diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6806 @@
+{
+  "best_metric": 14.6241,
+  "best_model_checkpoint": "ckpt_mt5/google/mt5-small/wmt14_de_en/lr0.0005_e5/checkpoint-485000",
+  "epoch": 1.728598296664301,
+  "global_step": 485000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0005,
+      "loss": 4.2906,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0005,
+      "loss": 3.3621,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005,
+      "loss": 3.1838,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005,
+      "loss": 3.1029,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005,
+      "loss": 2.9995,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005,
+      "loss": 2.9416,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005,
+      "loss": 2.9255,
+      "step": 3500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005,
+      "loss": 2.8504,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.7921,
+      "step": 4500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.785,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02,
+      "eval_bleu": 8.6689,
+      "eval_gen_len": 17.3423,
+      "eval_loss": 2.1870176792144775,
+      "eval_runtime": 74.9779,
+      "eval_samples_per_second": 40.012,
+      "eval_steps_per_second": 2.507,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.7676,
+      "step": 5500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.7304,
+      "step": 6000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.6996,
+      "step": 6500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.687,
+      "step": 7000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005,
+      "loss": 2.6594,
+      "step": 7500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005,
+      "loss": 2.6131,
+      "step": 8000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005,
+      "loss": 2.6093,
+      "step": 8500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005,
+      "loss": 2.6242,
+      "step": 9000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005,
+      "loss": 2.59,
+      "step": 9500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005,
+      "loss": 2.5956,
+      "step": 10000
+    },
+    {
+      "epoch": 0.04,
+      "eval_bleu": 9.9836,
+      "eval_gen_len": 17.241,
+      "eval_loss": 2.0619494915008545,
+      "eval_runtime": 75.5756,
+      "eval_samples_per_second": 39.695,
+      "eval_steps_per_second": 2.488,
+      "step": 10000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005,
+      "loss": 2.5955,
+      "step": 10500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005,
+      "loss": 2.5197,
+      "step": 11000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005,
+      "loss": 2.5671,
+      "step": 11500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005,
+      "loss": 2.5201,
+      "step": 12000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005,
+      "loss": 2.5214,
+      "step": 12500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005,
+      "loss": 2.523,
+      "step": 13000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005,
+      "loss": 2.4906,
+      "step": 13500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005,
+      "loss": 2.4856,
+      "step": 14000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005,
+      "loss": 2.4681,
+      "step": 14500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005,
+      "loss": 2.4763,
+      "step": 15000
+    },
+    {
+      "epoch": 0.05,
+      "eval_bleu": 10.67,
+      "eval_gen_len": 17.2197,
+      "eval_loss": 1.9743248224258423,
+      "eval_runtime": 79.6829,
+      "eval_samples_per_second": 37.649,
+      "eval_steps_per_second": 2.359,
+      "step": 15000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005,
+      "loss": 2.4721,
+      "step": 15500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005,
+      "loss": 2.4426,
+      "step": 16000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005,
+      "loss": 2.4512,
+      "step": 16500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005,
+      "loss": 2.4263,
+      "step": 17000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005,
+      "loss": 2.4351,
+      "step": 17500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005,
+      "loss": 2.4482,
+      "step": 18000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005,
+      "loss": 2.4204,
+      "step": 18500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005,
+      "loss": 2.4438,
+      "step": 19000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005,
+      "loss": 2.3884,
+      "step": 19500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005,
+      "loss": 2.3764,
+      "step": 20000
+    },
+    {
+      "epoch": 0.07,
+      "eval_bleu": 11.1967,
+      "eval_gen_len": 17.296,
+      "eval_loss": 1.922694206237793,
+      "eval_runtime": 77.6785,
+      "eval_samples_per_second": 38.621,
+      "eval_steps_per_second": 2.42,
+      "step": 20000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005,
+      "loss": 2.3651,
+      "step": 20500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005,
+      "loss": 2.3791,
+      "step": 21000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005,
+      "loss": 2.3891,
+      "step": 21500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005,
+      "loss": 2.3799,
+      "step": 22000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005,
+      "loss": 2.3572,
+      "step": 22500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005,
+      "loss": 2.3239,
+      "step": 23000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005,
+      "loss": 2.3508,
+      "step": 23500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005,
+      "loss": 2.3593,
+      "step": 24000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005,
+      "loss": 2.3267,
+      "step": 24500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005,
+      "loss": 2.329,
+      "step": 25000
+    },
+    {
+      "epoch": 0.09,
+      "eval_bleu": 11.5527,
+      "eval_gen_len": 17.196,
+      "eval_loss": 1.8880093097686768,
+      "eval_runtime": 76.9144,
+      "eval_samples_per_second": 39.004,
+      "eval_steps_per_second": 2.444,
+      "step": 25000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005,
+      "loss": 2.3136,
+      "step": 25500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005,
+      "loss": 2.3021,
+      "step": 26000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005,
+      "loss": 2.3095,
+      "step": 26500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005,
+      "loss": 2.3081,
+      "step": 27000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005,
+      "loss": 2.3129,
+      "step": 27500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005,
+      "loss": 2.2957,
+      "step": 28000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005,
+      "loss": 2.2801,
+      "step": 28500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005,
+      "loss": 2.3218,
+      "step": 29000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005,
+      "loss": 2.3056,
+      "step": 29500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005,
+      "loss": 2.3077,
+      "step": 30000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bleu": 11.6106,
+      "eval_gen_len": 17.2203,
+      "eval_loss": 1.847371220588684,
+      "eval_runtime": 77.5071,
+      "eval_samples_per_second": 38.706,
+      "eval_steps_per_second": 2.426,
+      "step": 30000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005,
+      "loss": 2.2749,
+      "step": 30500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005,
+      "loss": 2.2909,
+      "step": 31000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005,
+      "loss": 2.2818,
+      "step": 31500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005,
+      "loss": 2.2445,
+      "step": 32000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005,
+      "loss": 2.2738,
+      "step": 32500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005,
+      "loss": 2.2694,
+      "step": 33000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005,
+      "loss": 2.2831,
+      "step": 33500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005,
+      "loss": 2.2465,
+      "step": 34000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005,
+      "loss": 2.2417,
+      "step": 34500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005,
+      "loss": 2.2464,
+      "step": 35000
+    },
+    {
+      "epoch": 0.12,
+      "eval_bleu": 12.0534,
+      "eval_gen_len": 17.2567,
+      "eval_loss": 1.818135380744934,
+      "eval_runtime": 77.3917,
+      "eval_samples_per_second": 38.764,
+      "eval_steps_per_second": 2.429,
+      "step": 35000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005,
+      "loss": 2.2738,
+      "step": 35500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005,
+      "loss": 2.262,
+      "step": 36000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005,
+      "loss": 2.2466,
+      "step": 36500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005,
+      "loss": 2.2159,
+      "step": 37000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005,
+      "loss": 2.25,
+      "step": 37500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005,
+      "loss": 2.2642,
+      "step": 38000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005,
+      "loss": 2.2274,
+      "step": 38500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005,
+      "loss": 2.2393,
+      "step": 39000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005,
+      "loss": 2.2304,
+      "step": 39500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005,
+      "loss": 2.2375,
+      "step": 40000
+    },
+    {
+      "epoch": 0.14,
+      "eval_bleu": 12.1559,
+      "eval_gen_len": 17.216,
+      "eval_loss": 1.7900797128677368,
+      "eval_runtime": 78.0422,
+      "eval_samples_per_second": 38.441,
+      "eval_steps_per_second": 2.409,
+      "step": 40000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005,
+      "loss": 2.2212,
+      "step": 40500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005,
+      "loss": 2.206,
+      "step": 41000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005,
+      "loss": 2.208,
+      "step": 41500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005,
+      "loss": 2.2029,
+      "step": 42000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005,
+      "loss": 2.1751,
+      "step": 42500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005,
+      "loss": 2.184,
+      "step": 43000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005,
+      "loss": 2.1939,
+      "step": 43500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005,
+      "loss": 2.2117,
+      "step": 44000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005,
+      "loss": 2.2107,
+      "step": 44500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005,
+      "loss": 2.18,
+      "step": 45000
+    },
+    {
+      "epoch": 0.16,
+      "eval_bleu": 12.1248,
+      "eval_gen_len": 17.284,
+      "eval_loss": 1.7793049812316895,
+      "eval_runtime": 76.4511,
+      "eval_samples_per_second": 39.241,
+      "eval_steps_per_second": 2.459,
+      "step": 45000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005,
+      "loss": 2.2427,
+      "step": 45500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005,
+      "loss": 2.1951,
+      "step": 46000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005,
+      "loss": 2.1968,
+      "step": 46500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005,
+      "loss": 2.1945,
+      "step": 47000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005,
+      "loss": 2.2012,
+      "step": 47500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005,
+      "loss": 2.1959,
+      "step": 48000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005,
+      "loss": 2.1812,
+      "step": 48500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005,
+      "loss": 2.1749,
+      "step": 49000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005,
+      "loss": 2.1464,
+      "step": 49500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005,
+      "loss": 2.1738,
+      "step": 50000
+    },
+    {
+      "epoch": 0.18,
+      "eval_bleu": 12.3069,
+      "eval_gen_len": 17.1913,
+      "eval_loss": 1.7640517950057983,
+      "eval_runtime": 75.3864,
+      "eval_samples_per_second": 39.795,
+      "eval_steps_per_second": 2.494,
+      "step": 50000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005,
+      "loss": 2.1766,
+      "step": 50500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005,
+      "loss": 2.1622,
+      "step": 51000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005,
+      "loss": 2.1651,
+      "step": 51500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005,
+      "loss": 2.1622,
+      "step": 52000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005,
+      "loss": 2.1705,
+      "step": 52500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005,
+      "loss": 2.1765,
+      "step": 53000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005,
+      "loss": 2.1721,
+      "step": 53500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005,
+      "loss": 2.153,
+      "step": 54000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005,
+      "loss": 2.1421,
+      "step": 54500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 2.1476,
+      "step": 55000
+    },
+    {
+      "epoch": 0.2,
+      "eval_bleu": 12.3162,
+      "eval_gen_len": 17.2753,
+      "eval_loss": 1.7403764724731445,
+      "eval_runtime": 75.7862,
+      "eval_samples_per_second": 39.585,
+      "eval_steps_per_second": 2.481,
+      "step": 55000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 2.1788,
+      "step": 55500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 2.14,
+      "step": 56000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 2.1712,
+      "step": 56500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 2.1841,
+      "step": 57000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 2.1401,
+      "step": 57500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 2.1408,
+      "step": 58000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 2.1622,
+      "step": 58500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 2.1415,
+      "step": 59000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 2.1263,
+      "step": 59500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 2.1225,
+      "step": 60000
+    },
+    {
+      "epoch": 0.21,
+      "eval_bleu": 12.7127,
+      "eval_gen_len": 17.236,
+      "eval_loss": 1.7263902425765991,
+      "eval_runtime": 75.5523,
+      "eval_samples_per_second": 39.708,
+      "eval_steps_per_second": 2.488,
+      "step": 60000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 2.1242,
+      "step": 60500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 2.1397,
+      "step": 61000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 2.1143,
+      "step": 61500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 2.0984,
+      "step": 62000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 2.0944,
+      "step": 62500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 2.0971,
+      "step": 63000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 2.0958,
+      "step": 63500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 2.1482,
+      "step": 64000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 2.1305,
+      "step": 64500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 2.1116,
+      "step": 65000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bleu": 12.5727,
+      "eval_gen_len": 17.2863,
+      "eval_loss": 1.7269318103790283,
+      "eval_runtime": 76.2078,
+      "eval_samples_per_second": 39.366,
+      "eval_steps_per_second": 2.467,
+      "step": 65000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 2.1068,
+      "step": 65500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 2.1216,
+      "step": 66000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 2.1071,
+      "step": 66500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 2.0968,
+      "step": 67000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 2.1156,
+      "step": 67500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 2.1056,
+      "step": 68000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 2.0921,
+      "step": 68500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 2.1117,
+      "step": 69000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 2.0971,
+      "step": 69500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 2.092,
+      "step": 70000
+    },
+    {
+      "epoch": 0.25,
+      "eval_bleu": 12.5954,
+      "eval_gen_len": 17.2973,
+      "eval_loss": 1.7027376890182495,
+      "eval_runtime": 78.3436,
+      "eval_samples_per_second": 38.293,
+      "eval_steps_per_second": 2.4,
+      "step": 70000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 2.1107,
+      "step": 70500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 2.1055,
+      "step": 71000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 2.0887,
+      "step": 71500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 2.0878,
+      "step": 72000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 2.1093,
+      "step": 72500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 2.1082,
+      "step": 73000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 2.0952,
+      "step": 73500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 2.0695,
+      "step": 74000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0833,
+      "step": 74500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0755,
+      "step": 75000
+    },
+    {
+      "epoch": 0.27,
+      "eval_bleu": 12.8258,
+      "eval_gen_len": 17.2463,
+      "eval_loss": 1.6987178325653076,
+      "eval_runtime": 75.184,
+      "eval_samples_per_second": 39.902,
+      "eval_steps_per_second": 2.501,
+      "step": 75000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0794,
+      "step": 75500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0812,
+      "step": 76000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0791,
+      "step": 76500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0994,
+      "step": 77000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 2.0818,
+      "step": 77500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 2.0925,
+      "step": 78000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 2.0857,
+      "step": 78500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 2.0813,
+      "step": 79000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 2.0503,
+      "step": 79500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 2.0881,
+      "step": 80000
+    },
+    {
+      "epoch": 0.29,
+      "eval_bleu": 12.6752,
+      "eval_gen_len": 17.206,
+      "eval_loss": 1.6814645528793335,
+      "eval_runtime": 76.9222,
+      "eval_samples_per_second": 39.0,
+      "eval_steps_per_second": 2.444,
+      "step": 80000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 2.08,
+      "step": 80500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 2.0712,
+      "step": 81000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 2.0535,
+      "step": 81500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 2.0656,
+      "step": 82000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 2.077,
+      "step": 82500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 2.0606,
+      "step": 83000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 2.0517,
+      "step": 83500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 2.0699,
+      "step": 84000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 2.0894,
+      "step": 84500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 2.0526,
+      "step": 85000
+    },
+    {
+      "epoch": 0.3,
+      "eval_bleu": 13.0075,
+      "eval_gen_len": 17.24,
+      "eval_loss": 1.670270562171936,
+      "eval_runtime": 76.5437,
+      "eval_samples_per_second": 39.193,
+      "eval_steps_per_second": 2.456,
+      "step": 85000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 2.0639,
+      "step": 85500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 2.0713,
+      "step": 86000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 2.0675,
+      "step": 86500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 2.0394,
+      "step": 87000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 2.0445,
+      "step": 87500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 2.0401,
+      "step": 88000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 2.0642,
+      "step": 88500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 2.0263,
+      "step": 89000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 2.042,
+      "step": 89500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 2.0486,
+      "step": 90000
+    },
+    {
+      "epoch": 0.32,
+      "eval_bleu": 12.7186,
+      "eval_gen_len": 17.268,
+      "eval_loss": 1.6650505065917969,
+      "eval_runtime": 78.4806,
+      "eval_samples_per_second": 38.226,
+      "eval_steps_per_second": 2.395,
+      "step": 90000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 2.0436,
+      "step": 90500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 2.0531,
+      "step": 91000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 2.0461,
+      "step": 91500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 2.032,
+      "step": 92000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 2.0601,
+      "step": 92500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 2.06,
+      "step": 93000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 2.0491,
+      "step": 93500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 2.031,
+      "step": 94000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 2.0546,
+      "step": 94500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 2.033,
+      "step": 95000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bleu": 12.8054,
+      "eval_gen_len": 17.2767,
+      "eval_loss": 1.6645996570587158,
+      "eval_runtime": 77.2454,
+      "eval_samples_per_second": 38.837,
+      "eval_steps_per_second": 2.434,
+      "step": 95000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 2.0392,
+      "step": 95500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 2.0509,
+      "step": 96000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 2.0402,
+      "step": 96500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 2.0187,
+      "step": 97000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 2.0427,
+      "step": 97500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 2.0376,
+      "step": 98000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 2.0295,
+      "step": 98500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 2.0315,
+      "step": 99000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 2.0259,
+      "step": 99500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.9848,
+      "step": 100000
+    },
+    {
+      "epoch": 0.36,
+      "eval_bleu": 12.6947,
+      "eval_gen_len": 17.2777,
+      "eval_loss": 1.6519057750701904,
+      "eval_runtime": 75.7484,
+      "eval_samples_per_second": 39.605,
+      "eval_steps_per_second": 2.482,
+      "step": 100000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 2.0076,
+      "step": 100500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 2.0286,
+      "step": 101000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 2.0144,
+      "step": 101500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 2.0311,
+      "step": 102000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 2.0224,
+      "step": 102500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 2.0228,
+      "step": 103000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 2.0037,
+      "step": 103500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 2.0163,
+      "step": 104000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 2.0306,
+      "step": 104500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 2.0358,
+      "step": 105000
+    },
+    {
+      "epoch": 0.37,
+      "eval_bleu": 12.9065,
+      "eval_gen_len": 17.245,
+      "eval_loss": 1.6500033140182495,
+      "eval_runtime": 75.4006,
+      "eval_samples_per_second": 39.787,
+      "eval_steps_per_second": 2.493,
+      "step": 105000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 2.011,
+      "step": 105500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 2.0144,
+      "step": 106000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 2.0148,
+      "step": 106500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 2.0073,
+      "step": 107000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 2.0157,
+      "step": 107500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 2.0113,
+      "step": 108000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 2.0247,
+      "step": 108500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 2.027,
+      "step": 109000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 2.0195,
+      "step": 109500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 2.0008,
+      "step": 110000
+    },
+    {
+      "epoch": 0.39,
+      "eval_bleu": 13.1637,
+      "eval_gen_len": 17.2417,
+      "eval_loss": 1.646502137184143,
+      "eval_runtime": 77.1052,
+      "eval_samples_per_second": 38.908,
+      "eval_steps_per_second": 2.438,
+      "step": 110000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 2.0277,
+      "step": 110500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 2.0209,
+      "step": 111000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 2.0031,
+      "step": 111500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.9677,
+      "step": 112000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 2.009,
+      "step": 112500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 2.008,
+      "step": 113000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.9737,
+      "step": 113500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.9751,
+      "step": 114000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.9786,
+      "step": 114500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 2.0042,
+      "step": 115000
+    },
+    {
+      "epoch": 0.41,
+      "eval_bleu": 13.1984,
+      "eval_gen_len": 17.2503,
+      "eval_loss": 1.640678882598877,
+      "eval_runtime": 75.1382,
+      "eval_samples_per_second": 39.926,
+      "eval_steps_per_second": 2.502,
+      "step": 115000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 2.0001,
+      "step": 115500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.9932,
+      "step": 116000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 2.0072,
+      "step": 116500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 2.013,
+      "step": 117000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.9862,
+      "step": 117500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.9812,
+      "step": 118000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 2.0053,
+      "step": 118500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.9839,
+      "step": 119000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 2.0013,
+      "step": 119500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 2.0024,
+      "step": 120000
+    },
+    {
+      "epoch": 0.43,
+      "eval_bleu": 13.3466,
+      "eval_gen_len": 17.2173,
+      "eval_loss": 1.6259303092956543,
+      "eval_runtime": 75.3493,
+      "eval_samples_per_second": 39.815,
+      "eval_steps_per_second": 2.495,
+      "step": 120000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.9748,
+      "step": 120500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.9906,
+      "step": 121000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.9794,
+      "step": 121500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.9941,
+      "step": 122000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.9691,
+      "step": 122500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.9779,
+      "step": 123000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.9621,
+      "step": 123500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.9845,
+      "step": 124000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.9749,
+      "step": 124500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 2.0088,
+      "step": 125000
+    },
+    {
+      "epoch": 0.45,
+      "eval_bleu": 13.1189,
+      "eval_gen_len": 17.275,
+      "eval_loss": 1.6257858276367188,
+      "eval_runtime": 77.4713,
+      "eval_samples_per_second": 38.724,
+      "eval_steps_per_second": 2.427,
+      "step": 125000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 2.001,
+      "step": 125500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 2.0006,
+      "step": 126000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.9929,
+      "step": 126500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.9701,
+      "step": 127000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.9567,
+      "step": 127500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.9862,
+      "step": 128000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.967,
+      "step": 128500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.9643,
+      "step": 129000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.9735,
+      "step": 129500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.9863,
+      "step": 130000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bleu": 13.254,
+      "eval_gen_len": 17.289,
+      "eval_loss": 1.621120810508728,
+      "eval_runtime": 77.7893,
+      "eval_samples_per_second": 38.566,
+      "eval_steps_per_second": 2.417,
+      "step": 130000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.9613,
+      "step": 130500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.9403,
+      "step": 131000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.9848,
+      "step": 131500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.9784,
+      "step": 132000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.9878,
+      "step": 132500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.9622,
+      "step": 133000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.9696,
+      "step": 133500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.9702,
+      "step": 134000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 2.0051,
+      "step": 134500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.9406,
+      "step": 135000
+    },
+    {
+      "epoch": 0.48,
+      "eval_bleu": 13.0814,
+      "eval_gen_len": 17.2627,
+      "eval_loss": 1.6086177825927734,
+      "eval_runtime": 54.2921,
+      "eval_samples_per_second": 55.257,
+      "eval_steps_per_second": 3.463,
+      "step": 135000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.9463,
+      "step": 135500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.9744,
+      "step": 136000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.9768,
+      "step": 136500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.9781,
+      "step": 137000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.9686,
+      "step": 137500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.9826,
+      "step": 138000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.977,
+      "step": 138500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.9606,
+      "step": 139000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.9742,
+      "step": 139500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.9417,
+      "step": 140000
+    },
+    {
+      "epoch": 0.5,
+      "eval_bleu": 13.331,
+      "eval_gen_len": 17.243,
+      "eval_loss": 1.6007747650146484,
+      "eval_runtime": 55.0026,
+      "eval_samples_per_second": 54.543,
+      "eval_steps_per_second": 3.418,
+      "step": 140000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.9637,
+      "step": 140500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.9671,
+      "step": 141000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.9851,
+      "step": 141500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.9976,
+      "step": 142000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.9584,
+      "step": 142500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.9365,
+      "step": 143000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.9545,
+      "step": 143500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.9691,
+      "step": 144000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9197,
+      "step": 144500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9541,
+      "step": 145000
+    },
+    {
+      "epoch": 0.52,
+      "eval_bleu": 13.242,
+      "eval_gen_len": 17.283,
+      "eval_loss": 1.5940744876861572,
+      "eval_runtime": 54.5168,
+      "eval_samples_per_second": 55.029,
+      "eval_steps_per_second": 3.448,
+      "step": 145000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9906,
+      "step": 145500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9543,
+      "step": 146000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9844,
+      "step": 146500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9487,
+      "step": 147000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9823,
+      "step": 147500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9511,
+      "step": 148000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9233,
+      "step": 148500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9802,
+      "step": 149000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9584,
+      "step": 149500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9459,
+      "step": 150000
+    },
+    {
+      "epoch": 0.53,
+      "eval_bleu": 13.3553,
+      "eval_gen_len": 17.2607,
+      "eval_loss": 1.5883177518844604,
+      "eval_runtime": 57.005,
+      "eval_samples_per_second": 52.627,
+      "eval_steps_per_second": 3.298,
+      "step": 150000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9688,
+      "step": 150500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9732,
+      "step": 151000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9384,
+      "step": 151500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9899,
+      "step": 152000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9326,
+      "step": 152500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.9565,
+      "step": 153000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.9639,
+      "step": 153500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.9812,
+      "step": 154000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.9342,
+      "step": 154500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.9771,
+      "step": 155000
+    },
+    {
+      "epoch": 0.55,
+      "eval_bleu": 13.369,
+      "eval_gen_len": 17.278,
+      "eval_loss": 1.5940983295440674,
+      "eval_runtime": 56.3807,
+      "eval_samples_per_second": 53.21,
+      "eval_steps_per_second": 3.334,
+      "step": 155000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.9474,
+      "step": 155500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9577,
+      "step": 156000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9527,
+      "step": 156500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9492,
+      "step": 157000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9415,
+      "step": 157500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9463,
+      "step": 158000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.952,
+      "step": 158500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.9597,
+      "step": 159000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.9485,
+      "step": 159500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.9259,
+      "step": 160000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bleu": 13.4966,
+      "eval_gen_len": 17.2653,
+      "eval_loss": 1.586137056350708,
+      "eval_runtime": 54.7672,
+      "eval_samples_per_second": 54.777,
+      "eval_steps_per_second": 3.433,
+      "step": 160000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.933,
+      "step": 160500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.9394,
+      "step": 161000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.9685,
+      "step": 161500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.9489,
+      "step": 162000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.9291,
+      "step": 162500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.9038,
+      "step": 163000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.946,
+      "step": 163500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.9304,
+      "step": 164000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.9534,
+      "step": 164500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.9195,
+      "step": 165000
+    },
+    {
+      "epoch": 0.59,
+      "eval_bleu": 13.2786,
+      "eval_gen_len": 17.274,
+      "eval_loss": 1.5792673826217651,
+      "eval_runtime": 58.2997,
+      "eval_samples_per_second": 51.458,
+      "eval_steps_per_second": 3.225,
+      "step": 165000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.9385,
+      "step": 165500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.954,
+      "step": 166000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.9365,
+      "step": 166500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9571,
+      "step": 167000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9512,
+      "step": 167500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9368,
+      "step": 168000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9336,
+      "step": 168500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9229,
+      "step": 169000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9441,
+      "step": 169500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9082,
+      "step": 170000
+    },
+    {
+      "epoch": 0.61,
+      "eval_bleu": 13.3953,
+      "eval_gen_len": 17.2657,
+      "eval_loss": 1.5857409238815308,
+      "eval_runtime": 54.8476,
+      "eval_samples_per_second": 54.697,
+      "eval_steps_per_second": 3.428,
+      "step": 170000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9234,
+      "step": 170500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9353,
+      "step": 171000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9284,
+      "step": 171500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9159,
+      "step": 172000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9318,
+      "step": 172500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.9199,
+      "step": 173000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.9289,
+      "step": 173500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.9217,
+      "step": 174000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.9114,
+      "step": 174500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.9494,
+      "step": 175000
+    },
+    {
+      "epoch": 0.62,
+      "eval_bleu": 13.4004,
+      "eval_gen_len": 17.273,
+      "eval_loss": 1.5758062601089478,
+      "eval_runtime": 57.1903,
+      "eval_samples_per_second": 52.456,
+      "eval_steps_per_second": 3.287,
+      "step": 175000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9277,
+      "step": 175500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9157,
+      "step": 176000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9444,
+      "step": 176500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.889,
+      "step": 177000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9238,
+      "step": 177500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9507,
+      "step": 178000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.9373,
+      "step": 178500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.9132,
+      "step": 179000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.9406,
+      "step": 179500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.9249,
+      "step": 180000
+    },
+    {
+      "epoch": 0.64,
+      "eval_bleu": 13.6283,
+      "eval_gen_len": 17.2537,
+      "eval_loss": 1.575814962387085,
+      "eval_runtime": 57.0331,
+      "eval_samples_per_second": 52.601,
+      "eval_steps_per_second": 3.296,
+      "step": 180000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.8947,
+      "step": 180500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.9171,
+      "step": 181000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.9421,
+      "step": 181500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.9266,
+      "step": 182000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.9309,
+      "step": 182500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.8987,
+      "step": 183000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.9232,
+      "step": 183500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.9324,
+      "step": 184000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.9208,
+      "step": 184500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.8935,
+      "step": 185000
+    },
+    {
+      "epoch": 0.66,
+      "eval_bleu": 13.3788,
+      "eval_gen_len": 17.2483,
+      "eval_loss": 1.5698202848434448,
+      "eval_runtime": 55.6032,
+      "eval_samples_per_second": 53.954,
+      "eval_steps_per_second": 3.381,
+      "step": 185000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.8974,
+      "step": 185500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.9344,
+      "step": 186000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.9235,
+      "step": 186500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.9248,
+      "step": 187000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.9153,
+      "step": 187500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.937,
+      "step": 188000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.9145,
+      "step": 188500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.9141,
+      "step": 189000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.912,
+      "step": 189500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.9016,
+      "step": 190000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bleu": 13.2865,
+      "eval_gen_len": 17.2393,
+      "eval_loss": 1.5736759901046753,
+      "eval_runtime": 57.1459,
+      "eval_samples_per_second": 52.497,
+      "eval_steps_per_second": 3.29,
+      "step": 190000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.9082,
+      "step": 190500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.9126,
+      "step": 191000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.9127,
+      "step": 191500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.9173,
+      "step": 192000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.89,
+      "step": 192500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.9111,
+      "step": 193000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.8965,
+      "step": 193500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.8887,
+      "step": 194000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.9098,
+      "step": 194500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9234,
+      "step": 195000
+    },
+    {
+      "epoch": 0.7,
+      "eval_bleu": 13.6446,
+      "eval_gen_len": 17.2447,
+      "eval_loss": 1.556485652923584,
+      "eval_runtime": 56.7404,
+      "eval_samples_per_second": 52.872,
+      "eval_steps_per_second": 3.313,
+      "step": 195000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9286,
+      "step": 195500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9046,
+      "step": 196000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9172,
+      "step": 196500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.893,
+      "step": 197000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9275,
+      "step": 197500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.9002,
+      "step": 198000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.9166,
+      "step": 198500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.9,
+      "step": 199000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.9095,
+      "step": 199500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.9281,
+      "step": 200000
+    },
+    {
+      "epoch": 0.71,
+      "eval_bleu": 13.5403,
+      "eval_gen_len": 17.2677,
+      "eval_loss": 1.5595364570617676,
+      "eval_runtime": 55.8042,
+      "eval_samples_per_second": 53.759,
+      "eval_steps_per_second": 3.369,
+      "step": 200000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.9253,
+      "step": 200500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.8837,
+      "step": 201000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.9171,
+      "step": 201500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.9334,
+      "step": 202000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.9035,
+      "step": 202500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.9163,
+      "step": 203000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.9123,
+      "step": 203500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.871,
+      "step": 204000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.9119,
+      "step": 204500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.8896,
+      "step": 205000
+    },
+    {
+      "epoch": 0.73,
+      "eval_bleu": 13.61,
+      "eval_gen_len": 17.2883,
+      "eval_loss": 1.5544843673706055,
+      "eval_runtime": 57.0258,
+      "eval_samples_per_second": 52.608,
+      "eval_steps_per_second": 3.297,
+      "step": 205000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.8814,
+      "step": 205500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.8872,
+      "step": 206000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.8972,
+      "step": 206500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.9093,
+      "step": 207000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.898,
+      "step": 207500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.8848,
+      "step": 208000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.902,
+      "step": 208500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.8747,
+      "step": 209000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.9155,
+      "step": 209500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.8883,
+      "step": 210000
+    },
+    {
+      "epoch": 0.75,
+      "eval_bleu": 13.6382,
+      "eval_gen_len": 17.247,
+      "eval_loss": 1.5548808574676514,
+      "eval_runtime": 56.334,
+      "eval_samples_per_second": 53.254,
+      "eval_steps_per_second": 3.337,
+      "step": 210000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.9005,
+      "step": 210500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.9068,
+      "step": 211000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.9015,
+      "step": 211500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.8854,
+      "step": 212000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.8845,
+      "step": 212500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.8758,
+      "step": 213000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.9153,
+      "step": 213500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.8902,
+      "step": 214000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.875,
+      "step": 214500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8857,
+      "step": 215000
+    },
+    {
+      "epoch": 0.77,
+      "eval_bleu": 13.6471,
+      "eval_gen_len": 17.292,
+      "eval_loss": 1.553587555885315,
+      "eval_runtime": 57.7895,
+      "eval_samples_per_second": 51.913,
+      "eval_steps_per_second": 3.253,
+      "step": 215000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8922,
+      "step": 215500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8902,
+      "step": 216000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8988,
+      "step": 216500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.888,
+      "step": 217000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.8961,
+      "step": 217500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.8943,
+      "step": 218000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.9052,
+      "step": 218500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.8944,
+      "step": 219000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.8785,
+      "step": 219500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.865,
+      "step": 220000
+    },
+    {
+      "epoch": 0.78,
+      "eval_bleu": 13.918,
+      "eval_gen_len": 17.2647,
+      "eval_loss": 1.5439625978469849,
+      "eval_runtime": 54.6362,
+      "eval_samples_per_second": 54.909,
+      "eval_steps_per_second": 3.441,
+      "step": 220000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8849,
+      "step": 220500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8883,
+      "step": 221000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.862,
+      "step": 221500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8825,
+      "step": 222000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8815,
+      "step": 222500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8997,
+      "step": 223000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.8845,
+      "step": 223500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.8823,
+      "step": 224000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.8904,
+      "step": 224500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.8789,
+      "step": 225000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bleu": 13.9272,
+      "eval_gen_len": 17.2373,
+      "eval_loss": 1.5431175231933594,
+      "eval_runtime": 55.9058,
+      "eval_samples_per_second": 53.662,
+      "eval_steps_per_second": 3.363,
+      "step": 225000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.8795,
+      "step": 225500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8962,
+      "step": 226000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8963,
+      "step": 226500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8925,
+      "step": 227000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8666,
+      "step": 227500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.9047,
+      "step": 228000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8597,
+      "step": 228500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8897,
+      "step": 229000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8853,
+      "step": 229500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8904,
+      "step": 230000
+    },
+    {
+      "epoch": 0.82,
+      "eval_bleu": 13.7333,
+      "eval_gen_len": 17.261,
+      "eval_loss": 1.5343542098999023,
+      "eval_runtime": 55.826,
+      "eval_samples_per_second": 53.738,
+      "eval_steps_per_second": 3.368,
+      "step": 230000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8789,
+      "step": 230500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8872,
+      "step": 231000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8881,
+      "step": 231500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8674,
+      "step": 232000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8963,
+      "step": 232500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8907,
+      "step": 233000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8523,
+      "step": 233500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8699,
+      "step": 234000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8758,
+      "step": 234500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8653,
+      "step": 235000
+    },
+    {
+      "epoch": 0.84,
+      "eval_bleu": 13.7137,
+      "eval_gen_len": 17.2947,
+      "eval_loss": 1.5360989570617676,
+      "eval_runtime": 55.1916,
+      "eval_samples_per_second": 54.356,
+      "eval_steps_per_second": 3.406,
+      "step": 235000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8844,
+      "step": 235500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8897,
+      "step": 236000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8493,
+      "step": 236500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8544,
+      "step": 237000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.8903,
+      "step": 237500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.8902,
+      "step": 238000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.8679,
+      "step": 238500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.8794,
+      "step": 239000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.8974,
+      "step": 239500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8564,
+      "step": 240000
+    },
+    {
+      "epoch": 0.86,
+      "eval_bleu": 13.833,
+      "eval_gen_len": 17.231,
+      "eval_loss": 1.537398099899292,
+      "eval_runtime": 57.0147,
+      "eval_samples_per_second": 52.618,
+      "eval_steps_per_second": 3.297,
+      "step": 240000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.877,
+      "step": 240500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8711,
+      "step": 241000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8803,
+      "step": 241500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8849,
+      "step": 242000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8686,
+      "step": 242500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8793,
+      "step": 243000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8489,
+      "step": 243500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8603,
+      "step": 244000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8865,
+      "step": 244500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.878,
+      "step": 245000
+    },
+    {
+      "epoch": 0.87,
+      "eval_bleu": 13.9611,
+      "eval_gen_len": 17.2507,
+      "eval_loss": 1.5389394760131836,
+      "eval_runtime": 56.3954,
+      "eval_samples_per_second": 53.196,
+      "eval_steps_per_second": 3.334,
+      "step": 245000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8825,
+      "step": 245500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8495,
+      "step": 246000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8383,
+      "step": 246500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8827,
+      "step": 247000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8777,
+      "step": 247500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8593,
+      "step": 248000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.847,
+      "step": 248500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.853,
+      "step": 249000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.8495,
+      "step": 249500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.8768,
+      "step": 250000
+    },
+    {
+      "epoch": 0.89,
+      "eval_bleu": 13.9268,
+      "eval_gen_len": 17.26,
+      "eval_loss": 1.5289517641067505,
+      "eval_runtime": 56.5525,
+      "eval_samples_per_second": 53.048,
+      "eval_steps_per_second": 3.324,
+      "step": 250000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.8782,
+      "step": 250500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.8691,
+      "step": 251000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8416,
+      "step": 251500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.864,
+      "step": 252000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8687,
+      "step": 252500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8708,
+      "step": 253000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8867,
+      "step": 253500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.8646,
+      "step": 254000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.8713,
+      "step": 254500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.8699,
+      "step": 255000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bleu": 13.5624,
+      "eval_gen_len": 17.2623,
+      "eval_loss": 1.525820255279541,
+      "eval_runtime": 55.7292,
+      "eval_samples_per_second": 53.832,
+      "eval_steps_per_second": 3.373,
+      "step": 255000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.8889,
+      "step": 255500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.8737,
+      "step": 256000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.8729,
+      "step": 256500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8692,
+      "step": 257000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8613,
+      "step": 257500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8633,
+      "step": 258000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8492,
+      "step": 258500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8764,
+      "step": 259000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8567,
+      "step": 259500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8481,
+      "step": 260000
+    },
+    {
+      "epoch": 0.93,
+      "eval_bleu": 13.9528,
+      "eval_gen_len": 17.2313,
+      "eval_loss": 1.519413948059082,
+      "eval_runtime": 57.4399,
+      "eval_samples_per_second": 52.229,
+      "eval_steps_per_second": 3.273,
+      "step": 260000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8616,
+      "step": 260500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8378,
+      "step": 261000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8842,
+      "step": 261500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8682,
+      "step": 262000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8482,
+      "step": 262500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8511,
+      "step": 263000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8554,
+      "step": 263500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8719,
+      "step": 264000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8626,
+      "step": 264500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8584,
+      "step": 265000
+    },
+    {
+      "epoch": 0.94,
+      "eval_bleu": 13.7878,
+      "eval_gen_len": 17.2677,
+      "eval_loss": 1.521083950996399,
+      "eval_runtime": 55.1256,
+      "eval_samples_per_second": 54.421,
+      "eval_steps_per_second": 3.41,
+      "step": 265000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8635,
+      "step": 265500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8573,
+      "step": 266000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8443,
+      "step": 266500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8369,
+      "step": 267000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8836,
+      "step": 267500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8786,
+      "step": 268000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8461,
+      "step": 268500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8581,
+      "step": 269000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8826,
+      "step": 269500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8695,
+      "step": 270000
+    },
+    {
+      "epoch": 0.96,
+      "eval_bleu": 13.8741,
+      "eval_gen_len": 17.2333,
+      "eval_loss": 1.517186164855957,
+      "eval_runtime": 56.0257,
+      "eval_samples_per_second": 53.547,
+      "eval_steps_per_second": 3.356,
+      "step": 270000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8485,
+      "step": 270500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8609,
+      "step": 271000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8416,
+      "step": 271500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8761,
+      "step": 272000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8427,
+      "step": 272500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8336,
+      "step": 273000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8519,
+      "step": 273500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8467,
+      "step": 274000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8673,
+      "step": 274500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8586,
+      "step": 275000
+    },
+    {
+      "epoch": 0.98,
+      "eval_bleu": 14.0129,
+      "eval_gen_len": 17.2533,
+      "eval_loss": 1.517733097076416,
+      "eval_runtime": 55.5206,
+      "eval_samples_per_second": 54.034,
+      "eval_steps_per_second": 3.386,
+      "step": 275000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8393,
+      "step": 275500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8353,
+      "step": 276000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8582,
+      "step": 276500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.859,
+      "step": 277000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8649,
+      "step": 277500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8587,
+      "step": 278000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8392,
+      "step": 278500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8426,
+      "step": 279000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.8671,
+      "step": 279500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.8394,
+      "step": 280000
+    },
+    {
+      "epoch": 1.0,
+      "eval_bleu": 13.8641,
+      "eval_gen_len": 17.2733,
+      "eval_loss": 1.5193395614624023,
+      "eval_runtime": 55.516,
+      "eval_samples_per_second": 54.038,
+      "eval_steps_per_second": 3.386,
+      "step": 280000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.863,
+      "step": 280500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.8139,
+      "step": 281000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.8369,
+      "step": 281500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005,
+      "loss": 1.825,
+      "step": 282000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005,
+      "loss": 1.8105,
+      "step": 282500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005,
+      "loss": 1.8025,
+      "step": 283000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005,
+      "loss": 1.8121,
+      "step": 283500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005,
+      "loss": 1.8097,
+      "step": 284000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005,
+      "loss": 1.7995,
+      "step": 284500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005,
+      "loss": 1.8393,
+      "step": 285000
+    },
+    {
+      "epoch": 1.02,
+      "eval_bleu": 13.925,
+      "eval_gen_len": 17.2527,
+      "eval_loss": 1.5177398920059204,
+      "eval_runtime": 69.2297,
+      "eval_samples_per_second": 43.334,
+      "eval_steps_per_second": 2.716,
+      "step": 285000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005,
+      "loss": 1.8165,
+      "step": 285500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005,
+      "loss": 1.813,
+      "step": 286000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005,
+      "loss": 1.8205,
+      "step": 286500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005,
+      "loss": 1.829,
+      "step": 287000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005,
+      "loss": 1.8152,
+      "step": 287500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005,
+      "loss": 1.813,
+      "step": 288000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7925,
+      "step": 288500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005,
+      "loss": 1.8215,
+      "step": 289000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005,
+      "loss": 1.8105,
+      "step": 289500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7858,
+      "step": 290000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bleu": 13.9196,
+      "eval_gen_len": 17.269,
+      "eval_loss": 1.5100245475769043,
+      "eval_runtime": 58.7884,
+      "eval_samples_per_second": 51.03,
+      "eval_steps_per_second": 3.198,
+      "step": 290000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005,
+      "loss": 1.7698,
+      "step": 290500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005,
+      "loss": 1.8412,
+      "step": 291000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005,
+      "loss": 1.8213,
+      "step": 291500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005,
+      "loss": 1.8206,
+      "step": 292000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005,
+      "loss": 1.8101,
+      "step": 292500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005,
+      "loss": 1.8249,
+      "step": 293000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0005,
+      "loss": 1.8056,
+      "step": 293500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0005,
+      "loss": 1.8104,
+      "step": 294000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0005,
+      "loss": 1.8186,
+      "step": 294500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0005,
+      "loss": 1.7779,
+      "step": 295000
+    },
+    {
+      "epoch": 1.05,
+      "eval_bleu": 14.0234,
+      "eval_gen_len": 17.26,
+      "eval_loss": 1.5147415399551392,
+      "eval_runtime": 58.2281,
+      "eval_samples_per_second": 51.522,
+      "eval_steps_per_second": 3.229,
+      "step": 295000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0005,
+      "loss": 1.8025,
+      "step": 295500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0005,
+      "loss": 1.8255,
+      "step": 296000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0005,
+      "loss": 1.8153,
+      "step": 296500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0005,
+      "loss": 1.8065,
+      "step": 297000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0005,
+      "loss": 1.819,
+      "step": 297500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0005,
+      "loss": 1.8067,
+      "step": 298000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0005,
+      "loss": 1.8297,
+      "step": 298500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005,
+      "loss": 1.8232,
+      "step": 299000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005,
+      "loss": 1.8453,
+      "step": 299500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005,
+      "loss": 1.8086,
+      "step": 300000
+    },
+    {
+      "epoch": 1.07,
+      "eval_bleu": 13.8363,
+      "eval_gen_len": 17.2507,
+      "eval_loss": 1.5098841190338135,
+      "eval_runtime": 58.1386,
+      "eval_samples_per_second": 51.601,
+      "eval_steps_per_second": 3.234,
+      "step": 300000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005,
+      "loss": 1.8276,
+      "step": 300500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005,
+      "loss": 1.8265,
+      "step": 301000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005,
+      "loss": 1.8416,
+      "step": 301500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0005,
+      "loss": 1.8034,
+      "step": 302000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0005,
+      "loss": 1.7922,
+      "step": 302500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0005,
+      "loss": 1.8206,
+      "step": 303000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0005,
+      "loss": 1.8396,
+      "step": 303500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0005,
+      "loss": 1.8131,
+      "step": 304000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0005,
+      "loss": 1.8094,
+      "step": 304500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0005,
+      "loss": 1.8102,
+      "step": 305000
+    },
+    {
+      "epoch": 1.09,
+      "eval_bleu": 13.6674,
+      "eval_gen_len": 17.2643,
+      "eval_loss": 1.5067949295043945,
+      "eval_runtime": 58.0621,
+      "eval_samples_per_second": 51.669,
+      "eval_steps_per_second": 3.238,
+      "step": 305000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0005,
+      "loss": 1.8084,
+      "step": 305500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0005,
+      "loss": 1.8105,
+      "step": 306000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0005,
+      "loss": 1.8377,
+      "step": 306500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0005,
+      "loss": 1.7707,
+      "step": 307000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0005,
+      "loss": 1.8084,
+      "step": 307500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0005,
+      "loss": 1.8171,
+      "step": 308000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0005,
+      "loss": 1.8041,
+      "step": 308500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0005,
+      "loss": 1.7884,
+      "step": 309000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0005,
+      "loss": 1.8105,
+      "step": 309500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0005,
+      "loss": 1.8245,
+      "step": 310000
+    },
+    {
+      "epoch": 1.1,
+      "eval_bleu": 14.0697,
+      "eval_gen_len": 17.255,
+      "eval_loss": 1.5014456510543823,
+      "eval_runtime": 57.3567,
+      "eval_samples_per_second": 52.304,
+      "eval_steps_per_second": 3.278,
+      "step": 310000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0005,
+      "loss": 1.8095,
+      "step": 310500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0005,
+      "loss": 1.8078,
+      "step": 311000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0005,
+      "loss": 1.8002,
+      "step": 311500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0005,
+      "loss": 1.83,
+      "step": 312000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0005,
+      "loss": 1.8021,
+      "step": 312500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7793,
+      "step": 313000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0005,
+      "loss": 1.8273,
+      "step": 313500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7922,
+      "step": 314000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7867,
+      "step": 314500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0005,
+      "loss": 1.8321,
+      "step": 315000
+    },
+    {
+      "epoch": 1.12,
+      "eval_bleu": 14.1202,
+      "eval_gen_len": 17.2523,
+      "eval_loss": 1.499343752861023,
+      "eval_runtime": 59.1203,
+      "eval_samples_per_second": 50.744,
+      "eval_steps_per_second": 3.18,
+      "step": 315000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0005,
+      "loss": 1.8448,
+      "step": 315500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0005,
+      "loss": 1.8384,
+      "step": 316000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0005,
+      "loss": 1.8288,
+      "step": 316500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0005,
+      "loss": 1.8087,
+      "step": 317000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0005,
+      "loss": 1.8365,
+      "step": 317500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0005,
+      "loss": 1.8009,
+      "step": 318000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005,
+      "loss": 1.825,
+      "step": 318500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8186,
+      "step": 319000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8139,
+      "step": 319500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8002,
+      "step": 320000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bleu": 14.1518,
+      "eval_gen_len": 17.2397,
+      "eval_loss": 1.503738284111023,
+      "eval_runtime": 58.6225,
+      "eval_samples_per_second": 51.175,
+      "eval_steps_per_second": 3.207,
+      "step": 320000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8041,
+      "step": 320500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8179,
+      "step": 321000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0005,
+      "loss": 1.8251,
+      "step": 321500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0005,
+      "loss": 1.7919,
+      "step": 322000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0005,
+      "loss": 1.8136,
+      "step": 322500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0005,
+      "loss": 1.8143,
+      "step": 323000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0005,
+      "loss": 1.8115,
+      "step": 323500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0005,
+      "loss": 1.8046,
+      "step": 324000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0005,
+      "loss": 1.7949,
+      "step": 324500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0005,
+      "loss": 1.8312,
+      "step": 325000
+    },
+    {
+      "epoch": 1.16,
+      "eval_bleu": 13.9421,
+      "eval_gen_len": 17.249,
+      "eval_loss": 1.4959828853607178,
+      "eval_runtime": 59.1165,
+      "eval_samples_per_second": 50.747,
+      "eval_steps_per_second": 3.18,
+      "step": 325000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0005,
+      "loss": 1.808,
+      "step": 325500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0005,
+      "loss": 1.7676,
+      "step": 326000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0005,
+      "loss": 1.8051,
+      "step": 326500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8024,
+      "step": 327000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8423,
+      "step": 327500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8016,
+      "step": 328000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8092,
+      "step": 328500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0005,
+      "loss": 1.7687,
+      "step": 329000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8044,
+      "step": 329500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0005,
+      "loss": 1.8064,
+      "step": 330000
+    },
+    {
+      "epoch": 1.18,
+      "eval_bleu": 14.0486,
+      "eval_gen_len": 17.2337,
+      "eval_loss": 1.5011292695999146,
+      "eval_runtime": 57.2914,
+      "eval_samples_per_second": 52.364,
+      "eval_steps_per_second": 3.281,
+      "step": 330000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0005,
+      "loss": 1.7806,
+      "step": 330500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0005,
+      "loss": 1.8256,
+      "step": 331000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0005,
+      "loss": 1.7879,
+      "step": 331500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0005,
+      "loss": 1.7673,
+      "step": 332000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0005,
+      "loss": 1.7937,
+      "step": 332500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0005,
+      "loss": 1.7745,
+      "step": 333000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0005,
+      "loss": 1.8203,
+      "step": 333500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0005,
+      "loss": 1.7992,
+      "step": 334000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0005,
+      "loss": 1.8331,
+      "step": 334500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0005,
+      "loss": 1.8028,
+      "step": 335000
+    },
+    {
+      "epoch": 1.19,
+      "eval_bleu": 14.0331,
+      "eval_gen_len": 17.248,
+      "eval_loss": 1.4933398962020874,
+      "eval_runtime": 58.3359,
+      "eval_samples_per_second": 51.426,
+      "eval_steps_per_second": 3.223,
+      "step": 335000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0005,
+      "loss": 1.8095,
+      "step": 335500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0005,
+      "loss": 1.7758,
+      "step": 336000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0005,
+      "loss": 1.8345,
+      "step": 336500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0005,
+      "loss": 1.7921,
+      "step": 337000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0005,
+      "loss": 1.8323,
+      "step": 337500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0005,
+      "loss": 1.8243,
+      "step": 338000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0005,
+      "loss": 1.8437,
+      "step": 338500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0005,
+      "loss": 1.8125,
+      "step": 339000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7893,
+      "step": 339500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0005,
+      "loss": 1.8115,
+      "step": 340000
+    },
+    {
+      "epoch": 1.21,
+      "eval_bleu": 14.0689,
+      "eval_gen_len": 17.2523,
+      "eval_loss": 1.4908339977264404,
+      "eval_runtime": 57.6505,
+      "eval_samples_per_second": 52.038,
+      "eval_steps_per_second": 3.261,
+      "step": 340000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0005,
+      "loss": 1.8192,
+      "step": 340500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0005,
+      "loss": 1.8279,
+      "step": 341000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0005,
+      "loss": 1.7859,
+      "step": 341500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0005,
+      "loss": 1.8074,
+      "step": 342000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0005,
+      "loss": 1.7908,
+      "step": 342500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0005,
+      "loss": 1.783,
+      "step": 343000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0005,
+      "loss": 1.8118,
+      "step": 343500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0005,
+      "loss": 1.7951,
+      "step": 344000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0005,
+      "loss": 1.8099,
+      "step": 344500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0005,
+      "loss": 1.7992,
+      "step": 345000
+    },
+    {
+      "epoch": 1.23,
+      "eval_bleu": 13.9915,
+      "eval_gen_len": 17.2617,
+      "eval_loss": 1.4889838695526123,
+      "eval_runtime": 55.755,
+      "eval_samples_per_second": 53.807,
+      "eval_steps_per_second": 3.372,
+      "step": 345000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0005,
+      "loss": 1.8256,
+      "step": 345500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0005,
+      "loss": 1.8071,
+      "step": 346000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0005,
+      "loss": 1.8093,
+      "step": 346500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7783,
+      "step": 347000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7999,
+      "step": 347500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0005,
+      "loss": 1.8231,
+      "step": 348000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0005,
+      "loss": 1.803,
+      "step": 348500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0005,
+      "loss": 1.8121,
+      "step": 349000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7942,
+      "step": 349500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7732,
+      "step": 350000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bleu": 13.8037,
+      "eval_gen_len": 17.2967,
+      "eval_loss": 1.4908981323242188,
+      "eval_runtime": 58.6857,
+      "eval_samples_per_second": 51.12,
+      "eval_steps_per_second": 3.204,
+      "step": 350000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005,
+      "loss": 1.8,
+      "step": 350500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7652,
+      "step": 351000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7952,
+      "step": 351500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005,
+      "loss": 1.794,
+      "step": 352000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0005,
+      "loss": 1.8055,
+      "step": 352500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0005,
+      "loss": 1.8129,
+      "step": 353000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0005,
+      "loss": 1.8024,
+      "step": 353500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0005,
+      "loss": 1.804,
+      "step": 354000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0005,
+      "loss": 1.7791,
+      "step": 354500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0005,
+      "loss": 1.797,
+      "step": 355000
+    },
+    {
+      "epoch": 1.27,
+      "eval_bleu": 14.0526,
+      "eval_gen_len": 17.244,
+      "eval_loss": 1.488826870918274,
+      "eval_runtime": 56.593,
+      "eval_samples_per_second": 53.01,
+      "eval_steps_per_second": 3.322,
+      "step": 355000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0005,
+      "loss": 1.7846,
+      "step": 355500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0005,
+      "loss": 1.8094,
+      "step": 356000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0005,
+      "loss": 1.8021,
+      "step": 356500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0005,
+      "loss": 1.8118,
+      "step": 357000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0005,
+      "loss": 1.808,
+      "step": 357500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0005,
+      "loss": 1.7633,
+      "step": 358000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0005,
+      "loss": 1.7864,
+      "step": 358500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0005,
+      "loss": 1.7865,
+      "step": 359000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0005,
+      "loss": 1.8052,
+      "step": 359500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0005,
+      "loss": 1.7971,
+      "step": 360000
+    },
+    {
+      "epoch": 1.28,
+      "eval_bleu": 13.9767,
+      "eval_gen_len": 17.2483,
+      "eval_loss": 1.4857127666473389,
+      "eval_runtime": 68.0402,
+      "eval_samples_per_second": 44.092,
+      "eval_steps_per_second": 2.763,
+      "step": 360000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0005,
+      "loss": 1.7821,
+      "step": 360500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0005,
+      "loss": 1.7603,
+      "step": 361000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0005,
+      "loss": 1.7785,
+      "step": 361500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0005,
+      "loss": 1.7866,
+      "step": 362000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0005,
+      "loss": 1.8029,
+      "step": 362500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0005,
+      "loss": 1.8106,
+      "step": 363000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7781,
+      "step": 363500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0005,
+      "loss": 1.8185,
+      "step": 364000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7951,
+      "step": 364500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7982,
+      "step": 365000
+    },
+    {
+      "epoch": 1.3,
+      "eval_bleu": 14.2541,
+      "eval_gen_len": 17.2673,
+      "eval_loss": 1.4883339405059814,
+      "eval_runtime": 57.438,
+      "eval_samples_per_second": 52.23,
+      "eval_steps_per_second": 3.273,
+      "step": 365000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7834,
+      "step": 365500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7825,
+      "step": 366000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0005,
+      "loss": 1.7801,
+      "step": 366500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0005,
+      "loss": 1.7887,
+      "step": 367000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0005,
+      "loss": 1.8146,
+      "step": 367500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0005,
+      "loss": 1.8067,
+      "step": 368000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0005,
+      "loss": 1.7944,
+      "step": 368500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0005,
+      "loss": 1.789,
+      "step": 369000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0005,
+      "loss": 1.8147,
+      "step": 369500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0005,
+      "loss": 1.8009,
+      "step": 370000
+    },
+    {
+      "epoch": 1.32,
+      "eval_bleu": 13.9439,
+      "eval_gen_len": 17.2537,
+      "eval_loss": 1.4910156726837158,
+      "eval_runtime": 57.6598,
+      "eval_samples_per_second": 52.029,
+      "eval_steps_per_second": 3.261,
+      "step": 370000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0005,
+      "loss": 1.8013,
+      "step": 370500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0005,
+      "loss": 1.7743,
+      "step": 371000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0005,
+      "loss": 1.7997,
+      "step": 371500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0005,
+      "loss": 1.8086,
+      "step": 372000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7805,
+      "step": 372500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7833,
+      "step": 373000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0005,
+      "loss": 1.8107,
+      "step": 373500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0005,
+      "loss": 1.8051,
+      "step": 374000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7952,
+      "step": 374500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0005,
+      "loss": 1.8034,
+      "step": 375000
+    },
+    {
+      "epoch": 1.34,
+      "eval_bleu": 14.0659,
+      "eval_gen_len": 17.262,
+      "eval_loss": 1.4841067790985107,
+      "eval_runtime": 56.7073,
+      "eval_samples_per_second": 52.903,
+      "eval_steps_per_second": 3.315,
+      "step": 375000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7786,
+      "step": 375500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0005,
+      "loss": 1.8033,
+      "step": 376000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7906,
+      "step": 376500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7716,
+      "step": 377000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7754,
+      "step": 377500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0005,
+      "loss": 1.8103,
+      "step": 378000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0005,
+      "loss": 1.8032,
+      "step": 378500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7887,
+      "step": 379000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7988,
+      "step": 379500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7749,
+      "step": 380000
+    },
+    {
+      "epoch": 1.35,
+      "eval_bleu": 14.1403,
+      "eval_gen_len": 17.2243,
+      "eval_loss": 1.4869729280471802,
+      "eval_runtime": 57.3527,
+      "eval_samples_per_second": 52.308,
+      "eval_steps_per_second": 3.278,
+      "step": 380000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0005,
+      "loss": 1.7827,
+      "step": 380500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0005,
+      "loss": 1.7826,
+      "step": 381000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0005,
+      "loss": 1.7988,
+      "step": 381500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0005,
+      "loss": 1.8066,
+      "step": 382000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0005,
+      "loss": 1.7609,
+      "step": 382500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7997,
+      "step": 383000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7933,
+      "step": 383500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7993,
+      "step": 384000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005,
+      "loss": 1.8265,
+      "step": 384500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7996,
+      "step": 385000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bleu": 14.05,
+      "eval_gen_len": 17.2807,
+      "eval_loss": 1.4809935092926025,
+      "eval_runtime": 58.3627,
+      "eval_samples_per_second": 51.403,
+      "eval_steps_per_second": 3.221,
+      "step": 385000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7988,
+      "step": 385500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005,
+      "loss": 1.7904,
+      "step": 386000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005,
+      "loss": 1.7789,
+      "step": 386500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005,
+      "loss": 1.791,
+      "step": 387000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005,
+      "loss": 1.8078,
+      "step": 387500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005,
+      "loss": 1.7871,
+      "step": 388000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005,
+      "loss": 1.7951,
+      "step": 388500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7803,
+      "step": 389000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7769,
+      "step": 389500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7845,
+      "step": 390000
+    },
+    {
+      "epoch": 1.39,
+      "eval_bleu": 14.0637,
+      "eval_gen_len": 17.2767,
+      "eval_loss": 1.4824107885360718,
+      "eval_runtime": 59.4114,
+      "eval_samples_per_second": 50.495,
+      "eval_steps_per_second": 3.164,
+      "step": 390000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7962,
+      "step": 390500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7869,
+      "step": 391000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005,
+      "loss": 1.8016,
+      "step": 391500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005,
+      "loss": 1.7827,
+      "step": 392000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005,
+      "loss": 1.7577,
+      "step": 392500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005,
+      "loss": 1.7957,
+      "step": 393000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005,
+      "loss": 1.808,
+      "step": 393500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005,
+      "loss": 1.8029,
+      "step": 394000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005,
+      "loss": 1.8014,
+      "step": 394500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005,
+      "loss": 1.7882,
+      "step": 395000
+    },
+    {
+      "epoch": 1.41,
+      "eval_bleu": 13.933,
+      "eval_gen_len": 17.267,
+      "eval_loss": 1.477520227432251,
+      "eval_runtime": 57.9935,
+      "eval_samples_per_second": 51.73,
+      "eval_steps_per_second": 3.242,
+      "step": 395000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005,
+      "loss": 1.7853,
+      "step": 395500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005,
+      "loss": 1.7708,
+      "step": 396000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005,
+      "loss": 1.8081,
+      "step": 396500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005,
+      "loss": 1.77,
+      "step": 397000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005,
+      "loss": 1.7606,
+      "step": 397500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005,
+      "loss": 1.806,
+      "step": 398000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005,
+      "loss": 1.7997,
+      "step": 398500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005,
+      "loss": 1.7805,
+      "step": 399000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005,
+      "loss": 1.8046,
+      "step": 399500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7863,
+      "step": 400000
+    },
+    {
+      "epoch": 1.43,
+      "eval_bleu": 14.1466,
+      "eval_gen_len": 17.23,
+      "eval_loss": 1.479000210762024,
+      "eval_runtime": 63.7211,
+      "eval_samples_per_second": 47.08,
+      "eval_steps_per_second": 2.95,
+      "step": 400000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005,
+      "loss": 1.794,
+      "step": 400500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7924,
+      "step": 401000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7948,
+      "step": 401500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7797,
+      "step": 402000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7688,
+      "step": 402500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005,
+      "loss": 1.8014,
+      "step": 403000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7573,
+      "step": 403500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7894,
+      "step": 404000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7481,
+      "step": 404500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7667,
+      "step": 405000
+    },
+    {
+      "epoch": 1.44,
+      "eval_bleu": 14.1007,
+      "eval_gen_len": 17.2963,
+      "eval_loss": 1.472170352935791,
+      "eval_runtime": 57.3007,
+      "eval_samples_per_second": 52.355,
+      "eval_steps_per_second": 3.281,
+      "step": 405000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005,
+      "loss": 1.7474,
+      "step": 405500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005,
+      "loss": 1.8064,
+      "step": 406000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005,
+      "loss": 1.7628,
+      "step": 406500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005,
+      "loss": 1.7922,
+      "step": 407000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005,
+      "loss": 1.793,
+      "step": 407500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005,
+      "loss": 1.7827,
+      "step": 408000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005,
+      "loss": 1.8075,
+      "step": 408500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005,
+      "loss": 1.7674,
+      "step": 409000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005,
+      "loss": 1.7685,
+      "step": 409500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005,
+      "loss": 1.8137,
+      "step": 410000
+    },
+    {
+      "epoch": 1.46,
+      "eval_bleu": 13.8737,
+      "eval_gen_len": 17.2367,
+      "eval_loss": 1.4769831895828247,
+      "eval_runtime": 56.7931,
+      "eval_samples_per_second": 52.823,
+      "eval_steps_per_second": 3.31,
+      "step": 410000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005,
+      "loss": 1.7656,
+      "step": 410500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005,
+      "loss": 1.7856,
+      "step": 411000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005,
+      "loss": 1.795,
+      "step": 411500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005,
+      "loss": 1.7595,
+      "step": 412000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005,
+      "loss": 1.769,
+      "step": 412500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005,
+      "loss": 1.7841,
+      "step": 413000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005,
+      "loss": 1.7911,
+      "step": 413500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005,
+      "loss": 1.7899,
+      "step": 414000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005,
+      "loss": 1.7467,
+      "step": 414500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005,
+      "loss": 1.8101,
+      "step": 415000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bleu": 14.0735,
+      "eval_gen_len": 17.2617,
+      "eval_loss": 1.4739737510681152,
+      "eval_runtime": 58.2304,
+      "eval_samples_per_second": 51.519,
+      "eval_steps_per_second": 3.229,
+      "step": 415000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005,
+      "loss": 1.7823,
+      "step": 415500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005,
+      "loss": 1.7847,
+      "step": 416000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005,
+      "loss": 1.795,
+      "step": 416500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005,
+      "loss": 1.752,
+      "step": 417000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005,
+      "loss": 1.7738,
+      "step": 417500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005,
+      "loss": 1.794,
+      "step": 418000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005,
+      "loss": 1.7677,
+      "step": 418500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005,
+      "loss": 1.7672,
+      "step": 419000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005,
+      "loss": 1.7875,
+      "step": 419500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005,
+      "loss": 1.78,
+      "step": 420000
+    },
+    {
+      "epoch": 1.5,
+      "eval_bleu": 14.0199,
+      "eval_gen_len": 17.256,
+      "eval_loss": 1.4736225605010986,
+      "eval_runtime": 57.9404,
+      "eval_samples_per_second": 51.777,
+      "eval_steps_per_second": 3.245,
+      "step": 420000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005,
+      "loss": 1.7597,
+      "step": 420500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005,
+      "loss": 1.7742,
+      "step": 421000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005,
+      "loss": 1.771,
+      "step": 421500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005,
+      "loss": 1.759,
+      "step": 422000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7733,
+      "step": 422500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7776,
+      "step": 423000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7722,
+      "step": 423500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005,
+      "loss": 1.765,
+      "step": 424000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7692,
+      "step": 424500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7797,
+      "step": 425000
+    },
+    {
+      "epoch": 1.51,
+      "eval_bleu": 13.9873,
+      "eval_gen_len": 17.2477,
+      "eval_loss": 1.4714795351028442,
+      "eval_runtime": 58.3783,
+      "eval_samples_per_second": 51.389,
+      "eval_steps_per_second": 3.22,
+      "step": 425000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7897,
+      "step": 425500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7991,
+      "step": 426000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005,
+      "loss": 1.772,
+      "step": 426500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7809,
+      "step": 427000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7575,
+      "step": 427500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005,
+      "loss": 1.753,
+      "step": 428000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7875,
+      "step": 428500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7566,
+      "step": 429000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7924,
+      "step": 429500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005,
+      "loss": 1.775,
+      "step": 430000
+    },
+    {
+      "epoch": 1.53,
+      "eval_bleu": 14.361,
+      "eval_gen_len": 17.2647,
+      "eval_loss": 1.4696695804595947,
+      "eval_runtime": 60.0244,
+      "eval_samples_per_second": 49.98,
+      "eval_steps_per_second": 3.132,
+      "step": 430000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7852,
+      "step": 430500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7911,
+      "step": 431000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7905,
+      "step": 431500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7749,
+      "step": 432000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7974,
+      "step": 432500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7941,
+      "step": 433000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7899,
+      "step": 433500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7701,
+      "step": 434000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7809,
+      "step": 434500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7782,
+      "step": 435000
+    },
+    {
+      "epoch": 1.55,
+      "eval_bleu": 14.0776,
+      "eval_gen_len": 17.2567,
+      "eval_loss": 1.4726725816726685,
+      "eval_runtime": 58.09,
+      "eval_samples_per_second": 51.644,
+      "eval_steps_per_second": 3.236,
+      "step": 435000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7696,
+      "step": 435500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7899,
+      "step": 436000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005,
+      "loss": 1.8065,
+      "step": 436500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005,
+      "loss": 1.7821,
+      "step": 437000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005,
+      "loss": 1.7941,
+      "step": 437500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005,
+      "loss": 1.7931,
+      "step": 438000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005,
+      "loss": 1.7563,
+      "step": 438500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005,
+      "loss": 1.7817,
+      "step": 439000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7871,
+      "step": 439500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7646,
+      "step": 440000
+    },
+    {
+      "epoch": 1.57,
+      "eval_bleu": 14.0668,
+      "eval_gen_len": 17.2107,
+      "eval_loss": 1.4652223587036133,
+      "eval_runtime": 57.2394,
+      "eval_samples_per_second": 52.411,
+      "eval_steps_per_second": 3.284,
+      "step": 440000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005,
+      "loss": 1.786,
+      "step": 440500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005,
+      "loss": 1.754,
+      "step": 441000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7615,
+      "step": 441500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7547,
+      "step": 442000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7768,
+      "step": 442500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7578,
+      "step": 443000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7654,
+      "step": 443500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7801,
+      "step": 444000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7595,
+      "step": 444500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7911,
+      "step": 445000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bleu": 14.1315,
+      "eval_gen_len": 17.2263,
+      "eval_loss": 1.4709407091140747,
+      "eval_runtime": 57.9162,
+      "eval_samples_per_second": 51.799,
+      "eval_steps_per_second": 3.246,
+      "step": 445000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7791,
+      "step": 445500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7904,
+      "step": 446000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7612,
+      "step": 446500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7528,
+      "step": 447000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7453,
+      "step": 447500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005,
+      "loss": 1.7646,
+      "step": 448000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005,
+      "loss": 1.7598,
+      "step": 448500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005,
+      "loss": 1.7728,
+      "step": 449000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005,
+      "loss": 1.7657,
+      "step": 449500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005,
+      "loss": 1.7759,
+      "step": 450000
+    },
+    {
+      "epoch": 1.6,
+      "eval_bleu": 14.2901,
+      "eval_gen_len": 17.2527,
+      "eval_loss": 1.4628537893295288,
+      "eval_runtime": 57.2566,
+      "eval_samples_per_second": 52.396,
+      "eval_steps_per_second": 3.283,
+      "step": 450000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7606,
+      "step": 450500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7559,
+      "step": 451000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7758,
+      "step": 451500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7988,
+      "step": 452000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7472,
+      "step": 452500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005,
+      "loss": 1.774,
+      "step": 453000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7806,
+      "step": 453500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7657,
+      "step": 454000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7892,
+      "step": 454500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7972,
+      "step": 455000
+    },
+    {
+      "epoch": 1.62,
+      "eval_bleu": 14.2148,
+      "eval_gen_len": 17.2593,
+      "eval_loss": 1.4607665538787842,
+      "eval_runtime": 55.1996,
+      "eval_samples_per_second": 54.348,
+      "eval_steps_per_second": 3.406,
+      "step": 455000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7785,
+      "step": 455500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005,
+      "loss": 1.7538,
+      "step": 456000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005,
+      "loss": 1.7823,
+      "step": 456500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005,
+      "loss": 1.7772,
+      "step": 457000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005,
+      "loss": 1.7788,
+      "step": 457500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005,
+      "loss": 1.7597,
+      "step": 458000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005,
+      "loss": 1.7643,
+      "step": 458500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005,
+      "loss": 1.7757,
+      "step": 459000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005,
+      "loss": 1.764,
+      "step": 459500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005,
+      "loss": 1.7492,
+      "step": 460000
+    },
+    {
+      "epoch": 1.64,
+      "eval_bleu": 14.353,
+      "eval_gen_len": 17.271,
+      "eval_loss": 1.462286114692688,
+      "eval_runtime": 57.7586,
+      "eval_samples_per_second": 51.94,
+      "eval_steps_per_second": 3.255,
+      "step": 460000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005,
+      "loss": 1.765,
+      "step": 460500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005,
+      "loss": 1.7635,
+      "step": 461000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005,
+      "loss": 1.748,
+      "step": 461500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005,
+      "loss": 1.7941,
+      "step": 462000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005,
+      "loss": 1.7539,
+      "step": 462500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005,
+      "loss": 1.7672,
+      "step": 463000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005,
+      "loss": 1.792,
+      "step": 463500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005,
+      "loss": 1.7553,
+      "step": 464000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7461,
+      "step": 464500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7649,
+      "step": 465000
+    },
+    {
+      "epoch": 1.66,
+      "eval_bleu": 14.3656,
+      "eval_gen_len": 17.2963,
+      "eval_loss": 1.4579209089279175,
+      "eval_runtime": 55.4147,
+      "eval_samples_per_second": 54.137,
+      "eval_steps_per_second": 3.393,
+      "step": 465000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7659,
+      "step": 465500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7563,
+      "step": 466000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7759,
+      "step": 466500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7803,
+      "step": 467000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0005,
+      "loss": 1.7909,
+      "step": 467500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0005,
+      "loss": 1.758,
+      "step": 468000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0005,
+      "loss": 1.762,
+      "step": 468500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0005,
+      "loss": 1.7665,
+      "step": 469000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0005,
+      "loss": 1.7645,
+      "step": 469500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0005,
+      "loss": 1.757,
+      "step": 470000
+    },
+    {
+      "epoch": 1.68,
+      "eval_bleu": 14.1494,
+      "eval_gen_len": 17.2347,
+      "eval_loss": 1.4613909721374512,
+      "eval_runtime": 56.9605,
+      "eval_samples_per_second": 52.668,
+      "eval_steps_per_second": 3.301,
+      "step": 470000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7621,
+      "step": 470500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7538,
+      "step": 471000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0005,
+      "loss": 1.797,
+      "step": 471500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7722,
+      "step": 472000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7654,
+      "step": 472500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7723,
+      "step": 473000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7514,
+      "step": 473500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7586,
+      "step": 474000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7557,
+      "step": 474500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7518,
+      "step": 475000
+    },
+    {
+      "epoch": 1.69,
+      "eval_bleu": 14.2583,
+      "eval_gen_len": 17.259,
+      "eval_loss": 1.460300087928772,
+      "eval_runtime": 63.109,
+      "eval_samples_per_second": 47.537,
+      "eval_steps_per_second": 2.979,
+      "step": 475000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7781,
+      "step": 475500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7716,
+      "step": 476000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7518,
+      "step": 476500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7746,
+      "step": 477000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7607,
+      "step": 477500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7493,
+      "step": 478000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7524,
+      "step": 478500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7483,
+      "step": 479000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0005,
+      "loss": 1.762,
+      "step": 479500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7504,
+      "step": 480000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bleu": 14.2568,
+      "eval_gen_len": 17.264,
+      "eval_loss": 1.455437183380127,
+      "eval_runtime": 65.2807,
+      "eval_samples_per_second": 45.955,
+      "eval_steps_per_second": 2.88,
+      "step": 480000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7848,
+      "step": 480500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7731,
+      "step": 481000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7421,
+      "step": 481500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005,
+      "loss": 1.765,
+      "step": 482000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7516,
+      "step": 482500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7634,
+      "step": 483000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7578,
+      "step": 483500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7699,
+      "step": 484000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7659,
+      "step": 484500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7747,
+      "step": 485000
+    },
+    {
+      "epoch": 1.73,
+      "eval_bleu": 14.6241,
+      "eval_gen_len": 17.2663,
+      "eval_loss": 1.4529153108596802,
+      "eval_runtime": 55.9317,
+      "eval_samples_per_second": 53.637,
+      "eval_steps_per_second": 3.361,
+      "step": 485000
+    }
+  ],
+  "max_steps": 1409000,
+  "num_train_epochs": 5,
+  "total_flos": 7.347877541793485e+17,
+  "trial_name": null,
+  "trial_params": null
+}