3187,
+      "step": 309500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.8880408163265306e-05,
+      "loss": 1.3232,
+      "step": 309600
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.8860000000000004e-05,
+      "loss": 1.3144,
+      "step": 309700
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.8839591836734696e-05,
+      "loss": 1.3214,
+      "step": 309800
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.881918367346939e-05,
+      "loss": 1.3186,
+      "step": 309900
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.8798775510204086e-05,
+      "loss": 1.3225,
+      "step": 310000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.877836734693878e-05,
+      "loss": 1.3161,
+      "step": 310100
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.875795918367347e-05,
+      "loss": 1.3167,
+      "step": 310200
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.873755102040817e-05,
+      "loss": 1.3172,
+      "step": 310300
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.871714285714286e-05,
+      "loss": 1.3186,
+      "step": 310400
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.8696938775510206e-05,
+      "loss": 1.3218,
+      "step": 310500
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.86765306122449e-05,
+      "loss": 1.3192,
+      "step": 310600
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.865612244897959e-05,
+      "loss": 1.31,
+      "step": 310700
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.863571428571429e-05,
+      "loss": 1.3186,
+      "step": 310800
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.861530612244898e-05,
+      "loss": 1.3096,
+      "step": 310900
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.859489795918367e-05,
+      "loss": 1.3186,
+      "step": 311000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.857448979591837e-05,
+      "loss": 1.3208,
+      "step": 311100
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.855408163265306e-05,
+      "loss": 1.3205,
+      "step": 311200
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.853367346938775e-05,
+      "loss": 1.3187,
+      "step": 311300
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.85134693877551e-05,
+      "loss": 1.3121,
+      "step": 311400
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.84930612244898e-05,
+      "loss": 1.3198,
+      "step": 311500
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.847265306122449e-05,
+      "loss": 1.319,
+      "step": 311600
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.845224489795918e-05,
+      "loss": 1.3173,
+      "step": 311700
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.843183673469388e-05,
+      "loss": 1.3244,
+      "step": 311800
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.841142857142858e-05,
+      "loss": 1.315,
+      "step": 311900
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.839102040816327e-05,
+      "loss": 1.3205,
+      "step": 312000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.837061224489796e-05,
+      "loss": 1.3151,
+      "step": 312100
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.835020408163266e-05,
+      "loss": 1.3139,
+      "step": 312200
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.832979591836735e-05,
+      "loss": 1.3142,
+      "step": 312300
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.830938775510204e-05,
+      "loss": 1.3106,
+      "step": 312400
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.8288979591836735e-05,
+      "loss": 1.3155,
+      "step": 312500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.826857142857143e-05,
+      "loss": 1.3141,
+      "step": 312600
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.8248163265306125e-05,
+      "loss": 1.3135,
+      "step": 312700
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.8227755102040816e-05,
+      "loss": 1.3199,
+      "step": 312800
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.8207346938775514e-05,
+      "loss": 1.3079,
+      "step": 312900
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.8186938775510206e-05,
+      "loss": 1.3156,
+      "step": 313000
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.81665306122449e-05,
+      "loss": 1.3202,
+      "step": 313100
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.8146122448979596e-05,
+      "loss": 1.3177,
+      "step": 313200
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.812571428571429e-05,
+      "loss": 1.3097,
+      "step": 313300
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.810530612244898e-05,
+      "loss": 1.3173,
+      "step": 313400
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.808489795918367e-05,
+      "loss": 1.3215,
+      "step": 313500
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.806448979591837e-05,
+      "loss": 1.3134,
+      "step": 313600
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.804408163265306e-05,
+      "loss": 1.3152,
+      "step": 313700
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.802367346938775e-05,
+      "loss": 1.3109,
+      "step": 313800
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.800326530612245e-05,
+      "loss": 1.3119,
+      "step": 313900
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.798285714285714e-05,
+      "loss": 1.3127,
+      "step": 314000
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.7962448979591834e-05,
+      "loss": 1.3182,
+      "step": 314100
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.794204081632653e-05,
+      "loss": 1.3185,
+      "step": 314200
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.792163265306123e-05,
+      "loss": 1.3087,
+      "step": 314300
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.790122448979592e-05,
+      "loss": 1.3229,
+      "step": 314400
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.7880816326530614e-05,
+      "loss": 1.3167,
+      "step": 314500
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.786040816326531e-05,
+      "loss": 1.3142,
+      "step": 314600
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.7840000000000004e-05,
+      "loss": 1.3124,
+      "step": 314700
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.781979591836735e-05,
+      "loss": 1.3226,
+      "step": 314800
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.779938775510204e-05,
+      "loss": 1.3105,
+      "step": 314900
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.777897959183674e-05,
+      "loss": 1.3145,
+      "step": 315000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.775857142857143e-05,
+      "loss": 1.3144,
+      "step": 315100
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.773836734693878e-05,
+      "loss": 1.3147,
+      "step": 315200
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.771795918367347e-05,
+      "loss": 1.3084,
+      "step": 315300
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.769755102040817e-05,
+      "loss": 1.3124,
+      "step": 315400
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.767714285714286e-05,
+      "loss": 1.3171,
+      "step": 315500
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.7656734693877553e-05,
+      "loss": 1.3134,
+      "step": 315600
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.7636326530612245e-05,
+      "loss": 1.3104,
+      "step": 315700
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.761591836734694e-05,
+      "loss": 1.3171,
+      "step": 315800
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.7595510204081635e-05,
+      "loss": 1.3114,
+      "step": 315900
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.7575102040816327e-05,
+      "loss": 1.3132,
+      "step": 316000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.7554693877551025e-05,
+      "loss": 1.3143,
+      "step": 316100
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.7534285714285716e-05,
+      "loss": 1.3085,
+      "step": 316200
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.751387755102041e-05,
+      "loss": 1.3163,
+      "step": 316300
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.7493469387755106e-05,
+      "loss": 1.3171,
+      "step": 316400
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.74730612244898e-05,
+      "loss": 1.3019,
+      "step": 316500
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.745265306122449e-05,
+      "loss": 1.3107,
+      "step": 316600
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.743224489795918e-05,
+      "loss": 1.314,
+      "step": 316700
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.741183673469388e-05,
+      "loss": 1.3068,
+      "step": 316800
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.739142857142857e-05,
+      "loss": 1.3174,
+      "step": 316900
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.737102040816326e-05,
+      "loss": 1.3094,
+      "step": 317000
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.735061224489796e-05,
+      "loss": 1.3182,
+      "step": 317100
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.733020408163265e-05,
+      "loss": 1.3152,
+      "step": 317200
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.7309795918367344e-05,
+      "loss": 1.3123,
+      "step": 317300
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.728938775510204e-05,
+      "loss": 1.3154,
+      "step": 317400
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.726897959183674e-05,
+      "loss": 1.3166,
+      "step": 317500
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.724857142857143e-05,
+      "loss": 1.3154,
+      "step": 317600
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.7228163265306124e-05,
+      "loss": 1.3135,
+      "step": 317700
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.720775510204082e-05,
+      "loss": 1.3085,
+      "step": 317800
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.7187346938775514e-05,
+      "loss": 1.3059,
+      "step": 317900
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.7166938775510206e-05,
+      "loss": 1.311,
+      "step": 318000
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.7146530612244904e-05,
+      "loss": 1.3134,
+      "step": 318100
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.7126122448979596e-05,
+      "loss": 1.311,
+      "step": 318200
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.710571428571429e-05,
+      "loss": 1.3127,
+      "step": 318300
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.708530612244898e-05,
+      "loss": 1.3133,
+      "step": 318400
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.706489795918368e-05,
+      "loss": 1.3104,
+      "step": 318500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.704448979591837e-05,
+      "loss": 1.3058,
+      "step": 318600
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.702408163265306e-05,
+      "loss": 1.3101,
+      "step": 318700
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.700367346938776e-05,
+      "loss": 1.3102,
+      "step": 318800
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.698326530612245e-05,
+      "loss": 1.3169,
+      "step": 318900
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.696285714285714e-05,
+      "loss": 1.3097,
+      "step": 319000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.694244897959184e-05,
+      "loss": 1.3084,
+      "step": 319100
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.692204081632653e-05,
+      "loss": 1.3071,
+      "step": 319200
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.690163265306122e-05,
+      "loss": 1.3156,
+      "step": 319300
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.6881224489795915e-05,
+      "loss": 1.3128,
+      "step": 319400
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.686081632653061e-05,
+      "loss": 1.3094,
+      "step": 319500
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.684061224489796e-05,
+      "loss": 1.3172,
+      "step": 319600
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.682020408163265e-05,
+      "loss": 1.3105,
+      "step": 319700
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.6799795918367344e-05,
+      "loss": 1.3099,
+      "step": 319800
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.677938775510204e-05,
+      "loss": 1.3064,
+      "step": 319900
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.6758979591836734e-05,
+      "loss": 1.3207,
+      "step": 320000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.6738571428571426e-05,
+      "loss": 1.3059,
+      "step": 320100
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.6718163265306124e-05,
+      "loss": 1.3141,
+      "step": 320200
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.669775510204082e-05,
+      "loss": 1.3138,
+      "step": 320300
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.6677346938775514e-05,
+      "loss": 1.3121,
+      "step": 320400
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.6656938775510205e-05,
+      "loss": 1.3124,
+      "step": 320500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.6636530612244904e-05,
+      "loss": 1.3116,
+      "step": 320600
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.6616122448979595e-05,
+      "loss": 1.3113,
+      "step": 320700
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.659571428571429e-05,
+      "loss": 1.3055,
+      "step": 320800
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.6575306122448985e-05,
+      "loss": 1.3192,
+      "step": 320900
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.655489795918368e-05,
+      "loss": 1.3088,
+      "step": 321000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.653448979591837e-05,
+      "loss": 1.3121,
+      "step": 321100
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.651408163265306e-05,
+      "loss": 1.3061,
+      "step": 321200
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.649367346938776e-05,
+      "loss": 1.3089,
+      "step": 321300
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.647326530612245e-05,
+      "loss": 1.3049,
+      "step": 321400
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.645285714285714e-05,
+      "loss": 1.3,
+      "step": 321500
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.643244897959184e-05,
+      "loss": 1.311,
+      "step": 321600
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.641204081632653e-05,
+      "loss": 1.3216,
+      "step": 321700
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.639163265306122e-05,
+      "loss": 1.3031,
+      "step": 321800
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.637142857142857e-05,
+      "loss": 1.3118,
+      "step": 321900
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.635102040816327e-05,
+      "loss": 1.3147,
+      "step": 322000
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.633061224489796e-05,
+      "loss": 1.3097,
+      "step": 322100
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.631020408163265e-05,
+      "loss": 1.3123,
+      "step": 322200
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.628979591836735e-05,
+      "loss": 1.3118,
+      "step": 322300
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.626938775510204e-05,
+      "loss": 1.3127,
+      "step": 322400
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.6248979591836734e-05,
+      "loss": 1.3042,
+      "step": 322500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.6228571428571425e-05,
+      "loss": 1.311,
+      "step": 322600
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.6208163265306124e-05,
+      "loss": 1.3126,
+      "step": 322700
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.6187755102040815e-05,
+      "loss": 1.3107,
+      "step": 322800
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.616734693877551e-05,
+      "loss": 1.3204,
+      "step": 322900
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.6146938775510205e-05,
+      "loss": 1.3122,
+      "step": 323000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.6126530612244904e-05,
+      "loss": 1.3043,
+      "step": 323100
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.6106122448979595e-05,
+      "loss": 1.3094,
+      "step": 323200
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.608571428571429e-05,
+      "loss": 1.3133,
+      "step": 323300
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.6065306122448985e-05,
+      "loss": 1.3099,
+      "step": 323400
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.604489795918368e-05,
+      "loss": 1.3089,
+      "step": 323500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.602448979591837e-05,
+      "loss": 1.3064,
+      "step": 323600
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.600408163265307e-05,
+      "loss": 1.301,
+      "step": 323700
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.598367346938776e-05,
+      "loss": 1.3134,
+      "step": 323800
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.596326530612245e-05,
+      "loss": 1.3116,
+      "step": 323900
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.594285714285714e-05,
+      "loss": 1.3117,
+      "step": 324000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.592244897959184e-05,
+      "loss": 1.3162,
+      "step": 324100
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.590204081632653e-05,
+      "loss": 1.3099,
+      "step": 324200
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.588163265306122e-05,
+      "loss": 1.315,
+      "step": 324300
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.586122448979592e-05,
+      "loss": 1.3109,
+      "step": 324400
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.584081632653061e-05,
+      "loss": 1.3116,
+      "step": 324500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.5820408163265304e-05,
+      "loss": 1.3056,
+      "step": 324600
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.58e-05,
+      "loss": 1.3085,
+      "step": 324700
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.577979591836735e-05,
+      "loss": 1.3095,
+      "step": 324800
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.575938775510204e-05,
+      "loss": 1.3091,
+      "step": 324900
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.5738979591836734e-05,
+      "loss": 1.3172,
+      "step": 325000
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.571857142857143e-05,
+      "loss": 1.3093,
+      "step": 325100
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.5698163265306124e-05,
+      "loss": 1.3083,
+      "step": 325200
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.5677755102040815e-05,
+      "loss": 1.3097,
+      "step": 325300
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.565734693877551e-05,
+      "loss": 1.3098,
+      "step": 325400
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.5636938775510205e-05,
+      "loss": 1.2987,
+      "step": 325500
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.56165306122449e-05,
+      "loss": 1.3139,
+      "step": 325600
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.559612244897959e-05,
+      "loss": 1.3119,
+      "step": 325700
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.557571428571429e-05,
+      "loss": 1.3087,
+      "step": 325800
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.5555306122448985e-05,
+      "loss": 1.3107,
+      "step": 325900
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.5534897959183677e-05,
+      "loss": 1.3146,
+      "step": 326000
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.551448979591837e-05,
+      "loss": 1.3141,
+      "step": 326100
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.5494081632653066e-05,
+      "loss": 1.3091,
+      "step": 326200
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.547367346938776e-05,
+      "loss": 1.3104,
+      "step": 326300
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.545326530612245e-05,
+      "loss": 1.3057,
+      "step": 326400
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.543285714285715e-05,
+      "loss": 1.3078,
+      "step": 326500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.541244897959184e-05,
+      "loss": 1.3033,
+      "step": 326600
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.539204081632653e-05,
+      "loss": 1.3005,
+      "step": 326700
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.537163265306123e-05,
+      "loss": 1.3044,
+      "step": 326800
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.535122448979592e-05,
+      "loss": 1.3139,
+      "step": 326900
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.533081632653061e-05,
+      "loss": 1.3003,
+      "step": 327000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.5310408163265304e-05,
+      "loss": 1.2999,
+      "step": 327100
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.529e-05,
+      "loss": 1.3091,
+      "step": 327200
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.5269591836734694e-05,
+      "loss": 1.306,
+      "step": 327300
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.5249183673469386e-05,
+      "loss": 1.3001,
+      "step": 327400
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.5228775510204084e-05,
+      "loss": 1.3039,
+      "step": 327500
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.5208367346938776e-05,
+      "loss": 1.3114,
+      "step": 327600
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.518795918367347e-05,
+      "loss": 1.3011,
+      "step": 327700
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.5167755102040815e-05,
+      "loss": 1.31,
+      "step": 327800
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.514734693877551e-05,
+      "loss": 1.3121,
+      "step": 327900
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.5126938775510205e-05,
+      "loss": 1.3061,
+      "step": 328000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.5106530612244896e-05,
+      "loss": 1.303,
+      "step": 328100
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.5086122448979595e-05,
+      "loss": 1.309,
+      "step": 328200
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.5065714285714286e-05,
+      "loss": 1.3098,
+      "step": 328300
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.504530612244898e-05,
+      "loss": 1.3101,
+      "step": 328400
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.5024897959183676e-05,
+      "loss": 1.3033,
+      "step": 328500
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.5004489795918375e-05,
+      "loss": 1.308,
+      "step": 328600
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.4984081632653066e-05,
+      "loss": 1.3056,
+      "step": 328700
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.496367346938776e-05,
+      "loss": 1.3109,
+      "step": 328800
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.494326530612245e-05,
+      "loss": 1.3041,
+      "step": 328900
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.492285714285715e-05,
+      "loss": 1.3021,
+      "step": 329000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.490244897959184e-05,
+      "loss": 1.306,
+      "step": 329100
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.488204081632653e-05,
+      "loss": 1.3007,
+      "step": 329200
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.486163265306123e-05,
+      "loss": 1.3063,
+      "step": 329300
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.484122448979592e-05,
+      "loss": 1.3068,
+      "step": 329400
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.482081632653061e-05,
+      "loss": 1.3096,
+      "step": 329500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.480040816326531e-05,
+      "loss": 1.3119,
+      "step": 329600
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.478e-05,
+      "loss": 1.3046,
+      "step": 329700
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.4759591836734694e-05,
+      "loss": 1.3103,
+      "step": 329800
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.4739183673469386e-05,
+      "loss": 1.3052,
+      "step": 329900
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.4718775510204084e-05,
+      "loss": 1.3097,
+      "step": 330000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.4698367346938776e-05,
+      "loss": 1.3139,
+      "step": 330100
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.467795918367347e-05,
+      "loss": 1.3062,
+      "step": 330200
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.4657551020408165e-05,
+      "loss": 1.3089,
+      "step": 330300
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.463714285714286e-05,
+      "loss": 1.3098,
+      "step": 330400
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.4616938775510205e-05,
+      "loss": 1.3033,
+      "step": 330500
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.4596530612244896e-05,
+      "loss": 1.307,
+      "step": 330600
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.4576122448979595e-05,
+      "loss": 1.3047,
+      "step": 330700
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.4555714285714286e-05,
+      "loss": 1.3163,
+      "step": 330800
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.453530612244898e-05,
+      "loss": 1.3105,
+      "step": 330900
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.4514897959183676e-05,
+      "loss": 1.3024,
+      "step": 331000
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.449448979591837e-05,
+      "loss": 1.2983,
+      "step": 331100
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.447408163265306e-05,
+      "loss": 1.3086,
+      "step": 331200
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.445367346938776e-05,
+      "loss": 1.301,
+      "step": 331300
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.4433265306122456e-05,
+      "loss": 1.3044,
+      "step": 331400
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.441285714285715e-05,
+      "loss": 1.3011,
+      "step": 331500
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.439244897959184e-05,
+      "loss": 1.3021,
+      "step": 331600
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.437204081632653e-05,
+      "loss": 1.3072,
+      "step": 331700
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.435163265306123e-05,
+      "loss": 1.2992,
+      "step": 331800
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.433122448979592e-05,
+      "loss": 1.2948,
+      "step": 331900
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.431081632653061e-05,
+      "loss": 1.3023,
+      "step": 332000
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.429040816326531e-05,
+      "loss": 1.2973,
+      "step": 332100
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.427e-05,
+      "loss": 1.308,
+      "step": 332200
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.4249591836734694e-05,
+      "loss": 1.2983,
+      "step": 332300
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.422918367346939e-05,
+      "loss": 1.3004,
+      "step": 332400
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.4208775510204084e-05,
+      "loss": 1.2989,
+      "step": 332500
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.4188367346938775e-05,
+      "loss": 1.298,
+      "step": 332600
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.416795918367347e-05,
+      "loss": 1.2955,
+      "step": 332700
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.4147551020408165e-05,
+      "loss": 1.3023,
+      "step": 332800
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.412714285714286e-05,
+      "loss": 1.3,
+      "step": 332900
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.410673469387755e-05,
+      "loss": 1.2944,
+      "step": 333000
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.408632653061225e-05,
+      "loss": 1.3052,
+      "step": 333100
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.406591836734694e-05,
+      "loss": 1.3006,
+      "step": 333200
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.4045714285714286e-05,
+      "loss": 1.2956,
+      "step": 333300
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.402530612244898e-05,
+      "loss": 1.3024,
+      "step": 333400
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.4004897959183676e-05,
+      "loss": 1.3043,
+      "step": 333500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.398448979591837e-05,
+      "loss": 1.2999,
+      "step": 333600
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.396408163265306e-05,
+      "loss": 1.3052,
+      "step": 333700
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.394367346938776e-05,
+      "loss": 1.2954,
+      "step": 333800
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.392326530612245e-05,
+      "loss": 1.3117,
+      "step": 333900
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.390285714285714e-05,
+      "loss": 1.3017,
+      "step": 334000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.388244897959184e-05,
+      "loss": 1.3048,
+      "step": 334100
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.386204081632654e-05,
+      "loss": 1.299,
+      "step": 334200
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.384163265306123e-05,
+      "loss": 1.3038,
+      "step": 334300
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.382122448979592e-05,
+      "loss": 1.3089,
+      "step": 334400
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.380081632653061e-05,
+      "loss": 1.2962,
+      "step": 334500
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.378040816326531e-05,
+      "loss": 1.2973,
+      "step": 334600
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.376e-05,
+      "loss": 1.3028,
+      "step": 334700
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.3739591836734694e-05,
+      "loss": 1.2994,
+      "step": 334800
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.371918367346939e-05,
+      "loss": 1.2988,
+      "step": 334900
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.3698775510204084e-05,
+      "loss": 1.301,
+      "step": 335000
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.3678367346938775e-05,
+      "loss": 1.3051,
+      "step": 335100
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.3657959183673474e-05,
+      "loss": 1.3061,
+      "step": 335200
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.3637551020408165e-05,
+      "loss": 1.3036,
+      "step": 335300
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.361714285714286e-05,
+      "loss": 1.3026,
+      "step": 335400
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.3596734693877555e-05,
+      "loss": 1.3025,
+      "step": 335500
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.357632653061225e-05,
+      "loss": 1.2965,
+      "step": 335600
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.355591836734694e-05,
+      "loss": 1.2959,
+      "step": 335700
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.353551020408163e-05,
+      "loss": 1.2977,
+      "step": 335800
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.351510204081633e-05,
+      "loss": 1.305,
+      "step": 335900
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.349469387755102e-05,
+      "loss": 1.3095,
+      "step": 336000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.347428571428571e-05,
+      "loss": 1.2972,
+      "step": 336100
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.345387755102041e-05,
+      "loss": 1.293,
+      "step": 336200
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.343346938775511e-05,
+      "loss": 1.3013,
+      "step": 336300
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.34130612244898e-05,
+      "loss": 1.2942,
+      "step": 336400
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.339265306122449e-05,
+      "loss": 1.3028,
+      "step": 336500
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.337224489795919e-05,
+      "loss": 1.2994,
+      "step": 336600
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.335183673469388e-05,
+      "loss": 1.2935,
+      "step": 336700
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.333142857142857e-05,
+      "loss": 1.2984,
+      "step": 336800
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.331102040816327e-05,
+      "loss": 1.293,
+      "step": 336900
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.329061224489796e-05,
+      "loss": 1.2963,
+      "step": 337000
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.3270204081632654e-05,
+      "loss": 1.2949,
+      "step": 337100
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.3249795918367346e-05,
+      "loss": 1.3018,
+      "step": 337200
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.3229387755102044e-05,
+      "loss": 1.301,
+      "step": 337300
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.320918367346939e-05,
+      "loss": 1.2994,
+      "step": 337400
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.318877551020408e-05,
+      "loss": 1.2977,
+      "step": 337500
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.3168367346938775e-05,
+      "loss": 1.2964,
+      "step": 337600
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.314795918367347e-05,
+      "loss": 1.3011,
+      "step": 337700
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.3127551020408165e-05,
+      "loss": 1.2955,
+      "step": 337800
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.3107142857142856e-05,
+      "loss": 1.3005,
+      "step": 337900
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.3086734693877555e-05,
+      "loss": 1.2973,
+      "step": 338000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.3066326530612246e-05,
+      "loss": 1.2966,
+      "step": 338100
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.304591836734694e-05,
+      "loss": 1.309,
+      "step": 338200
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.3025510204081636e-05,
+      "loss": 1.301,
+      "step": 338300
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.300510204081633e-05,
+      "loss": 1.2924,
+      "step": 338400
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.298469387755102e-05,
+      "loss": 1.2964,
+      "step": 338500
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.296428571428571e-05,
+      "loss": 1.3052,
+      "step": 338600
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.294387755102041e-05,
+      "loss": 1.2924,
+      "step": 338700
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.29234693877551e-05,
+      "loss": 1.3044,
+      "step": 338800
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.290306122448979e-05,
+      "loss": 1.2986,
+      "step": 338900
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.288265306122449e-05,
+      "loss": 1.2973,
+      "step": 339000
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.286224489795919e-05,
+      "loss": 1.2934,
+      "step": 339100
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.284183673469388e-05,
+      "loss": 1.2973,
+      "step": 339200
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.282142857142857e-05,
+      "loss": 1.3004,
+      "step": 339300
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.280102040816327e-05,
+      "loss": 1.3021,
+      "step": 339400
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.278061224489796e-05,
+      "loss": 1.2899,
+      "step": 339500
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.276040816326531e-05,
+      "loss": 1.2916,
+      "step": 339600
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.274e-05,
+      "loss": 1.2915,
+      "step": 339700
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.27195918367347e-05,
+      "loss": 1.2867,
+      "step": 339800
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.269918367346939e-05,
+      "loss": 1.2958,
+      "step": 339900
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.267877551020408e-05,
+      "loss": 1.2957,
+      "step": 340000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.265836734693878e-05,
+      "loss": 1.2919,
+      "step": 340100
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.263795918367347e-05,
+      "loss": 1.2972,
+      "step": 340200
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.2617551020408165e-05,
+      "loss": 1.2976,
+      "step": 340300
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.2597142857142856e-05,
+      "loss": 1.2939,
+      "step": 340400
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.2576734693877555e-05,
+      "loss": 1.2917,
+      "step": 340500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.2556326530612246e-05,
+      "loss": 1.2998,
+      "step": 340600
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.253591836734694e-05,
+      "loss": 1.3076,
+      "step": 340700
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.2515510204081636e-05,
+      "loss": 1.3019,
+      "step": 340800
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.2495306122448984e-05,
+      "loss": 1.2993,
+      "step": 340900
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.2474897959183675e-05,
+      "loss": 1.3,
+      "step": 341000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.245448979591837e-05,
+      "loss": 1.3019,
+      "step": 341100
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.2434081632653065e-05,
+      "loss": 1.297,
+      "step": 341200
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.241367346938776e-05,
+      "loss": 1.2942,
+      "step": 341300
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.239326530612245e-05,
+      "loss": 1.3009,
+      "step": 341400
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.237285714285715e-05,
+      "loss": 1.2914,
+      "step": 341500
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.235244897959184e-05,
+      "loss": 1.3011,
+      "step": 341600
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.233204081632653e-05,
+      "loss": 1.2988,
+      "step": 341700
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.231163265306122e-05,
+      "loss": 1.2929,
+      "step": 341800
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.229122448979592e-05,
+      "loss": 1.2994,
+      "step": 341900
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.227081632653061e-05,
+      "loss": 1.2908,
+      "step": 342000
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.22504081632653e-05,
+      "loss": 1.296,
+      "step": 342100
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.223e-05,
+      "loss": 1.3053,
+      "step": 342200
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.220959183673469e-05,
+      "loss": 1.2978,
+      "step": 342300
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.218918367346939e-05,
+      "loss": 1.291,
+      "step": 342400
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.216877551020408e-05,
+      "loss": 1.2957,
+      "step": 342500
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.214836734693878e-05,
+      "loss": 1.2954,
+      "step": 342600
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.212816326530612e-05,
+      "loss": 1.2906,
+      "step": 342700
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.2107755102040814e-05,
+      "loss": 1.303,
+      "step": 342800
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.208734693877551e-05,
+      "loss": 1.294,
+      "step": 342900
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.206693877551021e-05,
+      "loss": 1.2978,
+      "step": 343000
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.20465306122449e-05,
+      "loss": 1.3041,
+      "step": 343100
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.2026122448979594e-05,
+      "loss": 1.2926,
+      "step": 343200
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.200571428571429e-05,
+      "loss": 1.2953,
+      "step": 343300
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.1985306122448984e-05,
+      "loss": 1.2961,
+      "step": 343400
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.1964897959183675e-05,
+      "loss": 1.2883,
+      "step": 343500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.194448979591837e-05,
+      "loss": 1.2871,
+      "step": 343600
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.1924081632653065e-05,
+      "loss": 1.2943,
+      "step": 343700
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.190367346938776e-05,
+      "loss": 1.2927,
+      "step": 343800
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.188326530612245e-05,
+      "loss": 1.296,
+      "step": 343900
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.1862857142857147e-05,
+      "loss": 1.3018,
+      "step": 344000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.184244897959184e-05,
+      "loss": 1.2908,
+      "step": 344100
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.182204081632653e-05,
+      "loss": 1.2863,
+      "step": 344200
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.180163265306123e-05,
+      "loss": 1.2922,
+      "step": 344300
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.178122448979592e-05,
+      "loss": 1.2941,
+      "step": 344400
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.176081632653061e-05,
+      "loss": 1.2915,
+      "step": 344500
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.17404081632653e-05,
+      "loss": 1.2908,
+      "step": 344600
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.172e-05,
+      "loss": 1.2966,
+      "step": 344700
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.169959183673469e-05,
+      "loss": 1.2973,
+      "step": 344800
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.1679183673469384e-05,
+      "loss": 1.2879,
+      "step": 344900
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.165877551020408e-05,
+      "loss": 1.297,
+      "step": 345000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.163836734693878e-05,
+      "loss": 1.2983,
+      "step": 345100
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.161795918367347e-05,
+      "loss": 1.295,
+      "step": 345200
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.1597551020408164e-05,
+      "loss": 1.291,
+      "step": 345300
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.157714285714286e-05,
+      "loss": 1.2889,
+      "step": 345400
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.1556734693877554e-05,
+      "loss": 1.2996,
+      "step": 345500
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.1536326530612246e-05,
+      "loss": 1.2922,
+      "step": 345600
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.1515918367346944e-05,
+      "loss": 1.2941,
+      "step": 345700
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.1495510204081636e-05,
+      "loss": 1.2859,
+      "step": 345800
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.147510204081633e-05,
+      "loss": 1.2844,
+      "step": 345900
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.1454693877551026e-05,
+      "loss": 1.2953,
+      "step": 346000
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.143428571428572e-05,
+      "loss": 1.2956,
+      "step": 346100
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.141387755102041e-05,
+      "loss": 1.29,
+      "step": 346200
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.13934693877551e-05,
+      "loss": 1.2898,
+      "step": 346300
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.13730612244898e-05,
+      "loss": 1.2926,
+      "step": 346400
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.135265306122449e-05,
+      "loss": 1.2953,
+      "step": 346500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.133224489795918e-05,
+      "loss": 1.2867,
+      "step": 346600
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.131183673469388e-05,
+      "loss": 1.288,
+      "step": 346700
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.129142857142857e-05,
+      "loss": 1.2917,
+      "step": 346800
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.1271020408163264e-05,
+      "loss": 1.2945,
+      "step": 346900
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.125061224489796e-05,
+      "loss": 1.2923,
+      "step": 347000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.1230204081632653e-05,
+      "loss": 1.2895,
+      "step": 347100
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.1209795918367345e-05,
+      "loss": 1.2969,
+      "step": 347200
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.118938775510204e-05,
+      "loss": 1.3009,
+      "step": 347300
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.116897959183674e-05,
+      "loss": 1.2961,
+      "step": 347400
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.114857142857143e-05,
+      "loss": 1.2925,
+      "step": 347500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.1128163265306125e-05,
+      "loss": 1.294,
+      "step": 347600
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.1107755102040817e-05,
+      "loss": 1.2896,
+      "step": 347700
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.1087346938775515e-05,
+      "loss": 1.2941,
+      "step": 347800
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.106714285714286e-05,
+      "loss": 1.2905,
+      "step": 347900
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.1046734693877554e-05,
+      "loss": 1.2898,
+      "step": 348000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.1026326530612246e-05,
+      "loss": 1.2947,
+      "step": 348100
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.1005918367346944e-05,
+      "loss": 1.2925,
+      "step": 348200
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.0985510204081636e-05,
+      "loss": 1.2983,
+      "step": 348300
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.096510204081633e-05,
+      "loss": 1.292,
+      "step": 348400
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.0944693877551026e-05,
+      "loss": 1.2885,
+      "step": 348500
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.092428571428572e-05,
+      "loss": 1.3001,
+      "step": 348600
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.090387755102041e-05,
+      "loss": 1.2919,
+      "step": 348700
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.088346938775511e-05,
+      "loss": 1.292,
+      "step": 348800
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.08630612244898e-05,
+      "loss": 1.2971,
+      "step": 348900
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.084265306122449e-05,
+      "loss": 1.2983,
+      "step": 349000
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.082224489795918e-05,
+      "loss": 1.2907,
+      "step": 349100
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.080183673469388e-05,
+      "loss": 1.2945,
+      "step": 349200
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.078142857142857e-05,
+      "loss": 1.2935,
+      "step": 349300
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.076102040816326e-05,
+      "loss": 1.2901,
+      "step": 349400
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.074061224489796e-05,
+      "loss": 1.2951,
+      "step": 349500
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.072020408163265e-05,
+      "loss": 1.2873,
+      "step": 349600
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.0699795918367345e-05,
+      "loss": 1.2834,
+      "step": 349700
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.067938775510204e-05,
+      "loss": 1.293,
+      "step": 349800
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.0658979591836735e-05,
+      "loss": 1.2874,
+      "step": 349900
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.063877551020408e-05,
+      "loss": 1.2897,
+      "step": 350000
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.0618367346938774e-05,
+      "loss": 1.2959,
+      "step": 350100
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.059795918367347e-05,
+      "loss": 1.293,
+      "step": 350200
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.0577551020408164e-05,
+      "loss": 1.2925,
+      "step": 350300
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.0557142857142855e-05,
+      "loss": 1.285,
+      "step": 350400
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.053673469387755e-05,
+      "loss": 1.2925,
+      "step": 350500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.0516326530612242e-05,
+      "loss": 1.2962,
+      "step": 350600
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.0495918367346944e-05,
+      "loss": 1.2846,
+      "step": 350700
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.0475510204081635e-05,
+      "loss": 1.2923,
+      "step": 350800
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.045510204081633e-05,
+      "loss": 1.2885,
+      "step": 350900
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.0434693877551025e-05,
+      "loss": 1.285,
+      "step": 351000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.0414285714285717e-05,
+      "loss": 1.2884,
+      "step": 351100
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.0393877551020412e-05,
+      "loss": 1.2922,
+      "step": 351200
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.0373469387755103e-05,
+      "loss": 1.2928,
+      "step": 351300
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.03530612244898e-05,
+      "loss": 1.2952,
+      "step": 351400
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.0332653061224493e-05,
+      "loss": 1.2903,
+      "step": 351500
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.0312244897959185e-05,
+      "loss": 1.292,
+      "step": 351600
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.029183673469388e-05,
+      "loss": 1.2913,
+      "step": 351700
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.027142857142857e-05,
+      "loss": 1.2856,
+      "step": 351800
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.0251020408163266e-05,
+      "loss": 1.2961,
+      "step": 351900
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.023061224489796e-05,
+      "loss": 1.2861,
+      "step": 352000
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.0210204081632653e-05,
+      "loss": 1.2824,
+      "step": 352100
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.0189795918367348e-05,
+      "loss": 1.2957,
+      "step": 352200
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.016938775510204e-05,
+      "loss": 1.29,
+      "step": 352300
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.0148979591836735e-05,
+      "loss": 1.2917,
+      "step": 352400
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.012857142857143e-05,
+      "loss": 1.295,
+      "step": 352500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.010816326530612e-05,
+      "loss": 1.2948,
+      "step": 352600
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.0087755102040816e-05,
+      "loss": 1.2868,
+      "step": 352700
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.0067346938775508e-05,
+      "loss": 1.2895,
+      "step": 352800
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.0046938775510203e-05,
+      "loss": 1.2917,
+      "step": 352900
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.00265306122449e-05,
+      "loss": 1.2942,
+      "step": 353000
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.0006122448979596e-05,
+      "loss": 1.2845,
+      "step": 353100
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.9985714285714288e-05,
+      "loss": 1.2915,
+      "step": 353200
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.9965306122448983e-05,
+      "loss": 1.2916,
+      "step": 353300
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.9944897959183678e-05,
+      "loss": 1.2915,
+      "step": 353400
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.992448979591837e-05,
+      "loss": 1.2914,
+      "step": 353500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.9904081632653064e-05,
+      "loss": 1.2877,
+      "step": 353600
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.988367346938776e-05,
+      "loss": 1.2855,
+      "step": 353700
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.986326530612245e-05,
+      "loss": 1.2848,
+      "step": 353800
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.9842857142857146e-05,
+      "loss": 1.2856,
+      "step": 353900
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.9822448979591837e-05,
+      "loss": 1.286,
+      "step": 354000
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.9802040816326532e-05,
+      "loss": 1.2871,
+      "step": 354100
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.9781632653061227e-05,
+      "loss": 1.2923,
+      "step": 354200
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.9761428571428575e-05,
+      "loss": 1.2878,
+      "step": 354300
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.9741020408163266e-05,
+      "loss": 1.2877,
+      "step": 354400
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.972061224489796e-05,
+      "loss": 1.2887,
+      "step": 354500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.9700204081632653e-05,
+      "loss": 1.2844,
+      "step": 354600
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.9679795918367348e-05,
+      "loss": 1.2941,
+      "step": 354700
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.9659387755102043e-05,
+      "loss": 1.2875,
+      "step": 354800
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.9638979591836734e-05,
+      "loss": 1.2861,
+      "step": 354900
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.961857142857143e-05,
+      "loss": 1.2888,
+      "step": 355000
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.959816326530612e-05,
+      "loss": 1.2912,
+      "step": 355100
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.9577755102040816e-05,
+      "loss": 1.2893,
+      "step": 355200
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.9557551020408163e-05,
+      "loss": 1.2963,
+      "step": 355300
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.953714285714286e-05,
+      "loss": 1.2893,
+      "step": 355400
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.951673469387755e-05,
+      "loss": 1.2925,
+      "step": 355500
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.9496326530612245e-05,
+      "loss": 1.2886,
+      "step": 355600
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.947591836734694e-05,
+      "loss": 1.2839,
+      "step": 355700
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.945551020408163e-05,
+      "loss": 1.2912,
+      "step": 355800
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.9435102040816327e-05,
+      "loss": 1.2853,
+      "step": 355900
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.9414693877551018e-05,
+      "loss": 1.2883,
+      "step": 356000
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.9394285714285713e-05,
+      "loss": 1.2924,
+      "step": 356100
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.937387755102041e-05,
+      "loss": 1.2887,
+      "step": 356200
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.9353469387755106e-05,
+      "loss": 1.2871,
+      "step": 356300
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.9333061224489798e-05,
+      "loss": 1.2898,
+      "step": 356400
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.9312653061224493e-05,
+      "loss": 1.2974,
+      "step": 356500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.9292244897959188e-05,
+      "loss": 1.2853,
+      "step": 356600
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.927183673469388e-05,
+      "loss": 1.2917,
+      "step": 356700
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.9251428571428575e-05,
+      "loss": 1.2844,
+      "step": 356800
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.9231020408163266e-05,
+      "loss": 1.2825,
+      "step": 356900
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.921061224489796e-05,
+      "loss": 1.2787,
+      "step": 357000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.9190204081632656e-05,
+      "loss": 1.2816,
+      "step": 357100
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.9169795918367348e-05,
+      "loss": 1.2835,
+      "step": 357200
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.9149387755102043e-05,
+      "loss": 1.288,
+      "step": 357300
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.9128979591836734e-05,
+      "loss": 1.2838,
+      "step": 357400
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.910857142857143e-05,
+      "loss": 1.2791,
+      "step": 357500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.9088163265306124e-05,
+      "loss": 1.2919,
+      "step": 357600
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.9067755102040816e-05,
+      "loss": 1.2882,
+      "step": 357700
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.904734693877551e-05,
+      "loss": 1.2906,
+      "step": 357800
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.9026938775510206e-05,
+      "loss": 1.2883,
+      "step": 357900
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.9006530612244897e-05,
+      "loss": 1.2857,
+      "step": 358000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.8986122448979592e-05,
+      "loss": 1.2797,
+      "step": 358100
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.8965714285714284e-05,
+      "loss": 1.2811,
+      "step": 358200
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.894551020408163e-05,
+      "loss": 1.2908,
+      "step": 358300
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.8925102040816326e-05,
+      "loss": 1.2872,
+      "step": 358400
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.890469387755102e-05,
+      "loss": 1.2767,
+      "step": 358500
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.8884285714285713e-05,
+      "loss": 1.2941,
+      "step": 358600
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.8863877551020408e-05,
+      "loss": 1.2854,
+      "step": 358700
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.88434693877551e-05,
+      "loss": 1.2763,
+      "step": 358800
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.8823061224489794e-05,
+      "loss": 1.2837,
+      "step": 358900
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.8802653061224493e-05,
+      "loss": 1.278,
+      "step": 359000
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.8782448979591837e-05,
+      "loss": 1.2906,
+      "step": 359100
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.876204081632653e-05,
+      "loss": 1.2836,
+      "step": 359200
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.8741632653061224e-05,
+      "loss": 1.2908,
+      "step": 359300
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.872122448979592e-05,
+      "loss": 1.2887,
+      "step": 359400
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.8700816326530617e-05,
+      "loss": 1.2855,
+      "step": 359500
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.868040816326531e-05,
+      "loss": 1.2855,
+      "step": 359600
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.8660000000000003e-05,
+      "loss": 1.2858,
+      "step": 359700
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.86395918367347e-05,
+      "loss": 1.284,
+      "step": 359800
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.861918367346939e-05,
+      "loss": 1.2785,
+      "step": 359900
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.8598775510204085e-05,
+      "loss": 1.2803,
+      "step": 360000
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.8578367346938777e-05,
+      "loss": 1.2821,
+      "step": 360100
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.855795918367347e-05,
+      "loss": 1.2841,
+      "step": 360200
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.8537551020408166e-05,
+      "loss": 1.28,
+      "step": 360300
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.8517142857142858e-05,
+      "loss": 1.2831,
+      "step": 360400
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.8496734693877553e-05,
+      "loss": 1.287,
+      "step": 360500
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.8476326530612245e-05,
+      "loss": 1.2871,
+      "step": 360600
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.845591836734694e-05,
+      "loss": 1.2758,
+      "step": 360700
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.8435510204081635e-05,
+      "loss": 1.2892,
+      "step": 360800
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.8415102040816326e-05,
+      "loss": 1.2854,
+      "step": 360900
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.839469387755102e-05,
+      "loss": 1.2857,
+      "step": 361000
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.8374285714285713e-05,
+      "loss": 1.2808,
+      "step": 361100
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.8353877551020408e-05,
+      "loss": 1.2867,
+      "step": 361200
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.8333469387755103e-05,
+      "loss": 1.2836,
+      "step": 361300
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.8313061224489794e-05,
+      "loss": 1.29,
+      "step": 361400
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.829265306122449e-05,
+      "loss": 1.28,
+      "step": 361500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.8272244897959184e-05,
+      "loss": 1.2846,
+      "step": 361600
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.8251836734693876e-05,
+      "loss": 1.283,
+      "step": 361700
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.8231428571428574e-05,
+      "loss": 1.2807,
+      "step": 361800
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.821102040816327e-05,
+      "loss": 1.2994,
+      "step": 361900
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.8190612244897964e-05,
+      "loss": 1.2895,
+      "step": 362000
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.8170204081632656e-05,
+      "loss": 1.2841,
+      "step": 362100
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.814979591836735e-05,
+      "loss": 1.2795,
+      "step": 362200
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.8129387755102042e-05,
+      "loss": 1.2858,
+      "step": 362300
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.8108979591836737e-05,
+      "loss": 1.2847,
+      "step": 362400
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.8088571428571432e-05,
+      "loss": 1.2844,
+      "step": 362500
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.8068163265306124e-05,
+      "loss": 1.2881,
+      "step": 362600
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.804775510204082e-05,
+      "loss": 1.2865,
+      "step": 362700
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.802734693877551e-05,
+      "loss": 1.2924,
+      "step": 362800
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.8006938775510205e-05,
+      "loss": 1.2798,
+      "step": 362900
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.79865306122449e-05,
+      "loss": 1.2857,
+      "step": 363000
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.7966122448979592e-05,
+      "loss": 1.2842,
+      "step": 363100
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.7945714285714287e-05,
+      "loss": 1.2828,
+      "step": 363200
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.792530612244898e-05,
+      "loss": 1.2823,
+      "step": 363300
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.7904897959183673e-05,
+      "loss": 1.2853,
+      "step": 363400
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.788448979591837e-05,
+      "loss": 1.2882,
+      "step": 363500
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.786408163265306e-05,
+      "loss": 1.2871,
+      "step": 363600
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.7843673469387755e-05,
+      "loss": 1.2749,
+      "step": 363700
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.7823265306122446e-05,
+      "loss": 1.2753,
+      "step": 363800
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.780285714285714e-05,
+      "loss": 1.277,
+      "step": 363900
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.778244897959184e-05,
+      "loss": 1.2835,
+      "step": 364000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.7762040816326535e-05,
+      "loss": 1.2791,
+      "step": 364100
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.774163265306123e-05,
+      "loss": 1.287,
+      "step": 364200
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.772122448979592e-05,
+      "loss": 1.2695,
+      "step": 364300
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.7700816326530616e-05,
+      "loss": 1.2803,
+      "step": 364400
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.7680408163265308e-05,
+      "loss": 1.2806,
+      "step": 364500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 1.2846,
+      "step": 364600
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7639591836734698e-05,
+      "loss": 1.2824,
+      "step": 364700
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.761918367346939e-05,
+      "loss": 1.2764,
+      "step": 364800
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7598775510204084e-05,
+      "loss": 1.2811,
+      "step": 364900
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7578367346938776e-05,
+      "loss": 1.2876,
+      "step": 365000
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.755795918367347e-05,
+      "loss": 1.2791,
+      "step": 365100
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7537551020408166e-05,
+      "loss": 1.2867,
+      "step": 365200
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7517142857142857e-05,
+      "loss": 1.2803,
+      "step": 365300
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7496734693877552e-05,
+      "loss": 1.2844,
+      "step": 365400
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.74765306122449e-05,
+      "loss": 1.287,
+      "step": 365500
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.745612244897959e-05,
+      "loss": 1.2741,
+      "step": 365600
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.7435714285714287e-05,
+      "loss": 1.2776,
+      "step": 365700
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.7415510204081634e-05,
+      "loss": 1.2867,
+      "step": 365800
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.739510204081633e-05,
+      "loss": 1.2806,
+      "step": 365900
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.737469387755102e-05,
+      "loss": 1.2823,
+      "step": 366000
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.7354285714285716e-05,
+      "loss": 1.2802,
+      "step": 366100
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.733387755102041e-05,
+      "loss": 1.2765,
+      "step": 366200
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.7313469387755102e-05,
+      "loss": 1.2814,
+      "step": 366300
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.7293061224489797e-05,
+      "loss": 1.2769,
+      "step": 366400
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.7272857142857145e-05,
+      "loss": 1.2855,
+      "step": 366500
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.7252448979591836e-05,
+      "loss": 1.2811,
+      "step": 366600
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.723204081632653e-05,
+      "loss": 1.2833,
+      "step": 366700
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.7211632653061226e-05,
+      "loss": 1.2842,
+      "step": 366800
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.7191224489795918e-05,
+      "loss": 1.2856,
+      "step": 366900
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.7170816326530613e-05,
+      "loss": 1.2788,
+      "step": 367000
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.7150408163265308e-05,
+      "loss": 1.276,
+      "step": 367100
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.713e-05,
+      "loss": 1.2867,
+      "step": 367200
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.7109591836734694e-05,
+      "loss": 1.2834,
+      "step": 367300
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.7089183673469386e-05,
+      "loss": 1.2718,
+      "step": 367400
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.706877551020408e-05,
+      "loss": 1.2791,
+      "step": 367500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.7048367346938776e-05,
+      "loss": 1.2779,
+      "step": 367600
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.7027959183673468e-05,
+      "loss": 1.2858,
+      "step": 367700
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.7007551020408166e-05,
+      "loss": 1.286,
+      "step": 367800
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.698714285714286e-05,
+      "loss": 1.2821,
+      "step": 367900
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.6966734693877556e-05,
+      "loss": 1.2819,
+      "step": 368000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.6946326530612247e-05,
+      "loss": 1.2838,
+      "step": 368100
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.6925918367346942e-05,
+      "loss": 1.2796,
+      "step": 368200
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.6905510204081634e-05,
+      "loss": 1.2813,
+      "step": 368300
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.688510204081633e-05,
+      "loss": 1.2715,
+      "step": 368400
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.6864693877551024e-05,
+      "loss": 1.2788,
+      "step": 368500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.6844285714285715e-05,
+      "loss": 1.2862,
+      "step": 368600
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.682387755102041e-05,
+      "loss": 1.2743,
+      "step": 368700
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.6803469387755102e-05,
+      "loss": 1.28,
+      "step": 368800
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.6783061224489797e-05,
+      "loss": 1.2773,
+      "step": 368900
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.6762653061224492e-05,
+      "loss": 1.2807,
+      "step": 369000
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.6742244897959184e-05,
+      "loss": 1.2764,
+      "step": 369100
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.672183673469388e-05,
+      "loss": 1.2843,
+      "step": 369200
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.670142857142857e-05,
+      "loss": 1.2796,
+      "step": 369300
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.6681020408163265e-05,
+      "loss": 1.2856,
+      "step": 369400
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.666061224489796e-05,
+      "loss": 1.2846,
+      "step": 369500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.664020408163265e-05,
+      "loss": 1.282,
+      "step": 369600
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.6619795918367347e-05,
+      "loss": 1.2746,
+      "step": 369700
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.6599387755102038e-05,
+      "loss": 1.2804,
+      "step": 369800
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.6578979591836733e-05,
+      "loss": 1.2804,
+      "step": 369900
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.6558571428571428e-05,
+      "loss": 1.2805,
+      "step": 370000
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.6538163265306127e-05,
+      "loss": 1.2806,
+      "step": 370100
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.651775510204082e-05,
+      "loss": 1.2758,
+      "step": 370200
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.6497346938775513e-05,
+      "loss": 1.2783,
+      "step": 370300
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.6476938775510208e-05,
+      "loss": 1.2883,
+      "step": 370400
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.64565306122449e-05,
+      "loss": 1.2812,
+      "step": 370500
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.6436122448979595e-05,
+      "loss": 1.2784,
+      "step": 370600
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.641571428571429e-05,
+      "loss": 1.278,
+      "step": 370700
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.639530612244898e-05,
+      "loss": 1.2798,
+      "step": 370800
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.6374897959183676e-05,
+      "loss": 1.2777,
+      "step": 370900
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.6354489795918368e-05,
+      "loss": 1.2721,
+      "step": 371000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.6334081632653063e-05,
+      "loss": 1.2754,
+      "step": 371100
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.6313673469387758e-05,
+      "loss": 1.2762,
+      "step": 371200
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.629326530612245e-05,
+      "loss": 1.2827,
+      "step": 371300
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.6273061224489797e-05,
+      "loss": 1.2791,
+      "step": 371400
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.6252653061224492e-05,
+      "loss": 1.277,
+      "step": 371500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.6232244897959183e-05,
+      "loss": 1.28,
+      "step": 371600
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.621183673469388e-05,
+      "loss": 1.2834,
+      "step": 371700
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.6191428571428573e-05,
+      "loss": 1.2721,
+      "step": 371800
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.6171020408163265e-05,
+      "loss": 1.2729,
+      "step": 371900
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.615061224489796e-05,
+      "loss": 1.2694,
+      "step": 372000
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.6130204081632655e-05,
+      "loss": 1.2768,
+      "step": 372100
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.6109795918367346e-05,
+      "loss": 1.277,
+      "step": 372200
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.608938775510204e-05,
+      "loss": 1.2818,
+      "step": 372300
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.6068979591836733e-05,
+      "loss": 1.2887,
+      "step": 372400
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.6048571428571428e-05,
+      "loss": 1.2802,
+      "step": 372500
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.6028163265306123e-05,
+      "loss": 1.2816,
+      "step": 372600
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.6007755102040815e-05,
+      "loss": 1.2844,
+      "step": 372700
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.5987346938775513e-05,
+      "loss": 1.2803,
+      "step": 372800
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.5966938775510208e-05,
+      "loss": 1.2753,
+      "step": 372900
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.5946530612244903e-05,
+      "loss": 1.2785,
+      "step": 373000
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.5926122448979594e-05,
+      "loss": 1.2811,
+      "step": 373100
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.590571428571429e-05,
+      "loss": 1.2787,
+      "step": 373200
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.588530612244898e-05,
+      "loss": 1.2811,
+      "step": 373300
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.5864897959183676e-05,
+      "loss": 1.2761,
+      "step": 373400
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.584448979591837e-05,
+      "loss": 1.2775,
+      "step": 373500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.5824081632653062e-05,
+      "loss": 1.2855,
+      "step": 373600
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.5803673469387757e-05,
+      "loss": 1.2823,
+      "step": 373700
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.578326530612245e-05,
+      "loss": 1.2706,
+      "step": 373800
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.5762857142857144e-05,
+      "loss": 1.2717,
+      "step": 373900
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.574244897959184e-05,
+      "loss": 1.2783,
+      "step": 374000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.572204081632653e-05,
+      "loss": 1.2737,
+      "step": 374100
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.5701632653061226e-05,
+      "loss": 1.2832,
+      "step": 374200
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.5681224489795917e-05,
+      "loss": 1.2828,
+      "step": 374300
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.5660816326530612e-05,
+      "loss": 1.2812,
+      "step": 374400
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.5640408163265307e-05,
+      "loss": 1.2722,
+      "step": 374500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.562e-05,
+      "loss": 1.2758,
+      "step": 374600
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.5599591836734694e-05,
+      "loss": 1.2754,
+      "step": 374700
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.5579183673469385e-05,
+      "loss": 1.2757,
+      "step": 374800
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.555877551020408e-05,
+      "loss": 1.284,
+      "step": 374900
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.5538367346938775e-05,
+      "loss": 1.2697,
+      "step": 375000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.5517959183673474e-05,
+      "loss": 1.2736,
+      "step": 375100
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.5497755102040814e-05,
+      "loss": 1.2727,
+      "step": 375200
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.547734693877551e-05,
+      "loss": 1.2769,
+      "step": 375300
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.5456938775510204e-05,
+      "loss": 1.2802,
+      "step": 375400
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.5436530612244896e-05,
+      "loss": 1.2828,
+      "step": 375500
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.5416122448979594e-05,
+      "loss": 1.2714,
+      "step": 375600
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.539571428571429e-05,
+      "loss": 1.2751,
+      "step": 375700
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.5375306122448984e-05,
+      "loss": 1.2703,
+      "step": 375800
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.5354897959183676e-05,
+      "loss": 1.2795,
+      "step": 375900
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.533448979591837e-05,
+      "loss": 1.2723,
+      "step": 376000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.5314285714285718e-05,
+      "loss": 1.2766,
+      "step": 376100
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.5293877551020413e-05,
+      "loss": 1.273,
+      "step": 376200
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.5273469387755105e-05,
+      "loss": 1.2848,
+      "step": 376300
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.52530612244898e-05,
+      "loss": 1.2706,
+      "step": 376400
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.523265306122449e-05,
+      "loss": 1.2772,
+      "step": 376500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.5212244897959186e-05,
+      "loss": 1.2788,
+      "step": 376600
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.519183673469388e-05,
+      "loss": 1.2772,
+      "step": 376700
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.5171428571428573e-05,
+      "loss": 1.2746,
+      "step": 376800
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.5151020408163268e-05,
+      "loss": 1.2796,
+      "step": 376900
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.513061224489796e-05,
+      "loss": 1.2695,
+      "step": 377000
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.5110204081632654e-05,
+      "loss": 1.2724,
+      "step": 377100
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.508979591836735e-05,
+      "loss": 1.2707,
+      "step": 377200
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.506938775510204e-05,
+      "loss": 1.2725,
+      "step": 377300
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.5048979591836736e-05,
+      "loss": 1.2809,
+      "step": 377400
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.5028571428571428e-05,
+      "loss": 1.2766,
+      "step": 377500
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.5008163265306123e-05,
+      "loss": 1.2769,
+      "step": 377600
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.4987755102040818e-05,
+      "loss": 1.2773,
+      "step": 377700
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.4967346938775512e-05,
+      "loss": 1.2796,
+      "step": 377800
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.4946938775510207e-05,
+      "loss": 1.2747,
+      "step": 377900
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.49265306122449e-05,
+      "loss": 1.2734,
+      "step": 378000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.4906122448979594e-05,
+      "loss": 1.2695,
+      "step": 378100
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.4885714285714286e-05,
+      "loss": 1.273,
+      "step": 378200
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.486530612244898e-05,
+      "loss": 1.2782,
+      "step": 378300
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.4844897959183676e-05,
+      "loss": 1.2723,
+      "step": 378400
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.4824489795918367e-05,
+      "loss": 1.2732,
+      "step": 378500
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.4804081632653062e-05,
+      "loss": 1.2817,
+      "step": 378600
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.4783673469387754e-05,
+      "loss": 1.274,
+      "step": 378700
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.476326530612245e-05,
+      "loss": 1.2809,
+      "step": 378800
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4742857142857147e-05,
+      "loss": 1.2733,
+      "step": 378900
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.472244897959184e-05,
+      "loss": 1.2722,
+      "step": 379000
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4702040816326534e-05,
+      "loss": 1.2782,
+      "step": 379100
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4681632653061225e-05,
+      "loss": 1.2677,
+      "step": 379200
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.466122448979592e-05,
+      "loss": 1.2856,
+      "step": 379300
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4640816326530615e-05,
+      "loss": 1.2705,
+      "step": 379400
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4620408163265307e-05,
+      "loss": 1.2663,
+      "step": 379500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.46e-05,
+      "loss": 1.2757,
+      "step": 379600
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.4579591836734693e-05,
+      "loss": 1.2639,
+      "step": 379700
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.4559183673469388e-05,
+      "loss": 1.2714,
+      "step": 379800
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.4538775510204083e-05,
+      "loss": 1.2703,
+      "step": 379900
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.4518367346938775e-05,
+      "loss": 1.2746,
+      "step": 380000
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.4497959183673473e-05,
+      "loss": 1.2744,
+      "step": 380100
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.4477551020408165e-05,
+      "loss": 1.2748,
+      "step": 380200
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.445714285714286e-05,
+      "loss": 1.2734,
+      "step": 380300
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.443673469387755e-05,
+      "loss": 1.2663,
+      "step": 380400
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.4416326530612246e-05,
+      "loss": 1.2649,
+      "step": 380500
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.439591836734694e-05,
+      "loss": 1.2684,
+      "step": 380600
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.4375510204081633e-05,
+      "loss": 1.2642,
+      "step": 380700
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.4355102040816328e-05,
+      "loss": 1.2774,
+      "step": 380800
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.433469387755102e-05,
+      "loss": 1.2657,
+      "step": 380900
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.4314285714285714e-05,
+      "loss": 1.2709,
+      "step": 381000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.429387755102041e-05,
+      "loss": 1.2718,
+      "step": 381100
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.4273469387755104e-05,
+      "loss": 1.2666,
+      "step": 381200
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.42530612244898e-05,
+      "loss": 1.2714,
+      "step": 381300
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.4232857142857143e-05,
+      "loss": 1.2711,
+      "step": 381400
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.421244897959184e-05,
+      "loss": 1.2745,
+      "step": 381500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.419204081632653e-05,
+      "loss": 1.2711,
+      "step": 381600
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.417163265306123e-05,
+      "loss": 1.2681,
+      "step": 381700
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.415122448979592e-05,
+      "loss": 1.2698,
+      "step": 381800
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.4130816326530615e-05,
+      "loss": 1.2719,
+      "step": 381900
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.4110408163265306e-05,
+      "loss": 1.2721,
+      "step": 382000
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.409e-05,
+      "loss": 1.2688,
+      "step": 382100
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.4069591836734696e-05,
+      "loss": 1.2674,
+      "step": 382200
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.4049183673469388e-05,
+      "loss": 1.267,
+      "step": 382300
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.4028775510204083e-05,
+      "loss": 1.2732,
+      "step": 382400
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.4008367346938775e-05,
+      "loss": 1.2744,
+      "step": 382500
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.398795918367347e-05,
+      "loss": 1.271,
+      "step": 382600
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3967551020408164e-05,
+      "loss": 1.2749,
+      "step": 382700
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.394714285714286e-05,
+      "loss": 1.2704,
+      "step": 382800
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3926734693877554e-05,
+      "loss": 1.2655,
+      "step": 382900
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.3906326530612246e-05,
+      "loss": 1.2684,
+      "step": 383000
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.388591836734694e-05,
+      "loss": 1.2765,
+      "step": 383100
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.3865510204081633e-05,
+      "loss": 1.2726,
+      "step": 383200
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.3845102040816328e-05,
+      "loss": 1.2724,
+      "step": 383300
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.3824693877551023e-05,
+      "loss": 1.2608,
+      "step": 383400
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.3804285714285714e-05,
+      "loss": 1.2764,
+      "step": 383500
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.378387755102041e-05,
+      "loss": 1.2758,
+      "step": 383600
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.37634693877551e-05,
+      "loss": 1.2716,
+      "step": 383700
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.3743061224489796e-05,
+      "loss": 1.2664,
+      "step": 383800
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.372265306122449e-05,
+      "loss": 1.2686,
+      "step": 383900
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.3702244897959186e-05,
+      "loss": 1.2711,
+      "step": 384000
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.368183673469388e-05,
+      "loss": 1.2676,
+      "step": 384100
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.3661428571428572e-05,
+      "loss": 1.2675,
+      "step": 384200
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.3641020408163267e-05,
+      "loss": 1.2674,
+      "step": 384300
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.3620612244897962e-05,
+      "loss": 1.2728,
+      "step": 384400
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.3600204081632654e-05,
+      "loss": 1.2676,
+      "step": 384500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.357979591836735e-05,
+      "loss": 1.2728,
+      "step": 384600
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.355938775510204e-05,
+      "loss": 1.2663,
+      "step": 384700
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.3538979591836735e-05,
+      "loss": 1.2724,
+      "step": 384800
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.351857142857143e-05,
+      "loss": 1.2703,
+      "step": 384900
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.3498163265306122e-05,
+      "loss": 1.2651,
+      "step": 385000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.347775510204082e-05,
+      "loss": 1.2631,
+      "step": 385100
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.345734693877551e-05,
+      "loss": 1.2661,
+      "step": 385200
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.3436938775510207e-05,
+      "loss": 1.27,
+      "step": 385300
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.3416530612244898e-05,
+      "loss": 1.2687,
+      "step": 385400
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.3396122448979593e-05,
+      "loss": 1.2703,
+      "step": 385500
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.337591836734694e-05,
+      "loss": 1.27,
+      "step": 385600
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.3355510204081636e-05,
+      "loss": 1.2641,
+      "step": 385700
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.3335102040816327e-05,
+      "loss": 1.2655,
+      "step": 385800
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.3314693877551022e-05,
+      "loss": 1.2691,
+      "step": 385900
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.3294285714285717e-05,
+      "loss": 1.2696,
+      "step": 386000
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.327387755102041e-05,
+      "loss": 1.2634,
+      "step": 386100
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.3253469387755104e-05,
+      "loss": 1.2582,
+      "step": 386200
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.3233061224489795e-05,
+      "loss": 1.2701,
+      "step": 386300
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.3212857142857143e-05,
+      "loss": 1.2726,
+      "step": 386400
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.3192448979591838e-05,
+      "loss": 1.2661,
+      "step": 386500
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.3172244897959186e-05,
+      "loss": 1.2686,
+      "step": 386600
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.315183673469388e-05,
+      "loss": 1.2653,
+      "step": 386700
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.3131428571428572e-05,
+      "loss": 1.2711,
+      "step": 386800
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.3111020408163267e-05,
+      "loss": 1.2652,
+      "step": 386900
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.3090612244897962e-05,
+      "loss": 1.2709,
+      "step": 387000
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.3070204081632654e-05,
+      "loss": 1.2628,
+      "step": 387100
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.304979591836735e-05,
+      "loss": 1.2654,
+      "step": 387200
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.302938775510204e-05,
+      "loss": 1.2686,
+      "step": 387300
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.3008979591836735e-05,
+      "loss": 1.2716,
+      "step": 387400
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.298857142857143e-05,
+      "loss": 1.2732,
+      "step": 387500
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2968163265306122e-05,
+      "loss": 1.2631,
+      "step": 387600
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.294775510204082e-05,
+      "loss": 1.2714,
+      "step": 387700
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.292734693877551e-05,
+      "loss": 1.2724,
+      "step": 387800
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2906938775510207e-05,
+      "loss": 1.2706,
+      "step": 387900
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2886530612244898e-05,
+      "loss": 1.2693,
+      "step": 388000
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2866122448979593e-05,
+      "loss": 1.2707,
+      "step": 388100
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2845714285714288e-05,
+      "loss": 1.2726,
+      "step": 388200
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.282530612244898e-05,
+      "loss": 1.2703,
+      "step": 388300
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2804897959183675e-05,
+      "loss": 1.2675,
+      "step": 388400
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2784489795918366e-05,
+      "loss": 1.2677,
+      "step": 388500
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.276408163265306e-05,
+      "loss": 1.273,
+      "step": 388600
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2743673469387756e-05,
+      "loss": 1.2677,
+      "step": 388700
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2723265306122448e-05,
+      "loss": 1.2606,
+      "step": 388800
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2702857142857146e-05,
+      "loss": 1.2746,
+      "step": 388900
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2682448979591838e-05,
+      "loss": 1.2675,
+      "step": 389000
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2662040816326533e-05,
+      "loss": 1.2648,
+      "step": 389100
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2641632653061228e-05,
+      "loss": 1.2704,
+      "step": 389200
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.262122448979592e-05,
+      "loss": 1.2692,
+      "step": 389300
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2600816326530614e-05,
+      "loss": 1.2691,
+      "step": 389400
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2580408163265306e-05,
+      "loss": 1.2689,
+      "step": 389500
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.256e-05,
+      "loss": 1.2681,
+      "step": 389600
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2539591836734696e-05,
+      "loss": 1.261,
+      "step": 389700
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2519183673469387e-05,
+      "loss": 1.2676,
+      "step": 389800
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2498775510204082e-05,
+      "loss": 1.2589,
+      "step": 389900
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2478367346938777e-05,
+      "loss": 1.2675,
+      "step": 390000
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2457959183673472e-05,
+      "loss": 1.2669,
+      "step": 390100
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2437551020408164e-05,
+      "loss": 1.265,
+      "step": 390200
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.241714285714286e-05,
+      "loss": 1.2673,
+      "step": 390300
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2396734693877554e-05,
+      "loss": 1.2676,
+      "step": 390400
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2376326530612245e-05,
+      "loss": 1.2659,
+      "step": 390500
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2356122448979593e-05,
+      "loss": 1.2664,
+      "step": 390600
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2335714285714288e-05,
+      "loss": 1.2708,
+      "step": 390700
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.231530612244898e-05,
+      "loss": 1.2633,
+      "step": 390800
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2294897959183675e-05,
+      "loss": 1.2667,
+      "step": 390900
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.227448979591837e-05,
+      "loss": 1.2632,
+      "step": 391000
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.225408163265306e-05,
+      "loss": 1.2655,
+      "step": 391100
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2233673469387756e-05,
+      "loss": 1.2609,
+      "step": 391200
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.221326530612245e-05,
+      "loss": 1.2611,
+      "step": 391300
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2192857142857143e-05,
+      "loss": 1.2707,
+      "step": 391400
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2172448979591838e-05,
+      "loss": 1.2668,
+      "step": 391500
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2152040816326533e-05,
+      "loss": 1.2655,
+      "step": 391600
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2131632653061228e-05,
+      "loss": 1.264,
+      "step": 391700
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.211122448979592e-05,
+      "loss": 1.2739,
+      "step": 391800
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2090816326530614e-05,
+      "loss": 1.2578,
+      "step": 391900
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.207040816326531e-05,
+      "loss": 1.2735,
+      "step": 392000
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.205e-05,
+      "loss": 1.263,
+      "step": 392100
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2029591836734696e-05,
+      "loss": 1.2637,
+      "step": 392200
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2009183673469387e-05,
+      "loss": 1.2771,
+      "step": 392300
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.1988775510204082e-05,
+      "loss": 1.2645,
+      "step": 392400
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.1968367346938777e-05,
+      "loss": 1.2677,
+      "step": 392500
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.194795918367347e-05,
+      "loss": 1.2612,
+      "step": 392600
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.1927551020408164e-05,
+      "loss": 1.2616,
+      "step": 392700
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.190714285714286e-05,
+      "loss": 1.2549,
+      "step": 392800
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.1886734693877554e-05,
+      "loss": 1.2601,
+      "step": 392900
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.1866326530612245e-05,
+      "loss": 1.2708,
+      "step": 393000
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.184591836734694e-05,
+      "loss": 1.2736,
+      "step": 393100
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.1825510204081635e-05,
+      "loss": 1.2712,
+      "step": 393200
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.1805102040816327e-05,
+      "loss": 1.2668,
+      "step": 393300
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.1784693877551022e-05,
+      "loss": 1.2512,
+      "step": 393400
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.1764285714285713e-05,
+      "loss": 1.2725,
+      "step": 393500
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.1743877551020408e-05,
+      "loss": 1.2645,
+      "step": 393600
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.1723469387755103e-05,
+      "loss": 1.2671,
+      "step": 393700
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.1703061224489795e-05,
+      "loss": 1.2671,
+      "step": 393800
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.1682653061224493e-05,
+      "loss": 1.2667,
+      "step": 393900
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.1662244897959185e-05,
+      "loss": 1.2628,
+      "step": 394000
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.1642040816326532e-05,
+      "loss": 1.2645,
+      "step": 394100
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.1621632653061224e-05,
+      "loss": 1.2584,
+      "step": 394200
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.160122448979592e-05,
+      "loss": 1.2632,
+      "step": 394300
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.1580816326530614e-05,
+      "loss": 1.2646,
+      "step": 394400
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.156040816326531e-05,
+      "loss": 1.2643,
+      "step": 394500
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.154e-05,
+      "loss": 1.2643,
+      "step": 394600
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.1519591836734695e-05,
+      "loss": 1.2687,
+      "step": 394700
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.149918367346939e-05,
+      "loss": 1.2673,
+      "step": 394800
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.1478775510204082e-05,
+      "loss": 1.2607,
+      "step": 394900
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.1458367346938777e-05,
+      "loss": 1.2668,
+      "step": 395000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.143795918367347e-05,
+      "loss": 1.2635,
+      "step": 395100
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.1417551020408163e-05,
+      "loss": 1.2705,
+      "step": 395200
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.139714285714286e-05,
+      "loss": 1.2689,
+      "step": 395300
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.137673469387755e-05,
+      "loss": 1.2594,
+      "step": 395400
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.135632653061225e-05,
+      "loss": 1.2576,
+      "step": 395500
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.133591836734694e-05,
+      "loss": 1.2606,
+      "step": 395600
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.1315510204081635e-05,
+      "loss": 1.2633,
+      "step": 395700
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.1295102040816327e-05,
+      "loss": 1.257,
+      "step": 395800
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.127469387755102e-05,
+      "loss": 1.2553,
+      "step": 395900
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.1254285714285716e-05,
+      "loss": 1.2625,
+      "step": 396000
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.1233877551020408e-05,
+      "loss": 1.2609,
+      "step": 396100
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.1213673469387756e-05,
+      "loss": 1.2607,
+      "step": 396200
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.119326530612245e-05,
+      "loss": 1.2657,
+      "step": 396300
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.1172857142857146e-05,
+      "loss": 1.2626,
+      "step": 396400
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.1152448979591837e-05,
+      "loss": 1.27,
+      "step": 396500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.1132040816326532e-05,
+      "loss": 1.263,
+      "step": 396600
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.1111632653061224e-05,
+      "loss": 1.2595,
+      "step": 396700
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.109122448979592e-05,
+      "loss": 1.2583,
+      "step": 396800
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.1070816326530614e-05,
+      "loss": 1.2562,
+      "step": 396900
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.1050408163265305e-05,
+      "loss": 1.2682,
+      "step": 397000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.103e-05,
+      "loss": 1.2611,
+      "step": 397100
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.1009591836734695e-05,
+      "loss": 1.2685,
+      "step": 397200
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.098918367346939e-05,
+      "loss": 1.2673,
+      "step": 397300
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.0968775510204082e-05,
+      "loss": 1.2659,
+      "step": 397400
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.0948367346938777e-05,
+      "loss": 1.2589,
+      "step": 397500
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.0927959183673472e-05,
+      "loss": 1.2624,
+      "step": 397600
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.0907551020408163e-05,
+      "loss": 1.2723,
+      "step": 397700
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.0887142857142858e-05,
+      "loss": 1.258,
+      "step": 397800
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.086673469387755e-05,
+      "loss": 1.2599,
+      "step": 397900
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.0846326530612245e-05,
+      "loss": 1.2705,
+      "step": 398000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.082591836734694e-05,
+      "loss": 1.2635,
+      "step": 398100
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.080551020408163e-05,
+      "loss": 1.2548,
+      "step": 398200
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.078510204081633e-05,
+      "loss": 1.2636,
+      "step": 398300
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.076469387755102e-05,
+      "loss": 1.2629,
+      "step": 398400
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.0744285714285716e-05,
+      "loss": 1.2608,
+      "step": 398500
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.072387755102041e-05,
+      "loss": 1.2602,
+      "step": 398600
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.0703673469387755e-05,
+      "loss": 1.2627,
+      "step": 398700
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.068326530612245e-05,
+      "loss": 1.2626,
+      "step": 398800
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.0662857142857145e-05,
+      "loss": 1.2648,
+      "step": 398900
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.0642448979591837e-05,
+      "loss": 1.2617,
+      "step": 399000
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.0622040816326532e-05,
+      "loss": 1.2639,
+      "step": 399100
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.0601632653061227e-05,
+      "loss": 1.2627,
+      "step": 399200
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.058122448979592e-05,
+      "loss": 1.26,
+      "step": 399300
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.0560816326530613e-05,
+      "loss": 1.262,
+      "step": 399400
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.0540408163265305e-05,
+      "loss": 1.2644,
+      "step": 399500
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.052e-05,
+      "loss": 1.2592,
+      "step": 399600
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.0499591836734695e-05,
+      "loss": 1.2655,
+      "step": 399700
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.0479183673469387e-05,
+      "loss": 1.2594,
+      "step": 399800
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.045877551020408e-05,
+      "loss": 1.2693,
+      "step": 399900
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.0438367346938777e-05,
+      "loss": 1.258,
+      "step": 400000
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.041795918367347e-05,
+      "loss": 1.2611,
+      "step": 400100
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.0397551020408166e-05,
+      "loss": 1.2568,
+      "step": 400200
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.0377142857142858e-05,
+      "loss": 1.2601,
+      "step": 400300
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.0356734693877553e-05,
+      "loss": 1.2633,
+      "step": 400400
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.0336326530612245e-05,
+      "loss": 1.264,
+      "step": 400500
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.031591836734694e-05,
+      "loss": 1.259,
+      "step": 400600
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.0295510204081635e-05,
+      "loss": 1.2678,
+      "step": 400700
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.0275102040816326e-05,
+      "loss": 1.2579,
+      "step": 400800
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.025469387755102e-05,
+      "loss": 1.2642,
+      "step": 400900
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.0234285714285713e-05,
+      "loss": 1.2625,
+      "step": 401000
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.021408163265306e-05,
+      "loss": 1.2733,
+      "step": 401100
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.0193673469387755e-05,
+      "loss": 1.2567,
+      "step": 401200
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.017326530612245e-05,
+      "loss": 1.2723,
+      "step": 401300
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.0152857142857142e-05,
+      "loss": 1.2624,
+      "step": 401400
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.0132448979591837e-05,
+      "loss": 1.2674,
+      "step": 401500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.0112040816326532e-05,
+      "loss": 1.259,
+      "step": 401600
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.0091632653061227e-05,
+      "loss": 1.2631,
+      "step": 401700
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.0071224489795922e-05,
+      "loss": 1.2579,
+      "step": 401800
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.0051020408163266e-05,
+      "loss": 1.2579,
+      "step": 401900
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.0030612244897957e-05,
+      "loss": 1.268,
+      "step": 402000
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.0010204081632656e-05,
+      "loss": 1.2664,
+      "step": 402100
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.9989795918367347e-05,
+      "loss": 1.2609,
+      "step": 402200
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.9969387755102042e-05,
+      "loss": 1.2514,
+      "step": 402300
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.9948979591836737e-05,
+      "loss": 1.2603,
+      "step": 402400
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.992857142857143e-05,
+      "loss": 1.2527,
+      "step": 402500
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.9908163265306124e-05,
+      "loss": 1.2599,
+      "step": 402600
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.9887755102040816e-05,
+      "loss": 1.2546,
+      "step": 402700
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.986734693877551e-05,
+      "loss": 1.2524,
+      "step": 402800
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.9846938775510205e-05,
+      "loss": 1.263,
+      "step": 402900
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.9826530612244897e-05,
+      "loss": 1.2614,
+      "step": 403000
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.9806122448979592e-05,
+      "loss": 1.2598,
+      "step": 403100
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.9785714285714287e-05,
+      "loss": 1.2624,
+      "step": 403200
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.9765306122448982e-05,
+      "loss": 1.2539,
+      "step": 403300
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.9744897959183677e-05,
+      "loss": 1.2634,
+      "step": 403400
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.972448979591837e-05,
+      "loss": 1.261,
+      "step": 403500
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.9704081632653063e-05,
+      "loss": 1.2612,
+      "step": 403600
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.9683673469387755e-05,
+      "loss": 1.2622,
+      "step": 403700
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.966326530612245e-05,
+      "loss": 1.2593,
+      "step": 403800
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 1.259,
+      "step": 403900
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.9622448979591837e-05,
+      "loss": 1.2549,
+      "step": 404000
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.960204081632653e-05,
+      "loss": 1.2541,
+      "step": 404100
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.9581632653061223e-05,
+      "loss": 1.2608,
+      "step": 404200
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.9561224489795918e-05,
+      "loss": 1.2674,
+      "step": 404300
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.9540816326530613e-05,
+      "loss": 1.2603,
+      "step": 404400
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.9520408163265308e-05,
+      "loss": 1.2555,
+      "step": 404500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 1.2582,
+      "step": 404600
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.9479591836734695e-05,
+      "loss": 1.2605,
+      "step": 404700
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.945918367346939e-05,
+      "loss": 1.26,
+      "step": 404800
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.943877551020408e-05,
+      "loss": 1.2607,
+      "step": 404900
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.9418367346938776e-05,
+      "loss": 1.256,
+      "step": 405000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.939795918367347e-05,
+      "loss": 1.2628,
+      "step": 405100
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.9377551020408163e-05,
+      "loss": 1.2578,
+      "step": 405200
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.9357142857142858e-05,
+      "loss": 1.2588,
+      "step": 405300
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.933673469387755e-05,
+      "loss": 1.2634,
+      "step": 405400
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.9316326530612248e-05,
+      "loss": 1.2607,
+      "step": 405500
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.929591836734694e-05,
+      "loss": 1.255,
+      "step": 405600
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.9275510204081634e-05,
+      "loss": 1.2534,
+      "step": 405700
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.925510204081633e-05,
+      "loss": 1.2635,
+      "step": 405800
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.923469387755102e-05,
+      "loss": 1.2602,
+      "step": 405900
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.9214285714285716e-05,
+      "loss": 1.2696,
+      "step": 406000
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.9193877551020407e-05,
+      "loss": 1.2571,
+      "step": 406100
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.9173469387755102e-05,
+      "loss": 1.2476,
+      "step": 406200
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.9153061224489797e-05,
+      "loss": 1.2572,
+      "step": 406300
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.913265306122449e-05,
+      "loss": 1.2598,
+      "step": 406400
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.9112244897959184e-05,
+      "loss": 1.2498,
+      "step": 406500
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.909183673469388e-05,
+      "loss": 1.2602,
+      "step": 406600
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.9071428571428574e-05,
+      "loss": 1.2536,
+      "step": 406700
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.905102040816327e-05,
+      "loss": 1.2625,
+      "step": 406800
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.903061224489796e-05,
+      "loss": 1.2586,
+      "step": 406900
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.9010204081632655e-05,
+      "loss": 1.2504,
+      "step": 407000
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.8989795918367347e-05,
+      "loss": 1.2601,
+      "step": 407100
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.8969387755102042e-05,
+      "loss": 1.258,
+      "step": 407200
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.8948979591836737e-05,
+      "loss": 1.259,
+      "step": 407300
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.892857142857143e-05,
+      "loss": 1.2561,
+      "step": 407400
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.8908163265306123e-05,
+      "loss": 1.2664,
+      "step": 407500
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.8887755102040815e-05,
+      "loss": 1.263,
+      "step": 407600
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.886734693877551e-05,
+      "loss": 1.2656,
+      "step": 407700
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.8846938775510205e-05,
+      "loss": 1.2585,
+      "step": 407800
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.8826734693877552e-05,
+      "loss": 1.261,
+      "step": 407900
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.8806326530612244e-05,
+      "loss": 1.2562,
+      "step": 408000
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.878591836734694e-05,
+      "loss": 1.2523,
+      "step": 408100
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.876551020408163e-05,
+      "loss": 1.2603,
+      "step": 408200
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.874510204081633e-05,
+      "loss": 1.261,
+      "step": 408300
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.8724693877551024e-05,
+      "loss": 1.2556,
+      "step": 408400
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.8704285714285716e-05,
+      "loss": 1.2585,
+      "step": 408500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.868387755102041e-05,
+      "loss": 1.2595,
+      "step": 408600
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.8663673469387755e-05,
+      "loss": 1.2546,
+      "step": 408700
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.864326530612245e-05,
+      "loss": 1.2516,
+      "step": 408800
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.8622857142857145e-05,
+      "loss": 1.2588,
+      "step": 408900
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.860244897959184e-05,
+      "loss": 1.2629,
+      "step": 409000
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.858204081632653e-05,
+      "loss": 1.2602,
+      "step": 409100
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.8561632653061226e-05,
+      "loss": 1.2556,
+      "step": 409200
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.8541224489795918e-05,
+      "loss": 1.2625,
+      "step": 409300
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.8520816326530613e-05,
+      "loss": 1.2564,
+      "step": 409400
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.8500408163265308e-05,
+      "loss": 1.2562,
+      "step": 409500
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.848e-05,
+      "loss": 1.2538,
+      "step": 409600
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.8459591836734694e-05,
+      "loss": 1.2593,
+      "step": 409700
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.8439183673469386e-05,
+      "loss": 1.2531,
+      "step": 409800
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.8418775510204084e-05,
+      "loss": 1.2618,
+      "step": 409900
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.8398367346938776e-05,
+      "loss": 1.255,
+      "step": 410000
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.837795918367347e-05,
+      "loss": 1.255,
+      "step": 410100
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.8357551020408166e-05,
+      "loss": 1.2565,
+      "step": 410200
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.8337142857142857e-05,
+      "loss": 1.25,
+      "step": 410300
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.8316734693877552e-05,
+      "loss": 1.2487,
+      "step": 410400
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.8296326530612247e-05,
+      "loss": 1.2644,
+      "step": 410500
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.827591836734694e-05,
+      "loss": 1.2544,
+      "step": 410600
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.8255510204081634e-05,
+      "loss": 1.2551,
+      "step": 410700
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.8235102040816325e-05,
+      "loss": 1.2503,
+      "step": 410800
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.821469387755102e-05,
+      "loss": 1.2652,
+      "step": 410900
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.8194285714285715e-05,
+      "loss": 1.2552,
+      "step": 411000
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.817387755102041e-05,
+      "loss": 1.2582,
+      "step": 411100
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.8153469387755105e-05,
+      "loss": 1.2576,
+      "step": 411200
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.8133061224489797e-05,
+      "loss": 1.2581,
+      "step": 411300
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.8112653061224492e-05,
+      "loss": 1.2441,
+      "step": 411400
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.8092244897959183e-05,
+      "loss": 1.2618,
+      "step": 411500
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.807183673469388e-05,
+      "loss": 1.2513,
+      "step": 411600
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.8051428571428573e-05,
+      "loss": 1.2618,
+      "step": 411700
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.8031020408163265e-05,
+      "loss": 1.2569,
+      "step": 411800
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.801061224489796e-05,
+      "loss": 1.2483,
+      "step": 411900
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.799020408163265e-05,
+      "loss": 1.2642,
+      "step": 412000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.7969795918367346e-05,
+      "loss": 1.2622,
+      "step": 412100
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.794938775510204e-05,
+      "loss": 1.2524,
+      "step": 412200
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.7928979591836736e-05,
+      "loss": 1.2497,
+      "step": 412300
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.790857142857143e-05,
+      "loss": 1.2536,
+      "step": 412400
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.7888163265306123e-05,
+      "loss": 1.2561,
+      "step": 412500
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.7867755102040818e-05,
+      "loss": 1.2552,
+      "step": 412600
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.7847551020408165e-05,
+      "loss": 1.2535,
+      "step": 412700
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.782714285714286e-05,
+      "loss": 1.2486,
+      "step": 412800
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.7806734693877552e-05,
+      "loss": 1.2536,
+      "step": 412900
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.7786326530612247e-05,
+      "loss": 1.2563,
+      "step": 413000
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.776591836734694e-05,
+      "loss": 1.2471,
+      "step": 413100
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.7745510204081634e-05,
+      "loss": 1.2577,
+      "step": 413200
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.772530612244898e-05,
+      "loss": 1.2535,
+      "step": 413300
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.7704897959183676e-05,
+      "loss": 1.247,
+      "step": 413400
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.7684489795918368e-05,
+      "loss": 1.2554,
+      "step": 413500
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.7664081632653063e-05,
+      "loss": 1.2545,
+      "step": 413600
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.7643673469387754e-05,
+      "loss": 1.26,
+      "step": 413700
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.762326530612245e-05,
+      "loss": 1.2521,
+      "step": 413800
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.7602857142857144e-05,
+      "loss": 1.2505,
+      "step": 413900
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.7582448979591836e-05,
+      "loss": 1.2529,
+      "step": 414000
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.756204081632653e-05,
+      "loss": 1.2488,
+      "step": 414100
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.7541632653061226e-05,
+      "loss": 1.2591,
+      "step": 414200
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.752122448979592e-05,
+      "loss": 1.2518,
+      "step": 414300
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.7500816326530616e-05,
+      "loss": 1.2535,
+      "step": 414400
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.7480408163265307e-05,
+      "loss": 1.2596,
+      "step": 414500
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.7460000000000002e-05,
+      "loss": 1.2469,
+      "step": 414600
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.7439591836734694e-05,
+      "loss": 1.257,
+      "step": 414700
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.741918367346939e-05,
+      "loss": 1.2585,
+      "step": 414800
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.7398775510204084e-05,
+      "loss": 1.2557,
+      "step": 414900
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.7378367346938775e-05,
+      "loss": 1.2614,
+      "step": 415000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.735795918367347e-05,
+      "loss": 1.259,
+      "step": 415100
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.7337551020408162e-05,
+      "loss": 1.2583,
+      "step": 415200
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.731734693877551e-05,
+      "loss": 1.25,
+      "step": 415300
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.7296938775510204e-05,
+      "loss": 1.2558,
+      "step": 415400
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.72765306122449e-05,
+      "loss": 1.2593,
+      "step": 415500
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.725612244897959e-05,
+      "loss": 1.2559,
+      "step": 415600
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.7235714285714286e-05,
+      "loss": 1.2592,
+      "step": 415700
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.7215306122448978e-05,
+      "loss": 1.2569,
+      "step": 415800
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.7194897959183676e-05,
+      "loss": 1.2492,
+      "step": 415900
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.717448979591837e-05,
+      "loss": 1.2555,
+      "step": 416000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.7154081632653062e-05,
+      "loss": 1.26,
+      "step": 416100
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.7133673469387757e-05,
+      "loss": 1.2585,
+      "step": 416200
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.711326530612245e-05,
+      "loss": 1.251,
+      "step": 416300
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.7092857142857144e-05,
+      "loss": 1.2587,
+      "step": 416400
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.707244897959184e-05,
+      "loss": 1.2496,
+      "step": 416500
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.705204081632653e-05,
+      "loss": 1.2498,
+      "step": 416600
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.7031632653061226e-05,
+      "loss": 1.2545,
+      "step": 416700
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.7011224489795917e-05,
+      "loss": 1.2533,
+      "step": 416800
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.6990816326530612e-05,
+      "loss": 1.2556,
+      "step": 416900
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.6970408163265307e-05,
+      "loss": 1.2513,
+      "step": 417000
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.6950000000000002e-05,
+      "loss": 1.2494,
+      "step": 417100
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.6929591836734697e-05,
+      "loss": 1.257,
+      "step": 417200
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.690918367346939e-05,
+      "loss": 1.2473,
+      "step": 417300
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.6888775510204084e-05,
+      "loss": 1.2577,
+      "step": 417400
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.6868367346938775e-05,
+      "loss": 1.251,
+      "step": 417500
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.684795918367347e-05,
+      "loss": 1.2515,
+      "step": 417600
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.6827551020408165e-05,
+      "loss": 1.2551,
+      "step": 417700
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.6807142857142857e-05,
+      "loss": 1.2595,
+      "step": 417800
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.678673469387755e-05,
+      "loss": 1.2507,
+      "step": 417900
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.6766326530612243e-05,
+      "loss": 1.2466,
+      "step": 418000
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.6745918367346938e-05,
+      "loss": 1.2532,
+      "step": 418100
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.6725714285714286e-05,
+      "loss": 1.241,
+      "step": 418200
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.670530612244898e-05,
+      "loss": 1.2578,
+      "step": 418300
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.6684897959183672e-05,
+      "loss": 1.2522,
+      "step": 418400
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.6664489795918367e-05,
+      "loss": 1.2526,
+      "step": 418500
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.6644081632653062e-05,
+      "loss": 1.2487,
+      "step": 418600
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.6623673469387757e-05,
+      "loss": 1.2462,
+      "step": 418700
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.6603265306122452e-05,
+      "loss": 1.2541,
+      "step": 418800
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.6582857142857144e-05,
+      "loss": 1.2481,
+      "step": 418900
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.656244897959184e-05,
+      "loss": 1.2549,
+      "step": 419000
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.654204081632653e-05,
+      "loss": 1.252,
+      "step": 419100
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.6521632653061225e-05,
+      "loss": 1.2552,
+      "step": 419200
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.650122448979592e-05,
+      "loss": 1.2527,
+      "step": 419300
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.6480816326530612e-05,
+      "loss": 1.2525,
+      "step": 419400
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.6460408163265307e-05,
+      "loss": 1.2543,
+      "step": 419500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.644e-05,
+      "loss": 1.2489,
+      "step": 419600
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.6419591836734693e-05,
+      "loss": 1.2588,
+      "step": 419700
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.639918367346939e-05,
+      "loss": 1.2491,
+      "step": 419800
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.6378775510204083e-05,
+      "loss": 1.2531,
+      "step": 419900
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.635836734693878e-05,
+      "loss": 1.2512,
+      "step": 420000
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.633795918367347e-05,
+      "loss": 1.2617,
+      "step": 420100
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.6317551020408165e-05,
+      "loss": 1.252,
+      "step": 420200
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.6297142857142856e-05,
+      "loss": 1.2526,
+      "step": 420300
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.627673469387755e-05,
+      "loss": 1.2501,
+      "step": 420400
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.6256326530612246e-05,
+      "loss": 1.25,
+      "step": 420500
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.6235918367346938e-05,
+      "loss": 1.2583,
+      "step": 420600
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.6215510204081633e-05,
+      "loss": 1.2578,
+      "step": 420700
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.6195102040816325e-05,
+      "loss": 1.2493,
+      "step": 420800
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.617469387755102e-05,
+      "loss": 1.2547,
+      "step": 420900
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.6154285714285718e-05,
+      "loss": 1.2509,
+      "step": 421000
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.613387755102041e-05,
+      "loss": 1.2547,
+      "step": 421100
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.6113469387755104e-05,
+      "loss": 1.2496,
+      "step": 421200
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.6093061224489796e-05,
+      "loss": 1.2533,
+      "step": 421300
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.607265306122449e-05,
+      "loss": 1.2485,
+      "step": 421400
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.6052244897959186e-05,
+      "loss": 1.246,
+      "step": 421500
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.6031836734693878e-05,
+      "loss": 1.2499,
+      "step": 421600
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.6011428571428573e-05,
+      "loss": 1.2493,
+      "step": 421700
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.5991020408163264e-05,
+      "loss": 1.2488,
+      "step": 421800
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.597061224489796e-05,
+      "loss": 1.2498,
+      "step": 421900
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.5950204081632654e-05,
+      "loss": 1.2426,
+      "step": 422000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.592979591836735e-05,
+      "loss": 1.2524,
+      "step": 422100
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.5909387755102044e-05,
+      "loss": 1.2535,
+      "step": 422200
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.5889183673469388e-05,
+      "loss": 1.2581,
+      "step": 422300
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.586877551020408e-05,
+      "loss": 1.252,
+      "step": 422400
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.5848367346938775e-05,
+      "loss": 1.2556,
+      "step": 422500
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.582795918367347e-05,
+      "loss": 1.252,
+      "step": 422600
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.5807551020408165e-05,
+      "loss": 1.2431,
+      "step": 422700
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.578714285714286e-05,
+      "loss": 1.2529,
+      "step": 422800
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.576673469387755e-05,
+      "loss": 1.2541,
+      "step": 422900
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.57465306122449e-05,
+      "loss": 1.2466,
+      "step": 423000
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.5726122448979594e-05,
+      "loss": 1.252,
+      "step": 423100
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.570571428571429e-05,
+      "loss": 1.2509,
+      "step": 423200
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.568530612244898e-05,
+      "loss": 1.2527,
+      "step": 423300
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.5664897959183675e-05,
+      "loss": 1.2534,
+      "step": 423400
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.5644489795918367e-05,
+      "loss": 1.2526,
+      "step": 423500
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.5624081632653062e-05,
+      "loss": 1.2479,
+      "step": 423600
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.5603673469387757e-05,
+      "loss": 1.2484,
+      "step": 423700
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.558326530612245e-05,
+      "loss": 1.2543,
+      "step": 423800
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.5562857142857143e-05,
+      "loss": 1.2522,
+      "step": 423900
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.5542448979591835e-05,
+      "loss": 1.2415,
+      "step": 424000
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.552204081632653e-05,
+      "loss": 1.2497,
+      "step": 424100
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.5501632653061225e-05,
+      "loss": 1.2466,
+      "step": 424200
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.548122448979592e-05,
+      "loss": 1.2482,
+      "step": 424300
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.5460816326530615e-05,
+      "loss": 1.2536,
+      "step": 424400
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.5440408163265306e-05,
+      "loss": 1.254,
+      "step": 424500
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.542e-05,
+      "loss": 1.2451,
+      "step": 424600
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.5399591836734696e-05,
+      "loss": 1.2566,
+      "step": 424700
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.5379183673469388e-05,
+      "loss": 1.2447,
+      "step": 424800
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.5358775510204083e-05,
+      "loss": 1.2508,
+      "step": 424900
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.5338367346938775e-05,
+      "loss": 1.2528,
+      "step": 425000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.531795918367347e-05,
+      "loss": 1.2482,
+      "step": 425100
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.5297551020408164e-05,
+      "loss": 1.2503,
+      "step": 425200
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.5277142857142856e-05,
+      "loss": 1.2447,
+      "step": 425300
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.5256734693877553e-05,
+      "loss": 1.2446,
+      "step": 425400
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.5236326530612246e-05,
+      "loss": 1.2468,
+      "step": 425500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.5215918367346941e-05,
+      "loss": 1.2499,
+      "step": 425600
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.5195510204081634e-05,
+      "loss": 1.24,
+      "step": 425700
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.5175102040816328e-05,
+      "loss": 1.2441,
+      "step": 425800
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.515469387755102e-05,
+      "loss": 1.2437,
+      "step": 425900
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.5134285714285714e-05,
+      "loss": 1.2485,
+      "step": 426000
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.5113877551020409e-05,
+      "loss": 1.2501,
+      "step": 426100
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.5093469387755102e-05,
+      "loss": 1.249,
+      "step": 426200
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.5073061224489796e-05,
+      "loss": 1.2437,
+      "step": 426300
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.5052653061224489e-05,
+      "loss": 1.244,
+      "step": 426400
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.5032244897959186e-05,
+      "loss": 1.2462,
+      "step": 426500
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.5011836734693879e-05,
+      "loss": 1.2543,
+      "step": 426600
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.4991428571428572e-05,
+      "loss": 1.2522,
+      "step": 426700
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.4971020408163267e-05,
+      "loss": 1.2431,
+      "step": 426800
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.495061224489796e-05,
+      "loss": 1.2455,
+      "step": 426900
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.4930408163265308e-05,
+      "loss": 1.2441,
+      "step": 427000
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.4910000000000001e-05,
+      "loss": 1.2413,
+      "step": 427100
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.4889591836734696e-05,
+      "loss": 1.2441,
+      "step": 427200
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.486918367346939e-05,
+      "loss": 1.2464,
+      "step": 427300
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.4848775510204083e-05,
+      "loss": 1.2408,
+      "step": 427400
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.4828367346938776e-05,
+      "loss": 1.2466,
+      "step": 427500
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.480795918367347e-05,
+      "loss": 1.2478,
+      "step": 427600
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.4787551020408164e-05,
+      "loss": 1.2459,
+      "step": 427700
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.4767142857142858e-05,
+      "loss": 1.2411,
+      "step": 427800
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.4746734693877551e-05,
+      "loss": 1.2473,
+      "step": 427900
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.4726326530612244e-05,
+      "loss": 1.2414,
+      "step": 428000
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.470591836734694e-05,
+      "loss": 1.2479,
+      "step": 428100
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.4685510204081634e-05,
+      "loss": 1.2459,
+      "step": 428200
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.4665102040816327e-05,
+      "loss": 1.2477,
+      "step": 428300
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.4644693877551022e-05,
+      "loss": 1.2413,
+      "step": 428400
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.4624285714285716e-05,
+      "loss": 1.2492,
+      "step": 428500
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.4603877551020409e-05,
+      "loss": 1.2411,
+      "step": 428600
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.4583469387755102e-05,
+      "loss": 1.2534,
+      "step": 428700
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.4563061224489795e-05,
+      "loss": 1.2448,
+      "step": 428800
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.454265306122449e-05,
+      "loss": 1.2504,
+      "step": 428900
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.4522244897959184e-05,
+      "loss": 1.2511,
+      "step": 429000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.4501836734693877e-05,
+      "loss": 1.2464,
+      "step": 429100
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.448142857142857e-05,
+      "loss": 1.2422,
+      "step": 429200
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.446122448979592e-05,
+      "loss": 1.2493,
+      "step": 429300
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.4440816326530613e-05,
+      "loss": 1.2444,
+      "step": 429400
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.4420408163265306e-05,
+      "loss": 1.2438,
+      "step": 429500
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.44e-05,
+      "loss": 1.244,
+      "step": 429600
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.4379591836734693e-05,
+      "loss": 1.2438,
+      "step": 429700
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.435918367346939e-05,
+      "loss": 1.2469,
+      "step": 429800
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.4338775510204083e-05,
+      "loss": 1.2465,
+      "step": 429900
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.4318367346938778e-05,
+      "loss": 1.2478,
+      "step": 430000
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.429795918367347e-05,
+      "loss": 1.2451,
+      "step": 430100
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.4277551020408164e-05,
+      "loss": 1.2459,
+      "step": 430200
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.4257142857142857e-05,
+      "loss": 1.2414,
+      "step": 430300
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.423673469387755e-05,
+      "loss": 1.2511,
+      "step": 430400
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.4216326530612246e-05,
+      "loss": 1.2425,
+      "step": 430500
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.4195918367346939e-05,
+      "loss": 1.2427,
+      "step": 430600
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.4175510204081632e-05,
+      "loss": 1.2444,
+      "step": 430700
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.4155102040816325e-05,
+      "loss": 1.2515,
+      "step": 430800
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.4134693877551022e-05,
+      "loss": 1.2433,
+      "step": 430900
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.4114285714285715e-05,
+      "loss": 1.2427,
+      "step": 431000
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.409387755102041e-05,
+      "loss": 1.2457,
+      "step": 431100
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.4073469387755104e-05,
+      "loss": 1.2503,
+      "step": 431200
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.4053265306122448e-05,
+      "loss": 1.2489,
+      "step": 431300
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.4032857142857145e-05,
+      "loss": 1.2492,
+      "step": 431400
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.4012448979591838e-05,
+      "loss": 1.2427,
+      "step": 431500
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.3992040816326533e-05,
+      "loss": 1.2501,
+      "step": 431600
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.3971632653061226e-05,
+      "loss": 1.2509,
+      "step": 431700
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.395122448979592e-05,
+      "loss": 1.2448,
+      "step": 431800
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.3930816326530613e-05,
+      "loss": 1.2509,
+      "step": 431900
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.3910408163265306e-05,
+      "loss": 1.2475,
+      "step": 432000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.389e-05,
+      "loss": 1.2423,
+      "step": 432100
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.3869591836734694e-05,
+      "loss": 1.2425,
+      "step": 432200
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.3849183673469387e-05,
+      "loss": 1.2463,
+      "step": 432300
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.382877551020408e-05,
+      "loss": 1.2475,
+      "step": 432400
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.3808367346938777e-05,
+      "loss": 1.2456,
+      "step": 432500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.378795918367347e-05,
+      "loss": 1.2465,
+      "step": 432600
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3767551020408166e-05,
+      "loss": 1.241,
+      "step": 432700
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3747142857142859e-05,
+      "loss": 1.2522,
+      "step": 432800
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3726734693877552e-05,
+      "loss": 1.2393,
+      "step": 432900
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.3706326530612245e-05,
+      "loss": 1.2474,
+      "step": 433000
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3685918367346939e-05,
+      "loss": 1.2451,
+      "step": 433100
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3665510204081634e-05,
+      "loss": 1.2423,
+      "step": 433200
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3645306122448981e-05,
+      "loss": 1.2449,
+      "step": 433300
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3624897959183675e-05,
+      "loss": 1.2491,
+      "step": 433400
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.3604489795918368e-05,
+      "loss": 1.2475,
+      "step": 433500
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.3584081632653061e-05,
+      "loss": 1.246,
+      "step": 433600
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.3563673469387756e-05,
+      "loss": 1.2427,
+      "step": 433700
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.354326530612245e-05,
+      "loss": 1.2478,
+      "step": 433800
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.3522857142857143e-05,
+      "loss": 1.2471,
+      "step": 433900
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.3502448979591836e-05,
+      "loss": 1.2414,
+      "step": 434000
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.348204081632653e-05,
+      "loss": 1.239,
+      "step": 434100
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.3461632653061226e-05,
+      "loss": 1.2398,
+      "step": 434200
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.344122448979592e-05,
+      "loss": 1.247,
+      "step": 434300
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.3420816326530614e-05,
+      "loss": 1.2393,
+      "step": 434400
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.3400408163265307e-05,
+      "loss": 1.2441,
+      "step": 434500
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.338e-05,
+      "loss": 1.2365,
+      "step": 434600
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.3359591836734694e-05,
+      "loss": 1.2406,
+      "step": 434700
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.3339183673469389e-05,
+      "loss": 1.2336,
+      "step": 434800
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.3318775510204082e-05,
+      "loss": 1.2418,
+      "step": 434900
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.3298367346938775e-05,
+      "loss": 1.2416,
+      "step": 435000
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.3277959183673469e-05,
+      "loss": 1.2439,
+      "step": 435100
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.3257551020408162e-05,
+      "loss": 1.245,
+      "step": 435200
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.3237142857142859e-05,
+      "loss": 1.2494,
+      "step": 435300
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.3216938775510205e-05,
+      "loss": 1.239,
+      "step": 435400
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.3196530612244898e-05,
+      "loss": 1.2371,
+      "step": 435500
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.3176122448979591e-05,
+      "loss": 1.2432,
+      "step": 435600
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.3155714285714284e-05,
+      "loss": 1.2382,
+      "step": 435700
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.3135306122448981e-05,
+      "loss": 1.2415,
+      "step": 435800
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.3114897959183674e-05,
+      "loss": 1.2455,
+      "step": 435900
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.309448979591837e-05,
+      "loss": 1.2492,
+      "step": 436000
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.3074081632653063e-05,
+      "loss": 1.2541,
+      "step": 436100
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.3053673469387756e-05,
+      "loss": 1.2342,
+      "step": 436200
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.3033265306122449e-05,
+      "loss": 1.2483,
+      "step": 436300
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.3012857142857144e-05,
+      "loss": 1.2407,
+      "step": 436400
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.2992448979591837e-05,
+      "loss": 1.2419,
+      "step": 436500
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.297204081632653e-05,
+      "loss": 1.2389,
+      "step": 436600
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.2951632653061224e-05,
+      "loss": 1.2445,
+      "step": 436700
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.2931224489795917e-05,
+      "loss": 1.2438,
+      "step": 436800
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.2910816326530614e-05,
+      "loss": 1.2435,
+      "step": 436900
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.2890408163265307e-05,
+      "loss": 1.2434,
+      "step": 437000
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.2870000000000002e-05,
+      "loss": 1.2368,
+      "step": 437100
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.2849591836734695e-05,
+      "loss": 1.2445,
+      "step": 437200
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.2829183673469389e-05,
+      "loss": 1.2442,
+      "step": 437300
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.2808775510204082e-05,
+      "loss": 1.2458,
+      "step": 437400
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.278857142857143e-05,
+      "loss": 1.247,
+      "step": 437500
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.2768163265306125e-05,
+      "loss": 1.2343,
+      "step": 437600
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.2747755102040818e-05,
+      "loss": 1.2412,
+      "step": 437700
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.2727346938775511e-05,
+      "loss": 1.2437,
+      "step": 437800
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.2706938775510204e-05,
+      "loss": 1.2473,
+      "step": 437900
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.2686530612244898e-05,
+      "loss": 1.2436,
+      "step": 438000
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.2666122448979593e-05,
+      "loss": 1.2375,
+      "step": 438100
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.2645714285714286e-05,
+      "loss": 1.25,
+      "step": 438200
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.262530612244898e-05,
+      "loss": 1.2376,
+      "step": 438300
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.2604897959183672e-05,
+      "loss": 1.2371,
+      "step": 438400
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.2584489795918367e-05,
+      "loss": 1.248,
+      "step": 438500
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.2564081632653062e-05,
+      "loss": 1.242,
+      "step": 438600
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.2543673469387757e-05,
+      "loss": 1.2369,
+      "step": 438700
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.252326530612245e-05,
+      "loss": 1.2444,
+      "step": 438800
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.2502857142857144e-05,
+      "loss": 1.2433,
+      "step": 438900
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.2482448979591837e-05,
+      "loss": 1.2451,
+      "step": 439000
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.246204081632653e-05,
+      "loss": 1.243,
+      "step": 439100
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.2441632653061225e-05,
+      "loss": 1.2411,
+      "step": 439200
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.2421224489795919e-05,
+      "loss": 1.236,
+      "step": 439300
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.2400816326530614e-05,
+      "loss": 1.2418,
+      "step": 439400
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.2380408163265307e-05,
+      "loss": 1.2419,
+      "step": 439500
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.2360204081632653e-05,
+      "loss": 1.2371,
+      "step": 439600
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.2339795918367348e-05,
+      "loss": 1.249,
+      "step": 439700
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.2319591836734694e-05,
+      "loss": 1.25,
+      "step": 439800
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.2299183673469389e-05,
+      "loss": 1.2408,
+      "step": 439900
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.2278775510204082e-05,
+      "loss": 1.2405,
+      "step": 440000
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.2258367346938775e-05,
+      "loss": 1.239,
+      "step": 440100
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.223795918367347e-05,
+      "loss": 1.2454,
+      "step": 440200
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.2217551020408163e-05,
+      "loss": 1.2353,
+      "step": 440300
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.2197142857142858e-05,
+      "loss": 1.2421,
+      "step": 440400
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.2176734693877552e-05,
+      "loss": 1.2458,
+      "step": 440500
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.2156326530612245e-05,
+      "loss": 1.2406,
+      "step": 440600
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.213591836734694e-05,
+      "loss": 1.2481,
+      "step": 440700
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.2115510204081633e-05,
+      "loss": 1.2404,
+      "step": 440800
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.2095102040816327e-05,
+      "loss": 1.2408,
+      "step": 440900
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.2074693877551022e-05,
+      "loss": 1.2416,
+      "step": 441000
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.2054285714285715e-05,
+      "loss": 1.2391,
+      "step": 441100
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.2033877551020408e-05,
+      "loss": 1.2511,
+      "step": 441200
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.2013469387755103e-05,
+      "loss": 1.2394,
+      "step": 441300
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.1993061224489796e-05,
+      "loss": 1.2472,
+      "step": 441400
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.197265306122449e-05,
+      "loss": 1.2344,
+      "step": 441500
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.1952244897959185e-05,
+      "loss": 1.2471,
+      "step": 441600
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.1931836734693878e-05,
+      "loss": 1.2389,
+      "step": 441700
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.1911428571428573e-05,
+      "loss": 1.2424,
+      "step": 441800
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.1891020408163266e-05,
+      "loss": 1.2401,
+      "step": 441900
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.187061224489796e-05,
+      "loss": 1.2402,
+      "step": 442000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.1850204081632654e-05,
+      "loss": 1.239,
+      "step": 442100
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.1829795918367348e-05,
+      "loss": 1.2472,
+      "step": 442200
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.1809387755102041e-05,
+      "loss": 1.2464,
+      "step": 442300
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.1788979591836736e-05,
+      "loss": 1.2418,
+      "step": 442400
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.176857142857143e-05,
+      "loss": 1.2398,
+      "step": 442500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.1748163265306122e-05,
+      "loss": 1.2416,
+      "step": 442600
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.1727755102040817e-05,
+      "loss": 1.2363,
+      "step": 442700
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.170734693877551e-05,
+      "loss": 1.2457,
+      "step": 442800
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.1686938775510204e-05,
+      "loss": 1.2488,
+      "step": 442900
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.1666530612244899e-05,
+      "loss": 1.2418,
+      "step": 443000
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.1646122448979592e-05,
+      "loss": 1.2437,
+      "step": 443100
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.1625714285714285e-05,
+      "loss": 1.2398,
+      "step": 443200
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.160530612244898e-05,
+      "loss": 1.2464,
+      "step": 443300
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.1584897959183674e-05,
+      "loss": 1.2376,
+      "step": 443400
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.1564489795918369e-05,
+      "loss": 1.247,
+      "step": 443500
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.1544081632653062e-05,
+      "loss": 1.2382,
+      "step": 443600
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.1523673469387755e-05,
+      "loss": 1.2397,
+      "step": 443700
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.150326530612245e-05,
+      "loss": 1.2359,
+      "step": 443800
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.1483061224489796e-05,
+      "loss": 1.2326,
+      "step": 443900
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.1462653061224491e-05,
+      "loss": 1.2398,
+      "step": 444000
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.1442244897959184e-05,
+      "loss": 1.243,
+      "step": 444100
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.1421836734693878e-05,
+      "loss": 1.2349,
+      "step": 444200
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.1401428571428573e-05,
+      "loss": 1.2441,
+      "step": 444300
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.1381020408163266e-05,
+      "loss": 1.2426,
+      "step": 444400
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.136061224489796e-05,
+      "loss": 1.2412,
+      "step": 444500
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.1340204081632654e-05,
+      "loss": 1.2404,
+      "step": 444600
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.1319795918367347e-05,
+      "loss": 1.2443,
+      "step": 444700
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.129938775510204e-05,
+      "loss": 1.2444,
+      "step": 444800
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.1278979591836736e-05,
+      "loss": 1.2364,
+      "step": 444900
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.1258571428571429e-05,
+      "loss": 1.2353,
+      "step": 445000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.1238163265306122e-05,
+      "loss": 1.236,
+      "step": 445100
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.1217755102040817e-05,
+      "loss": 1.2363,
+      "step": 445200
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.119734693877551e-05,
+      "loss": 1.2415,
+      "step": 445300
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.1176938775510204e-05,
+      "loss": 1.2381,
+      "step": 445400
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.1156530612244899e-05,
+      "loss": 1.2474,
+      "step": 445500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.1136122448979592e-05,
+      "loss": 1.249,
+      "step": 445600
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.1115714285714287e-05,
+      "loss": 1.2396,
+      "step": 445700
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.109530612244898e-05,
+      "loss": 1.2368,
+      "step": 445800
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.1074897959183674e-05,
+      "loss": 1.2354,
+      "step": 445900
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.1054489795918369e-05,
+      "loss": 1.2342,
+      "step": 446000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.1034081632653062e-05,
+      "loss": 1.2404,
+      "step": 446100
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.1013673469387755e-05,
+      "loss": 1.2393,
+      "step": 446200
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.0993469387755103e-05,
+      "loss": 1.2364,
+      "step": 446300
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.0973265306122449e-05,
+      "loss": 1.234,
+      "step": 446400
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.0952857142857144e-05,
+      "loss": 1.2337,
+      "step": 446500
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.0932448979591837e-05,
+      "loss": 1.2395,
+      "step": 446600
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.0912040816326532e-05,
+      "loss": 1.2408,
+      "step": 446700
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.0891632653061225e-05,
+      "loss": 1.2434,
+      "step": 446800
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.0871224489795918e-05,
+      "loss": 1.2354,
+      "step": 446900
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.0850816326530613e-05,
+      "loss": 1.2333,
+      "step": 447000
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.0830408163265307e-05,
+      "loss": 1.2321,
+      "step": 447100
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.081e-05,
+      "loss": 1.2326,
+      "step": 447200
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.0789591836734695e-05,
+      "loss": 1.2303,
+      "step": 447300
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.0769183673469388e-05,
+      "loss": 1.2372,
+      "step": 447400
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.0748775510204081e-05,
+      "loss": 1.2471,
+      "step": 447500
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.0728367346938776e-05,
+      "loss": 1.237,
+      "step": 447600
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.070795918367347e-05,
+      "loss": 1.2402,
+      "step": 447700
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.0687551020408165e-05,
+      "loss": 1.2388,
+      "step": 447800
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.0667142857142858e-05,
+      "loss": 1.2264,
+      "step": 447900
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.0646734693877551e-05,
+      "loss": 1.239,
+      "step": 448000
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.0626326530612244e-05,
+      "loss": 1.2327,
+      "step": 448100
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.060591836734694e-05,
+      "loss": 1.2407,
+      "step": 448200
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.0585510204081633e-05,
+      "loss": 1.2407,
+      "step": 448300
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.0565102040816328e-05,
+      "loss": 1.2353,
+      "step": 448400
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.0544693877551021e-05,
+      "loss": 1.2456,
+      "step": 448500
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.0524285714285714e-05,
+      "loss": 1.2371,
+      "step": 448600
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.050387755102041e-05,
+      "loss": 1.2392,
+      "step": 448700
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.0483469387755102e-05,
+      "loss": 1.2384,
+      "step": 448800
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.0463061224489797e-05,
+      "loss": 1.2366,
+      "step": 448900
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.044265306122449e-05,
+      "loss": 1.2367,
+      "step": 449000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.0422244897959184e-05,
+      "loss": 1.2388,
+      "step": 449100
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.0401836734693877e-05,
+      "loss": 1.2501,
+      "step": 449200
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.0381428571428572e-05,
+      "loss": 1.2356,
+      "step": 449300
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.0361020408163266e-05,
+      "loss": 1.2407,
+      "step": 449400
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.034061224489796e-05,
+      "loss": 1.2405,
+      "step": 449500
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.0320204081632654e-05,
+      "loss": 1.2381,
+      "step": 449600
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.0299795918367347e-05,
+      "loss": 1.2382,
+      "step": 449700
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.027938775510204e-05,
+      "loss": 1.2388,
+      "step": 449800
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.0258979591836735e-05,
+      "loss": 1.2367,
+      "step": 449900
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.0238571428571429e-05,
+      "loss": 1.2336,
+      "step": 450000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.0218163265306124e-05,
+      "loss": 1.239,
+      "step": 450100
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.0197755102040817e-05,
+      "loss": 1.2423,
+      "step": 450200
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.017734693877551e-05,
+      "loss": 1.2413,
+      "step": 450300
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.0157142857142858e-05,
+      "loss": 1.2273,
+      "step": 450400
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.0136734693877551e-05,
+      "loss": 1.2404,
+      "step": 450500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.0116326530612246e-05,
+      "loss": 1.2397,
+      "step": 450600
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.009591836734694e-05,
+      "loss": 1.2398,
+      "step": 450700
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.0075510204081632e-05,
+      "loss": 1.237,
+      "step": 450800
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.0055102040816327e-05,
+      "loss": 1.238,
+      "step": 450900
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.003469387755102e-05,
+      "loss": 1.2335,
+      "step": 451000
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.0014285714285716e-05,
+      "loss": 1.2384,
+      "step": 451100
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 9.993877551020409e-06,
+      "loss": 1.2306,
+      "step": 451200
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 9.973469387755102e-06,
+      "loss": 1.2378,
+      "step": 451300
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 9.953061224489796e-06,
+      "loss": 1.2386,
+      "step": 451400
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 9.93265306122449e-06,
+      "loss": 1.2336,
+      "step": 451500
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 9.912244897959184e-06,
+      "loss": 1.2373,
+      "step": 451600
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 9.891836734693879e-06,
+      "loss": 1.2421,
+      "step": 451700
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 9.871428571428572e-06,
+      "loss": 1.249,
+      "step": 451800
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 9.851020408163265e-06,
+      "loss": 1.2417,
+      "step": 451900
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 9.830612244897959e-06,
+      "loss": 1.232,
+      "step": 452000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 9.810204081632654e-06,
+      "loss": 1.2359,
+      "step": 452100
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 9.789795918367347e-06,
+      "loss": 1.2358,
+      "step": 452200
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 9.769387755102042e-06,
+      "loss": 1.236,
+      "step": 452300
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 9.748979591836735e-06,
+      "loss": 1.2347,
+      "step": 452400
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 9.728571428571428e-06,
+      "loss": 1.2311,
+      "step": 452500
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 9.708163265306123e-06,
+      "loss": 1.2371,
+      "step": 452600
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 9.687959183673471e-06,
+      "loss": 1.2427,
+      "step": 452700
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 9.667551020408164e-06,
+      "loss": 1.2357,
+      "step": 452800
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 9.647142857142857e-06,
+      "loss": 1.2333,
+      "step": 452900
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 9.62673469387755e-06,
+      "loss": 1.2403,
+      "step": 453000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 9.606326530612246e-06,
+      "loss": 1.2368,
+      "step": 453100
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 9.585918367346939e-06,
+      "loss": 1.2322,
+      "step": 453200
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 9.565510204081634e-06,
+      "loss": 1.2443,
+      "step": 453300
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 9.545102040816327e-06,
+      "loss": 1.2406,
+      "step": 453400
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 9.52469387755102e-06,
+      "loss": 1.2337,
+      "step": 453500
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 9.504285714285714e-06,
+      "loss": 1.2353,
+      "step": 453600
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 9.483877551020409e-06,
+      "loss": 1.2402,
+      "step": 453700
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 9.463469387755102e-06,
+      "loss": 1.2325,
+      "step": 453800
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 9.443061224489797e-06,
+      "loss": 1.2299,
+      "step": 453900
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 9.42265306122449e-06,
+      "loss": 1.2384,
+      "step": 454000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 9.402244897959184e-06,
+      "loss": 1.2388,
+      "step": 454100
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 9.381836734693877e-06,
+      "loss": 1.234,
+      "step": 454200
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 9.361428571428572e-06,
+      "loss": 1.2338,
+      "step": 454300
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 9.341020408163267e-06,
+      "loss": 1.2374,
+      "step": 454400
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 9.32061224489796e-06,
+      "loss": 1.2315,
+      "step": 454500
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 9.300204081632653e-06,
+      "loss": 1.2373,
+      "step": 454600
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 9.28e-06,
+      "loss": 1.2415,
+      "step": 454700
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 9.259591836734694e-06,
+      "loss": 1.2385,
+      "step": 454800
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 9.23918367346939e-06,
+      "loss": 1.2395,
+      "step": 454900
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 9.218775510204082e-06,
+      "loss": 1.2405,
+      "step": 455000
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 9.198367346938776e-06,
+      "loss": 1.232,
+      "step": 455100
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 9.177959183673469e-06,
+      "loss": 1.2298,
+      "step": 455200
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 9.157551020408164e-06,
+      "loss": 1.231,
+      "step": 455300
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 9.137142857142857e-06,
+      "loss": 1.2352,
+      "step": 455400
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 9.116734693877552e-06,
+      "loss": 1.2358,
+      "step": 455500
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 9.096326530612246e-06,
+      "loss": 1.2365,
+      "step": 455600
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 9.075918367346939e-06,
+      "loss": 1.2348,
+      "step": 455700
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 9.055510204081632e-06,
+      "loss": 1.2391,
+      "step": 455800
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 9.035102040816327e-06,
+      "loss": 1.2354,
+      "step": 455900
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 9.014693877551022e-06,
+      "loss": 1.2323,
+      "step": 456000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 8.994285714285715e-06,
+      "loss": 1.231,
+      "step": 456100
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 8.973877551020409e-06,
+      "loss": 1.2351,
+      "step": 456200
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 8.953469387755102e-06,
+      "loss": 1.2354,
+      "step": 456300
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 8.933061224489795e-06,
+      "loss": 1.2342,
+      "step": 456400
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 8.91265306122449e-06,
+      "loss": 1.2367,
+      "step": 456500
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 8.892244897959185e-06,
+      "loss": 1.2297,
+      "step": 456600
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 8.871836734693878e-06,
+      "loss": 1.2352,
+      "step": 456700
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 8.851632653061224e-06,
+      "loss": 1.2217,
+      "step": 456800
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 8.831224489795918e-06,
+      "loss": 1.2358,
+      "step": 456900
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 8.810816326530612e-06,
+      "loss": 1.2382,
+      "step": 457000
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 8.790408163265307e-06,
+      "loss": 1.234,
+      "step": 457100
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 8.77e-06,
+      "loss": 1.2403,
+      "step": 457200
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 8.749591836734694e-06,
+      "loss": 1.2419,
+      "step": 457300
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 8.729183673469387e-06,
+      "loss": 1.2362,
+      "step": 457400
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 8.708775510204082e-06,
+      "loss": 1.2348,
+      "step": 457500
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 8.688367346938776e-06,
+      "loss": 1.2304,
+      "step": 457600
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 8.66795918367347e-06,
+      "loss": 1.2397,
+      "step": 457700
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 8.647551020408164e-06,
+      "loss": 1.2366,
+      "step": 457800
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 8.627142857142857e-06,
+      "loss": 1.2333,
+      "step": 457900
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 8.60673469387755e-06,
+      "loss": 1.2369,
+      "step": 458000
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 8.586326530612245e-06,
+      "loss": 1.2423,
+      "step": 458100
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 8.56591836734694e-06,
+      "loss": 1.2342,
+      "step": 458200
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 8.545510204081634e-06,
+      "loss": 1.2385,
+      "step": 458300
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 8.525102040816327e-06,
+      "loss": 1.2378,
+      "step": 458400
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 8.50469387755102e-06,
+      "loss": 1.2301,
+      "step": 458500
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.484285714285713e-06,
+      "loss": 1.2421,
+      "step": 458600
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.463877551020408e-06,
+      "loss": 1.2405,
+      "step": 458700
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.443673469387756e-06,
+      "loss": 1.2233,
+      "step": 458800
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.42326530612245e-06,
+      "loss": 1.2338,
+      "step": 458900
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.402857142857143e-06,
+      "loss": 1.2351,
+      "step": 459000
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.382448979591836e-06,
+      "loss": 1.2339,
+      "step": 459100
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.36204081632653e-06,
+      "loss": 1.2354,
+      "step": 459200
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.341632653061226e-06,
+      "loss": 1.2351,
+      "step": 459300
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.321224489795919e-06,
+      "loss": 1.2375,
+      "step": 459400
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.300816326530612e-06,
+      "loss": 1.2342,
+      "step": 459500
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.280408163265306e-06,
+      "loss": 1.2346,
+      "step": 459600
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.26e-06,
+      "loss": 1.2345,
+      "step": 459700
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.239591836734696e-06,
+      "loss": 1.235,
+      "step": 459800
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.219183673469389e-06,
+      "loss": 1.2278,
+      "step": 459900
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.198775510204082e-06,
+      "loss": 1.2402,
+      "step": 460000
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.178367346938775e-06,
+      "loss": 1.236,
+      "step": 460100
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.157959183673469e-06,
+      "loss": 1.2323,
+      "step": 460200
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.137551020408164e-06,
+      "loss": 1.2335,
+      "step": 460300
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.117346938775511e-06,
+      "loss": 1.2294,
+      "step": 460400
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.096938775510204e-06,
+      "loss": 1.2386,
+      "step": 460500
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.076530612244898e-06,
+      "loss": 1.2357,
+      "step": 460600
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.056122448979591e-06,
+      "loss": 1.2268,
+      "step": 460700
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.035714285714286e-06,
+      "loss": 1.2276,
+      "step": 460800
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.015306122448981e-06,
+      "loss": 1.2283,
+      "step": 460900
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.994897959183674e-06,
+      "loss": 1.2358,
+      "step": 461000
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.974489795918368e-06,
+      "loss": 1.2315,
+      "step": 461100
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.95408163265306e-06,
+      "loss": 1.2306,
+      "step": 461200
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.933673469387754e-06,
+      "loss": 1.2313,
+      "step": 461300
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.913265306122449e-06,
+      "loss": 1.2294,
+      "step": 461400
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.892857142857144e-06,
+      "loss": 1.2403,
+      "step": 461500
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.872448979591837e-06,
+      "loss": 1.2268,
+      "step": 461600
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.85204081632653e-06,
+      "loss": 1.2377,
+      "step": 461700
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.831632653061224e-06,
+      "loss": 1.2258,
+      "step": 461800
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.811224489795919e-06,
+      "loss": 1.2371,
+      "step": 461900
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.790816326530614e-06,
+      "loss": 1.2308,
+      "step": 462000
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.770408163265307e-06,
+      "loss": 1.2333,
+      "step": 462100
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.75e-06,
+      "loss": 1.2338,
+      "step": 462200
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.729591836734694e-06,
+      "loss": 1.232,
+      "step": 462300
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.709183673469387e-06,
+      "loss": 1.2401,
+      "step": 462400
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.688775510204082e-06,
+      "loss": 1.2345,
+      "step": 462500
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.668367346938777e-06,
+      "loss": 1.2323,
+      "step": 462600
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.64795918367347e-06,
+      "loss": 1.2383,
+      "step": 462700
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.627551020408163e-06,
+      "loss": 1.2403,
+      "step": 462800
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 7.6071428571428575e-06,
+      "loss": 1.2365,
+      "step": 462900
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 7.586734693877551e-06,
+      "loss": 1.2308,
+      "step": 463000
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 7.566326530612246e-06,
+      "loss": 1.2328,
+      "step": 463100
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 7.545918367346939e-06,
+      "loss": 1.2273,
+      "step": 463200
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 7.525510204081633e-06,
+      "loss": 1.2312,
+      "step": 463300
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 7.5051020408163264e-06,
+      "loss": 1.2311,
+      "step": 463400
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 7.4846938775510206e-06,
+      "loss": 1.2328,
+      "step": 463500
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 7.4642857142857155e-06,
+      "loss": 1.2314,
+      "step": 463600
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 7.443877551020409e-06,
+      "loss": 1.2309,
+      "step": 463700
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 7.423469387755103e-06,
+      "loss": 1.2321,
+      "step": 463800
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 7.403061224489796e-06,
+      "loss": 1.2434,
+      "step": 463900
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 7.3826530612244895e-06,
+      "loss": 1.2362,
+      "step": 464000
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 7.362244897959184e-06,
+      "loss": 1.2339,
+      "step": 464100
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 7.341836734693879e-06,
+      "loss": 1.2285,
+      "step": 464200
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 7.321428571428572e-06,
+      "loss": 1.2407,
+      "step": 464300
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 7.301020408163266e-06,
+      "loss": 1.2298,
+      "step": 464400
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 7.280816326530612e-06,
+      "loss": 1.2265,
+      "step": 464500
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 7.260408163265306e-06,
+      "loss": 1.2275,
+      "step": 464600
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 7.240000000000001e-06,
+      "loss": 1.2379,
+      "step": 464700
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 7.219591836734694e-06,
+      "loss": 1.2394,
+      "step": 464800
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 7.199183673469388e-06,
+      "loss": 1.2284,
+      "step": 464900
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 7.178775510204082e-06,
+      "loss": 1.234,
+      "step": 465000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 7.158367346938776e-06,
+      "loss": 1.2377,
+      "step": 465100
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.137959183673469e-06,
+      "loss": 1.223,
+      "step": 465200
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.117551020408164e-06,
+      "loss": 1.2339,
+      "step": 465300
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.097142857142858e-06,
+      "loss": 1.228,
+      "step": 465400
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.076938775510204e-06,
+      "loss": 1.2262,
+      "step": 465500
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.056530612244898e-06,
+      "loss": 1.2399,
+      "step": 465600
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 7.0361224489795915e-06,
+      "loss": 1.2311,
+      "step": 465700
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 7.0157142857142864e-06,
+      "loss": 1.2337,
+      "step": 465800
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 6.9953061224489806e-06,
+      "loss": 1.2323,
+      "step": 465900
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 6.974897959183674e-06,
+      "loss": 1.2335,
+      "step": 466000
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 6.954489795918367e-06,
+      "loss": 1.2329,
+      "step": 466100
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.934081632653061e-06,
+      "loss": 1.2296,
+      "step": 466200
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.913673469387756e-06,
+      "loss": 1.2313,
+      "step": 466300
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.8932653061224495e-06,
+      "loss": 1.2323,
+      "step": 466400
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.872857142857144e-06,
+      "loss": 1.2332,
+      "step": 466500
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.852448979591837e-06,
+      "loss": 1.2308,
+      "step": 466600
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.832040816326531e-06,
+      "loss": 1.2311,
+      "step": 466700
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.811632653061224e-06,
+      "loss": 1.2272,
+      "step": 466800
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.791224489795919e-06,
+      "loss": 1.2278,
+      "step": 466900
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.7708163265306125e-06,
+      "loss": 1.2325,
+      "step": 467000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 6.750408163265307e-06,
+      "loss": 1.2324,
+      "step": 467100
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 6.73e-06,
+      "loss": 1.2384,
+      "step": 467200
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 6.709591836734694e-06,
+      "loss": 1.23,
+      "step": 467300
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 6.689183673469387e-06,
+      "loss": 1.2333,
+      "step": 467400
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 6.668775510204082e-06,
+      "loss": 1.2248,
+      "step": 467500
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 6.6483673469387764e-06,
+      "loss": 1.2297,
+      "step": 467600
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 6.62795918367347e-06,
+      "loss": 1.235,
+      "step": 467700
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 6.607551020408163e-06,
+      "loss": 1.2267,
+      "step": 467800
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 6.587142857142857e-06,
+      "loss": 1.231,
+      "step": 467900
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 6.566734693877552e-06,
+      "loss": 1.2336,
+      "step": 468000
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 6.546326530612245e-06,
+      "loss": 1.2284,
+      "step": 468100
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 6.5259183673469395e-06,
+      "loss": 1.2295,
+      "step": 468200
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 6.505510204081633e-06,
+      "loss": 1.2237,
+      "step": 468300
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 6.485102040816327e-06,
+      "loss": 1.228,
+      "step": 468400
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 6.46469387755102e-06,
+      "loss": 1.231,
+      "step": 468500
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 6.444285714285715e-06,
+      "loss": 1.2329,
+      "step": 468600
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 6.423877551020408e-06,
+      "loss": 1.229,
+      "step": 468700
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 6.4034693877551025e-06,
+      "loss": 1.2292,
+      "step": 468800
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 6.383061224489796e-06,
+      "loss": 1.2316,
+      "step": 468900
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 6.36265306122449e-06,
+      "loss": 1.23,
+      "step": 469000
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 6.342244897959183e-06,
+      "loss": 1.232,
+      "step": 469100
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 6.321836734693878e-06,
+      "loss": 1.2288,
+      "step": 469200
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 6.301428571428572e-06,
+      "loss": 1.234,
+      "step": 469300
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 6.281224489795918e-06,
+      "loss": 1.2364,
+      "step": 469400
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 6.260816326530612e-06,
+      "loss": 1.2225,
+      "step": 469500
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 6.2404081632653065e-06,
+      "loss": 1.2374,
+      "step": 469600
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 6.22e-06,
+      "loss": 1.2283,
+      "step": 469700
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 6.199591836734695e-06,
+      "loss": 1.2271,
+      "step": 469800
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 6.179183673469388e-06,
+      "loss": 1.2299,
+      "step": 469900
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 6.158775510204082e-06,
+      "loss": 1.2249,
+      "step": 470000
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 6.138367346938776e-06,
+      "loss": 1.2313,
+      "step": 470100
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 6.1179591836734695e-06,
+      "loss": 1.2344,
+      "step": 470200
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 6.097551020408164e-06,
+      "loss": 1.2297,
+      "step": 470300
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 6.077142857142858e-06,
+      "loss": 1.2277,
+      "step": 470400
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 6.056734693877551e-06,
+      "loss": 1.2318,
+      "step": 470500
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 6.036326530612245e-06,
+      "loss": 1.2318,
+      "step": 470600
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 6.015918367346939e-06,
+      "loss": 1.2237,
+      "step": 470700
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.9955102040816326e-06,
+      "loss": 1.2307,
+      "step": 470800
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.975102040816327e-06,
+      "loss": 1.2302,
+      "step": 470900
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.954693877551021e-06,
+      "loss": 1.2313,
+      "step": 471000
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.934285714285714e-06,
+      "loss": 1.2205,
+      "step": 471100
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.913877551020409e-06,
+      "loss": 1.2286,
+      "step": 471200
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.893469387755102e-06,
+      "loss": 1.2247,
+      "step": 471300
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.873061224489796e-06,
+      "loss": 1.2393,
+      "step": 471400
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.852653061224491e-06,
+      "loss": 1.2316,
+      "step": 471500
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.832244897959184e-06,
+      "loss": 1.2247,
+      "step": 471600
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.811836734693878e-06,
+      "loss": 1.2227,
+      "step": 471700
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.791428571428572e-06,
+      "loss": 1.2309,
+      "step": 471800
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.771020408163265e-06,
+      "loss": 1.2262,
+      "step": 471900
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.7506122448979595e-06,
+      "loss": 1.2302,
+      "step": 472000
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.730204081632654e-06,
+      "loss": 1.2347,
+      "step": 472100
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.709795918367347e-06,
+      "loss": 1.2315,
+      "step": 472200
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.689387755102041e-06,
+      "loss": 1.2299,
+      "step": 472300
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.668979591836735e-06,
+      "loss": 1.2335,
+      "step": 472400
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.6485714285714285e-06,
+      "loss": 1.2314,
+      "step": 472500
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.628163265306123e-06,
+      "loss": 1.2308,
+      "step": 472600
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.607755102040817e-06,
+      "loss": 1.2312,
+      "step": 472700
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.58734693877551e-06,
+      "loss": 1.2307,
+      "step": 472800
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.566938775510205e-06,
+      "loss": 1.2278,
+      "step": 472900
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.546530612244898e-06,
+      "loss": 1.2265,
+      "step": 473000
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.526122448979592e-06,
+      "loss": 1.2263,
+      "step": 473100
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.5057142857142865e-06,
+      "loss": 1.2235,
+      "step": 473200
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.485510204081633e-06,
+      "loss": 1.2288,
+      "step": 473300
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.465102040816327e-06,
+      "loss": 1.2196,
+      "step": 473400
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.444693877551021e-06,
+      "loss": 1.2312,
+      "step": 473500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.424285714285715e-06,
+      "loss": 1.2283,
+      "step": 473600
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.403877551020409e-06,
+      "loss": 1.2309,
+      "step": 473700
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.383469387755102e-06,
+      "loss": 1.2182,
+      "step": 473800
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.363061224489796e-06,
+      "loss": 1.2239,
+      "step": 473900
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.34265306122449e-06,
+      "loss": 1.2318,
+      "step": 474000
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.322244897959184e-06,
+      "loss": 1.2261,
+      "step": 474100
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.301836734693878e-06,
+      "loss": 1.2312,
+      "step": 474200
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.281428571428572e-06,
+      "loss": 1.2369,
+      "step": 474300
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.261020408163265e-06,
+      "loss": 1.2279,
+      "step": 474400
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.240612244897959e-06,
+      "loss": 1.2224,
+      "step": 474500
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.2202040816326535e-06,
+      "loss": 1.231,
+      "step": 474600
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.199795918367347e-06,
+      "loss": 1.2239,
+      "step": 474700
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.179387755102041e-06,
+      "loss": 1.2269,
+      "step": 474800
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.158979591836735e-06,
+      "loss": 1.2187,
+      "step": 474900
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.138571428571429e-06,
+      "loss": 1.2271,
+      "step": 475000
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.118163265306123e-06,
+      "loss": 1.2337,
+      "step": 475100
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.0977551020408165e-06,
+      "loss": 1.229,
+      "step": 475200
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.077346938775511e-06,
+      "loss": 1.2337,
+      "step": 475300
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.056938775510205e-06,
+      "loss": 1.2321,
+      "step": 475400
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.036530612244898e-06,
+      "loss": 1.228,
+      "step": 475500
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.016122448979592e-06,
+      "loss": 1.2311,
+      "step": 475600
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.995714285714286e-06,
+      "loss": 1.2198,
+      "step": 475700
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.9753061224489796e-06,
+      "loss": 1.229,
+      "step": 475800
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.954897959183674e-06,
+      "loss": 1.2195,
+      "step": 475900
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.934489795918368e-06,
+      "loss": 1.2255,
+      "step": 476000
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.914081632653061e-06,
+      "loss": 1.2266,
+      "step": 476100
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.893673469387755e-06,
+      "loss": 1.225,
+      "step": 476200
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.873265306122449e-06,
+      "loss": 1.2273,
+      "step": 476300
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.852857142857143e-06,
+      "loss": 1.2315,
+      "step": 476400
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.832448979591837e-06,
+      "loss": 1.2277,
+      "step": 476500
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.812040816326531e-06,
+      "loss": 1.2338,
+      "step": 476600
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.791632653061225e-06,
+      "loss": 1.2265,
+      "step": 476700
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.771224489795919e-06,
+      "loss": 1.2267,
+      "step": 476800
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.750816326530612e-06,
+      "loss": 1.2231,
+      "step": 476900
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.7304081632653065e-06,
+      "loss": 1.2203,
+      "step": 477000
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.710000000000001e-06,
+      "loss": 1.2236,
+      "step": 477100
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.689591836734694e-06,
+      "loss": 1.2229,
+      "step": 477200
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.669183673469388e-06,
+      "loss": 1.2329,
+      "step": 477300
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.648979591836735e-06,
+      "loss": 1.2221,
+      "step": 477400
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.628571428571429e-06,
+      "loss": 1.2226,
+      "step": 477500
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.608163265306123e-06,
+      "loss": 1.2282,
+      "step": 477600
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.587755102040816e-06,
+      "loss": 1.2237,
+      "step": 477700
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.5673469387755104e-06,
+      "loss": 1.2311,
+      "step": 477800
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.5469387755102046e-06,
+      "loss": 1.2274,
+      "step": 477900
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.526530612244898e-06,
+      "loss": 1.2282,
+      "step": 478000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.506122448979592e-06,
+      "loss": 1.2281,
+      "step": 478100
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.485714285714286e-06,
+      "loss": 1.2275,
+      "step": 478200
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.465306122448979e-06,
+      "loss": 1.22,
+      "step": 478300
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4448979591836735e-06,
+      "loss": 1.2232,
+      "step": 478400
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.424489795918368e-06,
+      "loss": 1.2264,
+      "step": 478500
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.404081632653062e-06,
+      "loss": 1.2218,
+      "step": 478600
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.383673469387755e-06,
+      "loss": 1.2193,
+      "step": 478700
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.363265306122449e-06,
+      "loss": 1.2281,
+      "step": 478800
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.342857142857143e-06,
+      "loss": 1.2275,
+      "step": 478900
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.322448979591837e-06,
+      "loss": 1.2196,
+      "step": 479000
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.302040816326531e-06,
+      "loss": 1.2273,
+      "step": 479100
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.281632653061225e-06,
+      "loss": 1.2242,
+      "step": 479200
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.261224489795919e-06,
+      "loss": 1.2295,
+      "step": 479300
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.240816326530612e-06,
+      "loss": 1.2229,
+      "step": 479400
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.22061224489796e-06,
+      "loss": 1.2248,
+      "step": 479500
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.200204081632653e-06,
+      "loss": 1.2287,
+      "step": 479600
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.179795918367347e-06,
+      "loss": 1.228,
+      "step": 479700
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.159387755102041e-06,
+      "loss": 1.2262,
+      "step": 479800
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.138979591836735e-06,
+      "loss": 1.2226,
+      "step": 479900
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.118571428571429e-06,
+      "loss": 1.2266,
+      "step": 480000
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.098163265306123e-06,
+      "loss": 1.2271,
+      "step": 480100
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.077755102040817e-06,
+      "loss": 1.2317,
+      "step": 480200
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.05734693877551e-06,
+      "loss": 1.2212,
+      "step": 480300
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.036938775510204e-06,
+      "loss": 1.2195,
+      "step": 480400
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.0165306122448985e-06,
+      "loss": 1.2338,
+      "step": 480500
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.996122448979592e-06,
+      "loss": 1.2159,
+      "step": 480600
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.975714285714286e-06,
+      "loss": 1.2264,
+      "step": 480700
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.95530612244898e-06,
+      "loss": 1.229,
+      "step": 480800
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.934897959183673e-06,
+      "loss": 1.2207,
+      "step": 480900
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.914489795918367e-06,
+      "loss": 1.2257,
+      "step": 481000
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.8940816326530615e-06,
+      "loss": 1.2255,
+      "step": 481100
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.873673469387756e-06,
+      "loss": 1.2307,
+      "step": 481200
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.853265306122449e-06,
+      "loss": 1.2227,
+      "step": 481300
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.832857142857143e-06,
+      "loss": 1.2256,
+      "step": 481400
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.812448979591837e-06,
+      "loss": 1.2249,
+      "step": 481500
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.792244897959184e-06,
+      "loss": 1.2172,
+      "step": 481600
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.771836734693878e-06,
+      "loss": 1.222,
+      "step": 481700
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.7514285714285718e-06,
+      "loss": 1.2157,
+      "step": 481800
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.7310204081632655e-06,
+      "loss": 1.237,
+      "step": 481900
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.7106122448979596e-06,
+      "loss": 1.2277,
+      "step": 482000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.6902040816326533e-06,
+      "loss": 1.2335,
+      "step": 482100
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.669795918367347e-06,
+      "loss": 1.2252,
+      "step": 482200
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.649591836734694e-06,
+      "loss": 1.2236,
+      "step": 482300
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.629183673469388e-06,
+      "loss": 1.2247,
+      "step": 482400
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.608775510204082e-06,
+      "loss": 1.2235,
+      "step": 482500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.5883673469387757e-06,
+      "loss": 1.2295,
+      "step": 482600
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.5679591836734694e-06,
+      "loss": 1.2266,
+      "step": 482700
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.5475510204081635e-06,
+      "loss": 1.2223,
+      "step": 482800
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.527142857142857e-06,
+      "loss": 1.2246,
+      "step": 482900
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.506734693877551e-06,
+      "loss": 1.2246,
+      "step": 483000
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.4863265306122454e-06,
+      "loss": 1.2276,
+      "step": 483100
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.4659183673469387e-06,
+      "loss": 1.2216,
+      "step": 483200
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.4455102040816324e-06,
+      "loss": 1.2202,
+      "step": 483300
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.425102040816327e-06,
+      "loss": 1.2219,
+      "step": 483400
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.4046938775510207e-06,
+      "loss": 1.2339,
+      "step": 483500
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.384285714285714e-06,
+      "loss": 1.2185,
+      "step": 483600
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.3638775510204085e-06,
+      "loss": 1.2231,
+      "step": 483700
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.343469387755102e-06,
+      "loss": 1.2306,
+      "step": 483800
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.323061224489796e-06,
+      "loss": 1.2254,
+      "step": 483900
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.30265306122449e-06,
+      "loss": 1.2276,
+      "step": 484000
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.2822448979591837e-06,
+      "loss": 1.2308,
+      "step": 484100
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.261836734693878e-06,
+      "loss": 1.2268,
+      "step": 484200
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.2414285714285716e-06,
+      "loss": 1.2258,
+      "step": 484300
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.2212244897959183e-06,
+      "loss": 1.2246,
+      "step": 484400
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.2008163265306124e-06,
+      "loss": 1.2265,
+      "step": 484500
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.180408163265306e-06,
+      "loss": 1.2245,
+      "step": 484600
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.1600000000000007e-06,
+      "loss": 1.2169,
+      "step": 484700
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.139591836734694e-06,
+      "loss": 1.2282,
+      "step": 484800
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.119183673469388e-06,
+      "loss": 1.2285,
+      "step": 484900
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.0987755102040818e-06,
+      "loss": 1.223,
+      "step": 485000
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.078367346938776e-06,
+      "loss": 1.2171,
+      "step": 485100
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.0579591836734696e-06,
+      "loss": 1.2261,
+      "step": 485200
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.0375510204081633e-06,
+      "loss": 1.2176,
+      "step": 485300
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.0171428571428574e-06,
+      "loss": 1.2157,
+      "step": 485400
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.996734693877551e-06,
+      "loss": 1.2245,
+      "step": 485500
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.976326530612245e-06,
+      "loss": 1.2222,
+      "step": 485600
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.955918367346939e-06,
+      "loss": 1.2213,
+      "step": 485700
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.9355102040816326e-06,
+      "loss": 1.219,
+      "step": 485800
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.9151020408163268e-06,
+      "loss": 1.222,
+      "step": 485900
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.8946938775510205e-06,
+      "loss": 1.2206,
+      "step": 486000
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.8742857142857146e-06,
+      "loss": 1.2267,
+      "step": 486100
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.8538775510204083e-06,
+      "loss": 1.2256,
+      "step": 486200
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.833469387755102e-06,
+      "loss": 1.226,
+      "step": 486300
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.813061224489796e-06,
+      "loss": 1.2251,
+      "step": 486400
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.79265306122449e-06,
+      "loss": 1.2263,
+      "step": 486500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.772244897959184e-06,
+      "loss": 1.2272,
+      "step": 486600
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.7518367346938777e-06,
+      "loss": 1.2217,
+      "step": 486700
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.7314285714285718e-06,
+      "loss": 1.2232,
+      "step": 486800
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.7110204081632655e-06,
+      "loss": 1.2339,
+      "step": 486900
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.690612244897959e-06,
+      "loss": 1.2232,
+      "step": 487000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.6702040816326533e-06,
+      "loss": 1.2222,
+      "step": 487100
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.649795918367347e-06,
+      "loss": 1.2229,
+      "step": 487200
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.6293877551020407e-06,
+      "loss": 1.2266,
+      "step": 487300
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.608979591836735e-06,
+      "loss": 1.2234,
+      "step": 487400
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.5885714285714285e-06,
+      "loss": 1.2215,
+      "step": 487500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.5681632653061227e-06,
+      "loss": 1.2245,
+      "step": 487600
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.5477551020408164e-06,
+      "loss": 1.2196,
+      "step": 487700
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.5273469387755105e-06,
+      "loss": 1.2243,
+      "step": 487800
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.506938775510204e-06,
+      "loss": 1.2196,
+      "step": 487900
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.486530612244898e-06,
+      "loss": 1.2181,
+      "step": 488000
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.466122448979592e-06,
+      "loss": 1.2224,
+      "step": 488100
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.4457142857142857e-06,
+      "loss": 1.2226,
+      "step": 488200
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.42530612244898e-06,
+      "loss": 1.2301,
+      "step": 488300
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4048979591836735e-06,
+      "loss": 1.2223,
+      "step": 488400
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.3844897959183677e-06,
+      "loss": 1.2235,
+      "step": 488500
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.3642857142857144e-06,
+      "loss": 1.2188,
+      "step": 488600
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.3438775510204085e-06,
+      "loss": 1.2206,
+      "step": 488700
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.3234693877551022e-06,
+      "loss": 1.2249,
+      "step": 488800
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.303061224489796e-06,
+      "loss": 1.2195,
+      "step": 488900
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.28265306122449e-06,
+      "loss": 1.2179,
+      "step": 489000
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.2622448979591838e-06,
+      "loss": 1.2227,
+      "step": 489100
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.2418367346938775e-06,
+      "loss": 1.2185,
+      "step": 489200
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.2214285714285716e-06,
+      "loss": 1.2242,
+      "step": 489300
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.2010204081632657e-06,
+      "loss": 1.2216,
+      "step": 489400
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.180612244897959e-06,
+      "loss": 1.2285,
+      "step": 489500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.160204081632653e-06,
+      "loss": 1.2212,
+      "step": 489600
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.1397959183673472e-06,
+      "loss": 1.2219,
+      "step": 489700
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.119387755102041e-06,
+      "loss": 1.2212,
+      "step": 489800
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.0989795918367346e-06,
+      "loss": 1.2282,
+      "step": 489900
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.0785714285714288e-06,
+      "loss": 1.2213,
+      "step": 490000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.0581632653061225e-06,
+      "loss": 1.2181,
+      "step": 490100
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.037755102040816e-06,
+      "loss": 1.2207,
+      "step": 490200
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.0173469387755103e-06,
+      "loss": 1.2176,
+      "step": 490300
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.9969387755102044e-06,
+      "loss": 1.2252,
+      "step": 490400
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.9765306122448977e-06,
+      "loss": 1.2247,
+      "step": 490500
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.9563265306122453e-06,
+      "loss": 1.2245,
+      "step": 490600
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.9359183673469385e-06,
+      "loss": 1.2242,
+      "step": 490700
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.9155102040816327e-06,
+      "loss": 1.2224,
+      "step": 490800
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.8951020408163268e-06,
+      "loss": 1.2253,
+      "step": 490900
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.8746938775510203e-06,
+      "loss": 1.2197,
+      "step": 491000
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.8542857142857144e-06,
+      "loss": 1.2264,
+      "step": 491100
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.8338775510204083e-06,
+      "loss": 1.2218,
+      "step": 491200
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.8134693877551022e-06,
+      "loss": 1.2184,
+      "step": 491300
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.793061224489796e-06,
+      "loss": 1.2283,
+      "step": 491400
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.7726530612244899e-06,
+      "loss": 1.2203,
+      "step": 491500
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.7522448979591838e-06,
+      "loss": 1.2184,
+      "step": 491600
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.7318367346938775e-06,
+      "loss": 1.2223,
+      "step": 491700
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.7114285714285714e-06,
+      "loss": 1.2258,
+      "step": 491800
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.6910204081632655e-06,
+      "loss": 1.2317,
+      "step": 491900
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.6706122448979594e-06,
+      "loss": 1.2195,
+      "step": 492000
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.6502040816326531e-06,
+      "loss": 1.2249,
+      "step": 492100
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.629795918367347e-06,
+      "loss": 1.2281,
+      "step": 492200
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.609387755102041e-06,
+      "loss": 1.2272,
+      "step": 492300
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5889795918367346e-06,
+      "loss": 1.2195,
+      "step": 492400
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5685714285714286e-06,
+      "loss": 1.2191,
+      "step": 492500
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5481632653061227e-06,
+      "loss": 1.2325,
+      "step": 492600
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5277551020408164e-06,
+      "loss": 1.2208,
+      "step": 492700
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5075510204081633e-06,
+      "loss": 1.2266,
+      "step": 492800
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.4871428571428572e-06,
+      "loss": 1.2201,
+      "step": 492900
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.4667346938775512e-06,
+      "loss": 1.2213,
+      "step": 493000
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.446326530612245e-06,
+      "loss": 1.2284,
+      "step": 493100
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.426122448979592e-06,
+      "loss": 1.2221,
+      "step": 493200
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.4057142857142857e-06,
+      "loss": 1.2205,
+      "step": 493300
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.3853061224489796e-06,
+      "loss": 1.219,
+      "step": 493400
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.3648979591836735e-06,
+      "loss": 1.2211,
+      "step": 493500
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.3444897959183675e-06,
+      "loss": 1.2197,
+      "step": 493600
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.3240816326530614e-06,
+      "loss": 1.2257,
+      "step": 493700
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.303673469387755e-06,
+      "loss": 1.2174,
+      "step": 493800
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.283265306122449e-06,
+      "loss": 1.218,
+      "step": 493900
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.262857142857143e-06,
+      "loss": 1.2164,
+      "step": 494000
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.2424489795918368e-06,
+      "loss": 1.2246,
+      "step": 494100
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.2220408163265307e-06,
+      "loss": 1.2148,
+      "step": 494200
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.2016326530612244e-06,
+      "loss": 1.2194,
+      "step": 494300
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.1812244897959185e-06,
+      "loss": 1.2263,
+      "step": 494400
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.1608163265306123e-06,
+      "loss": 1.2212,
+      "step": 494500
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.1404081632653062e-06,
+      "loss": 1.2207,
+      "step": 494600
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.12e-06,
+      "loss": 1.2134,
+      "step": 494700
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.0995918367346938e-06,
+      "loss": 1.2241,
+      "step": 494800
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.079183673469388e-06,
+      "loss": 1.2178,
+      "step": 494900
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.0587755102040816e-06,
+      "loss": 1.2188,
+      "step": 495000
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.0383673469387755e-06,
+      "loss": 1.2221,
+      "step": 495100
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.0179591836734694e-06,
+      "loss": 1.2234,
+      "step": 495200
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 9.975510204081633e-07,
+      "loss": 1.226,
+      "step": 495300
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 9.771428571428573e-07,
+      "loss": 1.2343,
+      "step": 495400
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 9.56734693877551e-07,
+      "loss": 1.2268,
+      "step": 495500
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 9.36326530612245e-07,
+      "loss": 1.2226,
+      "step": 495600
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 9.159183673469388e-07,
+      "loss": 1.2204,
+      "step": 495700
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 8.955102040816328e-07,
+      "loss": 1.2191,
+      "step": 495800
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 8.751020408163266e-07,
+      "loss": 1.2223,
+      "step": 495900
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 8.546938775510204e-07,
+      "loss": 1.2241,
+      "step": 496000
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 8.342857142857143e-07,
+      "loss": 1.2242,
+      "step": 496100
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 8.138775510204081e-07,
+      "loss": 1.2202,
+      "step": 496200
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 7.934693877551022e-07,
+      "loss": 1.2285,
+      "step": 496300
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 7.73061224489796e-07,
+      "loss": 1.2218,
+      "step": 496400
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 7.526530612244899e-07,
+      "loss": 1.2196,
+      "step": 496500
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 7.322448979591837e-07,
+      "loss": 1.227,
+      "step": 496600
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 7.118367346938776e-07,
+      "loss": 1.2142,
+      "step": 496700
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 6.914285714285714e-07,
+      "loss": 1.225,
+      "step": 496800
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 6.710204081632653e-07,
+      "loss": 1.2231,
+      "step": 496900
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 6.506122448979592e-07,
+      "loss": 1.2213,
+      "step": 497000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 6.302040816326531e-07,
+      "loss": 1.2222,
+      "step": 497100
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 6.100000000000001e-07,
+      "loss": 1.2253,
+      "step": 497200
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 5.895918367346939e-07,
+      "loss": 1.2174,
+      "step": 497300
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.691836734693878e-07,
+      "loss": 1.2149,
+      "step": 497400
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.487755102040816e-07,
+      "loss": 1.2161,
+      "step": 497500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.283673469387755e-07,
+      "loss": 1.2249,
+      "step": 497600
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.079591836734694e-07,
+      "loss": 1.2213,
+      "step": 497700
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.875510204081632e-07,
+      "loss": 1.2238,
+      "step": 497800
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.6714285714285716e-07,
+      "loss": 1.2297,
+      "step": 497900
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.469387755102041e-07,
+      "loss": 1.2267,
+      "step": 498000
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.2653061224489797e-07,
+      "loss": 1.2237,
+      "step": 498100
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.061224489795919e-07,
+      "loss": 1.2292,
+      "step": 498200
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.8571428571428574e-07,
+      "loss": 1.2176,
+      "step": 498300
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.653061224489796e-07,
+      "loss": 1.223,
+      "step": 498400
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.4489795918367346e-07,
+      "loss": 1.2245,
+      "step": 498500
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.244897959183674e-07,
+      "loss": 1.2216,
+      "step": 498600
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.0408163265306124e-07,
+      "loss": 1.2207,
+      "step": 498700
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.836734693877551e-07,
+      "loss": 1.223,
+      "step": 498800
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.63265306122449e-07,
+      "loss": 1.2292,
+      "step": 498900
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.4285714285714287e-07,
+      "loss": 1.2224,
+      "step": 499000
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.2244897959183673e-07,
+      "loss": 1.2235,
+      "step": 499100
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.0204081632653064e-07,
+      "loss": 1.2154,
+      "step": 499200
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.816326530612245e-07,
+      "loss": 1.2203,
+      "step": 499300
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.6122448979591836e-07,
+      "loss": 1.2276,
+      "step": 499400
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4081632653061225e-07,
+      "loss": 1.2178,
+      "step": 499500
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.2040816326530614e-07,
+      "loss": 1.2197,
+      "step": 499600
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 1.2206,
+      "step": 499700
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 7.959183673469388e-08,
+      "loss": 1.2273,
+      "step": 499800
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 5.9183673469387755e-08,
+      "loss": 1.2239,
+      "step": 499900
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.8775510204081635e-08,
+      "loss": 1.2166,
+      "step": 500000
+    }
+  ],
+  "max_steps": 500000,
+  "num_train_epochs": 11,
+  "total_flos": 1.6845286741848883e+19,
+  "trial_name": null,
+  "trial_params": null
+}