diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30666 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 39.20953575909661,
+  "global_step": 500000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005,
+      "loss": 2.4473,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.3281,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.3071,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005,
+      "loss": 2.2586,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005,
+      "loss": 2.2505,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005,
+      "loss": 2.2206,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005,
+      "loss": 2.2542,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005,
+      "loss": 2.2092,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005,
+      "loss": 2.1569,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005,
+      "loss": 2.1698,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005,
+      "loss": 2.1675,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005,
+      "loss": 2.1551,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005,
+      "loss": 2.1548,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005,
+      "loss": 2.1338,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005,
+      "loss": 2.0949,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005,
+      "loss": 2.1218,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005,
+      "loss": 2.1147,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005,
+      "loss": 2.1143,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005,
+      "loss": 2.1062,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005,
+      "loss": 2.0921,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005,
+      "loss": 2.0893,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005,
+      "loss": 2.1006,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005,
+      "loss": 2.0757,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005,
+      "loss": 2.0428,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 2.059,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 2.0686,
+      "step": 2600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 2.057,
+      "step": 2700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 2.0589,
+      "step": 2800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 2.0715,
+      "step": 2900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 2.0384,
+      "step": 3000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 2.0547,
+      "step": 3100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 2.0339,
+      "step": 3200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 2.0184,
+      "step": 3300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0153,
+      "step": 3400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0178,
+      "step": 3500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 2.0025,
+      "step": 3600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 2.0144,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 2.0053,
+      "step": 3800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.99,
+      "step": 3900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.9965,
+      "step": 4000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 2.0051,
+      "step": 4100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 2.014,
+      "step": 4200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.973,
+      "step": 4300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 1.9985,
+      "step": 4400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 2.0279,
+      "step": 4500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.9903,
+      "step": 4600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 1.9699,
+      "step": 4700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.9834,
+      "step": 4800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.9638,
+      "step": 4900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.9783,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.9678,
+      "step": 5100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.9462,
+      "step": 5200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.9501,
+      "step": 5300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.9362,
+      "step": 5400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.976,
+      "step": 5500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.9635,
+      "step": 5600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.9401,
+      "step": 5700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.9454,
+      "step": 5800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.9544,
+      "step": 5900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.9394,
+      "step": 6000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.943,
+      "step": 6100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.9451,
+      "step": 6200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.9259,
+      "step": 6300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.937,
+      "step": 6400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.9343,
+      "step": 6500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9644,
+      "step": 6600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9195,
+      "step": 6700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9413,
+      "step": 6800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9294,
+      "step": 6900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.9241,
+      "step": 7000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9278,
+      "step": 7100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9228,
+      "step": 7200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.8982,
+      "step": 7300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.8915,
+      "step": 7400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.9206,
+      "step": 7500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9225,
+      "step": 7600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9032,
+      "step": 7700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9072,
+      "step": 7800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.9003,
+      "step": 7900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9222,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.8978,
+      "step": 8100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.8991,
+      "step": 8200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.8978,
+      "step": 8300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.878,
+      "step": 8400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.921,
+      "step": 8500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.878,
+      "step": 8600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.9043,
+      "step": 8700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.8767,
+      "step": 8800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.879,
+      "step": 8900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.9042,
+      "step": 9000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.8796,
+      "step": 9100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.8692,
+      "step": 9200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.8757,
+      "step": 9300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.8602,
+      "step": 9400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.8847,
+      "step": 9500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.8655,
+      "step": 9600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.8683,
+      "step": 9700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8654,
+      "step": 9800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.8712,
+      "step": 9900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.8636,
+      "step": 10000
+    },
+    {
+      "epoch": 0.78,
+      "eval_gen_len": 18.695153173832434,
+      "eval_loss": 2.3997342586517334,
+      "eval_rouge1": 31.2993,
+      "eval_rouge2": 10.395,
+      "eval_rougeL": 25.2713,
+      "eval_rougeLsum": 25.2611,
+      "eval_runtime": 363.1513,
+      "eval_samples_per_second": 31.191,
+      "eval_steps_per_second": 1.95,
+      "step": 10000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8802,
+      "step": 10100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.8608,
+      "step": 10200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8651,
+      "step": 10300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8559,
+      "step": 10400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8508,
+      "step": 10500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.872,
+      "step": 10600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8624,
+      "step": 10700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.8603,
+      "step": 10800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.85,
+      "step": 10900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8705,
+      "step": 11000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8259,
+      "step": 11100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8653,
+      "step": 11200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.8511,
+      "step": 11300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.8401,
+      "step": 11400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8473,
+      "step": 11500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.8077,
+      "step": 11600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8412,
+      "step": 11700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8516,
+      "step": 11800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8604,
+      "step": 11900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8361,
+      "step": 12000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8098,
+      "step": 12100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.839,
+      "step": 12200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8477,
+      "step": 12300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8398,
+      "step": 12400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8353,
+      "step": 12500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8528,
+      "step": 12600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.8446,
+      "step": 12700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.7852,
+      "step": 12800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005,
+      "loss": 1.7505,
+      "step": 12900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005,
+      "loss": 1.7305,
+      "step": 13000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7235,
+      "step": 13100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005,
+      "loss": 1.7329,
+      "step": 13200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005,
+      "loss": 1.7175,
+      "step": 13300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0005,
+      "loss": 1.7238,
+      "step": 13400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0005,
+      "loss": 1.7691,
+      "step": 13500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005,
+      "loss": 1.7501,
+      "step": 13600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005,
+      "loss": 1.7305,
+      "step": 13700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0005,
+      "loss": 1.7305,
+      "step": 13800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0005,
+      "loss": 1.7437,
+      "step": 13900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0005,
+      "loss": 1.7608,
+      "step": 14000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0005,
+      "loss": 1.7506,
+      "step": 14100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0005,
+      "loss": 1.7368,
+      "step": 14200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7633,
+      "step": 14300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0005,
+      "loss": 1.7264,
+      "step": 14400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005,
+      "loss": 1.7292,
+      "step": 14500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005,
+      "loss": 1.7532,
+      "step": 14600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0005,
+      "loss": 1.741,
+      "step": 14700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0005,
+      "loss": 1.7643,
+      "step": 14800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0005,
+      "loss": 1.7447,
+      "step": 14900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0005,
+      "loss": 1.7263,
+      "step": 15000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0005,
+      "loss": 1.7093,
+      "step": 15100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0005,
+      "loss": 1.73,
+      "step": 15200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0005,
+      "loss": 1.7178,
+      "step": 15300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7379,
+      "step": 15400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0005,
+      "loss": 1.7404,
+      "step": 15500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0005,
+      "loss": 1.7497,
+      "step": 15600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0005,
+      "loss": 1.7417,
+      "step": 15700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7136,
+      "step": 15800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7516,
+      "step": 15900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7399,
+      "step": 16000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0005,
+      "loss": 1.7556,
+      "step": 16100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0005,
+      "loss": 1.73,
+      "step": 16200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0005,
+      "loss": 1.742,
+      "step": 16300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0005,
+      "loss": 1.7195,
+      "step": 16400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0005,
+      "loss": 1.7433,
+      "step": 16500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7207,
+      "step": 16600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0005,
+      "loss": 1.7332,
+      "step": 16700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0005,
+      "loss": 1.7288,
+      "step": 16800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7386,
+      "step": 16900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7213,
+      "step": 17000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7384,
+      "step": 17100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7298,
+      "step": 17200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0005,
+      "loss": 1.7055,
+      "step": 17300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0005,
+      "loss": 1.7337,
+      "step": 17400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7299,
+      "step": 17500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005,
+      "loss": 1.7272,
+      "step": 17600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7242,
+      "step": 17700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005,
+      "loss": 1.7312,
+      "step": 17800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005,
+      "loss": 1.7404,
+      "step": 17900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005,
+      "loss": 1.7056,
+      "step": 18000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005,
+      "loss": 1.7389,
+      "step": 18100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7082,
+      "step": 18200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6982,
+      "step": 18300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005,
+      "loss": 1.732,
+      "step": 18400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005,
+      "loss": 1.7388,
+      "step": 18500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005,
+      "loss": 1.709,
+      "step": 18600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005,
+      "loss": 1.7028,
+      "step": 18700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005,
+      "loss": 1.7392,
+      "step": 18800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005,
+      "loss": 1.748,
+      "step": 18900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005,
+      "loss": 1.7054,
+      "step": 19000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005,
+      "loss": 1.7273,
+      "step": 19100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7165,
+      "step": 19200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7423,
+      "step": 19300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7209,
+      "step": 19400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7042,
+      "step": 19500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7183,
+      "step": 19600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7267,
+      "step": 19700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7182,
+      "step": 19800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005,
+      "loss": 1.7098,
+      "step": 19900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7148,
+      "step": 20000
+    },
+    {
+      "epoch": 1.57,
+      "eval_gen_len": 18.688443541979343,
+      "eval_loss": 2.347717523574829,
+      "eval_rouge1": 32.4357,
+      "eval_rouge2": 11.3106,
+      "eval_rougeL": 26.2749,
+      "eval_rougeLsum": 26.2677,
+      "eval_runtime": 345.5194,
+      "eval_samples_per_second": 32.783,
+      "eval_steps_per_second": 2.049,
+      "step": 20000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7099,
+      "step": 20100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7115,
+      "step": 20200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7282,
+      "step": 20300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005,
+      "loss": 1.7062,
+      "step": 20400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7236,
+      "step": 20500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7141,
+      "step": 20600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7273,
+      "step": 20700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005,
+      "loss": 1.7074,
+      "step": 20800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005,
+      "loss": 1.7209,
+      "step": 20900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005,
+      "loss": 1.7173,
+      "step": 21000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005,
+      "loss": 1.6918,
+      "step": 21100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0005,
+      "loss": 1.6768,
+      "step": 21200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0005,
+      "loss": 1.7153,
+      "step": 21300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0005,
+      "loss": 1.738,
+      "step": 21400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6996,
+      "step": 21500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7035,
+      "step": 21600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0005,
+      "loss": 1.6942,
+      "step": 21700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6875,
+      "step": 21800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005,
+      "loss": 1.697,
+      "step": 21900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6911,
+      "step": 22000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6806,
+      "step": 22100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0005,
+      "loss": 1.7186,
+      "step": 22200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0005,
+      "loss": 1.7074,
+      "step": 22300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0005,
+      "loss": 1.7027,
+      "step": 22400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0005,
+      "loss": 1.7251,
+      "step": 22500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0005,
+      "loss": 1.6959,
+      "step": 22600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0005,
+      "loss": 1.7194,
+      "step": 22700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0005,
+      "loss": 1.6768,
+      "step": 22800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0005,
+      "loss": 1.6864,
+      "step": 22900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7057,
+      "step": 23000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0005,
+      "loss": 1.702,
+      "step": 23100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0005,
+      "loss": 1.7105,
+      "step": 23200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0005,
+      "loss": 1.7015,
+      "step": 23300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0005,
+      "loss": 1.7065,
+      "step": 23400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6636,
+      "step": 23500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0005,
+      "loss": 1.6822,
+      "step": 23600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0005,
+      "loss": 1.7033,
+      "step": 23700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0005,
+      "loss": 1.7382,
+      "step": 23800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6929,
+      "step": 23900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0005,
+      "loss": 1.6973,
+      "step": 24000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6824,
+      "step": 24100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0005,
+      "loss": 1.7013,
+      "step": 24200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0005,
+      "loss": 1.7004,
+      "step": 24300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0005,
+      "loss": 1.6694,
+      "step": 24400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0005,
+      "loss": 1.6998,
+      "step": 24500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0005,
+      "loss": 1.683,
+      "step": 24600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0005,
+      "loss": 1.6925,
+      "step": 24700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0005,
+      "loss": 1.6659,
+      "step": 24800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0005,
+      "loss": 1.6905,
+      "step": 24900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6997,
+      "step": 25000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0005,
+      "loss": 1.6695,
+      "step": 25100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6833,
+      "step": 25200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6832,
+      "step": 25300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0005,
+      "loss": 1.7041,
+      "step": 25400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6848,
+      "step": 25500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5853,
+      "step": 25600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5959,
+      "step": 25700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005,
+      "loss": 1.6016,
+      "step": 25800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005,
+      "loss": 1.5721,
+      "step": 25900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005,
+      "loss": 1.5981,
+      "step": 26000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5845,
+      "step": 26100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005,
+      "loss": 1.6169,
+      "step": 26200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005,
+      "loss": 1.5921,
+      "step": 26300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005,
+      "loss": 1.6167,
+      "step": 26400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005,
+      "loss": 1.6052,
+      "step": 26500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005,
+      "loss": 1.6189,
+      "step": 26600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005,
+      "loss": 1.6091,
+      "step": 26700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005,
+      "loss": 1.6312,
+      "step": 26800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5886,
+      "step": 26900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005,
+      "loss": 1.6133,
+      "step": 27000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005,
+      "loss": 1.6036,
+      "step": 27100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5927,
+      "step": 27200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5909,
+      "step": 27300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005,
+      "loss": 1.6093,
+      "step": 27400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5991,
+      "step": 27500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5617,
+      "step": 27600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005,
+      "loss": 1.5947,
+      "step": 27700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005,
+      "loss": 1.6055,
+      "step": 27800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005,
+      "loss": 1.6014,
+      "step": 27900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005,
+      "loss": 1.6091,
+      "step": 28000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005,
+      "loss": 1.6327,
+      "step": 28100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005,
+      "loss": 1.6125,
+      "step": 28200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005,
+      "loss": 1.5928,
+      "step": 28300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005,
+      "loss": 1.608,
+      "step": 28400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5971,
+      "step": 28500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005,
+      "loss": 1.6227,
+      "step": 28600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005,
+      "loss": 1.6162,
+      "step": 28700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005,
+      "loss": 1.5908,
+      "step": 28800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005,
+      "loss": 1.6052,
+      "step": 28900
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005,
+      "loss": 1.6251,
+      "step": 29000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005,
+      "loss": 1.6072,
+      "step": 29100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005,
+      "loss": 1.598,
+      "step": 29200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005,
+      "loss": 1.6139,
+      "step": 29300
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6048,
+      "step": 29400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6331,
+      "step": 29500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005,
+      "loss": 1.6119,
+      "step": 29600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6086,
+      "step": 29700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005,
+      "loss": 1.6392,
+      "step": 29800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005,
+      "loss": 1.6186,
+      "step": 29900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5952,
+      "step": 30000
+    },
+    {
+      "epoch": 2.35,
+      "eval_gen_len": 18.797475059592124,
+      "eval_loss": 2.3347554206848145,
+      "eval_rouge1": 32.7907,
+      "eval_rouge2": 11.73,
+      "eval_rougeL": 26.6762,
+      "eval_rougeLsum": 26.6615,
+      "eval_runtime": 345.6461,
+      "eval_samples_per_second": 32.771,
+      "eval_steps_per_second": 2.048,
+      "step": 30000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005,
+      "loss": 1.604,
+      "step": 30100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005,
+      "loss": 1.5966,
+      "step": 30200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005,
+      "loss": 1.6006,
+      "step": 30300
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5954,
+      "step": 30400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5982,
+      "step": 30500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5859,
+      "step": 30600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5999,
+      "step": 30700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6032,
+      "step": 30800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5834,
+      "step": 30900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5798,
+      "step": 31000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005,
+      "loss": 1.5805,
+      "step": 31100
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005,
+      "loss": 1.6271,
+      "step": 31200
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005,
+      "loss": 1.5944,
+      "step": 31300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005,
+      "loss": 1.6272,
+      "step": 31400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005,
+      "loss": 1.6026,
+      "step": 31500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005,
+      "loss": 1.6003,
+      "step": 31600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5817,
+      "step": 31700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5677,
+      "step": 31800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005,
+      "loss": 1.578,
+      "step": 31900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6262,
+      "step": 32000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005,
+      "loss": 1.6017,
+      "step": 32100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005,
+      "loss": 1.62,
+      "step": 32200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005,
+      "loss": 1.623,
+      "step": 32300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005,
+      "loss": 1.594,
+      "step": 32400
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005,
+      "loss": 1.589,
+      "step": 32500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5782,
+      "step": 32600
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5941,
+      "step": 32700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005,
+      "loss": 1.6033,
+      "step": 32800
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005,
+      "loss": 1.6129,
+      "step": 32900
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005,
+      "loss": 1.6148,
+      "step": 33000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6021,
+      "step": 33100
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6022,
+      "step": 33200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5881,
+      "step": 33300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5908,
+      "step": 33400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005,
+      "loss": 1.6188,
+      "step": 33500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005,
+      "loss": 1.583,
+      "step": 33600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005,
+      "loss": 1.587,
+      "step": 33700
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5957,
+      "step": 33800
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005,
+      "loss": 1.6009,
+      "step": 33900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005,
+      "loss": 1.62,
+      "step": 34000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005,
+      "loss": 1.6047,
+      "step": 34100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005,
+      "loss": 1.5877,
+      "step": 34200
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6295,
+      "step": 34300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005,
+      "loss": 1.6012,
+      "step": 34400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6089,
+      "step": 34500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5831,
+      "step": 34600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005,
+      "loss": 1.6055,
+      "step": 34700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6082,
+      "step": 34800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5876,
+      "step": 34900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5792,
+      "step": 35000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5907,
+      "step": 35100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5866,
+      "step": 35200
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005,
+      "loss": 1.5791,
+      "step": 35300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5899,
+      "step": 35400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6027,
+      "step": 35500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005,
+      "loss": 1.6122,
+      "step": 35600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5758,
+      "step": 35700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5936,
+      "step": 35800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5834,
+      "step": 35900
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5836,
+      "step": 36000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005,
+      "loss": 1.6074,
+      "step": 36100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5874,
+      "step": 36200
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5947,
+      "step": 36300
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005,
+      "loss": 1.6059,
+      "step": 36400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005,
+      "loss": 1.6163,
+      "step": 36500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005,
+      "loss": 1.621,
+      "step": 36600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005,
+      "loss": 1.6084,
+      "step": 36700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005,
+      "loss": 1.577,
+      "step": 36800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6052,
+      "step": 36900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005,
+      "loss": 1.596,
+      "step": 37000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5978,
+      "step": 37100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5814,
+      "step": 37200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5706,
+      "step": 37300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5826,
+      "step": 37400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5973,
+      "step": 37500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005,
+      "loss": 1.5973,
+      "step": 37600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5915,
+      "step": 37700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005,
+      "loss": 1.566,
+      "step": 37800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005,
+      "loss": 1.5978,
+      "step": 37900
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6077,
+      "step": 38000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5832,
+      "step": 38100
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005,
+      "loss": 1.598,
+      "step": 38200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5534,
+      "step": 38300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5079,
+      "step": 38400
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5098,
+      "step": 38500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0005,
+      "loss": 1.5127,
+      "step": 38600
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0005,
+      "loss": 1.506,
+      "step": 38700
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0005,
+      "loss": 1.483,
+      "step": 38800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5151,
+      "step": 38900
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4912,
+      "step": 39000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5238,
+      "step": 39100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0005,
+      "loss": 1.4906,
+      "step": 39200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0005,
+      "loss": 1.5101,
+      "step": 39300
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0005,
+      "loss": 1.4918,
+      "step": 39400
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0005,
+      "loss": 1.5124,
+      "step": 39500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5103,
+      "step": 39600
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5291,
+      "step": 39700
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0005,
+      "loss": 1.5165,
+      "step": 39800
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4857,
+      "step": 39900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0005,
+      "loss": 1.517,
+      "step": 40000
+    },
+    {
+      "epoch": 3.14,
+      "eval_gen_len": 18.765516023660282,
+      "eval_loss": 2.329390525817871,
+      "eval_rouge1": 33.5163,
+      "eval_rouge2": 12.3458,
+      "eval_rougeL": 27.2141,
+      "eval_rougeLsum": 27.2023,
+      "eval_runtime": 346.5778,
+      "eval_samples_per_second": 32.682,
+      "eval_steps_per_second": 2.043,
+      "step": 40000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0005,
+      "loss": 1.541,
+      "step": 40100
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4803,
+      "step": 40200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5181,
+      "step": 40300
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0005,
+      "loss": 1.514,
+      "step": 40400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0005,
+      "loss": 1.5116,
+      "step": 40500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4864,
+      "step": 40600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0005,
+      "loss": 1.5472,
+      "step": 40700
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5104,
+      "step": 40800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4961,
+      "step": 40900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005,
+      "loss": 1.4963,
+      "step": 41000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005,
+      "loss": 1.523,
+      "step": 41100
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5071,
+      "step": 41200
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005,
+      "loss": 1.5086,
+      "step": 41300
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005,
+      "loss": 1.5413,
+      "step": 41400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005,
+      "loss": 1.5059,
+      "step": 41500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4852,
+      "step": 41600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5075,
+      "step": 41700
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005,
+      "loss": 1.5452,
+      "step": 41800
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005,
+      "loss": 1.519,
+      "step": 41900
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5199,
+      "step": 42000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005,
+      "loss": 1.5053,
+      "step": 42100
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005,
+      "loss": 1.5043,
+      "step": 42200
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5195,
+      "step": 42300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4911,
+      "step": 42400
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005,
+      "loss": 1.5378,
+      "step": 42500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005,
+      "loss": 1.5144,
+      "step": 42600
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5232,
+      "step": 42700
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5226,
+      "step": 42800
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5166,
+      "step": 42900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005,
+      "loss": 1.5221,
+      "step": 43000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5047,
+      "step": 43100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5198,
+      "step": 43200
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5085,
+      "step": 43300
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5374,
+      "step": 43400
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5015,
+      "step": 43500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5371,
+      "step": 43600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5167,
+      "step": 43700
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5109,
+      "step": 43800
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0005,
+      "loss": 1.5245,
+      "step": 43900
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0005,
+      "loss": 1.516,
+      "step": 44000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0005,
+      "loss": 1.5132,
+      "step": 44100
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0005,
+      "loss": 1.524,
+      "step": 44200
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5232,
+      "step": 44300
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4996,
+      "step": 44400
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5283,
+      "step": 44500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0005,
+      "loss": 1.5137,
+      "step": 44600
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5107,
+      "step": 44700
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5086,
+      "step": 44800
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5199,
+      "step": 44900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0005,
+      "loss": 1.4955,
+      "step": 45000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5341,
+      "step": 45100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5085,
+      "step": 45200
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0005,
+      "loss": 1.5202,
+      "step": 45300
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0005,
+      "loss": 1.533,
+      "step": 45400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0005,
+      "loss": 1.5463,
+      "step": 45500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5416,
+      "step": 45600
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0005,
+      "loss": 1.541,
+      "step": 45700
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0005,
+      "loss": 1.5129,
+      "step": 45800
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5249,
+      "step": 45900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5039,
+      "step": 46000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5128,
+      "step": 46100
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0005,
+      "loss": 1.51,
+      "step": 46200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0005,
+      "loss": 1.5107,
+      "step": 46300
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0005,
+      "loss": 1.5166,
+      "step": 46400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5162,
+      "step": 46500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5309,
+      "step": 46600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0005,
+      "loss": 1.5266,
+      "step": 46700
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5208,
+      "step": 46800
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0005,
+      "loss": 1.5088,
+      "step": 46900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5148,
+      "step": 47000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5374,
+      "step": 47100
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5115,
+      "step": 47200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5082,
+      "step": 47300
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0005,
+      "loss": 1.516,
+      "step": 47400
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0005,
+      "loss": 1.5106,
+      "step": 47500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5326,
+      "step": 47600
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5427,
+      "step": 47700
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5213,
+      "step": 47800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5051,
+      "step": 47900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5078,
+      "step": 48000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0005,
+      "loss": 1.5257,
+      "step": 48100
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0005,
+      "loss": 1.4985,
+      "step": 48200
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5218,
+      "step": 48300
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5171,
+      "step": 48400
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5148,
+      "step": 48500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0005,
+      "loss": 1.4931,
+      "step": 48600
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5306,
+      "step": 48700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5006,
+      "step": 48800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5175,
+      "step": 48900
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5317,
+      "step": 49000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5371,
+      "step": 49100
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0005,
+      "loss": 1.5198,
+      "step": 49200
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5099,
+      "step": 49300
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5223,
+      "step": 49400
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5273,
+      "step": 49500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5459,
+      "step": 49600
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0005,
+      "loss": 1.5136,
+      "step": 49700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5381,
+      "step": 49800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5148,
+      "step": 49900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5429,
+      "step": 50000
+    },
+    {
+      "epoch": 3.92,
+      "eval_gen_len": 18.583826255848855,
+      "eval_loss": 2.3042964935302734,
+      "eval_rouge1": 33.7669,
+      "eval_rouge2": 12.7128,
+      "eval_rougeL": 27.641,
+      "eval_rougeLsum": 27.6383,
+      "eval_runtime": 356.4473,
+      "eval_samples_per_second": 31.777,
+      "eval_steps_per_second": 1.986,
+      "step": 50000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5023,
+      "step": 50100
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0005,
+      "loss": 1.537,
+      "step": 50200
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5346,
+      "step": 50300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0005,
+      "loss": 1.5038,
+      "step": 50400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5142,
+      "step": 50500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0005,
+      "loss": 1.5001,
+      "step": 50600
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0005,
+      "loss": 1.515,
+      "step": 50700
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5021,
+      "step": 50800
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5119,
+      "step": 50900
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5179,
+      "step": 51000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4309,
+      "step": 51100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0005,
+      "loss": 1.4105,
+      "step": 51200
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0005,
+      "loss": 1.4356,
+      "step": 51300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4311,
+      "step": 51400
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4365,
+      "step": 51500
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4153,
+      "step": 51600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4293,
+      "step": 51700
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4669,
+      "step": 51800
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0005,
+      "loss": 1.437,
+      "step": 51900
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0005,
+      "loss": 1.4437,
+      "step": 52000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0005,
+      "loss": 1.4431,
+      "step": 52100
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0005,
+      "loss": 1.4264,
+      "step": 52200
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4178,
+      "step": 52300
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0005,
+      "loss": 1.4409,
+      "step": 52400
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4422,
+      "step": 52500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4344,
+      "step": 52600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4418,
+      "step": 52700
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4359,
+      "step": 52800
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4548,
+      "step": 52900
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4474,
+      "step": 53000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4544,
+      "step": 53100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0005,
+      "loss": 1.4201,
+      "step": 53200
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4328,
+      "step": 53300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4372,
+      "step": 53400
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0005,
+      "loss": 1.42,
+      "step": 53500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0005,
+      "loss": 1.4338,
+      "step": 53600
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4679,
+      "step": 53700
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0005,
+      "loss": 1.4378,
+      "step": 53800
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4466,
+      "step": 53900
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4482,
+      "step": 54000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0005,
+      "loss": 1.4473,
+      "step": 54100
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4631,
+      "step": 54200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4578,
+      "step": 54300
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4458,
+      "step": 54400
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4371,
+      "step": 54500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0005,
+      "loss": 1.439,
+      "step": 54600
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0005,
+      "loss": 1.4404,
+      "step": 54700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4423,
+      "step": 54800
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4443,
+      "step": 54900
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4579,
+      "step": 55000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4229,
+      "step": 55100
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0005,
+      "loss": 1.4595,
+      "step": 55200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4533,
+      "step": 55300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4332,
+      "step": 55400
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0005,
+      "loss": 1.4315,
+      "step": 55500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4342,
+      "step": 55600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4532,
+      "step": 55700
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0005,
+      "loss": 1.4271,
+      "step": 55800
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0005,
+      "loss": 1.4403,
+      "step": 55900
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0005,
+      "loss": 1.4561,
+      "step": 56000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0005,
+      "loss": 1.45,
+      "step": 56100
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4653,
+      "step": 56200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4416,
+      "step": 56300
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0005,
+      "loss": 1.4449,
+      "step": 56400
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4261,
+      "step": 56500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0005,
+      "loss": 1.4527,
+      "step": 56600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4751,
+      "step": 56700
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4417,
+      "step": 56800
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0005,
+      "loss": 1.4775,
+      "step": 56900
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0005,
+      "loss": 1.4505,
+      "step": 57000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4606,
+      "step": 57100
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0005,
+      "loss": 1.4669,
+      "step": 57200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0005,
+      "loss": 1.4589,
+      "step": 57300
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0005,
+      "loss": 1.4662,
+      "step": 57400
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0005,
+      "loss": 1.4423,
+      "step": 57500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0005,
+      "loss": 1.4523,
+      "step": 57600
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0005,
+      "loss": 1.462,
+      "step": 57700
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0005,
+      "loss": 1.4748,
+      "step": 57800
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0005,
+      "loss": 1.4717,
+      "step": 57900
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0005,
+      "loss": 1.4641,
+      "step": 58000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0005,
+      "loss": 1.4514,
+      "step": 58100
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0005,
+      "loss": 1.4559,
+      "step": 58200
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0005,
+      "loss": 1.4707,
+      "step": 58300
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0005,
+      "loss": 1.4649,
+      "step": 58400
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4394,
+      "step": 58500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0005,
+      "loss": 1.4809,
+      "step": 58600
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0005,
+      "loss": 1.4622,
+      "step": 58700
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0005,
+      "loss": 1.479,
+      "step": 58800
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0005,
+      "loss": 1.4401,
+      "step": 58900
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4464,
+      "step": 59000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4539,
+      "step": 59100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0005,
+      "loss": 1.4582,
+      "step": 59200
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0005,
+      "loss": 1.4579,
+      "step": 59300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0005,
+      "loss": 1.4641,
+      "step": 59400
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0005,
+      "loss": 1.4532,
+      "step": 59500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0005,
+      "loss": 1.4389,
+      "step": 59600
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0005,
+      "loss": 1.4542,
+      "step": 59700
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0005,
+      "loss": 1.473,
+      "step": 59800
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0005,
+      "loss": 1.4598,
+      "step": 59900
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0005,
+      "loss": 1.475,
+      "step": 60000
+    },
+    {
+      "epoch": 4.71,
+      "eval_gen_len": 18.74971307495365,
+      "eval_loss": 2.3022000789642334,
+      "eval_rouge1": 34.0539,
+      "eval_rouge2": 12.9764,
+      "eval_rougeL": 27.8823,
+      "eval_rougeLsum": 27.8695,
+      "eval_runtime": 353.7063,
+      "eval_samples_per_second": 32.024,
+      "eval_steps_per_second": 2.002,
+      "step": 60000
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0005,
+      "loss": 1.4611,
+      "step": 60100
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0005,
+      "loss": 1.4512,
+      "step": 60200
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0005,
+      "loss": 1.4714,
+      "step": 60300
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0005,
+      "loss": 1.4443,
+      "step": 60400
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0005,
+      "loss": 1.4727,
+      "step": 60500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0005,
+      "loss": 1.4707,
+      "step": 60600
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0005,
+      "loss": 1.4508,
+      "step": 60700
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0005,
+      "loss": 1.4455,
+      "step": 60800
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0005,
+      "loss": 1.4334,
+      "step": 60900
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0005,
+      "loss": 1.4428,
+      "step": 61000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0005,
+      "loss": 1.4751,
+      "step": 61100
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0005,
+      "loss": 1.4547,
+      "step": 61200
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0005,
+      "loss": 1.4461,
+      "step": 61300
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0005,
+      "loss": 1.4598,
+      "step": 61400
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0005,
+      "loss": 1.4487,
+      "step": 61500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0005,
+      "loss": 1.4386,
+      "step": 61600
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0005,
+      "loss": 1.4465,
+      "step": 61700
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0005,
+      "loss": 1.4423,
+      "step": 61800
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0005,
+      "loss": 1.4615,
+      "step": 61900
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0005,
+      "loss": 1.4242,
+      "step": 62000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0005,
+      "loss": 1.4547,
+      "step": 62100
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0005,
+      "loss": 1.4248,
+      "step": 62200
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0005,
+      "loss": 1.4664,
+      "step": 62300
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0005,
+      "loss": 1.4831,
+      "step": 62400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0005,
+      "loss": 1.4716,
+      "step": 62500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0005,
+      "loss": 1.482,
+      "step": 62600
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0005,
+      "loss": 1.4642,
+      "step": 62700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0005,
+      "loss": 1.4536,
+      "step": 62800
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0005,
+      "loss": 1.4631,
+      "step": 62900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0005,
+      "loss": 1.4642,
+      "step": 63000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0005,
+      "loss": 1.4832,
+      "step": 63100
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0005,
+      "loss": 1.4393,
+      "step": 63200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0005,
+      "loss": 1.4678,
+      "step": 63300
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0005,
+      "loss": 1.4476,
+      "step": 63400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0005,
+      "loss": 1.4689,
+      "step": 63500
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0005,
+      "loss": 1.4466,
+      "step": 63600
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0005,
+      "loss": 1.4605,
+      "step": 63700
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0005,
+      "loss": 1.431,
+      "step": 63800
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0005,
+      "loss": 1.3646,
+      "step": 63900
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0005,
+      "loss": 1.376,
+      "step": 64000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0005,
+      "loss": 1.3577,
+      "step": 64100
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0005,
+      "loss": 1.3447,
+      "step": 64200
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0005,
+      "loss": 1.3696,
+      "step": 64300
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4079,
+      "step": 64400
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0005,
+      "loss": 1.367,
+      "step": 64500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0005,
+      "loss": 1.3771,
+      "step": 64600
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0005,
+      "loss": 1.4109,
+      "step": 64700
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0005,
+      "loss": 1.38,
+      "step": 64800
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0005,
+      "loss": 1.3861,
+      "step": 64900
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0005,
+      "loss": 1.3871,
+      "step": 65000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0005,
+      "loss": 1.3557,
+      "step": 65100
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0005,
+      "loss": 1.3918,
+      "step": 65200
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0005,
+      "loss": 1.3841,
+      "step": 65300
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4053,
+      "step": 65400
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0005,
+      "loss": 1.3789,
+      "step": 65500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0005,
+      "loss": 1.3924,
+      "step": 65600
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0005,
+      "loss": 1.3657,
+      "step": 65700
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0005,
+      "loss": 1.3801,
+      "step": 65800
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0005,
+      "loss": 1.3889,
+      "step": 65900
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0005,
+      "loss": 1.3745,
+      "step": 66000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0005,
+      "loss": 1.3834,
+      "step": 66100
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0005,
+      "loss": 1.3891,
+      "step": 66200
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0005,
+      "loss": 1.3836,
+      "step": 66300
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0005,
+      "loss": 1.3732,
+      "step": 66400
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0005,
+      "loss": 1.3928,
+      "step": 66500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0005,
+      "loss": 1.3993,
+      "step": 66600
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0005,
+      "loss": 1.3795,
+      "step": 66700
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0005,
+      "loss": 1.4197,
+      "step": 66800
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0005,
+      "loss": 1.3677,
+      "step": 66900
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0005,
+      "loss": 1.3964,
+      "step": 67000
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0005,
+      "loss": 1.3717,
+      "step": 67100
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0005,
+      "loss": 1.3891,
+      "step": 67200
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0005,
+      "loss": 1.4038,
+      "step": 67300
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0005,
+      "loss": 1.3955,
+      "step": 67400
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0005,
+      "loss": 1.3901,
+      "step": 67500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0005,
+      "loss": 1.389,
+      "step": 67600
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4091,
+      "step": 67700
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0005,
+      "loss": 1.3881,
+      "step": 67800
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4026,
+      "step": 67900
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0005,
+      "loss": 1.3844,
+      "step": 68000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0005,
+      "loss": 1.3687,
+      "step": 68100
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0005,
+      "loss": 1.379,
+      "step": 68200
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0005,
+      "loss": 1.3795,
+      "step": 68300
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4076,
+      "step": 68400
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0005,
+      "loss": 1.3999,
+      "step": 68500
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0005,
+      "loss": 1.3924,
+      "step": 68600
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0005,
+      "loss": 1.4324,
+      "step": 68700
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0005,
+      "loss": 1.406,
+      "step": 68800
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0005,
+      "loss": 1.404,
+      "step": 68900
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4119,
+      "step": 69000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0005,
+      "loss": 1.4178,
+      "step": 69100
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4095,
+      "step": 69200
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0005,
+      "loss": 1.3986,
+      "step": 69300
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0005,
+      "loss": 1.3808,
+      "step": 69400
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0005,
+      "loss": 1.3877,
+      "step": 69500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0005,
+      "loss": 1.4135,
+      "step": 69600
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0005,
+      "loss": 1.3957,
+      "step": 69700
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0005,
+      "loss": 1.4108,
+      "step": 69800
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4068,
+      "step": 69900
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0005,
+      "loss": 1.3804,
+      "step": 70000
+    },
+    {
+      "epoch": 5.49,
+      "eval_gen_len": 18.74026661958153,
+      "eval_loss": 2.3253302574157715,
+      "eval_rouge1": 34.1086,
+      "eval_rouge2": 13.1195,
+      "eval_rougeL": 28.0176,
+      "eval_rougeLsum": 28.0186,
+      "eval_runtime": 348.2658,
+      "eval_samples_per_second": 32.524,
+      "eval_steps_per_second": 2.033,
+      "step": 70000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0005,
+      "loss": 1.3911,
+      "step": 70100
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0005,
+      "loss": 1.4149,
+      "step": 70200
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0005,
+      "loss": 1.3961,
+      "step": 70300
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0005,
+      "loss": 1.3899,
+      "step": 70400
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0005,
+      "loss": 1.4053,
+      "step": 70500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0005,
+      "loss": 1.4002,
+      "step": 70600
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0005,
+      "loss": 1.4142,
+      "step": 70700
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0005,
+      "loss": 1.4121,
+      "step": 70800
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0005,
+      "loss": 1.4094,
+      "step": 70900
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0005,
+      "loss": 1.3936,
+      "step": 71000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0005,
+      "loss": 1.4091,
+      "step": 71100
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0005,
+      "loss": 1.4007,
+      "step": 71200
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4159,
+      "step": 71300
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0005,
+      "loss": 1.3911,
+      "step": 71400
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0005,
+      "loss": 1.4275,
+      "step": 71500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0005,
+      "loss": 1.4008,
+      "step": 71600
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0005,
+      "loss": 1.4069,
+      "step": 71700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0005,
+      "loss": 1.3904,
+      "step": 71800
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0005,
+      "loss": 1.4138,
+      "step": 71900
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0005,
+      "loss": 1.3912,
+      "step": 72000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0005,
+      "loss": 1.3971,
+      "step": 72100
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0005,
+      "loss": 1.4153,
+      "step": 72200
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0005,
+      "loss": 1.3954,
+      "step": 72300
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0005,
+      "loss": 1.3777,
+      "step": 72400
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0005,
+      "loss": 1.4101,
+      "step": 72500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0005,
+      "loss": 1.4203,
+      "step": 72600
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0005,
+      "loss": 1.4302,
+      "step": 72700
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0005,
+      "loss": 1.3765,
+      "step": 72800
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0005,
+      "loss": 1.3977,
+      "step": 72900
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0005,
+      "loss": 1.3909,
+      "step": 73000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0005,
+      "loss": 1.3792,
+      "step": 73100
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0005,
+      "loss": 1.4065,
+      "step": 73200
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0005,
+      "loss": 1.3779,
+      "step": 73300
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0005,
+      "loss": 1.4049,
+      "step": 73400
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0005,
+      "loss": 1.4121,
+      "step": 73500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0005,
+      "loss": 1.4046,
+      "step": 73600
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0005,
+      "loss": 1.4133,
+      "step": 73700
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0005,
+      "loss": 1.4056,
+      "step": 73800
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0005,
+      "loss": 1.4131,
+      "step": 73900
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0005,
+      "loss": 1.4183,
+      "step": 74000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0005,
+      "loss": 1.39,
+      "step": 74100
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0005,
+      "loss": 1.3884,
+      "step": 74200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0005,
+      "loss": 1.3922,
+      "step": 74300
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0005,
+      "loss": 1.3974,
+      "step": 74400
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0005,
+      "loss": 1.3898,
+      "step": 74500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0005,
+      "loss": 1.422,
+      "step": 74600
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0005,
+      "loss": 1.4073,
+      "step": 74700
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0005,
+      "loss": 1.4182,
+      "step": 74800
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0005,
+      "loss": 1.404,
+      "step": 74900
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0005,
+      "loss": 1.404,
+      "step": 75000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0005,
+      "loss": 1.417,
+      "step": 75100
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0005,
+      "loss": 1.3935,
+      "step": 75200
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0005,
+      "loss": 1.4128,
+      "step": 75300
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0005,
+      "loss": 1.4266,
+      "step": 75400
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0005,
+      "loss": 1.4013,
+      "step": 75500
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0005,
+      "loss": 1.4308,
+      "step": 75600
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0005,
+      "loss": 1.3954,
+      "step": 75700
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0005,
+      "loss": 1.4102,
+      "step": 75800
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0005,
+      "loss": 1.4205,
+      "step": 75900
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0005,
+      "loss": 1.4083,
+      "step": 76000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0005,
+      "loss": 1.4017,
+      "step": 76100
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0005,
+      "loss": 1.4201,
+      "step": 76200
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0005,
+      "loss": 1.4072,
+      "step": 76300
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0005,
+      "loss": 1.4121,
+      "step": 76400
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0005,
+      "loss": 1.3649,
+      "step": 76500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0005,
+      "loss": 1.3496,
+      "step": 76600
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0005,
+      "loss": 1.3088,
+      "step": 76700
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0005,
+      "loss": 1.3273,
+      "step": 76800
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0005,
+      "loss": 1.3207,
+      "step": 76900
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0005,
+      "loss": 1.3231,
+      "step": 77000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0005,
+      "loss": 1.3061,
+      "step": 77100
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0005,
+      "loss": 1.3236,
+      "step": 77200
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0005,
+      "loss": 1.3353,
+      "step": 77300
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0005,
+      "loss": 1.3088,
+      "step": 77400
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0005,
+      "loss": 1.3163,
+      "step": 77500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0005,
+      "loss": 1.3123,
+      "step": 77600
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0005,
+      "loss": 1.3145,
+      "step": 77700
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0005,
+      "loss": 1.3152,
+      "step": 77800
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0005,
+      "loss": 1.3271,
+      "step": 77900
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0005,
+      "loss": 1.3517,
+      "step": 78000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0005,
+      "loss": 1.3451,
+      "step": 78100
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0005,
+      "loss": 1.3471,
+      "step": 78200
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0005,
+      "loss": 1.3285,
+      "step": 78300
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0005,
+      "loss": 1.3541,
+      "step": 78400
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0005,
+      "loss": 1.3361,
+      "step": 78500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0005,
+      "loss": 1.3424,
+      "step": 78600
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0005,
+      "loss": 1.3402,
+      "step": 78700
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0005,
+      "loss": 1.3305,
+      "step": 78800
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0005,
+      "loss": 1.3223,
+      "step": 78900
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0005,
+      "loss": 1.3367,
+      "step": 79000
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0005,
+      "loss": 1.3242,
+      "step": 79100
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0005,
+      "loss": 1.3428,
+      "step": 79200
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0005,
+      "loss": 1.3484,
+      "step": 79300
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0005,
+      "loss": 1.3515,
+      "step": 79400
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0005,
+      "loss": 1.3462,
+      "step": 79500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0005,
+      "loss": 1.3391,
+      "step": 79600
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0005,
+      "loss": 1.3436,
+      "step": 79700
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0005,
+      "loss": 1.3496,
+      "step": 79800
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0005,
+      "loss": 1.3447,
+      "step": 79900
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0005,
+      "loss": 1.3329,
+      "step": 80000
+    },
+    {
+      "epoch": 6.27,
+      "eval_gen_len": 18.70927871457579,
+      "eval_loss": 2.332681655883789,
+      "eval_rouge1": 34.2256,
+      "eval_rouge2": 13.2076,
+      "eval_rougeL": 28.0369,
+      "eval_rougeLsum": 28.0357,
+      "eval_runtime": 344.7634,
+      "eval_samples_per_second": 32.854,
+      "eval_steps_per_second": 2.054,
+      "step": 80000
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0005,
+      "loss": 1.351,
+      "step": 80100
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0005,
+      "loss": 1.3381,
+      "step": 80200
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0005,
+      "loss": 1.327,
+      "step": 80300
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0005,
+      "loss": 1.3345,
+      "step": 80400
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0005,
+      "loss": 1.3395,
+      "step": 80500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0005,
+      "loss": 1.3643,
+      "step": 80600
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0005,
+      "loss": 1.3277,
+      "step": 80700
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0005,
+      "loss": 1.3444,
+      "step": 80800
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0005,
+      "loss": 1.3393,
+      "step": 80900
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0005,
+      "loss": 1.3726,
+      "step": 81000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0005,
+      "loss": 1.3172,
+      "step": 81100
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0005,
+      "loss": 1.3653,
+      "step": 81200
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0005,
+      "loss": 1.3827,
+      "step": 81300
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0005,
+      "loss": 1.3105,
+      "step": 81400
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0005,
+      "loss": 1.3732,
+      "step": 81500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0005,
+      "loss": 1.3452,
+      "step": 81600
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0005,
+      "loss": 1.3516,
+      "step": 81700
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0005,
+      "loss": 1.3753,
+      "step": 81800
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0005,
+      "loss": 1.3545,
+      "step": 81900
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0005,
+      "loss": 1.3357,
+      "step": 82000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0005,
+      "loss": 1.3628,
+      "step": 82100
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0005,
+      "loss": 1.351,
+      "step": 82200
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0005,
+      "loss": 1.3438,
+      "step": 82300
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0005,
+      "loss": 1.3635,
+      "step": 82400
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0005,
+      "loss": 1.3506,
+      "step": 82500
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0005,
+      "loss": 1.3611,
+      "step": 82600
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0005,
+      "loss": 1.3467,
+      "step": 82700
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0005,
+      "loss": 1.3528,
+      "step": 82800
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0005,
+      "loss": 1.3513,
+      "step": 82900
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0005,
+      "loss": 1.3372,
+      "step": 83000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0005,
+      "loss": 1.3724,
+      "step": 83100
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0005,
+      "loss": 1.3554,
+      "step": 83200
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0005,
+      "loss": 1.3711,
+      "step": 83300
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0005,
+      "loss": 1.3534,
+      "step": 83400
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0005,
+      "loss": 1.3736,
+      "step": 83500
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0005,
+      "loss": 1.3265,
+      "step": 83600
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0005,
+      "loss": 1.3467,
+      "step": 83700
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0005,
+      "loss": 1.3601,
+      "step": 83800
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0005,
+      "loss": 1.3549,
+      "step": 83900
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0005,
+      "loss": 1.3584,
+      "step": 84000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0005,
+      "loss": 1.3577,
+      "step": 84100
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0005,
+      "loss": 1.3551,
+      "step": 84200
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0005,
+      "loss": 1.3676,
+      "step": 84300
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0005,
+      "loss": 1.3669,
+      "step": 84400
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0005,
+      "loss": 1.3832,
+      "step": 84500
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0005,
+      "loss": 1.3649,
+      "step": 84600
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0005,
+      "loss": 1.3812,
+      "step": 84700
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0005,
+      "loss": 1.3602,
+      "step": 84800
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0005,
+      "loss": 1.3428,
+      "step": 84900
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0005,
+      "loss": 1.3502,
+      "step": 85000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0005,
+      "loss": 1.354,
+      "step": 85100
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0005,
+      "loss": 1.3682,
+      "step": 85200
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0005,
+      "loss": 1.3206,
+      "step": 85300
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0005,
+      "loss": 1.3594,
+      "step": 85400
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0005,
+      "loss": 1.3724,
+      "step": 85500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0005,
+      "loss": 1.3418,
+      "step": 85600
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0005,
+      "loss": 1.3721,
+      "step": 85700
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0005,
+      "loss": 1.3535,
+      "step": 85800
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0005,
+      "loss": 1.3513,
+      "step": 85900
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0005,
+      "loss": 1.3545,
+      "step": 86000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0005,
+      "loss": 1.3537,
+      "step": 86100
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0005,
+      "loss": 1.3609,
+      "step": 86200
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0005,
+      "loss": 1.3591,
+      "step": 86300
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0005,
+      "loss": 1.3687,
+      "step": 86400
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0005,
+      "loss": 1.3531,
+      "step": 86500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0005,
+      "loss": 1.3581,
+      "step": 86600
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0005,
+      "loss": 1.3692,
+      "step": 86700
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0005,
+      "loss": 1.3517,
+      "step": 86800
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0005,
+      "loss": 1.3695,
+      "step": 86900
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0005,
+      "loss": 1.3506,
+      "step": 87000
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0005,
+      "loss": 1.3671,
+      "step": 87100
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0005,
+      "loss": 1.3768,
+      "step": 87200
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0005,
+      "loss": 1.3591,
+      "step": 87300
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0005,
+      "loss": 1.3556,
+      "step": 87400
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0005,
+      "loss": 1.3318,
+      "step": 87500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0005,
+      "loss": 1.3809,
+      "step": 87600
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0005,
+      "loss": 1.3655,
+      "step": 87700
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0005,
+      "loss": 1.3515,
+      "step": 87800
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0005,
+      "loss": 1.3745,
+      "step": 87900
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0005,
+      "loss": 1.3488,
+      "step": 88000
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0005,
+      "loss": 1.3714,
+      "step": 88100
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0005,
+      "loss": 1.3579,
+      "step": 88200
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0005,
+      "loss": 1.3671,
+      "step": 88300
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0005,
+      "loss": 1.3617,
+      "step": 88400
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0005,
+      "loss": 1.3514,
+      "step": 88500
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0005,
+      "loss": 1.3786,
+      "step": 88600
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0005,
+      "loss": 1.3782,
+      "step": 88700
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0005,
+      "loss": 1.3563,
+      "step": 88800
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0005,
+      "loss": 1.3643,
+      "step": 88900
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0005,
+      "loss": 1.3903,
+      "step": 89000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0005,
+      "loss": 1.3791,
+      "step": 89100
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0005,
+      "loss": 1.3554,
+      "step": 89200
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0005,
+      "loss": 1.3172,
+      "step": 89300
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0005,
+      "loss": 1.2948,
+      "step": 89400
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0005,
+      "loss": 1.2977,
+      "step": 89500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0005,
+      "loss": 1.284,
+      "step": 89600
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0005,
+      "loss": 1.2795,
+      "step": 89700
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0005,
+      "loss": 1.268,
+      "step": 89800
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.0005,
+      "loss": 1.2986,
+      "step": 89900
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0005,
+      "loss": 1.2728,
+      "step": 90000
+    },
+    {
+      "epoch": 7.06,
+      "eval_gen_len": 18.74291515847091,
+      "eval_loss": 2.339409112930298,
+      "eval_rouge1": 34.4263,
+      "eval_rouge2": 13.442,
+      "eval_rougeL": 28.2935,
+      "eval_rougeLsum": 28.2854,
+      "eval_runtime": 343.606,
+      "eval_samples_per_second": 32.965,
+      "eval_steps_per_second": 2.06,
+      "step": 90000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0005,
+      "loss": 1.2807,
+      "step": 90100
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0005,
+      "loss": 1.3081,
+      "step": 90200
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.0005,
+      "loss": 1.2923,
+      "step": 90300
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.0005,
+      "loss": 1.273,
+      "step": 90400
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.0005,
+      "loss": 1.2567,
+      "step": 90500
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.0005,
+      "loss": 1.2771,
+      "step": 90600
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.0005,
+      "loss": 1.2768,
+      "step": 90700
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0005,
+      "loss": 1.2704,
+      "step": 90800
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0005,
+      "loss": 1.2755,
+      "step": 90900
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0005,
+      "loss": 1.2813,
+      "step": 91000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0005,
+      "loss": 1.2972,
+      "step": 91100
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0005,
+      "loss": 1.2805,
+      "step": 91200
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0005,
+      "loss": 1.2757,
+      "step": 91300
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0005,
+      "loss": 1.2915,
+      "step": 91400
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0005,
+      "loss": 1.3053,
+      "step": 91500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0005,
+      "loss": 1.3232,
+      "step": 91600
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.0005,
+      "loss": 1.3037,
+      "step": 91700
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0005,
+      "loss": 1.2884,
+      "step": 91800
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0005,
+      "loss": 1.2971,
+      "step": 91900
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0005,
+      "loss": 1.3044,
+      "step": 92000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.0005,
+      "loss": 1.3029,
+      "step": 92100
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0005,
+      "loss": 1.2959,
+      "step": 92200
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.0005,
+      "loss": 1.3149,
+      "step": 92300
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0005,
+      "loss": 1.2849,
+      "step": 92400
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0005,
+      "loss": 1.3093,
+      "step": 92500
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0005,
+      "loss": 1.3339,
+      "step": 92600
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0005,
+      "loss": 1.3035,
+      "step": 92700
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0005,
+      "loss": 1.3038,
+      "step": 92800
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0005,
+      "loss": 1.3045,
+      "step": 92900
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0005,
+      "loss": 1.3056,
+      "step": 93000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0005,
+      "loss": 1.2988,
+      "step": 93100
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0005,
+      "loss": 1.3026,
+      "step": 93200
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0005,
+      "loss": 1.2925,
+      "step": 93300
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0005,
+      "loss": 1.2901,
+      "step": 93400
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.0005,
+      "loss": 1.2944,
+      "step": 93500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0005,
+      "loss": 1.3112,
+      "step": 93600
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0005,
+      "loss": 1.313,
+      "step": 93700
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0005,
+      "loss": 1.2896,
+      "step": 93800
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0005,
+      "loss": 1.2935,
+      "step": 93900
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0005,
+      "loss": 1.3067,
+      "step": 94000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0005,
+      "loss": 1.311,
+      "step": 94100
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0005,
+      "loss": 1.2834,
+      "step": 94200
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0005,
+      "loss": 1.2961,
+      "step": 94300
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0005,
+      "loss": 1.3183,
+      "step": 94400
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0005,
+      "loss": 1.3031,
+      "step": 94500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.0005,
+      "loss": 1.3242,
+      "step": 94600
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0005,
+      "loss": 1.3186,
+      "step": 94700
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0005,
+      "loss": 1.3099,
+      "step": 94800
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.0005,
+      "loss": 1.2844,
+      "step": 94900
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.0005,
+      "loss": 1.3044,
+      "step": 95000
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.0005,
+      "loss": 1.3048,
+      "step": 95100
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.0005,
+      "loss": 1.3194,
+      "step": 95200
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.0005,
+      "loss": 1.3178,
+      "step": 95300
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0005,
+      "loss": 1.3232,
+      "step": 95400
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0005,
+      "loss": 1.3172,
+      "step": 95500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0005,
+      "loss": 1.3211,
+      "step": 95600
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0005,
+      "loss": 1.3178,
+      "step": 95700
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.0005,
+      "loss": 1.3227,
+      "step": 95800
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0005,
+      "loss": 1.3077,
+      "step": 95900
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0005,
+      "loss": 1.3239,
+      "step": 96000
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0005,
+      "loss": 1.2918,
+      "step": 96100
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0005,
+      "loss": 1.3214,
+      "step": 96200
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.0005,
+      "loss": 1.2858,
+      "step": 96300
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0005,
+      "loss": 1.323,
+      "step": 96400
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0005,
+      "loss": 1.3028,
+      "step": 96500
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0005,
+      "loss": 1.3091,
+      "step": 96600
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0005,
+      "loss": 1.3236,
+      "step": 96700
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.0005,
+      "loss": 1.2892,
+      "step": 96800
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.0005,
+      "loss": 1.2942,
+      "step": 96900
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.0005,
+      "loss": 1.3217,
+      "step": 97000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.0005,
+      "loss": 1.3195,
+      "step": 97100
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0005,
+      "loss": 1.3238,
+      "step": 97200
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.0005,
+      "loss": 1.3113,
+      "step": 97300
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.0005,
+      "loss": 1.3075,
+      "step": 97400
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.0005,
+      "loss": 1.3338,
+      "step": 97500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.0005,
+      "loss": 1.3121,
+      "step": 97600
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.0005,
+      "loss": 1.3415,
+      "step": 97700
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0005,
+      "loss": 1.3144,
+      "step": 97800
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.0005,
+      "loss": 1.327,
+      "step": 97900
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0005,
+      "loss": 1.3231,
+      "step": 98000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0005,
+      "loss": 1.327,
+      "step": 98100
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.0005,
+      "loss": 1.2992,
+      "step": 98200
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.0005,
+      "loss": 1.3232,
+      "step": 98300
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0005,
+      "loss": 1.3222,
+      "step": 98400
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0005,
+      "loss": 1.3297,
+      "step": 98500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.0005,
+      "loss": 1.3063,
+      "step": 98600
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.0005,
+      "loss": 1.3341,
+      "step": 98700
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.0005,
+      "loss": 1.3312,
+      "step": 98800
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0005,
+      "loss": 1.3341,
+      "step": 98900
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0005,
+      "loss": 1.317,
+      "step": 99000
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.0005,
+      "loss": 1.3329,
+      "step": 99100
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.0005,
+      "loss": 1.3381,
+      "step": 99200
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.0005,
+      "loss": 1.3284,
+      "step": 99300
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.0005,
+      "loss": 1.3254,
+      "step": 99400
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.0005,
+      "loss": 1.3338,
+      "step": 99500
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.0005,
+      "loss": 1.3293,
+      "step": 99600
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.0005,
+      "loss": 1.3338,
+      "step": 99700
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.0005,
+      "loss": 1.3222,
+      "step": 99800
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.0005,
+      "loss": 1.3132,
+      "step": 99900
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.0005,
+      "loss": 1.3226,
+      "step": 100000
+    },
+    {
+      "epoch": 7.84,
+      "eval_gen_len": 18.71413436920632,
+      "eval_loss": 2.3180172443389893,
+      "eval_rouge1": 34.7026,
+      "eval_rouge2": 13.6524,
+      "eval_rougeL": 28.5004,
+      "eval_rougeLsum": 28.4899,
+      "eval_runtime": 346.1988,
+      "eval_samples_per_second": 32.718,
+      "eval_steps_per_second": 2.045,
+      "step": 100000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.0005,
+      "loss": 1.3376,
+      "step": 100100
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0005,
+      "loss": 1.3383,
+      "step": 100200
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0005,
+      "loss": 1.3239,
+      "step": 100300
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0005,
+      "loss": 1.3319,
+      "step": 100400
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.0005,
+      "loss": 1.3161,
+      "step": 100500
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.0005,
+      "loss": 1.3419,
+      "step": 100600
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0005,
+      "loss": 1.3262,
+      "step": 100700
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0005,
+      "loss": 1.3392,
+      "step": 100800
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0005,
+      "loss": 1.307,
+      "step": 100900
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.0005,
+      "loss": 1.3017,
+      "step": 101000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.0005,
+      "loss": 1.3419,
+      "step": 101100
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0005,
+      "loss": 1.3279,
+      "step": 101200
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0005,
+      "loss": 1.344,
+      "step": 101300
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.0005,
+      "loss": 1.3274,
+      "step": 101400
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.0005,
+      "loss": 1.331,
+      "step": 101500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.0005,
+      "loss": 1.3233,
+      "step": 101600
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.0005,
+      "loss": 1.3093,
+      "step": 101700
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.0005,
+      "loss": 1.3195,
+      "step": 101800
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0005,
+      "loss": 1.3373,
+      "step": 101900
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.0005,
+      "loss": 1.3105,
+      "step": 102000
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.0005,
+      "loss": 1.2658,
+      "step": 102100
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.0005,
+      "loss": 1.2354,
+      "step": 102200
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.0005,
+      "loss": 1.2282,
+      "step": 102300
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.0005,
+      "loss": 1.2475,
+      "step": 102400
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.0005,
+      "loss": 1.2326,
+      "step": 102500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.0005,
+      "loss": 1.2364,
+      "step": 102600
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.0005,
+      "loss": 1.2297,
+      "step": 102700
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.0005,
+      "loss": 1.2495,
+      "step": 102800
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.0005,
+      "loss": 1.2551,
+      "step": 102900
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.0005,
+      "loss": 1.2288,
+      "step": 103000
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.0005,
+      "loss": 1.2316,
+      "step": 103100
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.0005,
+      "loss": 1.2272,
+      "step": 103200
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.0005,
+      "loss": 1.2344,
+      "step": 103300
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.0005,
+      "loss": 1.2526,
+      "step": 103400
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.0005,
+      "loss": 1.2614,
+      "step": 103500
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.0005,
+      "loss": 1.2656,
+      "step": 103600
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.0005,
+      "loss": 1.2525,
+      "step": 103700
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.0005,
+      "loss": 1.2627,
+      "step": 103800
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.0005,
+      "loss": 1.2767,
+      "step": 103900
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.0005,
+      "loss": 1.244,
+      "step": 104000
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.0005,
+      "loss": 1.2445,
+      "step": 104100
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.0005,
+      "loss": 1.2767,
+      "step": 104200
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.0005,
+      "loss": 1.2529,
+      "step": 104300
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0005,
+      "loss": 1.2635,
+      "step": 104400
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0005,
+      "loss": 1.2651,
+      "step": 104500
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 0.0005,
+      "loss": 1.2541,
+      "step": 104600
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.0005,
+      "loss": 1.2532,
+      "step": 104700
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.0005,
+      "loss": 1.2615,
+      "step": 104800
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.0005,
+      "loss": 1.2415,
+      "step": 104900
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.0005,
+      "loss": 1.261,
+      "step": 105000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.0005,
+      "loss": 1.2472,
+      "step": 105100
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 0.0005,
+      "loss": 1.2539,
+      "step": 105200
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.0005,
+      "loss": 1.2555,
+      "step": 105300
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.0005,
+      "loss": 1.2804,
+      "step": 105400
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.0005,
+      "loss": 1.2768,
+      "step": 105500
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.0005,
+      "loss": 1.266,
+      "step": 105600
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.0005,
+      "loss": 1.266,
+      "step": 105700
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.0005,
+      "loss": 1.2677,
+      "step": 105800
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.0005,
+      "loss": 1.2573,
+      "step": 105900
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.0005,
+      "loss": 1.2606,
+      "step": 106000
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.0005,
+      "loss": 1.2601,
+      "step": 106100
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.0005,
+      "loss": 1.2811,
+      "step": 106200
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.0005,
+      "loss": 1.2713,
+      "step": 106300
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.0005,
+      "loss": 1.27,
+      "step": 106400
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 0.0005,
+      "loss": 1.2663,
+      "step": 106500
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.0005,
+      "loss": 1.2709,
+      "step": 106600
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 0.0005,
+      "loss": 1.2922,
+      "step": 106700
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0005,
+      "loss": 1.2709,
+      "step": 106800
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0005,
+      "loss": 1.2704,
+      "step": 106900
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.0005,
+      "loss": 1.261,
+      "step": 107000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.0005,
+      "loss": 1.2758,
+      "step": 107100
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.0005,
+      "loss": 1.2745,
+      "step": 107200
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.0005,
+      "loss": 1.2802,
+      "step": 107300
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.0005,
+      "loss": 1.2722,
+      "step": 107400
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.0005,
+      "loss": 1.262,
+      "step": 107500
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 0.0005,
+      "loss": 1.2662,
+      "step": 107600
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.0005,
+      "loss": 1.2709,
+      "step": 107700
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.0005,
+      "loss": 1.2977,
+      "step": 107800
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.0005,
+      "loss": 1.296,
+      "step": 107900
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.0005,
+      "loss": 1.2796,
+      "step": 108000
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.0005,
+      "loss": 1.2628,
+      "step": 108100
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.0005,
+      "loss": 1.2708,
+      "step": 108200
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 0.0005,
+      "loss": 1.2661,
+      "step": 108300
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.0005,
+      "loss": 1.2518,
+      "step": 108400
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 0.0005,
+      "loss": 1.2881,
+      "step": 108500
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.0005,
+      "loss": 1.2887,
+      "step": 108600
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.0005,
+      "loss": 1.2821,
+      "step": 108700
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 0.0005,
+      "loss": 1.282,
+      "step": 108800
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.0005,
+      "loss": 1.2672,
+      "step": 108900
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.0005,
+      "loss": 1.2563,
+      "step": 109000
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.0005,
+      "loss": 1.2637,
+      "step": 109100
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.0005,
+      "loss": 1.285,
+      "step": 109200
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.0005,
+      "loss": 1.2787,
+      "step": 109300
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 0.0005,
+      "loss": 1.2965,
+      "step": 109400
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.0005,
+      "loss": 1.2879,
+      "step": 109500
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.0005,
+      "loss": 1.2715,
+      "step": 109600
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 0.0005,
+      "loss": 1.2717,
+      "step": 109700
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 0.0005,
+      "loss": 1.3013,
+      "step": 109800
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 0.0005,
+      "loss": 1.2841,
+      "step": 109900
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.0005,
+      "loss": 1.2956,
+      "step": 110000
+    },
+    {
+      "epoch": 8.63,
+      "eval_gen_len": 18.680056502162973,
+      "eval_loss": 2.327761650085449,
+      "eval_rouge1": 34.8338,
+      "eval_rouge2": 13.7914,
+      "eval_rougeL": 28.7114,
+      "eval_rougeLsum": 28.7056,
+      "eval_runtime": 344.147,
+      "eval_samples_per_second": 32.913,
+      "eval_steps_per_second": 2.057,
+      "step": 110000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.0005,
+      "loss": 1.2873,
+      "step": 110100
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 0.0005,
+      "loss": 1.2802,
+      "step": 110200
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 0.0005,
+      "loss": 1.2734,
+      "step": 110300
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 0.0005,
+      "loss": 1.2922,
+      "step": 110400
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.0005,
+      "loss": 1.2602,
+      "step": 110500
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.0005,
+      "loss": 1.2733,
+      "step": 110600
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.0005,
+      "loss": 1.29,
+      "step": 110700
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 0.0005,
+      "loss": 1.2988,
+      "step": 110800
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.0005,
+      "loss": 1.2828,
+      "step": 110900
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.0005,
+      "loss": 1.2803,
+      "step": 111000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 0.0005,
+      "loss": 1.2651,
+      "step": 111100
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 0.0005,
+      "loss": 1.2762,
+      "step": 111200
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 0.0005,
+      "loss": 1.3007,
+      "step": 111300
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.0005,
+      "loss": 1.3065,
+      "step": 111400
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.0005,
+      "loss": 1.2941,
+      "step": 111500
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 0.0005,
+      "loss": 1.2835,
+      "step": 111600
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 0.0005,
+      "loss": 1.2879,
+      "step": 111700
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.0005,
+      "loss": 1.2951,
+      "step": 111800
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.0005,
+      "loss": 1.289,
+      "step": 111900
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.0005,
+      "loss": 1.3021,
+      "step": 112000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 0.0005,
+      "loss": 1.278,
+      "step": 112100
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.0005,
+      "loss": 1.3145,
+      "step": 112200
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.0005,
+      "loss": 1.3147,
+      "step": 112300
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.0005,
+      "loss": 1.2801,
+      "step": 112400
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 0.0005,
+      "loss": 1.2945,
+      "step": 112500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.0005,
+      "loss": 1.2865,
+      "step": 112600
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 0.0005,
+      "loss": 1.2919,
+      "step": 112700
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.0005,
+      "loss": 1.3269,
+      "step": 112800
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.0005,
+      "loss": 1.3069,
+      "step": 112900
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 0.0005,
+      "loss": 1.3089,
+      "step": 113000
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 0.0005,
+      "loss": 1.2979,
+      "step": 113100
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.0005,
+      "loss": 1.2858,
+      "step": 113200
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.0005,
+      "loss": 1.2684,
+      "step": 113300
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 0.0005,
+      "loss": 1.2868,
+      "step": 113400
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 0.0005,
+      "loss": 1.2744,
+      "step": 113500
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 0.0005,
+      "loss": 1.2957,
+      "step": 113600
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.0005,
+      "loss": 1.3137,
+      "step": 113700
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.0005,
+      "loss": 1.3029,
+      "step": 113800
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 0.0005,
+      "loss": 1.2701,
+      "step": 113900
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 0.0005,
+      "loss": 1.2671,
+      "step": 114000
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.0005,
+      "loss": 1.2999,
+      "step": 114100
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 0.0005,
+      "loss": 1.2827,
+      "step": 114200
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 0.0005,
+      "loss": 1.2958,
+      "step": 114300
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 0.0005,
+      "loss": 1.2641,
+      "step": 114400
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.0005,
+      "loss": 1.2787,
+      "step": 114500
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.0005,
+      "loss": 1.2903,
+      "step": 114600
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.0005,
+      "loss": 1.2938,
+      "step": 114700
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 0.0005,
+      "loss": 1.2768,
+      "step": 114800
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 0.0005,
+      "loss": 1.1876,
+      "step": 114900
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 0.0005,
+      "loss": 1.2012,
+      "step": 115000
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 0.0005,
+      "loss": 1.2075,
+      "step": 115100
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 0.0005,
+      "loss": 1.2316,
+      "step": 115200
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.0005,
+      "loss": 1.1978,
+      "step": 115300
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 0.0005,
+      "loss": 1.2175,
+      "step": 115400
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 0.0005,
+      "loss": 1.2097,
+      "step": 115500
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.0005,
+      "loss": 1.1848,
+      "step": 115600
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.0005,
+      "loss": 1.2143,
+      "step": 115700
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 0.0005,
+      "loss": 1.2262,
+      "step": 115800
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 0.0005,
+      "loss": 1.2187,
+      "step": 115900
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 0.0005,
+      "loss": 1.2057,
+      "step": 116000
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 0.0005,
+      "loss": 1.1871,
+      "step": 116100
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 0.0005,
+      "loss": 1.2168,
+      "step": 116200
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 0.0005,
+      "loss": 1.2171,
+      "step": 116300
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.0005,
+      "loss": 1.2214,
+      "step": 116400
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 0.0005,
+      "loss": 1.2076,
+      "step": 116500
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 0.0005,
+      "loss": 1.2064,
+      "step": 116600
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 0.0005,
+      "loss": 1.2093,
+      "step": 116700
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.0005,
+      "loss": 1.1972,
+      "step": 116800
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 0.0005,
+      "loss": 1.2101,
+      "step": 116900
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 0.0005,
+      "loss": 1.2405,
+      "step": 117000
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 0.0005,
+      "loss": 1.2157,
+      "step": 117100
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.0005,
+      "loss": 1.2268,
+      "step": 117200
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 0.0005,
+      "loss": 1.2173,
+      "step": 117300
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 0.0005,
+      "loss": 1.2324,
+      "step": 117400
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 0.0005,
+      "loss": 1.2277,
+      "step": 117500
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.0005,
+      "loss": 1.2212,
+      "step": 117600
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.0005,
+      "loss": 1.2272,
+      "step": 117700
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 0.0005,
+      "loss": 1.2437,
+      "step": 117800
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.0005,
+      "loss": 1.2423,
+      "step": 117900
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.0005,
+      "loss": 1.2392,
+      "step": 118000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 0.0005,
+      "loss": 1.2487,
+      "step": 118100
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 0.0005,
+      "loss": 1.2178,
+      "step": 118200
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.0005,
+      "loss": 1.2299,
+      "step": 118300
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.0005,
+      "loss": 1.2235,
+      "step": 118400
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 0.0005,
+      "loss": 1.2293,
+      "step": 118500
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 0.0005,
+      "loss": 1.2289,
+      "step": 118600
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 0.0005,
+      "loss": 1.226,
+      "step": 118700
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.0005,
+      "loss": 1.2305,
+      "step": 118800
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.0005,
+      "loss": 1.2195,
+      "step": 118900
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 0.0005,
+      "loss": 1.2574,
+      "step": 119000
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.0005,
+      "loss": 1.2401,
+      "step": 119100
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 0.0005,
+      "loss": 1.2439,
+      "step": 119200
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 0.0005,
+      "loss": 1.2165,
+      "step": 119300
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 0.0005,
+      "loss": 1.2318,
+      "step": 119400
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 0.0005,
+      "loss": 1.2358,
+      "step": 119500
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 0.0005,
+      "loss": 1.2421,
+      "step": 119600
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.0005,
+      "loss": 1.2324,
+      "step": 119700
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.0005,
+      "loss": 1.2332,
+      "step": 119800
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 0.0005,
+      "loss": 1.2377,
+      "step": 119900
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 0.0005,
+      "loss": 1.2568,
+      "step": 120000
+    },
+    {
+      "epoch": 9.41,
+      "eval_gen_len": 18.66372384567847,
+      "eval_loss": 2.3514394760131836,
+      "eval_rouge1": 34.7411,
+      "eval_rouge2": 13.8346,
+      "eval_rougeL": 28.6141,
+      "eval_rougeLsum": 28.6032,
+      "eval_runtime": 345.2724,
+      "eval_samples_per_second": 32.806,
+      "eval_steps_per_second": 2.051,
+      "step": 120000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 0.0005,
+      "loss": 1.2152,
+      "step": 120100
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.0005,
+      "loss": 1.2593,
+      "step": 120200
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.0005,
+      "loss": 1.2727,
+      "step": 120300
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 0.0005,
+      "loss": 1.2449,
+      "step": 120400
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 0.0005,
+      "loss": 1.2418,
+      "step": 120500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 0.0005,
+      "loss": 1.2555,
+      "step": 120600
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.0005,
+      "loss": 1.2419,
+      "step": 120700
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.0005,
+      "loss": 1.222,
+      "step": 120800
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 0.0005,
+      "loss": 1.2457,
+      "step": 120900
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 0.0005,
+      "loss": 1.2392,
+      "step": 121000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.0005,
+      "loss": 1.243,
+      "step": 121100
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.0005,
+      "loss": 1.2354,
+      "step": 121200
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 0.0005,
+      "loss": 1.2476,
+      "step": 121300
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 0.0005,
+      "loss": 1.2493,
+      "step": 121400
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 0.0005,
+      "loss": 1.2469,
+      "step": 121500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 0.0005,
+      "loss": 1.248,
+      "step": 121600
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 0.0005,
+      "loss": 1.2485,
+      "step": 121700
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 0.0005,
+      "loss": 1.2614,
+      "step": 121800
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 0.0005,
+      "loss": 1.2542,
+      "step": 121900
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.0005,
+      "loss": 1.2463,
+      "step": 122000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.0005,
+      "loss": 1.2421,
+      "step": 122100
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 0.0005,
+      "loss": 1.2599,
+      "step": 122200
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 0.0005,
+      "loss": 1.2507,
+      "step": 122300
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 0.0005,
+      "loss": 1.2517,
+      "step": 122400
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 0.0005,
+      "loss": 1.2524,
+      "step": 122500
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 0.0005,
+      "loss": 1.2492,
+      "step": 122600
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 0.0005,
+      "loss": 1.2445,
+      "step": 122700
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 0.0005,
+      "loss": 1.235,
+      "step": 122800
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 0.0005,
+      "loss": 1.2788,
+      "step": 122900
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.0005,
+      "loss": 1.2313,
+      "step": 123000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.0005,
+      "loss": 1.2594,
+      "step": 123100
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 0.0005,
+      "loss": 1.237,
+      "step": 123200
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 0.0005,
+      "loss": 1.2641,
+      "step": 123300
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.0005,
+      "loss": 1.2346,
+      "step": 123400
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.0005,
+      "loss": 1.2685,
+      "step": 123500
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 0.0005,
+      "loss": 1.2756,
+      "step": 123600
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 0.0005,
+      "loss": 1.2478,
+      "step": 123700
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.0005,
+      "loss": 1.2453,
+      "step": 123800
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 0.0005,
+      "loss": 1.2631,
+      "step": 123900
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 0.0005,
+      "loss": 1.2612,
+      "step": 124000
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 0.0005,
+      "loss": 1.2553,
+      "step": 124100
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.0005,
+      "loss": 1.2489,
+      "step": 124200
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 0.0005,
+      "loss": 1.2581,
+      "step": 124300
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 0.0005,
+      "loss": 1.2501,
+      "step": 124400
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 0.0005,
+      "loss": 1.2433,
+      "step": 124500
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 0.0005,
+      "loss": 1.2479,
+      "step": 124600
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 0.0005,
+      "loss": 1.2705,
+      "step": 124700
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 0.0005,
+      "loss": 1.2819,
+      "step": 124800
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 0.0005,
+      "loss": 1.2689,
+      "step": 124900
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 0.0005,
+      "loss": 1.2599,
+      "step": 125000
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 0.0005,
+      "loss": 1.2337,
+      "step": 125100
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 0.0005,
+      "loss": 1.2309,
+      "step": 125200
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.0005,
+      "loss": 1.2511,
+      "step": 125300
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.0005,
+      "loss": 1.2792,
+      "step": 125400
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 0.0005,
+      "loss": 1.2619,
+      "step": 125500
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 0.0005,
+      "loss": 1.2643,
+      "step": 125600
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 0.0005,
+      "loss": 1.24,
+      "step": 125700
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 0.0005,
+      "loss": 1.2424,
+      "step": 125800
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 0.0005,
+      "loss": 1.2288,
+      "step": 125900
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 0.0005,
+      "loss": 1.2395,
+      "step": 126000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 0.0005,
+      "loss": 1.2835,
+      "step": 126100
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.0005,
+      "loss": 1.2581,
+      "step": 126200
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.0005,
+      "loss": 1.2563,
+      "step": 126300
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 0.0005,
+      "loss": 1.2553,
+      "step": 126400
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.0005,
+      "loss": 1.26,
+      "step": 126500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 0.0005,
+      "loss": 1.2634,
+      "step": 126600
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 0.0005,
+      "loss": 1.2648,
+      "step": 126700
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 0.0005,
+      "loss": 1.2538,
+      "step": 126800
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 0.0005,
+      "loss": 1.2725,
+      "step": 126900
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 0.0005,
+      "loss": 1.2494,
+      "step": 127000
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 0.0005,
+      "loss": 1.2813,
+      "step": 127100
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 0.0005,
+      "loss": 1.2514,
+      "step": 127200
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 0.0005,
+      "loss": 1.2516,
+      "step": 127300
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.0005,
+      "loss": 1.2691,
+      "step": 127400
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0005,
+      "loss": 1.2495,
+      "step": 127500
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 0.0005,
+      "loss": 1.2004,
+      "step": 127600
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 0.0005,
+      "loss": 1.1841,
+      "step": 127700
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 0.0005,
+      "loss": 1.1587,
+      "step": 127800
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 0.0005,
+      "loss": 1.151,
+      "step": 127900
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 0.0005,
+      "loss": 1.1594,
+      "step": 128000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.0005,
+      "loss": 1.1961,
+      "step": 128100
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.0005,
+      "loss": 1.1642,
+      "step": 128200
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 0.0005,
+      "loss": 1.1716,
+      "step": 128300
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 0.0005,
+      "loss": 1.1898,
+      "step": 128400
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.0005,
+      "loss": 1.1968,
+      "step": 128500
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.0005,
+      "loss": 1.2105,
+      "step": 128600
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 0.0005,
+      "loss": 1.1601,
+      "step": 128700
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 0.0005,
+      "loss": 1.1855,
+      "step": 128800
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 0.0005,
+      "loss": 1.1889,
+      "step": 128900
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 0.0005,
+      "loss": 1.1638,
+      "step": 129000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 0.0005,
+      "loss": 1.1665,
+      "step": 129100
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.0005,
+      "loss": 1.191,
+      "step": 129200
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 0.0005,
+      "loss": 1.1814,
+      "step": 129300
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 0.0005,
+      "loss": 1.2109,
+      "step": 129400
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.0005,
+      "loss": 1.1878,
+      "step": 129500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.0005,
+      "loss": 1.1812,
+      "step": 129600
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.0005,
+      "loss": 1.19,
+      "step": 129700
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 0.0005,
+      "loss": 1.2061,
+      "step": 129800
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.0005,
+      "loss": 1.1939,
+      "step": 129900
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.0005,
+      "loss": 1.196,
+      "step": 130000
+    },
+    {
+      "epoch": 10.19,
+      "eval_gen_len": 18.68429416438598,
+      "eval_loss": 2.362765073776245,
+      "eval_rouge1": 35.0302,
+      "eval_rouge2": 13.9715,
+      "eval_rougeL": 28.7794,
+      "eval_rougeLsum": 28.7677,
+      "eval_runtime": 342.4044,
+      "eval_samples_per_second": 33.081,
+      "eval_steps_per_second": 2.068,
+      "step": 130000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 0.0005,
+      "loss": 1.196,
+      "step": 130100
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 0.0005,
+      "loss": 1.1993,
+      "step": 130200
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 0.0005,
+      "loss": 1.1875,
+      "step": 130300
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.0005,
+      "loss": 1.196,
+      "step": 130400
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.0005,
+      "loss": 1.2116,
+      "step": 130500
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 0.0005,
+      "loss": 1.1748,
+      "step": 130600
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1957,
+      "step": 130700
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 0.0005,
+      "loss": 1.2038,
+      "step": 130800
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1831,
+      "step": 130900
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1999,
+      "step": 131000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1911,
+      "step": 131100
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1899,
+      "step": 131200
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 0.0005,
+      "loss": 1.2048,
+      "step": 131300
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 0.0005,
+      "loss": 1.2188,
+      "step": 131400
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 0.0005,
+      "loss": 1.2149,
+      "step": 131500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 0.0005,
+      "loss": 1.2153,
+      "step": 131600
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 0.0005,
+      "loss": 1.2156,
+      "step": 131700
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1967,
+      "step": 131800
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.0005,
+      "loss": 1.2148,
+      "step": 131900
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 0.0005,
+      "loss": 1.211,
+      "step": 132000
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1971,
+      "step": 132100
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.0005,
+      "loss": 1.2019,
+      "step": 132200
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.0005,
+      "loss": 1.2078,
+      "step": 132300
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 0.0005,
+      "loss": 1.1958,
+      "step": 132400
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 0.0005,
+      "loss": 1.243,
+      "step": 132500
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 0.0005,
+      "loss": 1.2074,
+      "step": 132600
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 0.0005,
+      "loss": 1.2086,
+      "step": 132700
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 0.0005,
+      "loss": 1.2075,
+      "step": 132800
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 0.0005,
+      "loss": 1.2124,
+      "step": 132900
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 0.0005,
+      "loss": 1.2052,
+      "step": 133000
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1858,
+      "step": 133100
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1925,
+      "step": 133200
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1945,
+      "step": 133300
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 0.0005,
+      "loss": 1.2041,
+      "step": 133400
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 0.0005,
+      "loss": 1.22,
+      "step": 133500
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.0005,
+      "loss": 1.2024,
+      "step": 133600
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.0005,
+      "loss": 1.2076,
+      "step": 133700
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 0.0005,
+      "loss": 1.2013,
+      "step": 133800
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 0.0005,
+      "loss": 1.2254,
+      "step": 133900
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 0.0005,
+      "loss": 1.202,
+      "step": 134000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.0005,
+      "loss": 1.2101,
+      "step": 134100
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.0005,
+      "loss": 1.2049,
+      "step": 134200
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 0.0005,
+      "loss": 1.2248,
+      "step": 134300
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 0.0005,
+      "loss": 1.2321,
+      "step": 134400
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 0.0005,
+      "loss": 1.2106,
+      "step": 134500
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 0.0005,
+      "loss": 1.2166,
+      "step": 134600
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 0.0005,
+      "loss": 1.2197,
+      "step": 134700
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 0.0005,
+      "loss": 1.2132,
+      "step": 134800
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 0.0005,
+      "loss": 1.2318,
+      "step": 134900
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.0005,
+      "loss": 1.2189,
+      "step": 135000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.0005,
+      "loss": 1.2334,
+      "step": 135100
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 0.0005,
+      "loss": 1.2417,
+      "step": 135200
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 0.0005,
+      "loss": 1.2381,
+      "step": 135300
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 0.0005,
+      "loss": 1.2264,
+      "step": 135400
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 0.0005,
+      "loss": 1.2124,
+      "step": 135500
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 0.0005,
+      "loss": 1.2251,
+      "step": 135600
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 0.0005,
+      "loss": 1.2115,
+      "step": 135700
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.0005,
+      "loss": 1.2197,
+      "step": 135800
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.0005,
+      "loss": 1.2118,
+      "step": 135900
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.0005,
+      "loss": 1.2366,
+      "step": 136000
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 0.0005,
+      "loss": 1.2107,
+      "step": 136100
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 0.0005,
+      "loss": 1.2308,
+      "step": 136200
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 0.0005,
+      "loss": 1.2161,
+      "step": 136300
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 0.0005,
+      "loss": 1.2348,
+      "step": 136400
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 0.0005,
+      "loss": 1.207,
+      "step": 136500
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 0.0005,
+      "loss": 1.2388,
+      "step": 136600
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 0.0005,
+      "loss": 1.2165,
+      "step": 136700
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 0.0005,
+      "loss": 1.2283,
+      "step": 136800
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 0.0005,
+      "loss": 1.2366,
+      "step": 136900
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 0.0005,
+      "loss": 1.2192,
+      "step": 137000
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 0.0005,
+      "loss": 1.2077,
+      "step": 137100
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 0.0005,
+      "loss": 1.225,
+      "step": 137200
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.0005,
+      "loss": 1.2208,
+      "step": 137300
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.0005,
+      "loss": 1.2194,
+      "step": 137400
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 0.0005,
+      "loss": 1.2345,
+      "step": 137500
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 0.0005,
+      "loss": 1.2508,
+      "step": 137600
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 0.0005,
+      "loss": 1.2184,
+      "step": 137700
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 0.0005,
+      "loss": 1.247,
+      "step": 137800
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 0.0005,
+      "loss": 1.2342,
+      "step": 137900
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 0.0005,
+      "loss": 1.2214,
+      "step": 138000
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 0.0005,
+      "loss": 1.2064,
+      "step": 138100
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 0.0005,
+      "loss": 1.2336,
+      "step": 138200
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.0005,
+      "loss": 1.2518,
+      "step": 138300
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.0005,
+      "loss": 1.2242,
+      "step": 138400
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 0.0005,
+      "loss": 1.2257,
+      "step": 138500
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 0.0005,
+      "loss": 1.2038,
+      "step": 138600
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.0005,
+      "loss": 1.2265,
+      "step": 138700
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.0005,
+      "loss": 1.2268,
+      "step": 138800
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 0.0005,
+      "loss": 1.2488,
+      "step": 138900
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1833,
+      "step": 139000
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 0.0005,
+      "loss": 1.2291,
+      "step": 139100
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.0005,
+      "loss": 1.2266,
+      "step": 139200
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.0005,
+      "loss": 1.2037,
+      "step": 139300
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 0.0005,
+      "loss": 1.2575,
+      "step": 139400
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 0.0005,
+      "loss": 1.2288,
+      "step": 139500
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 0.0005,
+      "loss": 1.2285,
+      "step": 139600
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 0.0005,
+      "loss": 1.2516,
+      "step": 139700
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 0.0005,
+      "loss": 1.2349,
+      "step": 139800
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 0.0005,
+      "loss": 1.2487,
+      "step": 139900
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 0.0005,
+      "loss": 1.2228,
+      "step": 140000
+    },
+    {
+      "epoch": 10.98,
+      "eval_gen_len": 18.681733910126248,
+      "eval_loss": 2.3462131023406982,
+      "eval_rouge1": 35.2577,
+      "eval_rouge2": 14.1723,
+      "eval_rougeL": 28.9678,
+      "eval_rougeLsum": 28.9582,
+      "eval_runtime": 342.5375,
+      "eval_samples_per_second": 33.068,
+      "eval_steps_per_second": 2.067,
+      "step": 140000
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 0.0005,
+      "loss": 1.2395,
+      "step": 140100
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 0.0005,
+      "loss": 1.2426,
+      "step": 140200
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.0005,
+      "loss": 1.1879,
+      "step": 140300
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 0.0005,
+      "loss": 1.15,
+      "step": 140400
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 0.0005,
+      "loss": 1.1689,
+      "step": 140500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.0005,
+      "loss": 1.1181,
+      "step": 140600
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.0005,
+      "loss": 1.1402,
+      "step": 140700
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.0005,
+      "loss": 1.1483,
+      "step": 140800
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 0.0005,
+      "loss": 1.1526,
+      "step": 140900
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.0005,
+      "loss": 1.1349,
+      "step": 141000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.0005,
+      "loss": 1.1376,
+      "step": 141100
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 0.0005,
+      "loss": 1.1485,
+      "step": 141200
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 0.0005,
+      "loss": 1.1632,
+      "step": 141300
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 0.0005,
+      "loss": 1.1419,
+      "step": 141400
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.0005,
+      "loss": 1.1597,
+      "step": 141500
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.0005,
+      "loss": 1.1683,
+      "step": 141600
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 0.0005,
+      "loss": 1.1875,
+      "step": 141700
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 0.0005,
+      "loss": 1.1291,
+      "step": 141800
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 0.0005,
+      "loss": 1.1621,
+      "step": 141900
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 0.0005,
+      "loss": 1.1523,
+      "step": 142000
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 0.0005,
+      "loss": 1.1555,
+      "step": 142100
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 0.0005,
+      "loss": 1.1392,
+      "step": 142200
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 0.0005,
+      "loss": 1.1446,
+      "step": 142300
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 0.0005,
+      "loss": 1.1578,
+      "step": 142400
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 0.0005,
+      "loss": 1.1403,
+      "step": 142500
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.0005,
+      "loss": 1.15,
+      "step": 142600
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 0.0005,
+      "loss": 1.1473,
+      "step": 142700
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 0.0005,
+      "loss": 1.1652,
+      "step": 142800
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.0005,
+      "loss": 1.1748,
+      "step": 142900
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.0005,
+      "loss": 1.1487,
+      "step": 143000
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 0.0005,
+      "loss": 1.177,
+      "step": 143100
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 0.0005,
+      "loss": 1.1724,
+      "step": 143200
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 0.0005,
+      "loss": 1.1556,
+      "step": 143300
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1606,
+      "step": 143400
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1826,
+      "step": 143500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 0.0005,
+      "loss": 1.1488,
+      "step": 143600
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1783,
+      "step": 143700
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1688,
+      "step": 143800
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1695,
+      "step": 143900
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1735,
+      "step": 144000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1811,
+      "step": 144100
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 0.0005,
+      "loss": 1.1831,
+      "step": 144200
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1766,
+      "step": 144300
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1709,
+      "step": 144400
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 0.0005,
+      "loss": 1.1633,
+      "step": 144500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1802,
+      "step": 144600
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1757,
+      "step": 144700
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1612,
+      "step": 144800
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1808,
+      "step": 144900
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1823,
+      "step": 145000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 0.0005,
+      "loss": 1.1876,
+      "step": 145100
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1948,
+      "step": 145200
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.0005,
+      "loss": 1.183,
+      "step": 145300
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 0.0005,
+      "loss": 1.19,
+      "step": 145400
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1493,
+      "step": 145500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 0.0005,
+      "loss": 1.1708,
+      "step": 145600
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1942,
+      "step": 145700
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1929,
+      "step": 145800
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 0.0005,
+      "loss": 1.2045,
+      "step": 145900
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1818,
+      "step": 146000
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.0005,
+      "loss": 1.2014,
+      "step": 146100
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.0005,
+      "loss": 1.1809,
+      "step": 146200
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 0.0005,
+      "loss": 1.1725,
+      "step": 146300
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1694,
+      "step": 146400
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1675,
+      "step": 146500
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 0.0005,
+      "loss": 1.2055,
+      "step": 146600
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 0.0005,
+      "loss": 1.2039,
+      "step": 146700
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1823,
+      "step": 146800
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1643,
+      "step": 146900
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 0.0005,
+      "loss": 1.1949,
+      "step": 147000
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1884,
+      "step": 147100
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1693,
+      "step": 147200
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1765,
+      "step": 147300
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1672,
+      "step": 147400
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 0.0005,
+      "loss": 1.192,
+      "step": 147500
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1945,
+      "step": 147600
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1885,
+      "step": 147700
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1994,
+      "step": 147800
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 0.0005,
+      "loss": 1.1773,
+      "step": 147900
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 0.0005,
+      "loss": 1.209,
+      "step": 148000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1904,
+      "step": 148100
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 0.0005,
+      "loss": 1.204,
+      "step": 148200
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 0.0005,
+      "loss": 1.2284,
+      "step": 148300
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 0.0005,
+      "loss": 1.1981,
+      "step": 148400
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.0005,
+      "loss": 1.1934,
+      "step": 148500
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.0005,
+      "loss": 1.1706,
+      "step": 148600
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1944,
+      "step": 148700
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1932,
+      "step": 148800
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1723,
+      "step": 148900
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1924,
+      "step": 149000
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 0.0005,
+      "loss": 1.198,
+      "step": 149100
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1977,
+      "step": 149200
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 0.0005,
+      "loss": 1.2074,
+      "step": 149300
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 0.0005,
+      "loss": 1.1834,
+      "step": 149400
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 0.0005,
+      "loss": 1.1882,
+      "step": 149500
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 0.0005,
+      "loss": 1.1864,
+      "step": 149600
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1997,
+      "step": 149700
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 0.0005,
+      "loss": 1.198,
+      "step": 149800
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.0005,
+      "loss": 1.1838,
+      "step": 149900
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.0005,
+      "loss": 1.2029,
+      "step": 150000
+    },
+    {
+      "epoch": 11.76,
+      "eval_gen_len": 18.718636885318265,
+      "eval_loss": 2.359110116958618,
+      "eval_rouge1": 34.8446,
+      "eval_rouge2": 14.049,
+      "eval_rougeL": 28.7144,
+      "eval_rougeLsum": 28.7106,
+      "eval_runtime": 344.0253,
+      "eval_samples_per_second": 32.925,
+      "eval_steps_per_second": 2.058,
+      "step": 150000
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 0.0005,
+      "loss": 1.2122,
+      "step": 150100
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 0.0005,
+      "loss": 1.1915,
+      "step": 150200
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.0005,
+      "loss": 1.204,
+      "step": 150300
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.0005,
+      "loss": 1.2037,
+      "step": 150400
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 0.0005,
+      "loss": 1.1972,
+      "step": 150500
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 0.0005,
+      "loss": 1.2176,
+      "step": 150600
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 0.0005,
+      "loss": 1.2285,
+      "step": 150700
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.0005,
+      "loss": 1.2129,
+      "step": 150800
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.0005,
+      "loss": 1.2006,
+      "step": 150900
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 0.0005,
+      "loss": 1.2294,
+      "step": 151000
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 0.0005,
+      "loss": 1.2171,
+      "step": 151100
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.0005,
+      "loss": 1.2049,
+      "step": 151200
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.0005,
+      "loss": 1.1918,
+      "step": 151300
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1853,
+      "step": 151400
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1898,
+      "step": 151500
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 0.0005,
+      "loss": 1.2226,
+      "step": 151600
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 0.0005,
+      "loss": 1.2084,
+      "step": 151700
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1979,
+      "step": 151800
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 0.0005,
+      "loss": 1.2002,
+      "step": 151900
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 0.0005,
+      "loss": 1.2009,
+      "step": 152000
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 0.0005,
+      "loss": 1.2152,
+      "step": 152100
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.0005,
+      "loss": 1.2111,
+      "step": 152200
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.0005,
+      "loss": 1.2051,
+      "step": 152300
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.0005,
+      "loss": 1.2065,
+      "step": 152400
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 0.0005,
+      "loss": 1.2008,
+      "step": 152500
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.0005,
+      "loss": 1.2221,
+      "step": 152600
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.0005,
+      "loss": 1.2175,
+      "step": 152700
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 0.0005,
+      "loss": 1.2027,
+      "step": 152800
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 0.0005,
+      "loss": 1.1946,
+      "step": 152900
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 0.0005,
+      "loss": 1.2154,
+      "step": 153000
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.0005,
+      "loss": 1.1322,
+      "step": 153100
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.0005,
+      "loss": 1.1326,
+      "step": 153200
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 0.0005,
+      "loss": 1.129,
+      "step": 153300
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 0.0005,
+      "loss": 1.1352,
+      "step": 153400
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.0005,
+      "loss": 1.1344,
+      "step": 153500
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 0.0005,
+      "loss": 1.0996,
+      "step": 153600
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 0.0005,
+      "loss": 1.1338,
+      "step": 153700
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 0.0005,
+      "loss": 1.122,
+      "step": 153800
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 0.0005,
+      "loss": 1.1314,
+      "step": 153900
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 0.0005,
+      "loss": 1.1346,
+      "step": 154000
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 0.0005,
+      "loss": 1.1408,
+      "step": 154100
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 0.0005,
+      "loss": 1.127,
+      "step": 154200
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 0.0005,
+      "loss": 1.1311,
+      "step": 154300
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 0.0005,
+      "loss": 1.1393,
+      "step": 154400
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.0005,
+      "loss": 1.1483,
+      "step": 154500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.0005,
+      "loss": 1.1587,
+      "step": 154600
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 0.0005,
+      "loss": 1.1518,
+      "step": 154700
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 0.0005,
+      "loss": 1.1536,
+      "step": 154800
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.0005,
+      "loss": 1.1415,
+      "step": 154900
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.0005,
+      "loss": 1.1333,
+      "step": 155000
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 0.0005,
+      "loss": 1.1504,
+      "step": 155100
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 0.0005,
+      "loss": 1.1564,
+      "step": 155200
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 0.0005,
+      "loss": 1.1661,
+      "step": 155300
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.0005,
+      "loss": 1.1426,
+      "step": 155400
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.0005,
+      "loss": 1.1569,
+      "step": 155500
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 0.0005,
+      "loss": 1.1492,
+      "step": 155600
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 0.0005,
+      "loss": 1.1497,
+      "step": 155700
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.0005,
+      "loss": 1.1648,
+      "step": 155800
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 0.0005,
+      "loss": 1.1646,
+      "step": 155900
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 0.0005,
+      "loss": 1.152,
+      "step": 156000
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 0.0005,
+      "loss": 1.1823,
+      "step": 156100
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1476,
+      "step": 156200
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 0.0005,
+      "loss": 1.164,
+      "step": 156300
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 0.0005,
+      "loss": 1.1714,
+      "step": 156400
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1536,
+      "step": 156500
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1508,
+      "step": 156600
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1664,
+      "step": 156700
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1602,
+      "step": 156800
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1709,
+      "step": 156900
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.0005,
+      "loss": 1.1634,
+      "step": 157000
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1512,
+      "step": 157100
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 0.0005,
+      "loss": 1.1742,
+      "step": 157200
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 0.0005,
+      "loss": 1.175,
+      "step": 157300
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1651,
+      "step": 157400
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1466,
+      "step": 157500
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1837,
+      "step": 157600
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1443,
+      "step": 157700
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1536,
+      "step": 157800
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 0.0005,
+      "loss": 1.1573,
+      "step": 157900
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1771,
+      "step": 158000
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 0.0005,
+      "loss": 1.1854,
+      "step": 158100
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1495,
+      "step": 158200
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1569,
+      "step": 158300
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 0.0005,
+      "loss": 1.1653,
+      "step": 158400
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1752,
+      "step": 158500
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1757,
+      "step": 158600
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1979,
+      "step": 158700
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1784,
+      "step": 158800
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 0.0005,
+      "loss": 1.1607,
+      "step": 158900
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 0.0005,
+      "loss": 1.1937,
+      "step": 159000
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1799,
+      "step": 159100
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1793,
+      "step": 159200
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1743,
+      "step": 159300
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1639,
+      "step": 159400
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1779,
+      "step": 159500
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1658,
+      "step": 159600
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1748,
+      "step": 159700
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 0.0005,
+      "loss": 1.1649,
+      "step": 159800
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1988,
+      "step": 159900
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1803,
+      "step": 160000
+    },
+    {
+      "epoch": 12.55,
+      "eval_gen_len": 18.743091727730203,
+      "eval_loss": 2.3673198223114014,
+      "eval_rouge1": 35.1989,
+      "eval_rouge2": 14.1819,
+      "eval_rougeL": 29.066,
+      "eval_rougeLsum": 29.054,
+      "eval_runtime": 343.7666,
+      "eval_samples_per_second": 32.95,
+      "eval_steps_per_second": 2.06,
+      "step": 160000
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1638,
+      "step": 160100
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 0.0005,
+      "loss": 1.149,
+      "step": 160200
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1321,
+      "step": 160300
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1495,
+      "step": 160400
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1303,
+      "step": 160500
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1605,
+      "step": 160600
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 0.0005,
+      "loss": 1.1455,
+      "step": 160700
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1459,
+      "step": 160800
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 0.0005,
+      "loss": 1.1393,
+      "step": 160900
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1386,
+      "step": 161000
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1447,
+      "step": 161100
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 0.0005,
+      "loss": 1.145,
+      "step": 161200
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 0.0005,
+      "loss": 1.152,
+      "step": 161300
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1342,
+      "step": 161400
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1623,
+      "step": 161500
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1526,
+      "step": 161600
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1477,
+      "step": 161700
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 0.0005,
+      "loss": 1.1437,
+      "step": 161800
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1639,
+      "step": 161900
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1569,
+      "step": 162000
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 0.0005,
+      "loss": 1.1576,
+      "step": 162100
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 0.0005,
+      "loss": 1.1515,
+      "step": 162200
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 0.0005,
+      "loss": 1.1578,
+      "step": 162300
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1403,
+      "step": 162400
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1511,
+      "step": 162500
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 0.0005,
+      "loss": 1.1581,
+      "step": 162600
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 0.0005,
+      "loss": 1.1572,
+      "step": 162700
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 0.0005,
+      "loss": 1.1475,
+      "step": 162800
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 0.0005,
+      "loss": 1.159,
+      "step": 162900
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 0.0005,
+      "loss": 1.1562,
+      "step": 163000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 0.0005,
+      "loss": 1.1721,
+      "step": 163100
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 0.0005,
+      "loss": 1.164,
+      "step": 163200
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 0.0005,
+      "loss": 1.1522,
+      "step": 163300
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 0.0005,
+      "loss": 1.1563,
+      "step": 163400
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1595,
+      "step": 163500
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 0.0005,
+      "loss": 1.1642,
+      "step": 163600
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 0.0005,
+      "loss": 1.1569,
+      "step": 163700
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1536,
+      "step": 163800
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1575,
+      "step": 163900
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 0.0005,
+      "loss": 1.1597,
+      "step": 164000
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1452,
+      "step": 164100
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1616,
+      "step": 164200
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1667,
+      "step": 164300
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 0.0005,
+      "loss": 1.1595,
+      "step": 164400
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1741,
+      "step": 164500
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.0005,
+      "loss": 1.1855,
+      "step": 164600
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 0.0005,
+      "loss": 1.1701,
+      "step": 164700
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 0.0005,
+      "loss": 1.1808,
+      "step": 164800
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 0.0005,
+      "loss": 1.159,
+      "step": 164900
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 0.0005,
+      "loss": 1.1646,
+      "step": 165000
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 0.0005,
+      "loss": 1.1641,
+      "step": 165100
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 0.0005,
+      "loss": 1.1478,
+      "step": 165200
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 0.0005,
+      "loss": 1.1586,
+      "step": 165300
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 0.0005,
+      "loss": 1.1634,
+      "step": 165400
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 0.0005,
+      "loss": 1.1682,
+      "step": 165500
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 0.0005,
+      "loss": 1.1613,
+      "step": 165600
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 0.0005,
+      "loss": 1.1577,
+      "step": 165700
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 0.0005,
+      "loss": 1.1236,
+      "step": 165800
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 0.0005,
+      "loss": 1.1019,
+      "step": 165900
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 0.0005,
+      "loss": 1.1098,
+      "step": 166000
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 0.0005,
+      "loss": 1.094,
+      "step": 166100
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 0.0005,
+      "loss": 1.1069,
+      "step": 166200
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 0.0005,
+      "loss": 1.0924,
+      "step": 166300
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 0.0005,
+      "loss": 1.1061,
+      "step": 166400
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.0005,
+      "loss": 1.1109,
+      "step": 166500
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.0005,
+      "loss": 1.1294,
+      "step": 166600
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 0.0005,
+      "loss": 1.1029,
+      "step": 166700
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 0.0005,
+      "loss": 1.1201,
+      "step": 166800
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 0.0005,
+      "loss": 1.0957,
+      "step": 166900
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 0.0005,
+      "loss": 1.1034,
+      "step": 167000
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 0.0005,
+      "loss": 1.1175,
+      "step": 167100
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 0.0005,
+      "loss": 1.1181,
+      "step": 167200
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 0.0005,
+      "loss": 1.1114,
+      "step": 167300
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 0.0005,
+      "loss": 1.1111,
+      "step": 167400
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 0.0005,
+      "loss": 1.0998,
+      "step": 167500
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 0.0005,
+      "loss": 1.1189,
+      "step": 167600
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 0.0005,
+      "loss": 1.092,
+      "step": 167700
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 0.0005,
+      "loss": 1.1112,
+      "step": 167800
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 0.0005,
+      "loss": 1.1004,
+      "step": 167900
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 0.0005,
+      "loss": 1.1054,
+      "step": 168000
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 0.0005,
+      "loss": 1.1193,
+      "step": 168100
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 0.0005,
+      "loss": 1.1312,
+      "step": 168200
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 0.0005,
+      "loss": 1.128,
+      "step": 168300
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 0.0005,
+      "loss": 1.1258,
+      "step": 168400
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 0.0005,
+      "loss": 1.1162,
+      "step": 168500
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 0.0005,
+      "loss": 1.1088,
+      "step": 168600
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 0.0005,
+      "loss": 1.1106,
+      "step": 168700
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 0.0005,
+      "loss": 1.0878,
+      "step": 168800
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 0.0005,
+      "loss": 1.1264,
+      "step": 168900
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 0.0005,
+      "loss": 1.114,
+      "step": 169000
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 0.0005,
+      "loss": 1.1263,
+      "step": 169100
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1365,
+      "step": 169200
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1339,
+      "step": 169300
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1212,
+      "step": 169400
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 0.0005,
+      "loss": 1.129,
+      "step": 169500
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1214,
+      "step": 169600
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 0.0005,
+      "loss": 1.1283,
+      "step": 169700
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.0005,
+      "loss": 1.0951,
+      "step": 169800
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1302,
+      "step": 169900
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 0.0005,
+      "loss": 1.1123,
+      "step": 170000
+    },
+    {
+      "epoch": 13.33,
+      "eval_gen_len": 18.691621788646597,
+      "eval_loss": 2.402947187423706,
+      "eval_rouge1": 35.1849,
+      "eval_rouge2": 14.2186,
+      "eval_rougeL": 29.0353,
+      "eval_rougeLsum": 29.0203,
+      "eval_runtime": 343.1956,
+      "eval_samples_per_second": 33.005,
+      "eval_steps_per_second": 2.063,
+      "step": 170000
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1359,
+      "step": 170100
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1214,
+      "step": 170200
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1228,
+      "step": 170300
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1058,
+      "step": 170400
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1077,
+      "step": 170500
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 0.0005,
+      "loss": 1.1253,
+      "step": 170600
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1375,
+      "step": 170700
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1263,
+      "step": 170800
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 0.0005,
+      "loss": 1.1423,
+      "step": 170900
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 0.0005,
+      "loss": 1.141,
+      "step": 171000
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 0.0005,
+      "loss": 1.1288,
+      "step": 171100
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1449,
+      "step": 171200
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1116,
+      "step": 171300
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1132,
+      "step": 171400
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1256,
+      "step": 171500
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.0005,
+      "loss": 1.1448,
+      "step": 171600
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.0005,
+      "loss": 1.1284,
+      "step": 171700
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 0.0005,
+      "loss": 1.1418,
+      "step": 171800
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1365,
+      "step": 171900
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1445,
+      "step": 172000
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1358,
+      "step": 172100
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1433,
+      "step": 172200
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 0.0005,
+      "loss": 1.149,
+      "step": 172300
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 0.0005,
+      "loss": 1.136,
+      "step": 172400
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 0.0005,
+      "loss": 1.1626,
+      "step": 172500
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1296,
+      "step": 172600
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1324,
+      "step": 172700
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1233,
+      "step": 172800
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1472,
+      "step": 172900
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1348,
+      "step": 173000
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1338,
+      "step": 173100
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1341,
+      "step": 173200
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1679,
+      "step": 173300
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 0.0005,
+      "loss": 1.1462,
+      "step": 173400
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1452,
+      "step": 173500
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.0005,
+      "loss": 1.122,
+      "step": 173600
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 0.0005,
+      "loss": 1.1425,
+      "step": 173700
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1436,
+      "step": 173800
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.0005,
+      "loss": 1.1484,
+      "step": 173900
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.0005,
+      "loss": 1.1434,
+      "step": 174000
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 0.0005,
+      "loss": 1.137,
+      "step": 174100
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1412,
+      "step": 174200
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1312,
+      "step": 174300
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1483,
+      "step": 174400
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1517,
+      "step": 174500
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 0.0005,
+      "loss": 1.1333,
+      "step": 174600
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1375,
+      "step": 174700
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 0.0005,
+      "loss": 1.1216,
+      "step": 174800
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 0.0005,
+      "loss": 1.132,
+      "step": 174900
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 0.0005,
+      "loss": 1.1635,
+      "step": 175000
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 0.0005,
+      "loss": 1.1496,
+      "step": 175100
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1656,
+      "step": 175200
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 0.0005,
+      "loss": 1.1612,
+      "step": 175300
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 0.0005,
+      "loss": 1.1553,
+      "step": 175400
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 0.0005,
+      "loss": 1.1427,
+      "step": 175500
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 0.0005,
+      "loss": 1.1426,
+      "step": 175600
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 0.0005,
+      "loss": 1.123,
+      "step": 175700
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.0005,
+      "loss": 1.1331,
+      "step": 175800
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.0005,
+      "loss": 1.1252,
+      "step": 175900
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 0.0005,
+      "loss": 1.1554,
+      "step": 176000
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 0.0005,
+      "loss": 1.1659,
+      "step": 176100
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1598,
+      "step": 176200
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.0005,
+      "loss": 1.1493,
+      "step": 176300
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.0005,
+      "loss": 1.1486,
+      "step": 176400
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 0.0005,
+      "loss": 1.165,
+      "step": 176500
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.0005,
+      "loss": 1.162,
+      "step": 176600
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 0.0005,
+      "loss": 1.1494,
+      "step": 176700
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 0.0005,
+      "loss": 1.1545,
+      "step": 176800
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1265,
+      "step": 176900
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1291,
+      "step": 177000
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 0.0005,
+      "loss": 1.156,
+      "step": 177100
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1521,
+      "step": 177200
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1479,
+      "step": 177300
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 0.0005,
+      "loss": 1.1744,
+      "step": 177400
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 0.0005,
+      "loss": 1.1429,
+      "step": 177500
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 0.0005,
+      "loss": 1.1531,
+      "step": 177600
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 0.0005,
+      "loss": 1.1391,
+      "step": 177700
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 0.0005,
+      "loss": 1.184,
+      "step": 177800
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 0.0005,
+      "loss": 1.1511,
+      "step": 177900
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 0.0005,
+      "loss": 1.169,
+      "step": 178000
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 0.0005,
+      "loss": 1.1608,
+      "step": 178100
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 0.0005,
+      "loss": 1.157,
+      "step": 178200
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 0.0005,
+      "loss": 1.1458,
+      "step": 178300
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 0.0005,
+      "loss": 1.1811,
+      "step": 178400
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 0.0005,
+      "loss": 1.1609,
+      "step": 178500
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 0.0005,
+      "loss": 1.0836,
+      "step": 178600
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 0.0005,
+      "loss": 1.0551,
+      "step": 178700
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 0.0005,
+      "loss": 1.0806,
+      "step": 178800
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 0.0005,
+      "loss": 1.0879,
+      "step": 178900
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 0.0005,
+      "loss": 1.0807,
+      "step": 179000
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 0.0005,
+      "loss": 1.0772,
+      "step": 179100
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 0.0005,
+      "loss": 1.081,
+      "step": 179200
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 0.0005,
+      "loss": 1.0607,
+      "step": 179300
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 0.0005,
+      "loss": 1.0847,
+      "step": 179400
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 0.0005,
+      "loss": 1.0823,
+      "step": 179500
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 0.0005,
+      "loss": 1.0844,
+      "step": 179600
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 0.0005,
+      "loss": 1.0688,
+      "step": 179700
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 0.0005,
+      "loss": 1.0674,
+      "step": 179800
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 0.0005,
+      "loss": 1.0893,
+      "step": 179900
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.0005,
+      "loss": 1.0958,
+      "step": 180000
+    },
+    {
+      "epoch": 14.12,
+      "eval_gen_len": 18.657632206232893,
+      "eval_loss": 2.432025671005249,
+      "eval_rouge1": 34.8892,
+      "eval_rouge2": 14.0297,
+      "eval_rougeL": 28.7636,
+      "eval_rougeLsum": 28.7631,
+      "eval_runtime": 344.3087,
+      "eval_samples_per_second": 32.898,
+      "eval_steps_per_second": 2.056,
+      "step": 180000
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.0005,
+      "loss": 1.0977,
+      "step": 180100
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 0.0005,
+      "loss": 1.0785,
+      "step": 180200
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 0.0005,
+      "loss": 1.0967,
+      "step": 180300
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.0005,
+      "loss": 1.0603,
+      "step": 180400
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.0005,
+      "loss": 1.0986,
+      "step": 180500
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 0.0005,
+      "loss": 1.0738,
+      "step": 180600
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 0.0005,
+      "loss": 1.1258,
+      "step": 180700
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 0.0005,
+      "loss": 1.1022,
+      "step": 180800
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 0.0005,
+      "loss": 1.0924,
+      "step": 180900
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 0.0005,
+      "loss": 1.0958,
+      "step": 181000
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 0.0005,
+      "loss": 1.0834,
+      "step": 181100
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 0.0005,
+      "loss": 1.0942,
+      "step": 181200
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 0.0005,
+      "loss": 1.118,
+      "step": 181300
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 0.0005,
+      "loss": 1.0997,
+      "step": 181400
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 0.0005,
+      "loss": 1.0987,
+      "step": 181500
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.0005,
+      "loss": 1.1071,
+      "step": 181600
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 0.0005,
+      "loss": 1.0964,
+      "step": 181700
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.0005,
+      "loss": 1.1189,
+      "step": 181800
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.0005,
+      "loss": 1.0887,
+      "step": 181900
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1021,
+      "step": 182000
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 0.0005,
+      "loss": 1.0815,
+      "step": 182100
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1012,
+      "step": 182200
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.0005,
+      "loss": 1.129,
+      "step": 182300
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.0005,
+      "loss": 1.0999,
+      "step": 182400
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 0.0005,
+      "loss": 1.0838,
+      "step": 182500
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1064,
+      "step": 182600
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 0.0005,
+      "loss": 1.1099,
+      "step": 182700
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1062,
+      "step": 182800
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 0.0005,
+      "loss": 1.0973,
+      "step": 182900
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1286,
+      "step": 183000
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1003,
+      "step": 183100
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1231,
+      "step": 183200
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 0.0005,
+      "loss": 1.099,
+      "step": 183300
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 0.0005,
+      "loss": 1.0944,
+      "step": 183400
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 0.0005,
+      "loss": 1.0811,
+      "step": 183500
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 0.0005,
+      "loss": 1.0824,
+      "step": 183600
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1227,
+      "step": 183700
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1075,
+      "step": 183800
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 0.0005,
+      "loss": 1.1308,
+      "step": 183900
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 0.0005,
+      "loss": 1.0981,
+      "step": 184000
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0908,
+      "step": 184100
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1119,
+      "step": 184200
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1105,
+      "step": 184300
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0906,
+      "step": 184400
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 0.0005,
+      "loss": 1.1159,
+      "step": 184500
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1075,
+      "step": 184600
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1018,
+      "step": 184700
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1072,
+      "step": 184800
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1099,
+      "step": 184900
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1203,
+      "step": 185000
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1082,
+      "step": 185100
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1119,
+      "step": 185200
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 0.0005,
+      "loss": 1.125,
+      "step": 185300
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1,
+      "step": 185400
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 0.0005,
+      "loss": 1.117,
+      "step": 185500
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1313,
+      "step": 185600
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1077,
+      "step": 185700
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 0.0005,
+      "loss": 1.0911,
+      "step": 185800
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1025,
+      "step": 185900
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1294,
+      "step": 186000
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1237,
+      "step": 186100
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 0.0005,
+      "loss": 1.129,
+      "step": 186200
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1181,
+      "step": 186300
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.0005,
+      "loss": 1.1075,
+      "step": 186400
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1288,
+      "step": 186500
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1283,
+      "step": 186600
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 0.0005,
+      "loss": 1.1295,
+      "step": 186700
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.0005,
+      "loss": 1.0999,
+      "step": 186800
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1222,
+      "step": 186900
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1256,
+      "step": 187000
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1186,
+      "step": 187100
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1217,
+      "step": 187200
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 0.0005,
+      "loss": 1.1336,
+      "step": 187300
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1138,
+      "step": 187400
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1287,
+      "step": 187500
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 0.0005,
+      "loss": 1.1219,
+      "step": 187600
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 0.0005,
+      "loss": 1.1354,
+      "step": 187700
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 0.0005,
+      "loss": 1.1256,
+      "step": 187800
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 0.0005,
+      "loss": 1.113,
+      "step": 187900
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1323,
+      "step": 188000
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 0.0005,
+      "loss": 1.1357,
+      "step": 188100
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 0.0005,
+      "loss": 1.138,
+      "step": 188200
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 0.0005,
+      "loss": 1.1336,
+      "step": 188300
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 0.0005,
+      "loss": 1.1109,
+      "step": 188400
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 0.0005,
+      "loss": 1.1071,
+      "step": 188500
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 0.0005,
+      "loss": 1.1515,
+      "step": 188600
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 0.0005,
+      "loss": 1.1109,
+      "step": 188700
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 0.0005,
+      "loss": 1.1463,
+      "step": 188800
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 0.0005,
+      "loss": 1.1401,
+      "step": 188900
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1327,
+      "step": 189000
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 0.0005,
+      "loss": 1.1423,
+      "step": 189100
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 0.0005,
+      "loss": 1.125,
+      "step": 189200
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 0.0005,
+      "loss": 1.1287,
+      "step": 189300
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1205,
+      "step": 189400
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 0.0005,
+      "loss": 1.1175,
+      "step": 189500
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1428,
+      "step": 189600
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1161,
+      "step": 189700
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1274,
+      "step": 189800
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 0.0005,
+      "loss": 1.1579,
+      "step": 189900
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1264,
+      "step": 190000
+    },
+    {
+      "epoch": 14.9,
+      "eval_gen_len": 18.73020217180189,
+      "eval_loss": 2.38362717628479,
+      "eval_rouge1": 35.2446,
+      "eval_rouge2": 14.196,
+      "eval_rougeL": 28.9968,
+      "eval_rougeLsum": 28.9914,
+      "eval_runtime": 342.3996,
+      "eval_samples_per_second": 33.081,
+      "eval_steps_per_second": 2.068,
+      "step": 190000
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 0.0005,
+      "loss": 1.1146,
+      "step": 190100
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.0005,
+      "loss": 1.1429,
+      "step": 190200
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.0005,
+      "loss": 1.1314,
+      "step": 190300
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 0.0005,
+      "loss": 1.1577,
+      "step": 190400
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 0.0005,
+      "loss": 1.1414,
+      "step": 190500
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 0.0005,
+      "loss": 1.109,
+      "step": 190600
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 0.0005,
+      "loss": 1.1154,
+      "step": 190700
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 0.0005,
+      "loss": 1.129,
+      "step": 190800
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 0.0005,
+      "loss": 1.144,
+      "step": 190900
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 0.0005,
+      "loss": 1.1323,
+      "step": 191000
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 0.0005,
+      "loss": 1.155,
+      "step": 191100
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 0.0005,
+      "loss": 1.1309,
+      "step": 191200
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 0.0005,
+      "loss": 1.0932,
+      "step": 191300
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 0.0005,
+      "loss": 1.0808,
+      "step": 191400
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 0.0005,
+      "loss": 1.046,
+      "step": 191500
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 0.0005,
+      "loss": 1.0491,
+      "step": 191600
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 0.0005,
+      "loss": 1.0523,
+      "step": 191700
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.0005,
+      "loss": 1.072,
+      "step": 191800
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 0.0005,
+      "loss": 1.0484,
+      "step": 191900
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 0.0005,
+      "loss": 1.046,
+      "step": 192000
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 0.0005,
+      "loss": 1.0693,
+      "step": 192100
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 0.0005,
+      "loss": 1.0533,
+      "step": 192200
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 0.0005,
+      "loss": 1.0542,
+      "step": 192300
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 0.0005,
+      "loss": 1.0432,
+      "step": 192400
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.0005,
+      "loss": 1.0598,
+      "step": 192500
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.0005,
+      "loss": 1.0533,
+      "step": 192600
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 0.0005,
+      "loss": 1.0682,
+      "step": 192700
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 0.0005,
+      "loss": 1.0734,
+      "step": 192800
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.0005,
+      "loss": 1.0609,
+      "step": 192900
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.0005,
+      "loss": 1.0822,
+      "step": 193000
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 0.0005,
+      "loss": 1.0573,
+      "step": 193100
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 0.0005,
+      "loss": 1.0729,
+      "step": 193200
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 0.0005,
+      "loss": 1.0827,
+      "step": 193300
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 0.0005,
+      "loss": 1.083,
+      "step": 193400
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 0.0005,
+      "loss": 1.0814,
+      "step": 193500
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 0.0005,
+      "loss": 1.0638,
+      "step": 193600
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 0.0005,
+      "loss": 1.0634,
+      "step": 193700
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 0.0005,
+      "loss": 1.0851,
+      "step": 193800
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 0.0005,
+      "loss": 1.0631,
+      "step": 193900
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 0.0005,
+      "loss": 1.0872,
+      "step": 194000
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 0.0005,
+      "loss": 1.087,
+      "step": 194100
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 0.0005,
+      "loss": 1.0731,
+      "step": 194200
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 0.0005,
+      "loss": 1.0594,
+      "step": 194300
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 0.0005,
+      "loss": 1.041,
+      "step": 194400
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 0.0005,
+      "loss": 1.0599,
+      "step": 194500
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 0.0005,
+      "loss": 1.0616,
+      "step": 194600
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 0.0005,
+      "loss": 1.0786,
+      "step": 194700
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.0005,
+      "loss": 1.0772,
+      "step": 194800
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.0005,
+      "loss": 1.0876,
+      "step": 194900
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 0.0005,
+      "loss": 1.0908,
+      "step": 195000
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 0.0005,
+      "loss": 1.0711,
+      "step": 195100
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 0.0005,
+      "loss": 1.0557,
+      "step": 195200
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.0005,
+      "loss": 1.0476,
+      "step": 195300
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.0005,
+      "loss": 1.0742,
+      "step": 195400
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 0.0005,
+      "loss": 1.0806,
+      "step": 195500
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 0.0005,
+      "loss": 1.0555,
+      "step": 195600
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.0005,
+      "loss": 1.0759,
+      "step": 195700
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.0005,
+      "loss": 1.0776,
+      "step": 195800
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 0.0005,
+      "loss": 1.0914,
+      "step": 195900
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 0.0005,
+      "loss": 1.0939,
+      "step": 196000
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 0.0005,
+      "loss": 1.0655,
+      "step": 196100
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.0005,
+      "loss": 1.0871,
+      "step": 196200
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.0005,
+      "loss": 1.0839,
+      "step": 196300
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 0.0005,
+      "loss": 1.0851,
+      "step": 196400
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 0.0005,
+      "loss": 1.0742,
+      "step": 196500
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 0.0005,
+      "loss": 1.0946,
+      "step": 196600
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 0.0005,
+      "loss": 1.0919,
+      "step": 196700
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 0.0005,
+      "loss": 1.0594,
+      "step": 196800
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0874,
+      "step": 196900
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 0.0005,
+      "loss": 1.0856,
+      "step": 197000
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0925,
+      "step": 197100
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0887,
+      "step": 197200
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 0.0005,
+      "loss": 1.0745,
+      "step": 197300
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.0005,
+      "loss": 1.0864,
+      "step": 197400
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 0.0005,
+      "loss": 1.0766,
+      "step": 197500
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.0005,
+      "loss": 1.0996,
+      "step": 197600
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.0005,
+      "loss": 1.0983,
+      "step": 197700
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1033,
+      "step": 197800
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1132,
+      "step": 197900
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.0005,
+      "loss": 1.1001,
+      "step": 198000
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.0005,
+      "loss": 1.0912,
+      "step": 198100
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 0.0005,
+      "loss": 1.0939,
+      "step": 198200
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1,
+      "step": 198300
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1019,
+      "step": 198400
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1057,
+      "step": 198500
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1087,
+      "step": 198600
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 0.0005,
+      "loss": 1.0962,
+      "step": 198700
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 0.0005,
+      "loss": 1.0962,
+      "step": 198800
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 0.0005,
+      "loss": 1.0943,
+      "step": 198900
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 0.0005,
+      "loss": 1.0968,
+      "step": 199000
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1118,
+      "step": 199100
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 0.0005,
+      "loss": 1.0882,
+      "step": 199200
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 0.0005,
+      "loss": 1.088,
+      "step": 199300
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 0.0005,
+      "loss": 1.0895,
+      "step": 199400
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 0.0005,
+      "loss": 1.1213,
+      "step": 199500
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 0.0005,
+      "loss": 1.1064,
+      "step": 199600
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1179,
+      "step": 199700
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1243,
+      "step": 199800
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1109,
+      "step": 199900
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1161,
+      "step": 200000
+    },
+    {
+      "epoch": 15.68,
+      "eval_gen_len": 18.732762426061623,
+      "eval_loss": 2.4067530632019043,
+      "eval_rouge1": 35.3601,
+      "eval_rouge2": 14.3396,
+      "eval_rougeL": 29.1605,
+      "eval_rougeLsum": 29.1595,
+      "eval_runtime": 343.0369,
+      "eval_samples_per_second": 33.02,
+      "eval_steps_per_second": 2.064,
+      "step": 200000
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 0.0005,
+      "loss": 1.1047,
+      "step": 200100
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 0.0005,
+      "loss": 1.0996,
+      "step": 200200
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0988,
+      "step": 200300
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 0.0005,
+      "loss": 1.1002,
+      "step": 200400
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 0.0005,
+      "loss": 1.1066,
+      "step": 200500
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 0.0005,
+      "loss": 1.0973,
+      "step": 200600
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.0005,
+      "loss": 1.0974,
+      "step": 200700
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 0.0005,
+      "loss": 1.0902,
+      "step": 200800
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 0.0005,
+      "loss": 1.1074,
+      "step": 200900
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 0.0005,
+      "loss": 1.1035,
+      "step": 201000
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 0.0005,
+      "loss": 1.114,
+      "step": 201100
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 0.0005,
+      "loss": 1.1059,
+      "step": 201200
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 0.0005,
+      "loss": 1.1142,
+      "step": 201300
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0982,
+      "step": 201400
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0976,
+      "step": 201500
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 0.0005,
+      "loss": 1.0827,
+      "step": 201600
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 0.0005,
+      "loss": 1.118,
+      "step": 201700
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 0.0005,
+      "loss": 1.0929,
+      "step": 201800
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.0005,
+      "loss": 1.1065,
+      "step": 201900
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0972,
+      "step": 202000
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1277,
+      "step": 202100
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0867,
+      "step": 202200
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.0005,
+      "loss": 1.1086,
+      "step": 202300
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1308,
+      "step": 202400
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 0.0005,
+      "loss": 1.115,
+      "step": 202500
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 0.0005,
+      "loss": 1.1275,
+      "step": 202600
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1183,
+      "step": 202700
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1048,
+      "step": 202800
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 0.0005,
+      "loss": 1.1195,
+      "step": 202900
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 0.0005,
+      "loss": 1.1114,
+      "step": 203000
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.0005,
+      "loss": 1.1119,
+      "step": 203100
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0991,
+      "step": 203200
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 0.0005,
+      "loss": 1.1211,
+      "step": 203300
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0928,
+      "step": 203400
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 0.0005,
+      "loss": 1.1166,
+      "step": 203500
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 0.0005,
+      "loss": 1.1247,
+      "step": 203600
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 0.0005,
+      "loss": 1.1156,
+      "step": 203700
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 0.0005,
+      "loss": 1.1297,
+      "step": 203800
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0901,
+      "step": 203900
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 0.0005,
+      "loss": 1.1182,
+      "step": 204000
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.0005,
+      "loss": 1.0607,
+      "step": 204100
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.0005,
+      "loss": 1.027,
+      "step": 204200
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 0.0005,
+      "loss": 1.0254,
+      "step": 204300
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 0.0005,
+      "loss": 1.0366,
+      "step": 204400
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.0005,
+      "loss": 1.0353,
+      "step": 204500
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.0005,
+      "loss": 1.03,
+      "step": 204600
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 0.0005,
+      "loss": 1.0267,
+      "step": 204700
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 0.0005,
+      "loss": 1.0589,
+      "step": 204800
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 0.0005,
+      "loss": 1.044,
+      "step": 204900
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.0005,
+      "loss": 1.0634,
+      "step": 205000
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.0005,
+      "loss": 1.0341,
+      "step": 205100
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 0.0005,
+      "loss": 1.0392,
+      "step": 205200
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 0.0005,
+      "loss": 1.0415,
+      "step": 205300
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 0.0005,
+      "loss": 1.0326,
+      "step": 205400
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 0.0005,
+      "loss": 1.0315,
+      "step": 205500
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 0.0005,
+      "loss": 1.05,
+      "step": 205600
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 0.0005,
+      "loss": 1.0281,
+      "step": 205700
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 0.0005,
+      "loss": 1.0348,
+      "step": 205800
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.0005,
+      "loss": 1.0406,
+      "step": 205900
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.0005,
+      "loss": 1.0691,
+      "step": 206000
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 0.0005,
+      "loss": 1.0536,
+      "step": 206100
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 0.0005,
+      "loss": 1.0411,
+      "step": 206200
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 0.0005,
+      "loss": 1.0639,
+      "step": 206300
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.0005,
+      "loss": 1.0356,
+      "step": 206400
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.0005,
+      "loss": 1.07,
+      "step": 206500
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 0.0005,
+      "loss": 1.0525,
+      "step": 206600
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 0.0005,
+      "loss": 1.0434,
+      "step": 206700
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.0005,
+      "loss": 1.0605,
+      "step": 206800
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.0005,
+      "loss": 1.0554,
+      "step": 206900
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 0.0005,
+      "loss": 1.0467,
+      "step": 207000
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 0.0005,
+      "loss": 1.0378,
+      "step": 207100
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 0.0005,
+      "loss": 1.0699,
+      "step": 207200
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.0005,
+      "loss": 1.0577,
+      "step": 207300
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.0005,
+      "loss": 1.0271,
+      "step": 207400
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 0.0005,
+      "loss": 1.0292,
+      "step": 207500
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 0.0005,
+      "loss": 1.0541,
+      "step": 207600
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 0.0005,
+      "loss": 1.0584,
+      "step": 207700
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 0.0005,
+      "loss": 1.056,
+      "step": 207800
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 0.0005,
+      "loss": 1.0629,
+      "step": 207900
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 0.0005,
+      "loss": 1.0629,
+      "step": 208000
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 0.0005,
+      "loss": 1.0636,
+      "step": 208100
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 0.0005,
+      "loss": 1.0589,
+      "step": 208200
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 0.0005,
+      "loss": 1.066,
+      "step": 208300
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 0.0005,
+      "loss": 1.0703,
+      "step": 208400
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 0.0005,
+      "loss": 1.0861,
+      "step": 208500
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 0.0005,
+      "loss": 1.0717,
+      "step": 208600
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.0005,
+      "loss": 1.0534,
+      "step": 208700
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.0005,
+      "loss": 1.0496,
+      "step": 208800
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 0.0005,
+      "loss": 1.0954,
+      "step": 208900
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 0.0005,
+      "loss": 1.0616,
+      "step": 209000
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 0.0005,
+      "loss": 1.0744,
+      "step": 209100
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 0.0005,
+      "loss": 1.0587,
+      "step": 209200
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 0.0005,
+      "loss": 1.0927,
+      "step": 209300
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 0.0005,
+      "loss": 1.0739,
+      "step": 209400
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 0.0005,
+      "loss": 1.0573,
+      "step": 209500
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0688,
+      "step": 209600
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0621,
+      "step": 209700
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 0.0005,
+      "loss": 1.0803,
+      "step": 209800
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0593,
+      "step": 209900
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 0.0005,
+      "loss": 1.0635,
+      "step": 210000
+    },
+    {
+      "epoch": 16.47,
+      "eval_gen_len": 18.72428710161561,
+      "eval_loss": 2.420046806335449,
+      "eval_rouge1": 35.2643,
+      "eval_rouge2": 14.3476,
+      "eval_rougeL": 29.0944,
+      "eval_rougeLsum": 29.0995,
+      "eval_runtime": 342.8229,
+      "eval_samples_per_second": 33.04,
+      "eval_steps_per_second": 2.065,
+      "step": 210000
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 0.0005,
+      "loss": 1.0756,
+      "step": 210100
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 0.0005,
+      "loss": 1.0664,
+      "step": 210200
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 0.0005,
+      "loss": 1.0633,
+      "step": 210300
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 0.0005,
+      "loss": 1.0576,
+      "step": 210400
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 0.0005,
+      "loss": 1.0769,
+      "step": 210500
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 0.0005,
+      "loss": 1.075,
+      "step": 210600
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 0.0005,
+      "loss": 1.0745,
+      "step": 210700
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 0.0005,
+      "loss": 1.0794,
+      "step": 210800
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 0.0005,
+      "loss": 1.0714,
+      "step": 210900
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 0.0005,
+      "loss": 1.0805,
+      "step": 211000
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 0.0005,
+      "loss": 1.0741,
+      "step": 211100
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 0.0005,
+      "loss": 1.0728,
+      "step": 211200
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 0.0005,
+      "loss": 1.068,
+      "step": 211300
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 0.0005,
+      "loss": 1.0635,
+      "step": 211400
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.0005,
+      "loss": 1.065,
+      "step": 211500
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.0005,
+      "loss": 1.0566,
+      "step": 211600
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 0.0005,
+      "loss": 1.0866,
+      "step": 211700
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 0.0005,
+      "loss": 1.0751,
+      "step": 211800
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 0.0005,
+      "loss": 1.0654,
+      "step": 211900
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 0.0005,
+      "loss": 1.0855,
+      "step": 212000
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 0.0005,
+      "loss": 1.0628,
+      "step": 212100
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 0.0005,
+      "loss": 1.0968,
+      "step": 212200
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 0.0005,
+      "loss": 1.063,
+      "step": 212300
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 0.0005,
+      "loss": 1.0699,
+      "step": 212400
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 0.0005,
+      "loss": 1.0878,
+      "step": 212500
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 0.0005,
+      "loss": 1.0845,
+      "step": 212600
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 0.0005,
+      "loss": 1.071,
+      "step": 212700
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 0.0005,
+      "loss": 1.0747,
+      "step": 212800
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.0005,
+      "loss": 1.0946,
+      "step": 212900
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.0005,
+      "loss": 1.0802,
+      "step": 213000
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0749,
+      "step": 213100
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 0.0005,
+      "loss": 1.0836,
+      "step": 213200
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.0005,
+      "loss": 1.077,
+      "step": 213300
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.0005,
+      "loss": 1.1056,
+      "step": 213400
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1031,
+      "step": 213500
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 0.0005,
+      "loss": 1.106,
+      "step": 213600
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 0.0005,
+      "loss": 1.0814,
+      "step": 213700
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0765,
+      "step": 213800
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0846,
+      "step": 213900
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 0.0005,
+      "loss": 1.0821,
+      "step": 214000
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0836,
+      "step": 214100
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0871,
+      "step": 214200
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 0.0005,
+      "loss": 1.0805,
+      "step": 214300
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 0.0005,
+      "loss": 1.08,
+      "step": 214400
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 0.0005,
+      "loss": 1.0934,
+      "step": 214500
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 0.0005,
+      "loss": 1.0884,
+      "step": 214600
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0547,
+      "step": 214700
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0746,
+      "step": 214800
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 0.0005,
+      "loss": 1.079,
+      "step": 214900
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 0.0005,
+      "loss": 1.103,
+      "step": 215000
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 0.0005,
+      "loss": 1.0682,
+      "step": 215100
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1059,
+      "step": 215200
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1022,
+      "step": 215300
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0808,
+      "step": 215400
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1052,
+      "step": 215500
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0847,
+      "step": 215600
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0745,
+      "step": 215700
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.0005,
+      "loss": 1.13,
+      "step": 215800
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0783,
+      "step": 215900
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 0.0005,
+      "loss": 1.0946,
+      "step": 216000
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.0005,
+      "loss": 1.086,
+      "step": 216100
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0999,
+      "step": 216200
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0837,
+      "step": 216300
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0905,
+      "step": 216400
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0939,
+      "step": 216500
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0817,
+      "step": 216600
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 0.0005,
+      "loss": 1.1159,
+      "step": 216700
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 0.0005,
+      "loss": 1.0563,
+      "step": 216800
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 0.0005,
+      "loss": 1.0278,
+      "step": 216900
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 0.0005,
+      "loss": 1.0096,
+      "step": 217000
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 0.0005,
+      "loss": 1.0302,
+      "step": 217100
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 0.0005,
+      "loss": 1.0054,
+      "step": 217200
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 0.0005,
+      "loss": 1.0115,
+      "step": 217300
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 0.0005,
+      "loss": 1.013,
+      "step": 217400
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.0005,
+      "loss": 1.0184,
+      "step": 217500
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.0005,
+      "loss": 1.0257,
+      "step": 217600
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 0.0005,
+      "loss": 0.9953,
+      "step": 217700
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 0.0005,
+      "loss": 1.0177,
+      "step": 217800
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 0.0005,
+      "loss": 1.026,
+      "step": 217900
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.0005,
+      "loss": 1.0229,
+      "step": 218000
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.0005,
+      "loss": 1.0296,
+      "step": 218100
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 0.0005,
+      "loss": 1.0076,
+      "step": 218200
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 0.0005,
+      "loss": 1.0329,
+      "step": 218300
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.0005,
+      "loss": 1.0276,
+      "step": 218400
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.0005,
+      "loss": 1.0306,
+      "step": 218500
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 0.0005,
+      "loss": 1.0196,
+      "step": 218600
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 0.0005,
+      "loss": 1.0203,
+      "step": 218700
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 0.0005,
+      "loss": 1.021,
+      "step": 218800
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 0.0005,
+      "loss": 1.0275,
+      "step": 218900
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 0.0005,
+      "loss": 1.0293,
+      "step": 219000
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 0.0005,
+      "loss": 1.0421,
+      "step": 219100
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 0.0005,
+      "loss": 1.027,
+      "step": 219200
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 0.0005,
+      "loss": 1.0328,
+      "step": 219300
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 0.0005,
+      "loss": 1.0245,
+      "step": 219400
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 0.0005,
+      "loss": 1.0433,
+      "step": 219500
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 0.0005,
+      "loss": 1.0251,
+      "step": 219600
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 0.0005,
+      "loss": 1.0364,
+      "step": 219700
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 0.0005,
+      "loss": 1.0404,
+      "step": 219800
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 0.0005,
+      "loss": 1.0352,
+      "step": 219900
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 0.0005,
+      "loss": 1.0312,
+      "step": 220000
+    },
+    {
+      "epoch": 17.25,
+      "eval_gen_len": 18.717047761984638,
+      "eval_loss": 2.43268084526062,
+      "eval_rouge1": 35.5056,
+      "eval_rouge2": 14.5785,
+      "eval_rougeL": 29.2484,
+      "eval_rougeLsum": 29.2521,
+      "eval_runtime": 343.9116,
+      "eval_samples_per_second": 32.936,
+      "eval_steps_per_second": 2.059,
+      "step": 220000
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 0.0005,
+      "loss": 1.0377,
+      "step": 220100
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 0.0005,
+      "loss": 1.0271,
+      "step": 220200
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 0.0005,
+      "loss": 1.0351,
+      "step": 220300
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 0.0005,
+      "loss": 1.0499,
+      "step": 220400
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 0.0005,
+      "loss": 1.0376,
+      "step": 220500
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 0.0005,
+      "loss": 1.0376,
+      "step": 220600
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.0005,
+      "loss": 1.0462,
+      "step": 220700
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.0005,
+      "loss": 1.024,
+      "step": 220800
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 0.0005,
+      "loss": 1.0334,
+      "step": 220900
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 0.0005,
+      "loss": 1.0291,
+      "step": 221000
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 0.0005,
+      "loss": 1.037,
+      "step": 221100
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 0.0005,
+      "loss": 1.0465,
+      "step": 221200
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 0.0005,
+      "loss": 1.0459,
+      "step": 221300
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 0.0005,
+      "loss": 1.0302,
+      "step": 221400
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 0.0005,
+      "loss": 1.0508,
+      "step": 221500
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 0.0005,
+      "loss": 1.0357,
+      "step": 221600
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 0.0005,
+      "loss": 1.0265,
+      "step": 221700
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 0.0005,
+      "loss": 1.062,
+      "step": 221800
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 0.0005,
+      "loss": 1.0392,
+      "step": 221900
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 0.0005,
+      "loss": 1.055,
+      "step": 222000
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 0.0005,
+      "loss": 1.0305,
+      "step": 222100
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 0.0005,
+      "loss": 1.0596,
+      "step": 222200
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 0.0005,
+      "loss": 1.0147,
+      "step": 222300
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0573,
+      "step": 222400
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 0.0005,
+      "loss": 1.0502,
+      "step": 222500
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0507,
+      "step": 222600
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0257,
+      "step": 222700
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 0.0005,
+      "loss": 1.0551,
+      "step": 222800
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 0.0005,
+      "loss": 1.0579,
+      "step": 222900
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 0.0005,
+      "loss": 1.0459,
+      "step": 223000
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.0005,
+      "loss": 1.041,
+      "step": 223100
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.0005,
+      "loss": 1.0705,
+      "step": 223200
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 0.0005,
+      "loss": 1.0524,
+      "step": 223300
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 0.0005,
+      "loss": 1.0552,
+      "step": 223400
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.0005,
+      "loss": 1.0558,
+      "step": 223500
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.0005,
+      "loss": 1.0466,
+      "step": 223600
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 0.0005,
+      "loss": 1.0459,
+      "step": 223700
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 0.0005,
+      "loss": 1.0556,
+      "step": 223800
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 0.0005,
+      "loss": 1.0623,
+      "step": 223900
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.0005,
+      "loss": 1.0697,
+      "step": 224000
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.0005,
+      "loss": 1.0428,
+      "step": 224100
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 0.0005,
+      "loss": 1.0537,
+      "step": 224200
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 0.0005,
+      "loss": 1.0653,
+      "step": 224300
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 0.0005,
+      "loss": 1.0703,
+      "step": 224400
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.0005,
+      "loss": 1.0649,
+      "step": 224500
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.0005,
+      "loss": 1.062,
+      "step": 224600
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 0.0005,
+      "loss": 1.0796,
+      "step": 224700
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 0.0005,
+      "loss": 1.0423,
+      "step": 224800
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.0005,
+      "loss": 1.039,
+      "step": 224900
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.0005,
+      "loss": 1.0674,
+      "step": 225000
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.0005,
+      "loss": 1.0594,
+      "step": 225100
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 0.0005,
+      "loss": 1.0662,
+      "step": 225200
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 0.0005,
+      "loss": 1.0737,
+      "step": 225300
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.0005,
+      "loss": 1.064,
+      "step": 225400
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.0005,
+      "loss": 1.0769,
+      "step": 225500
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 0.0005,
+      "loss": 1.0544,
+      "step": 225600
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 0.0005,
+      "loss": 1.0469,
+      "step": 225700
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0653,
+      "step": 225800
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0664,
+      "step": 225900
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 0.0005,
+      "loss": 1.0693,
+      "step": 226000
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 0.0005,
+      "loss": 1.0555,
+      "step": 226100
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.0005,
+      "loss": 1.0667,
+      "step": 226200
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 0.0005,
+      "loss": 1.0664,
+      "step": 226300
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 0.0005,
+      "loss": 1.0606,
+      "step": 226400
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 0.0005,
+      "loss": 1.0692,
+      "step": 226500
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0626,
+      "step": 226600
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 0.0005,
+      "loss": 1.059,
+      "step": 226700
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0598,
+      "step": 226800
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0858,
+      "step": 226900
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0791,
+      "step": 227000
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 0.0005,
+      "loss": 1.0832,
+      "step": 227100
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.0005,
+      "loss": 1.064,
+      "step": 227200
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.0005,
+      "loss": 1.0677,
+      "step": 227300
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.0005,
+      "loss": 1.0612,
+      "step": 227400
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 0.0005,
+      "loss": 1.08,
+      "step": 227500
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 0.0005,
+      "loss": 1.0648,
+      "step": 227600
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0848,
+      "step": 227700
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0622,
+      "step": 227800
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 0.0005,
+      "loss": 1.0794,
+      "step": 227900
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.0005,
+      "loss": 1.0737,
+      "step": 228000
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0633,
+      "step": 228100
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 0.0005,
+      "loss": 1.0909,
+      "step": 228200
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 0.0005,
+      "loss": 1.0621,
+      "step": 228300
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0706,
+      "step": 228400
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 0.0005,
+      "loss": 1.0838,
+      "step": 228500
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0868,
+      "step": 228600
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0682,
+      "step": 228700
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 0.0005,
+      "loss": 1.0739,
+      "step": 228800
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0755,
+      "step": 228900
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0702,
+      "step": 229000
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0946,
+      "step": 229100
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0562,
+      "step": 229200
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0592,
+      "step": 229300
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0799,
+      "step": 229400
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 0.0005,
+      "loss": 1.0777,
+      "step": 229500
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 0.0005,
+      "loss": 1.0075,
+      "step": 229600
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 0.0005,
+      "loss": 0.9877,
+      "step": 229700
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 0.0005,
+      "loss": 0.9947,
+      "step": 229800
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 0.0005,
+      "loss": 1.0062,
+      "step": 229900
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.0005,
+      "loss": 0.9993,
+      "step": 230000
+    },
+    {
+      "epoch": 18.04,
+      "eval_gen_len": 18.69480003531385,
+      "eval_loss": 2.4594662189483643,
+      "eval_rouge1": 35.4197,
+      "eval_rouge2": 14.3682,
+      "eval_rougeL": 29.0924,
+      "eval_rougeLsum": 29.0945,
+      "eval_runtime": 342.3949,
+      "eval_samples_per_second": 33.082,
+      "eval_steps_per_second": 2.068,
+      "step": 230000
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.0005,
+      "loss": 0.9855,
+      "step": 230100
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 0.0005,
+      "loss": 1.0019,
+      "step": 230200
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 0.0005,
+      "loss": 1.008,
+      "step": 230300
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 0.0005,
+      "loss": 0.9993,
+      "step": 230400
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 0.0005,
+      "loss": 1.0155,
+      "step": 230500
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 0.0005,
+      "loss": 1.0059,
+      "step": 230600
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 0.0005,
+      "loss": 1.0119,
+      "step": 230700
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 0.0005,
+      "loss": 1.0197,
+      "step": 230800
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 0.0005,
+      "loss": 0.9936,
+      "step": 230900
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 0.0005,
+      "loss": 1.0079,
+      "step": 231000
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 0.0005,
+      "loss": 1.0032,
+      "step": 231100
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.0005,
+      "loss": 1.0169,
+      "step": 231200
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 0.0005,
+      "loss": 1.0071,
+      "step": 231300
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 0.0005,
+      "loss": 0.9963,
+      "step": 231400
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 0.0005,
+      "loss": 1.0187,
+      "step": 231500
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 0.0005,
+      "loss": 0.999,
+      "step": 231600
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 0.0005,
+      "loss": 1.0289,
+      "step": 231700
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 0.0005,
+      "loss": 1.0194,
+      "step": 231800
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 0.0005,
+      "loss": 1.0131,
+      "step": 231900
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 0.0005,
+      "loss": 1.0165,
+      "step": 232000
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 0.0005,
+      "loss": 0.99,
+      "step": 232100
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 0.0005,
+      "loss": 1.027,
+      "step": 232200
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.0005,
+      "loss": 1.0241,
+      "step": 232300
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.0005,
+      "loss": 0.9904,
+      "step": 232400
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 0.0005,
+      "loss": 1.0165,
+      "step": 232500
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 0.0005,
+      "loss": 0.9947,
+      "step": 232600
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 0.0005,
+      "loss": 1.0217,
+      "step": 232700
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 0.0005,
+      "loss": 1.0232,
+      "step": 232800
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 0.0005,
+      "loss": 1.017,
+      "step": 232900
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 0.0005,
+      "loss": 1.0289,
+      "step": 233000
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 0.0005,
+      "loss": 1.011,
+      "step": 233100
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 0.0005,
+      "loss": 1.0222,
+      "step": 233200
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.0005,
+      "loss": 1.0305,
+      "step": 233300
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.0005,
+      "loss": 1.0203,
+      "step": 233400
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.0005,
+      "loss": 1.0057,
+      "step": 233500
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 0.0005,
+      "loss": 1.0363,
+      "step": 233600
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 0.0005,
+      "loss": 1.0223,
+      "step": 233700
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 0.0005,
+      "loss": 1.0201,
+      "step": 233800
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 0.0005,
+      "loss": 1.0183,
+      "step": 233900
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 0.0005,
+      "loss": 1.0363,
+      "step": 234000
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 0.0005,
+      "loss": 1.0114,
+      "step": 234100
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.0005,
+      "loss": 1.0234,
+      "step": 234200
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.0005,
+      "loss": 1.0166,
+      "step": 234300
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 0.0005,
+      "loss": 1.053,
+      "step": 234400
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 0.0005,
+      "loss": 1.0298,
+      "step": 234500
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.0005,
+      "loss": 1.0401,
+      "step": 234600
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.0005,
+      "loss": 1.0265,
+      "step": 234700
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 0.0005,
+      "loss": 1.0171,
+      "step": 234800
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 0.0005,
+      "loss": 1.0059,
+      "step": 234900
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 0.0005,
+      "loss": 1.043,
+      "step": 235000
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0012,
+      "step": 235100
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0426,
+      "step": 235200
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 0.0005,
+      "loss": 1.0278,
+      "step": 235300
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0228,
+      "step": 235400
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 0.0005,
+      "loss": 1.0167,
+      "step": 235500
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.0005,
+      "loss": 1.0007,
+      "step": 235600
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.0005,
+      "loss": 1.0293,
+      "step": 235700
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 0.0005,
+      "loss": 1.0294,
+      "step": 235800
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 0.0005,
+      "loss": 1.0245,
+      "step": 235900
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.0005,
+      "loss": 1.0381,
+      "step": 236000
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.0005,
+      "loss": 1.0439,
+      "step": 236100
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 0.0005,
+      "loss": 1.0417,
+      "step": 236200
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 0.0005,
+      "loss": 1.0362,
+      "step": 236300
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 0.0005,
+      "loss": 1.035,
+      "step": 236400
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.0005,
+      "loss": 1.0478,
+      "step": 236500
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.0005,
+      "loss": 1.023,
+      "step": 236600
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 0.0005,
+      "loss": 1.0543,
+      "step": 236700
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 0.0005,
+      "loss": 1.0427,
+      "step": 236800
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 0.0005,
+      "loss": 1.0494,
+      "step": 236900
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.0005,
+      "loss": 1.0253,
+      "step": 237000
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.0005,
+      "loss": 1.074,
+      "step": 237100
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 0.0005,
+      "loss": 1.043,
+      "step": 237200
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 0.0005,
+      "loss": 1.0405,
+      "step": 237300
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 0.0005,
+      "loss": 1.0566,
+      "step": 237400
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 0.0005,
+      "loss": 1.037,
+      "step": 237500
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 0.0005,
+      "loss": 1.0321,
+      "step": 237600
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 0.0005,
+      "loss": 1.024,
+      "step": 237700
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 0.0005,
+      "loss": 1.0432,
+      "step": 237800
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.0005,
+      "loss": 1.0468,
+      "step": 237900
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.0005,
+      "loss": 1.0686,
+      "step": 238000
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 0.0005,
+      "loss": 1.0365,
+      "step": 238100
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 0.0005,
+      "loss": 1.0464,
+      "step": 238200
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 0.0005,
+      "loss": 1.0473,
+      "step": 238300
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.0005,
+      "loss": 1.0364,
+      "step": 238400
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.0005,
+      "loss": 1.0441,
+      "step": 238500
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0376,
+      "step": 238600
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 0.0005,
+      "loss": 1.0514,
+      "step": 238700
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.0005,
+      "loss": 1.076,
+      "step": 238800
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.0005,
+      "loss": 1.0455,
+      "step": 238900
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 0.0005,
+      "loss": 1.0466,
+      "step": 239000
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 0.0005,
+      "loss": 1.0354,
+      "step": 239100
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 0.0005,
+      "loss": 1.0522,
+      "step": 239200
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0421,
+      "step": 239300
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0795,
+      "step": 239400
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 0.0005,
+      "loss": 1.064,
+      "step": 239500
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0534,
+      "step": 239600
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0589,
+      "step": 239700
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0534,
+      "step": 239800
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 0.0005,
+      "loss": 1.0576,
+      "step": 239900
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 0.0005,
+      "loss": 1.0384,
+      "step": 240000
+    },
+    {
+      "epoch": 18.82,
+      "eval_gen_len": 18.7365586651364,
+      "eval_loss": 2.430091142654419,
+      "eval_rouge1": 35.7684,
+      "eval_rouge2": 14.6816,
+      "eval_rougeL": 29.457,
+      "eval_rougeLsum": 29.4604,
+      "eval_runtime": 342.3499,
+      "eval_samples_per_second": 33.086,
+      "eval_steps_per_second": 2.068,
+      "step": 240000
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 0.0005,
+      "loss": 1.035,
+      "step": 240100
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0457,
+      "step": 240200
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0503,
+      "step": 240300
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 0.0005,
+      "loss": 1.0483,
+      "step": 240400
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0436,
+      "step": 240500
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 0.0005,
+      "loss": 1.0271,
+      "step": 240600
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 0.0005,
+      "loss": 1.0503,
+      "step": 240700
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 0.0005,
+      "loss": 1.0274,
+      "step": 240800
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0678,
+      "step": 240900
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 0.0005,
+      "loss": 1.0538,
+      "step": 241000
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0525,
+      "step": 241100
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0766,
+      "step": 241200
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 0.0005,
+      "loss": 1.0735,
+      "step": 241300
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0399,
+      "step": 241400
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 0.0005,
+      "loss": 1.058,
+      "step": 241500
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0276,
+      "step": 241600
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0593,
+      "step": 241700
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0349,
+      "step": 241800
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0605,
+      "step": 241900
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0536,
+      "step": 242000
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 0.0005,
+      "loss": 1.075,
+      "step": 242100
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0602,
+      "step": 242200
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 0.0005,
+      "loss": 1.0765,
+      "step": 242300
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 0.0005,
+      "loss": 0.9739,
+      "step": 242400
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 0.0005,
+      "loss": 0.9734,
+      "step": 242500
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 0.0005,
+      "loss": 0.9762,
+      "step": 242600
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 0.0005,
+      "loss": 0.941,
+      "step": 242700
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 0.0005,
+      "loss": 1.0016,
+      "step": 242800
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 0.0005,
+      "loss": 1.0044,
+      "step": 242900
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 0.0005,
+      "loss": 0.9789,
+      "step": 243000
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 0.0005,
+      "loss": 0.9885,
+      "step": 243100
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 0.0005,
+      "loss": 0.9731,
+      "step": 243200
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 0.0005,
+      "loss": 0.9897,
+      "step": 243300
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 0.0005,
+      "loss": 0.9879,
+      "step": 243400
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 0.0005,
+      "loss": 0.9734,
+      "step": 243500
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 0.0005,
+      "loss": 0.9783,
+      "step": 243600
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 0.0005,
+      "loss": 0.9865,
+      "step": 243700
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 0.0005,
+      "loss": 0.9979,
+      "step": 243800
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 0.0005,
+      "loss": 1.0051,
+      "step": 243900
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 0.0005,
+      "loss": 1.0012,
+      "step": 244000
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 0.0005,
+      "loss": 0.9992,
+      "step": 244100
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 0.0005,
+      "loss": 1.0128,
+      "step": 244200
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 0.0005,
+      "loss": 0.9674,
+      "step": 244300
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 0.0005,
+      "loss": 0.9806,
+      "step": 244400
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 0.0005,
+      "loss": 1.008,
+      "step": 244500
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 0.0005,
+      "loss": 0.9852,
+      "step": 244600
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 0.0005,
+      "loss": 1.0044,
+      "step": 244700
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 0.0005,
+      "loss": 0.9852,
+      "step": 244800
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 0.0005,
+      "loss": 1.0055,
+      "step": 244900
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 0.0005,
+      "loss": 1.0117,
+      "step": 245000
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 0.0005,
+      "loss": 1.005,
+      "step": 245100
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 0.0005,
+      "loss": 0.9853,
+      "step": 245200
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 0.0005,
+      "loss": 1.0087,
+      "step": 245300
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 0.0005,
+      "loss": 0.992,
+      "step": 245400
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 0.0005,
+      "loss": 1.013,
+      "step": 245500
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 0.0005,
+      "loss": 0.9937,
+      "step": 245600
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 0.0005,
+      "loss": 1.0154,
+      "step": 245700
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 0.0005,
+      "loss": 0.9788,
+      "step": 245800
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 0.0005,
+      "loss": 1.0055,
+      "step": 245900
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 0.0005,
+      "loss": 1.0161,
+      "step": 246000
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 0.0005,
+      "loss": 1.0108,
+      "step": 246100
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9975,
+      "step": 246200
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9924,
+      "step": 246300
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 0.0005,
+      "loss": 0.9959,
+      "step": 246400
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 0.0005,
+      "loss": 0.9903,
+      "step": 246500
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 0.0005,
+      "loss": 1.0071,
+      "step": 246600
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 0.0005,
+      "loss": 1.011,
+      "step": 246700
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 0.0005,
+      "loss": 1.0049,
+      "step": 246800
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 0.0005,
+      "loss": 1.013,
+      "step": 246900
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 0.0005,
+      "loss": 1.0137,
+      "step": 247000
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 0.0005,
+      "loss": 1.0292,
+      "step": 247100
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.0005,
+      "loss": 1.0282,
+      "step": 247200
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.0005,
+      "loss": 1.0087,
+      "step": 247300
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 0.0005,
+      "loss": 1.0116,
+      "step": 247400
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 0.0005,
+      "loss": 1.0191,
+      "step": 247500
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 0.0005,
+      "loss": 1.0034,
+      "step": 247600
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 0.0005,
+      "loss": 0.9946,
+      "step": 247700
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 0.0005,
+      "loss": 1.0317,
+      "step": 247800
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0143,
+      "step": 247900
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 0.0005,
+      "loss": 1.0393,
+      "step": 248000
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0203,
+      "step": 248100
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0123,
+      "step": 248200
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 0.0005,
+      "loss": 1.0127,
+      "step": 248300
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 0.0005,
+      "loss": 1.017,
+      "step": 248400
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 0.0005,
+      "loss": 1.0104,
+      "step": 248500
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 0.0005,
+      "loss": 1.0166,
+      "step": 248600
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 0.0005,
+      "loss": 1.0118,
+      "step": 248700
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9998,
+      "step": 248800
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 0.0005,
+      "loss": 1.0286,
+      "step": 248900
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 0.0005,
+      "loss": 1.0186,
+      "step": 249000
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 0.0005,
+      "loss": 1.0343,
+      "step": 249100
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 0.0005,
+      "loss": 1.0138,
+      "step": 249200
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 0.0005,
+      "loss": 1.0173,
+      "step": 249300
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 0.0005,
+      "loss": 1.0029,
+      "step": 249400
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.0005,
+      "loss": 0.9903,
+      "step": 249500
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.0005,
+      "loss": 1.0233,
+      "step": 249600
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 0.0005,
+      "loss": 1.0258,
+      "step": 249700
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 0.0005,
+      "loss": 1.0016,
+      "step": 249800
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 0.0005,
+      "loss": 1.0227,
+      "step": 249900
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 0.0005,
+      "loss": 1.0359,
+      "step": 250000
+    },
+    {
+      "epoch": 19.6,
+      "eval_gen_len": 18.751302198287277,
+      "eval_loss": 2.4243288040161133,
+      "eval_rouge1": 35.3752,
+      "eval_rouge2": 14.4405,
+      "eval_rougeL": 29.2303,
+      "eval_rougeLsum": 29.2273,
+      "eval_runtime": 342.217,
+      "eval_samples_per_second": 33.099,
+      "eval_steps_per_second": 2.069,
+      "step": 250000
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 0.0005,
+      "loss": 1.0146,
+      "step": 250100
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 0.0005,
+      "loss": 1.0322,
+      "step": 250200
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 0.0005,
+      "loss": 1.0361,
+      "step": 250300
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 0.0005,
+      "loss": 1.0092,
+      "step": 250400
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 0.0005,
+      "loss": 1.0137,
+      "step": 250500
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 0.0005,
+      "loss": 1.0234,
+      "step": 250600
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 0.0005,
+      "loss": 1.0318,
+      "step": 250700
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 0.0005,
+      "loss": 1.0336,
+      "step": 250800
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 0.0005,
+      "loss": 1.0132,
+      "step": 250900
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 0.0005,
+      "loss": 1.0267,
+      "step": 251000
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 0.0005,
+      "loss": 1.018,
+      "step": 251100
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 0.0005,
+      "loss": 1.0231,
+      "step": 251200
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0028,
+      "step": 251300
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0123,
+      "step": 251400
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 0.0005,
+      "loss": 1.0257,
+      "step": 251500
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 0.0005,
+      "loss": 1.013,
+      "step": 251600
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 0.0005,
+      "loss": 1.0356,
+      "step": 251700
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 0.0005,
+      "loss": 1.0293,
+      "step": 251800
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 0.0005,
+      "loss": 1.0391,
+      "step": 251900
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 0.0005,
+      "loss": 1.0325,
+      "step": 252000
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0476,
+      "step": 252100
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 0.0005,
+      "loss": 1.0539,
+      "step": 252200
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0288,
+      "step": 252300
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0317,
+      "step": 252400
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0352,
+      "step": 252500
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 0.0005,
+      "loss": 1.0252,
+      "step": 252600
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 0.0005,
+      "loss": 1.0328,
+      "step": 252700
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 0.0005,
+      "loss": 1.0312,
+      "step": 252800
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 0.0005,
+      "loss": 1.0343,
+      "step": 252900
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0557,
+      "step": 253000
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 0.0005,
+      "loss": 1.019,
+      "step": 253100
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0414,
+      "step": 253200
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0328,
+      "step": 253300
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 0.0005,
+      "loss": 1.0358,
+      "step": 253400
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 0.0005,
+      "loss": 1.0557,
+      "step": 253500
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0232,
+      "step": 253600
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0468,
+      "step": 253700
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 0.0005,
+      "loss": 1.0587,
+      "step": 253800
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0396,
+      "step": 253900
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 0.0005,
+      "loss": 1.0713,
+      "step": 254000
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0377,
+      "step": 254100
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0257,
+      "step": 254200
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 0.0005,
+      "loss": 1.0425,
+      "step": 254300
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0259,
+      "step": 254400
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0423,
+      "step": 254500
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0305,
+      "step": 254600
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0248,
+      "step": 254700
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0344,
+      "step": 254800
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0227,
+      "step": 254900
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.0005,
+      "loss": 1.0442,
+      "step": 255000
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9903,
+      "step": 255100
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 0.0005,
+      "loss": 0.9586,
+      "step": 255200
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 0.0005,
+      "loss": 0.9437,
+      "step": 255300
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 0.0005,
+      "loss": 0.9741,
+      "step": 255400
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 0.0005,
+      "loss": 0.9624,
+      "step": 255500
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 0.0005,
+      "loss": 0.9602,
+      "step": 255600
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 0.0005,
+      "loss": 0.9789,
+      "step": 255700
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 0.0005,
+      "loss": 0.9414,
+      "step": 255800
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 0.0005,
+      "loss": 0.9911,
+      "step": 255900
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 0.0005,
+      "loss": 0.9909,
+      "step": 256000
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 0.0005,
+      "loss": 0.9742,
+      "step": 256100
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 0.0005,
+      "loss": 0.9853,
+      "step": 256200
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 0.0005,
+      "loss": 0.9748,
+      "step": 256300
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 0.0005,
+      "loss": 0.9825,
+      "step": 256400
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 0.0005,
+      "loss": 0.9797,
+      "step": 256500
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 0.0005,
+      "loss": 0.9804,
+      "step": 256600
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 0.0005,
+      "loss": 0.9567,
+      "step": 256700
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 0.0005,
+      "loss": 0.9746,
+      "step": 256800
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 0.0005,
+      "loss": 0.9547,
+      "step": 256900
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 0.0005,
+      "loss": 0.9787,
+      "step": 257000
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 0.0005,
+      "loss": 0.9768,
+      "step": 257100
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 0.0005,
+      "loss": 0.9795,
+      "step": 257200
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 0.0005,
+      "loss": 0.9594,
+      "step": 257300
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 0.0005,
+      "loss": 0.9685,
+      "step": 257400
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 0.0005,
+      "loss": 0.9883,
+      "step": 257500
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 0.0005,
+      "loss": 0.9843,
+      "step": 257600
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 0.0005,
+      "loss": 0.9766,
+      "step": 257700
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 0.0005,
+      "loss": 0.9899,
+      "step": 257800
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 0.0005,
+      "loss": 0.9939,
+      "step": 257900
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 0.0005,
+      "loss": 0.9716,
+      "step": 258000
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 0.0005,
+      "loss": 0.9831,
+      "step": 258100
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 0.0005,
+      "loss": 0.9773,
+      "step": 258200
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 0.0005,
+      "loss": 0.9931,
+      "step": 258300
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 0.0005,
+      "loss": 0.9704,
+      "step": 258400
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 0.0005,
+      "loss": 0.9973,
+      "step": 258500
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 0.0005,
+      "loss": 0.972,
+      "step": 258600
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 0.0005,
+      "loss": 0.9754,
+      "step": 258700
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 0.0005,
+      "loss": 0.9839,
+      "step": 258800
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 0.0005,
+      "loss": 0.9971,
+      "step": 258900
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9784,
+      "step": 259000
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 0.0005,
+      "loss": 0.9969,
+      "step": 259100
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 0.0005,
+      "loss": 0.974,
+      "step": 259200
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 0.0005,
+      "loss": 0.9913,
+      "step": 259300
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 0.0005,
+      "loss": 1.0049,
+      "step": 259400
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 0.0005,
+      "loss": 0.9758,
+      "step": 259500
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 0.0005,
+      "loss": 0.9923,
+      "step": 259600
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 0.0005,
+      "loss": 0.9768,
+      "step": 259700
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 0.0005,
+      "loss": 0.9961,
+      "step": 259800
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 0.0005,
+      "loss": 0.9883,
+      "step": 259900
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 0.0005,
+      "loss": 0.9967,
+      "step": 260000
+    },
+    {
+      "epoch": 20.39,
+      "eval_gen_len": 18.706630175686414,
+      "eval_loss": 2.452528476715088,
+      "eval_rouge1": 35.327,
+      "eval_rouge2": 14.3972,
+      "eval_rougeL": 29.2321,
+      "eval_rougeLsum": 29.2316,
+      "eval_runtime": 343.2241,
+      "eval_samples_per_second": 33.002,
+      "eval_steps_per_second": 2.063,
+      "step": 260000
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9936,
+      "step": 260100
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9856,
+      "step": 260200
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 0.0005,
+      "loss": 0.9917,
+      "step": 260300
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 0.0005,
+      "loss": 0.9936,
+      "step": 260400
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 0.0005,
+      "loss": 0.9985,
+      "step": 260500
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0116,
+      "step": 260600
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0058,
+      "step": 260700
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 0.0005,
+      "loss": 0.998,
+      "step": 260800
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0119,
+      "step": 260900
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 0.0005,
+      "loss": 1.0024,
+      "step": 261000
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 0.0005,
+      "loss": 0.9831,
+      "step": 261100
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 0.0005,
+      "loss": 0.998,
+      "step": 261200
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 0.0005,
+      "loss": 1.0103,
+      "step": 261300
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 0.0005,
+      "loss": 0.9915,
+      "step": 261400
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9759,
+      "step": 261500
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9999,
+      "step": 261600
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 0.0005,
+      "loss": 0.9837,
+      "step": 261700
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 0.0005,
+      "loss": 1.0255,
+      "step": 261800
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 0.0005,
+      "loss": 0.9982,
+      "step": 261900
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 0.0005,
+      "loss": 1.0194,
+      "step": 262000
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 0.0005,
+      "loss": 0.9831,
+      "step": 262100
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 0.0005,
+      "loss": 1.0041,
+      "step": 262200
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 0.0005,
+      "loss": 1.0116,
+      "step": 262300
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9942,
+      "step": 262400
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 0.0005,
+      "loss": 1.0259,
+      "step": 262500
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 0.0005,
+      "loss": 1.0009,
+      "step": 262600
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 0.0005,
+      "loss": 0.9986,
+      "step": 262700
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 0.0005,
+      "loss": 0.9902,
+      "step": 262800
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 0.0005,
+      "loss": 1.0143,
+      "step": 262900
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 0.0005,
+      "loss": 1.0043,
+      "step": 263000
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 0.0005,
+      "loss": 0.9882,
+      "step": 263100
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 0.0005,
+      "loss": 1.0132,
+      "step": 263200
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9972,
+      "step": 263300
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 0.0005,
+      "loss": 1.0014,
+      "step": 263400
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 0.0005,
+      "loss": 1.0263,
+      "step": 263500
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 0.0005,
+      "loss": 0.9992,
+      "step": 263600
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 0.0005,
+      "loss": 1.0181,
+      "step": 263700
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 0.0005,
+      "loss": 1.0116,
+      "step": 263800
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 0.0005,
+      "loss": 1.0202,
+      "step": 263900
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 0.0005,
+      "loss": 1.0037,
+      "step": 264000
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0129,
+      "step": 264100
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 0.0005,
+      "loss": 0.988,
+      "step": 264200
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 0.0005,
+      "loss": 1.0108,
+      "step": 264300
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 0.0005,
+      "loss": 1.007,
+      "step": 264400
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 0.0005,
+      "loss": 1.0244,
+      "step": 264500
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 0.0005,
+      "loss": 1.006,
+      "step": 264600
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 0.0005,
+      "loss": 1.0086,
+      "step": 264700
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0056,
+      "step": 264800
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 0.0005,
+      "loss": 0.9948,
+      "step": 264900
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9973,
+      "step": 265000
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0202,
+      "step": 265100
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0289,
+      "step": 265200
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 0.0005,
+      "loss": 1.004,
+      "step": 265300
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 0.0005,
+      "loss": 1.0016,
+      "step": 265400
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 0.0005,
+      "loss": 1.0172,
+      "step": 265500
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 0.0005,
+      "loss": 1.0211,
+      "step": 265600
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 0.0005,
+      "loss": 1.022,
+      "step": 265700
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0195,
+      "step": 265800
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 0.0005,
+      "loss": 1.0249,
+      "step": 265900
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0184,
+      "step": 266000
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 0.0005,
+      "loss": 1.0103,
+      "step": 266100
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 0.0005,
+      "loss": 1.0382,
+      "step": 266200
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 0.0005,
+      "loss": 1.01,
+      "step": 266300
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0226,
+      "step": 266400
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 0.0005,
+      "loss": 1.0214,
+      "step": 266500
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0136,
+      "step": 266600
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0151,
+      "step": 266700
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 0.0005,
+      "loss": 1.037,
+      "step": 266800
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0292,
+      "step": 266900
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 0.0005,
+      "loss": 1.0106,
+      "step": 267000
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0214,
+      "step": 267100
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0186,
+      "step": 267200
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0375,
+      "step": 267300
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0222,
+      "step": 267400
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0314,
+      "step": 267500
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0544,
+      "step": 267600
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0301,
+      "step": 267700
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 0.0005,
+      "loss": 1.0324,
+      "step": 267800
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 0.0005,
+      "loss": 0.9279,
+      "step": 267900
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 0.0005,
+      "loss": 0.9454,
+      "step": 268000
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 0.0005,
+      "loss": 0.9487,
+      "step": 268100
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 0.0005,
+      "loss": 0.9441,
+      "step": 268200
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 0.0005,
+      "loss": 0.9371,
+      "step": 268300
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 0.0005,
+      "loss": 0.9406,
+      "step": 268400
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 0.0005,
+      "loss": 0.9463,
+      "step": 268500
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 0.0005,
+      "loss": 0.9527,
+      "step": 268600
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 0.0005,
+      "loss": 0.9422,
+      "step": 268700
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 0.0005,
+      "loss": 0.9628,
+      "step": 268800
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 0.0005,
+      "loss": 0.9451,
+      "step": 268900
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 0.0005,
+      "loss": 0.9577,
+      "step": 269000
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 0.0005,
+      "loss": 0.9565,
+      "step": 269100
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 0.0005,
+      "loss": 0.9538,
+      "step": 269200
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 0.0005,
+      "loss": 0.9585,
+      "step": 269300
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 0.0005,
+      "loss": 0.9591,
+      "step": 269400
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 0.0005,
+      "loss": 0.9521,
+      "step": 269500
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 0.0005,
+      "loss": 0.9673,
+      "step": 269600
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 0.0005,
+      "loss": 0.9683,
+      "step": 269700
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 0.0005,
+      "loss": 0.9678,
+      "step": 269800
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 0.0005,
+      "loss": 0.9586,
+      "step": 269900
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 0.0005,
+      "loss": 0.9795,
+      "step": 270000
+    },
+    {
+      "epoch": 21.17,
+      "eval_gen_len": 18.693564050498807,
+      "eval_loss": 2.467229127883911,
+      "eval_rouge1": 35.4965,
+      "eval_rouge2": 14.6108,
+      "eval_rougeL": 29.35,
+      "eval_rougeLsum": 29.3506,
+      "eval_runtime": 343.7678,
+      "eval_samples_per_second": 32.95,
+      "eval_steps_per_second": 2.06,
+      "step": 270000
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 0.0005,
+      "loss": 0.9695,
+      "step": 270100
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 0.0005,
+      "loss": 0.9805,
+      "step": 270200
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 0.0005,
+      "loss": 0.9727,
+      "step": 270300
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 0.0005,
+      "loss": 0.9616,
+      "step": 270400
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 0.0005,
+      "loss": 0.9516,
+      "step": 270500
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 0.0005,
+      "loss": 0.9587,
+      "step": 270600
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 0.0005,
+      "loss": 0.9734,
+      "step": 270700
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 0.0005,
+      "loss": 0.986,
+      "step": 270800
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 0.0005,
+      "loss": 0.9704,
+      "step": 270900
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 0.0005,
+      "loss": 0.9682,
+      "step": 271000
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 0.0005,
+      "loss": 0.9564,
+      "step": 271100
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 0.0005,
+      "loss": 0.9675,
+      "step": 271200
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 0.0005,
+      "loss": 0.9716,
+      "step": 271300
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 0.0005,
+      "loss": 0.98,
+      "step": 271400
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 0.0005,
+      "loss": 0.9725,
+      "step": 271500
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 0.0005,
+      "loss": 0.9896,
+      "step": 271600
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9696,
+      "step": 271700
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9725,
+      "step": 271800
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 0.0005,
+      "loss": 0.9607,
+      "step": 271900
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 0.0005,
+      "loss": 0.9783,
+      "step": 272000
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 0.0005,
+      "loss": 0.9735,
+      "step": 272100
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 0.0005,
+      "loss": 0.9913,
+      "step": 272200
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 0.0005,
+      "loss": 0.97,
+      "step": 272300
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 0.0005,
+      "loss": 0.9813,
+      "step": 272400
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 0.0005,
+      "loss": 0.9614,
+      "step": 272500
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 0.0005,
+      "loss": 0.9659,
+      "step": 272600
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 0.0005,
+      "loss": 0.9846,
+      "step": 272700
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 0.0005,
+      "loss": 0.9823,
+      "step": 272800
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9821,
+      "step": 272900
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 0.0005,
+      "loss": 0.9821,
+      "step": 273000
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 0.0005,
+      "loss": 0.9652,
+      "step": 273100
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 0.0005,
+      "loss": 0.998,
+      "step": 273200
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 0.0005,
+      "loss": 0.9841,
+      "step": 273300
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 0.0005,
+      "loss": 0.9719,
+      "step": 273400
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 0.0005,
+      "loss": 0.9823,
+      "step": 273500
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 0.0005,
+      "loss": 0.9653,
+      "step": 273600
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 0.0005,
+      "loss": 0.9888,
+      "step": 273700
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 0.0005,
+      "loss": 0.985,
+      "step": 273800
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 0.0005,
+      "loss": 0.9825,
+      "step": 273900
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 0.0005,
+      "loss": 0.9731,
+      "step": 274000
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 0.0005,
+      "loss": 0.973,
+      "step": 274100
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 0.0005,
+      "loss": 0.9733,
+      "step": 274200
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9774,
+      "step": 274300
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 0.0005,
+      "loss": 0.9793,
+      "step": 274400
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 0.0005,
+      "loss": 0.9911,
+      "step": 274500
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 0.0005,
+      "loss": 0.9896,
+      "step": 274600
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 0.0005,
+      "loss": 0.9881,
+      "step": 274700
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 0.0005,
+      "loss": 0.9874,
+      "step": 274800
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 0.0005,
+      "loss": 0.9969,
+      "step": 274900
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 0.0005,
+      "loss": 0.9889,
+      "step": 275000
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 0.0005,
+      "loss": 0.9803,
+      "step": 275100
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9862,
+      "step": 275200
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 0.0005,
+      "loss": 1.0038,
+      "step": 275300
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 0.0005,
+      "loss": 0.9891,
+      "step": 275400
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 0.0005,
+      "loss": 0.9918,
+      "step": 275500
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 0.0005,
+      "loss": 1.0091,
+      "step": 275600
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 0.0005,
+      "loss": 0.9937,
+      "step": 275700
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 0.0005,
+      "loss": 0.9942,
+      "step": 275800
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 0.0005,
+      "loss": 0.9806,
+      "step": 275900
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 0.0005,
+      "loss": 0.9961,
+      "step": 276000
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9939,
+      "step": 276100
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 0.0005,
+      "loss": 1.003,
+      "step": 276200
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 0.0005,
+      "loss": 0.9841,
+      "step": 276300
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 0.0005,
+      "loss": 0.9813,
+      "step": 276400
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 0.0005,
+      "loss": 0.9821,
+      "step": 276500
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9906,
+      "step": 276600
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 0.0005,
+      "loss": 0.999,
+      "step": 276700
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0202,
+      "step": 276800
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 0.0005,
+      "loss": 0.9839,
+      "step": 276900
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 0.0005,
+      "loss": 0.9964,
+      "step": 277000
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 0.0005,
+      "loss": 1.0038,
+      "step": 277100
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 0.0005,
+      "loss": 1.0373,
+      "step": 277200
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 0.0005,
+      "loss": 0.9909,
+      "step": 277300
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 0.0005,
+      "loss": 1.002,
+      "step": 277400
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9934,
+      "step": 277500
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 0.0005,
+      "loss": 0.9949,
+      "step": 277600
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9884,
+      "step": 277700
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 0.0005,
+      "loss": 1.0014,
+      "step": 277800
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 0.0005,
+      "loss": 1.005,
+      "step": 277900
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0149,
+      "step": 278000
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 0.0005,
+      "loss": 0.9846,
+      "step": 278100
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 0.0005,
+      "loss": 1.0009,
+      "step": 278200
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 0.0005,
+      "loss": 1.0093,
+      "step": 278300
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 0.0005,
+      "loss": 0.9975,
+      "step": 278400
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0008,
+      "step": 278500
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 0.0005,
+      "loss": 0.9921,
+      "step": 278600
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0073,
+      "step": 278700
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 0.0005,
+      "loss": 0.9945,
+      "step": 278800
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 0.0005,
+      "loss": 0.9944,
+      "step": 278900
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 0.0005,
+      "loss": 1.0061,
+      "step": 279000
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0033,
+      "step": 279100
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0135,
+      "step": 279200
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 0.0005,
+      "loss": 1.0141,
+      "step": 279300
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9956,
+      "step": 279400
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 0.0005,
+      "loss": 0.9999,
+      "step": 279500
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0112,
+      "step": 279600
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 0.0005,
+      "loss": 0.987,
+      "step": 279700
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 0.0005,
+      "loss": 0.9968,
+      "step": 279800
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0061,
+      "step": 279900
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0227,
+      "step": 280000
+    },
+    {
+      "epoch": 21.96,
+      "eval_gen_len": 18.737794649951443,
+      "eval_loss": 2.455573081970215,
+      "eval_rouge1": 35.5589,
+      "eval_rouge2": 14.6062,
+      "eval_rougeL": 29.3551,
+      "eval_rougeLsum": 29.3506,
+      "eval_runtime": 342.0853,
+      "eval_samples_per_second": 33.112,
+      "eval_steps_per_second": 2.07,
+      "step": 280000
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0039,
+      "step": 280100
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 0.0005,
+      "loss": 0.9851,
+      "step": 280200
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0024,
+      "step": 280300
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 0.0005,
+      "loss": 1.028,
+      "step": 280400
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 0.0005,
+      "loss": 1.0062,
+      "step": 280500
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9581,
+      "step": 280600
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 0.0005,
+      "loss": 0.9343,
+      "step": 280700
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 0.0005,
+      "loss": 0.9261,
+      "step": 280800
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 0.0005,
+      "loss": 0.9291,
+      "step": 280900
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 0.0005,
+      "loss": 0.9247,
+      "step": 281000
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 0.0005,
+      "loss": 0.9282,
+      "step": 281100
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 0.0005,
+      "loss": 0.9436,
+      "step": 281200
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 0.0005,
+      "loss": 0.9216,
+      "step": 281300
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 0.0005,
+      "loss": 0.9422,
+      "step": 281400
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 0.0005,
+      "loss": 0.9206,
+      "step": 281500
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 0.0005,
+      "loss": 0.9291,
+      "step": 281600
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 0.0005,
+      "loss": 0.942,
+      "step": 281700
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 0.0005,
+      "loss": 0.9529,
+      "step": 281800
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 0.0005,
+      "loss": 0.9295,
+      "step": 281900
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 0.0005,
+      "loss": 0.9375,
+      "step": 282000
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 0.0005,
+      "loss": 0.9517,
+      "step": 282100
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 0.0005,
+      "loss": 0.9386,
+      "step": 282200
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 0.0005,
+      "loss": 0.9447,
+      "step": 282300
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 0.0005,
+      "loss": 0.9393,
+      "step": 282400
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 0.0005,
+      "loss": 0.9287,
+      "step": 282500
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 0.0005,
+      "loss": 0.9471,
+      "step": 282600
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 0.0005,
+      "loss": 0.9314,
+      "step": 282700
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 0.0005,
+      "loss": 0.9519,
+      "step": 282800
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 0.0005,
+      "loss": 0.9508,
+      "step": 282900
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 0.0005,
+      "loss": 0.9471,
+      "step": 283000
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 0.0005,
+      "loss": 0.9551,
+      "step": 283100
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 0.0005,
+      "loss": 0.9452,
+      "step": 283200
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 0.0005,
+      "loss": 0.939,
+      "step": 283300
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 0.0005,
+      "loss": 0.9668,
+      "step": 283400
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 0.0005,
+      "loss": 0.9455,
+      "step": 283500
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 0.0005,
+      "loss": 0.9514,
+      "step": 283600
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 0.0005,
+      "loss": 0.9538,
+      "step": 283700
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 0.0005,
+      "loss": 0.957,
+      "step": 283800
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 0.0005,
+      "loss": 0.9721,
+      "step": 283900
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 0.0005,
+      "loss": 0.95,
+      "step": 284000
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 0.0005,
+      "loss": 0.9582,
+      "step": 284100
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 0.0005,
+      "loss": 0.9633,
+      "step": 284200
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 0.0005,
+      "loss": 0.9668,
+      "step": 284300
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 0.0005,
+      "loss": 0.9461,
+      "step": 284400
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9372,
+      "step": 284500
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 0.0005,
+      "loss": 0.9617,
+      "step": 284600
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 0.0005,
+      "loss": 0.9605,
+      "step": 284700
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 0.0005,
+      "loss": 0.969,
+      "step": 284800
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 0.0005,
+      "loss": 0.968,
+      "step": 284900
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 0.0005,
+      "loss": 0.9552,
+      "step": 285000
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 0.0005,
+      "loss": 0.9512,
+      "step": 285100
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 0.0005,
+      "loss": 0.9615,
+      "step": 285200
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 0.0005,
+      "loss": 0.9674,
+      "step": 285300
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 0.0005,
+      "loss": 0.967,
+      "step": 285400
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 0.0005,
+      "loss": 0.9632,
+      "step": 285500
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9451,
+      "step": 285600
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9425,
+      "step": 285700
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 0.0005,
+      "loss": 0.9785,
+      "step": 285800
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 0.0005,
+      "loss": 0.9427,
+      "step": 285900
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 0.0005,
+      "loss": 0.9843,
+      "step": 286000
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 0.0005,
+      "loss": 0.9629,
+      "step": 286100
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 0.0005,
+      "loss": 0.9794,
+      "step": 286200
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 0.0005,
+      "loss": 0.9716,
+      "step": 286300
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 0.0005,
+      "loss": 0.97,
+      "step": 286400
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 0.0005,
+      "loss": 0.9727,
+      "step": 286500
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 0.0005,
+      "loss": 0.9604,
+      "step": 286600
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 0.0005,
+      "loss": 0.9535,
+      "step": 286700
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 0.0005,
+      "loss": 0.985,
+      "step": 286800
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 0.0005,
+      "loss": 0.975,
+      "step": 286900
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9729,
+      "step": 287000
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9601,
+      "step": 287100
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 0.0005,
+      "loss": 0.9857,
+      "step": 287200
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 0.0005,
+      "loss": 0.9483,
+      "step": 287300
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 0.0005,
+      "loss": 0.9885,
+      "step": 287400
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 0.0005,
+      "loss": 0.9737,
+      "step": 287500
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 0.0005,
+      "loss": 0.9759,
+      "step": 287600
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 0.0005,
+      "loss": 0.9752,
+      "step": 287700
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 0.0005,
+      "loss": 0.9863,
+      "step": 287800
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9626,
+      "step": 287900
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9761,
+      "step": 288000
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 0.0005,
+      "loss": 0.9795,
+      "step": 288100
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 0.0005,
+      "loss": 0.9566,
+      "step": 288200
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 0.0005,
+      "loss": 0.9886,
+      "step": 288300
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 0.0005,
+      "loss": 0.9908,
+      "step": 288400
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 0.0005,
+      "loss": 0.9809,
+      "step": 288500
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 0.0005,
+      "loss": 0.9773,
+      "step": 288600
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 0.0005,
+      "loss": 0.9723,
+      "step": 288700
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9736,
+      "step": 288800
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 0.0005,
+      "loss": 0.9768,
+      "step": 288900
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 0.0005,
+      "loss": 0.9986,
+      "step": 289000
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 0.0005,
+      "loss": 0.9831,
+      "step": 289100
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 0.0005,
+      "loss": 0.976,
+      "step": 289200
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9756,
+      "step": 289300
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9809,
+      "step": 289400
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 0.0005,
+      "loss": 0.9535,
+      "step": 289500
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0008,
+      "step": 289600
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 0.0005,
+      "loss": 0.9864,
+      "step": 289700
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 0.0005,
+      "loss": 0.9981,
+      "step": 289800
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 0.0005,
+      "loss": 1.0064,
+      "step": 289900
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 0.0005,
+      "loss": 0.9895,
+      "step": 290000
+    },
+    {
+      "epoch": 22.74,
+      "eval_gen_len": 18.690474088461198,
+      "eval_loss": 2.4708948135375977,
+      "eval_rouge1": 35.6683,
+      "eval_rouge2": 14.696,
+      "eval_rougeL": 29.5244,
+      "eval_rougeLsum": 29.516,
+      "eval_runtime": 342.2396,
+      "eval_samples_per_second": 33.097,
+      "eval_steps_per_second": 2.069,
+      "step": 290000
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 0.0005,
+      "loss": 0.9688,
+      "step": 290100
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9683,
+      "step": 290200
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 0.0005,
+      "loss": 0.9704,
+      "step": 290300
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 0.0005,
+      "loss": 0.963,
+      "step": 290400
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9841,
+      "step": 290500
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 0.0005,
+      "loss": 0.9654,
+      "step": 290600
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 0.0005,
+      "loss": 0.9909,
+      "step": 290700
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0001,
+      "step": 290800
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 0.0005,
+      "loss": 0.956,
+      "step": 290900
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 0.0005,
+      "loss": 0.9809,
+      "step": 291000
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 0.0005,
+      "loss": 0.9805,
+      "step": 291100
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 0.0005,
+      "loss": 0.9826,
+      "step": 291200
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 0.0005,
+      "loss": 0.9794,
+      "step": 291300
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 0.0005,
+      "loss": 0.9889,
+      "step": 291400
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 0.0005,
+      "loss": 0.9799,
+      "step": 291500
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 0.0005,
+      "loss": 1.0019,
+      "step": 291600
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 0.0005,
+      "loss": 0.9861,
+      "step": 291700
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 0.0005,
+      "loss": 0.9767,
+      "step": 291800
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 0.0005,
+      "loss": 0.9774,
+      "step": 291900
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 0.0005,
+      "loss": 0.9976,
+      "step": 292000
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9868,
+      "step": 292100
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9814,
+      "step": 292200
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 0.0005,
+      "loss": 1.0007,
+      "step": 292300
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 0.0005,
+      "loss": 0.986,
+      "step": 292400
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 0.0005,
+      "loss": 1.0014,
+      "step": 292500
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 0.0005,
+      "loss": 0.9739,
+      "step": 292600
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 0.0005,
+      "loss": 0.9743,
+      "step": 292700
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9843,
+      "step": 292800
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 0.0005,
+      "loss": 0.9942,
+      "step": 292900
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0039,
+      "step": 293000
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0217,
+      "step": 293100
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0022,
+      "step": 293200
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9824,
+      "step": 293300
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 0.0005,
+      "loss": 0.9044,
+      "step": 293400
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 0.0005,
+      "loss": 0.9098,
+      "step": 293500
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 0.0005,
+      "loss": 0.9264,
+      "step": 293600
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 0.0005,
+      "loss": 0.9192,
+      "step": 293700
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 0.0005,
+      "loss": 0.9024,
+      "step": 293800
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 0.0005,
+      "loss": 0.9148,
+      "step": 293900
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 0.0005,
+      "loss": 0.9014,
+      "step": 294000
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 0.0005,
+      "loss": 0.9089,
+      "step": 294100
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 0.0005,
+      "loss": 0.9291,
+      "step": 294200
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 0.0005,
+      "loss": 0.9165,
+      "step": 294300
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 0.0005,
+      "loss": 0.9362,
+      "step": 294400
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 0.0005,
+      "loss": 0.9204,
+      "step": 294500
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 0.0005,
+      "loss": 0.9362,
+      "step": 294600
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 0.0005,
+      "loss": 0.9093,
+      "step": 294700
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 0.0005,
+      "loss": 0.9178,
+      "step": 294800
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 0.0005,
+      "loss": 0.9327,
+      "step": 294900
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 0.0005,
+      "loss": 0.931,
+      "step": 295000
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 0.0005,
+      "loss": 0.934,
+      "step": 295100
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 0.0005,
+      "loss": 0.9421,
+      "step": 295200
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 0.0005,
+      "loss": 0.9302,
+      "step": 295300
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 0.0005,
+      "loss": 0.9291,
+      "step": 295400
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 0.0005,
+      "loss": 0.9308,
+      "step": 295500
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 0.0005,
+      "loss": 0.9408,
+      "step": 295600
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 0.0005,
+      "loss": 0.9283,
+      "step": 295700
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 0.0005,
+      "loss": 0.9234,
+      "step": 295800
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 0.0005,
+      "loss": 0.9234,
+      "step": 295900
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 0.0005,
+      "loss": 0.9442,
+      "step": 296000
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 0.0005,
+      "loss": 0.9326,
+      "step": 296100
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 0.0005,
+      "loss": 0.9359,
+      "step": 296200
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 0.0005,
+      "loss": 0.9191,
+      "step": 296300
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 0.0005,
+      "loss": 0.9371,
+      "step": 296400
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 0.0005,
+      "loss": 0.9316,
+      "step": 296500
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 0.0005,
+      "loss": 0.9369,
+      "step": 296600
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 0.0005,
+      "loss": 0.9274,
+      "step": 296700
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 0.0005,
+      "loss": 0.9196,
+      "step": 296800
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 0.0005,
+      "loss": 0.9472,
+      "step": 296900
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 0.0005,
+      "loss": 0.9214,
+      "step": 297000
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 0.0005,
+      "loss": 0.9558,
+      "step": 297100
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 0.0005,
+      "loss": 0.96,
+      "step": 297200
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9244,
+      "step": 297300
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 0.0005,
+      "loss": 0.9418,
+      "step": 297400
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 0.0005,
+      "loss": 0.9385,
+      "step": 297500
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 0.0005,
+      "loss": 0.9549,
+      "step": 297600
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 0.0005,
+      "loss": 0.9449,
+      "step": 297700
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 0.0005,
+      "loss": 0.965,
+      "step": 297800
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 0.0005,
+      "loss": 0.9288,
+      "step": 297900
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 0.0005,
+      "loss": 0.9475,
+      "step": 298000
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 0.0005,
+      "loss": 0.9444,
+      "step": 298100
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 0.0005,
+      "loss": 0.95,
+      "step": 298200
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 0.0005,
+      "loss": 0.9717,
+      "step": 298300
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9358,
+      "step": 298400
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 0.0005,
+      "loss": 0.9449,
+      "step": 298500
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 0.0005,
+      "loss": 0.9601,
+      "step": 298600
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 0.0005,
+      "loss": 0.954,
+      "step": 298700
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 0.0005,
+      "loss": 0.9564,
+      "step": 298800
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 0.0005,
+      "loss": 0.9391,
+      "step": 298900
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 0.0005,
+      "loss": 0.9517,
+      "step": 299000
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 0.0005,
+      "loss": 0.9769,
+      "step": 299100
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 0.0005,
+      "loss": 0.9563,
+      "step": 299200
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 0.0005,
+      "loss": 0.9446,
+      "step": 299300
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 0.0005,
+      "loss": 0.9597,
+      "step": 299400
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 0.0005,
+      "loss": 0.9432,
+      "step": 299500
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 0.0005,
+      "loss": 0.9546,
+      "step": 299600
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 0.0005,
+      "loss": 0.9512,
+      "step": 299700
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9726,
+      "step": 299800
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 0.0005,
+      "loss": 0.9428,
+      "step": 299900
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 0.0005,
+      "loss": 0.9506,
+      "step": 300000
+    },
+    {
+      "epoch": 23.53,
+      "eval_gen_len": 18.740001765692593,
+      "eval_loss": 2.510540246963501,
+      "eval_rouge1": 35.6195,
+      "eval_rouge2": 14.6871,
+      "eval_rougeL": 29.4076,
+      "eval_rougeLsum": 29.4076,
+      "eval_runtime": 342.9376,
+      "eval_samples_per_second": 33.029,
+      "eval_steps_per_second": 2.065,
+      "step": 300000
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 0.0005,
+      "loss": 0.9789,
+      "step": 300100
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 0.0005,
+      "loss": 0.9619,
+      "step": 300200
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 0.0005,
+      "loss": 0.961,
+      "step": 300300
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 0.0005,
+      "loss": 0.9542,
+      "step": 300400
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 0.0005,
+      "loss": 0.9636,
+      "step": 300500
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 0.0005,
+      "loss": 0.9378,
+      "step": 300600
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9712,
+      "step": 300700
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 0.0005,
+      "loss": 0.9617,
+      "step": 300800
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 0.0005,
+      "loss": 0.9551,
+      "step": 300900
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 0.0005,
+      "loss": 0.9565,
+      "step": 301000
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 0.0005,
+      "loss": 0.9667,
+      "step": 301100
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 0.0005,
+      "loss": 0.9635,
+      "step": 301200
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 0.0005,
+      "loss": 0.9572,
+      "step": 301300
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 0.0005,
+      "loss": 0.949,
+      "step": 301400
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 0.0005,
+      "loss": 0.964,
+      "step": 301500
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9588,
+      "step": 301600
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 0.0005,
+      "loss": 0.9772,
+      "step": 301700
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 0.0005,
+      "loss": 0.9718,
+      "step": 301800
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 0.0005,
+      "loss": 0.9754,
+      "step": 301900
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 0.0005,
+      "loss": 0.9525,
+      "step": 302000
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9535,
+      "step": 302100
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 0.0005,
+      "loss": 0.9448,
+      "step": 302200
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 0.0005,
+      "loss": 0.9672,
+      "step": 302300
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 0.0005,
+      "loss": 0.9591,
+      "step": 302400
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 0.0005,
+      "loss": 0.9598,
+      "step": 302500
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 0.0005,
+      "loss": 0.9723,
+      "step": 302600
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 0.0005,
+      "loss": 0.9555,
+      "step": 302700
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 0.0005,
+      "loss": 0.9657,
+      "step": 302800
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 0.0005,
+      "loss": 0.9562,
+      "step": 302900
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9602,
+      "step": 303000
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 0.0005,
+      "loss": 0.9468,
+      "step": 303100
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9736,
+      "step": 303200
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 0.0005,
+      "loss": 0.967,
+      "step": 303300
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 0.0005,
+      "loss": 0.9763,
+      "step": 303400
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 0.0005,
+      "loss": 0.969,
+      "step": 303500
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 0.0005,
+      "loss": 0.9665,
+      "step": 303600
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 0.0005,
+      "loss": 0.9804,
+      "step": 303700
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 0.0005,
+      "loss": 0.9462,
+      "step": 303800
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 0.0005,
+      "loss": 0.9625,
+      "step": 303900
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 0.0005,
+      "loss": 0.9743,
+      "step": 304000
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 0.0005,
+      "loss": 0.9707,
+      "step": 304100
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 0.0005,
+      "loss": 0.9713,
+      "step": 304200
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 0.0005,
+      "loss": 0.9709,
+      "step": 304300
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 0.0005,
+      "loss": 0.9839,
+      "step": 304400
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 0.0005,
+      "loss": 0.983,
+      "step": 304500
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 0.0005,
+      "loss": 0.9687,
+      "step": 304600
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 0.0005,
+      "loss": 0.9677,
+      "step": 304700
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 0.0005,
+      "loss": 0.9705,
+      "step": 304800
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9711,
+      "step": 304900
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 0.0005,
+      "loss": 0.9754,
+      "step": 305000
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 0.0005,
+      "loss": 0.9583,
+      "step": 305100
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 0.0005,
+      "loss": 0.9885,
+      "step": 305200
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 0.0005,
+      "loss": 0.9866,
+      "step": 305300
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 0.0005,
+      "loss": 0.976,
+      "step": 305400
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9722,
+      "step": 305500
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9872,
+      "step": 305600
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 0.0005,
+      "loss": 0.9762,
+      "step": 305700
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 0.0005,
+      "loss": 0.9781,
+      "step": 305800
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 0.0005,
+      "loss": 0.9819,
+      "step": 305900
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9992,
+      "step": 306000
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9518,
+      "step": 306100
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8936,
+      "step": 306200
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8861,
+      "step": 306300
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8966,
+      "step": 306400
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 0.0005,
+      "loss": 0.8952,
+      "step": 306500
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 0.0005,
+      "loss": 0.9118,
+      "step": 306600
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 0.0005,
+      "loss": 0.9187,
+      "step": 306700
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 0.0005,
+      "loss": 0.92,
+      "step": 306800
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 0.0005,
+      "loss": 0.889,
+      "step": 306900
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 0.0005,
+      "loss": 0.9044,
+      "step": 307000
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 0.0005,
+      "loss": 0.8956,
+      "step": 307100
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8973,
+      "step": 307200
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 0.0005,
+      "loss": 0.9062,
+      "step": 307300
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 0.0005,
+      "loss": 0.9122,
+      "step": 307400
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8848,
+      "step": 307500
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 0.0005,
+      "loss": 0.917,
+      "step": 307600
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 0.0005,
+      "loss": 0.9064,
+      "step": 307700
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 0.0005,
+      "loss": 0.9,
+      "step": 307800
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 0.0005,
+      "loss": 0.9087,
+      "step": 307900
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 0.0005,
+      "loss": 0.9075,
+      "step": 308000
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 0.0005,
+      "loss": 0.9061,
+      "step": 308100
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 0.0005,
+      "loss": 0.9002,
+      "step": 308200
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 0.0005,
+      "loss": 0.9232,
+      "step": 308300
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 0.0005,
+      "loss": 0.9232,
+      "step": 308400
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 0.0005,
+      "loss": 0.9097,
+      "step": 308500
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 0.0005,
+      "loss": 0.923,
+      "step": 308600
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 0.0005,
+      "loss": 0.9297,
+      "step": 308700
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 0.0005,
+      "loss": 0.9235,
+      "step": 308800
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 0.0005,
+      "loss": 0.9112,
+      "step": 308900
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 0.0005,
+      "loss": 0.9345,
+      "step": 309000
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 0.0005,
+      "loss": 0.9348,
+      "step": 309100
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 0.0005,
+      "loss": 0.921,
+      "step": 309200
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 0.0005,
+      "loss": 0.9245,
+      "step": 309300
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 0.0005,
+      "loss": 0.9134,
+      "step": 309400
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 0.0005,
+      "loss": 0.9355,
+      "step": 309500
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 0.0005,
+      "loss": 0.9352,
+      "step": 309600
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 0.0005,
+      "loss": 0.9359,
+      "step": 309700
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 0.0005,
+      "loss": 0.935,
+      "step": 309800
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 0.0005,
+      "loss": 0.9262,
+      "step": 309900
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9499,
+      "step": 310000
+    },
+    {
+      "epoch": 24.31,
+      "eval_gen_len": 18.726317648097467,
+      "eval_loss": 2.5155093669891357,
+      "eval_rouge1": 35.6119,
+      "eval_rouge2": 14.7476,
+      "eval_rougeL": 29.3748,
+      "eval_rougeLsum": 29.3711,
+      "eval_runtime": 343.5455,
+      "eval_samples_per_second": 32.971,
+      "eval_steps_per_second": 2.061,
+      "step": 310000
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 0.0005,
+      "loss": 0.9157,
+      "step": 310100
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 0.0005,
+      "loss": 0.9398,
+      "step": 310200
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 0.0005,
+      "loss": 0.9503,
+      "step": 310300
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 0.0005,
+      "loss": 0.9284,
+      "step": 310400
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 0.0005,
+      "loss": 0.9337,
+      "step": 310500
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 0.0005,
+      "loss": 0.9406,
+      "step": 310600
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 0.0005,
+      "loss": 0.9304,
+      "step": 310700
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 0.0005,
+      "loss": 0.9302,
+      "step": 310800
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 0.0005,
+      "loss": 0.9162,
+      "step": 310900
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 0.0005,
+      "loss": 0.9313,
+      "step": 311000
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9174,
+      "step": 311100
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9431,
+      "step": 311200
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 0.0005,
+      "loss": 0.93,
+      "step": 311300
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 0.0005,
+      "loss": 0.943,
+      "step": 311400
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 0.0005,
+      "loss": 0.9217,
+      "step": 311500
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 0.0005,
+      "loss": 0.9348,
+      "step": 311600
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 0.0005,
+      "loss": 0.9392,
+      "step": 311700
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 0.0005,
+      "loss": 0.9436,
+      "step": 311800
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 0.0005,
+      "loss": 0.9374,
+      "step": 311900
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 0.0005,
+      "loss": 0.9466,
+      "step": 312000
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 0.0005,
+      "loss": 0.95,
+      "step": 312100
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 0.0005,
+      "loss": 0.9483,
+      "step": 312200
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 0.0005,
+      "loss": 0.9408,
+      "step": 312300
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 0.0005,
+      "loss": 0.9281,
+      "step": 312400
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 0.0005,
+      "loss": 0.94,
+      "step": 312500
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9492,
+      "step": 312600
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 0.0005,
+      "loss": 0.9343,
+      "step": 312700
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 0.0005,
+      "loss": 0.9294,
+      "step": 312800
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 0.0005,
+      "loss": 0.9483,
+      "step": 312900
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 0.0005,
+      "loss": 0.9525,
+      "step": 313000
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 0.0005,
+      "loss": 0.9486,
+      "step": 313100
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 0.0005,
+      "loss": 0.9371,
+      "step": 313200
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 0.0005,
+      "loss": 0.9395,
+      "step": 313300
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9323,
+      "step": 313400
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9456,
+      "step": 313500
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 0.0005,
+      "loss": 0.9501,
+      "step": 313600
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 0.0005,
+      "loss": 0.948,
+      "step": 313700
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 0.0005,
+      "loss": 0.9544,
+      "step": 313800
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 0.0005,
+      "loss": 0.9495,
+      "step": 313900
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 0.0005,
+      "loss": 0.9236,
+      "step": 314000
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 0.0005,
+      "loss": 0.9539,
+      "step": 314100
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 0.0005,
+      "loss": 0.9539,
+      "step": 314200
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9588,
+      "step": 314300
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9363,
+      "step": 314400
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 0.0005,
+      "loss": 0.973,
+      "step": 314500
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 0.0005,
+      "loss": 0.9378,
+      "step": 314600
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 0.0005,
+      "loss": 0.9579,
+      "step": 314700
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9457,
+      "step": 314800
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9462,
+      "step": 314900
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 0.0005,
+      "loss": 0.9528,
+      "step": 315000
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 0.0005,
+      "loss": 0.9593,
+      "step": 315100
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 0.0005,
+      "loss": 0.9437,
+      "step": 315200
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 0.0005,
+      "loss": 0.9418,
+      "step": 315300
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 0.0005,
+      "loss": 0.9584,
+      "step": 315400
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 0.0005,
+      "loss": 0.9642,
+      "step": 315500
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 0.0005,
+      "loss": 0.9411,
+      "step": 315600
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9644,
+      "step": 315700
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9629,
+      "step": 315800
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 0.0005,
+      "loss": 0.9649,
+      "step": 315900
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 0.0005,
+      "loss": 0.959,
+      "step": 316000
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 0.0005,
+      "loss": 0.9554,
+      "step": 316100
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 0.0005,
+      "loss": 0.9371,
+      "step": 316200
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 0.0005,
+      "loss": 0.9448,
+      "step": 316300
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 0.0005,
+      "loss": 0.9506,
+      "step": 316400
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 0.0005,
+      "loss": 0.9453,
+      "step": 316500
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 0.0005,
+      "loss": 0.9538,
+      "step": 316600
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 0.0005,
+      "loss": 0.9692,
+      "step": 316700
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 0.0005,
+      "loss": 0.9621,
+      "step": 316800
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 0.0005,
+      "loss": 0.9688,
+      "step": 316900
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 0.0005,
+      "loss": 0.9607,
+      "step": 317000
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 0.0005,
+      "loss": 0.965,
+      "step": 317100
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 0.0005,
+      "loss": 0.9317,
+      "step": 317200
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 0.0005,
+      "loss": 0.964,
+      "step": 317300
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 0.0005,
+      "loss": 0.9664,
+      "step": 317400
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 0.0005,
+      "loss": 0.9636,
+      "step": 317500
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9622,
+      "step": 317600
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9588,
+      "step": 317700
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 0.0005,
+      "loss": 0.9647,
+      "step": 317800
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 0.0005,
+      "loss": 0.9625,
+      "step": 317900
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 0.0005,
+      "loss": 0.9411,
+      "step": 318000
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 0.0005,
+      "loss": 0.9632,
+      "step": 318100
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 0.0005,
+      "loss": 0.9674,
+      "step": 318200
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9758,
+      "step": 318300
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 0.0005,
+      "loss": 0.9692,
+      "step": 318400
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 0.0005,
+      "loss": 0.99,
+      "step": 318500
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 0.0005,
+      "loss": 0.9506,
+      "step": 318600
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 0.0005,
+      "loss": 0.9591,
+      "step": 318700
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9722,
+      "step": 318800
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8809,
+      "step": 318900
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 0.0005,
+      "loss": 0.9027,
+      "step": 319000
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8877,
+      "step": 319100
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8942,
+      "step": 319200
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 0.0005,
+      "loss": 0.8973,
+      "step": 319300
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8892,
+      "step": 319400
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8824,
+      "step": 319500
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 0.0005,
+      "loss": 0.907,
+      "step": 319600
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8954,
+      "step": 319700
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 0.0005,
+      "loss": 0.8879,
+      "step": 319800
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8861,
+      "step": 319900
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 0.0005,
+      "loss": 0.893,
+      "step": 320000
+    },
+    {
+      "epoch": 25.09,
+      "eval_gen_len": 18.719696300874016,
+      "eval_loss": 2.545152187347412,
+      "eval_rouge1": 35.5957,
+      "eval_rouge2": 14.6319,
+      "eval_rougeL": 29.3381,
+      "eval_rougeLsum": 29.3398,
+      "eval_runtime": 342.8631,
+      "eval_samples_per_second": 33.037,
+      "eval_steps_per_second": 2.065,
+      "step": 320000
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 0.0005,
+      "loss": 0.8925,
+      "step": 320100
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 0.0005,
+      "loss": 0.904,
+      "step": 320200
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8881,
+      "step": 320300
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8957,
+      "step": 320400
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8987,
+      "step": 320500
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8994,
+      "step": 320600
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 0.0005,
+      "loss": 0.9095,
+      "step": 320700
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 0.0005,
+      "loss": 0.9188,
+      "step": 320800
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 0.0005,
+      "loss": 0.9201,
+      "step": 320900
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 0.0005,
+      "loss": 0.8953,
+      "step": 321000
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8988,
+      "step": 321100
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 0.0005,
+      "loss": 0.9004,
+      "step": 321200
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8827,
+      "step": 321300
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 0.0005,
+      "loss": 0.9016,
+      "step": 321400
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8842,
+      "step": 321500
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 0.0005,
+      "loss": 0.9222,
+      "step": 321600
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 0.0005,
+      "loss": 0.9102,
+      "step": 321700
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 0.0005,
+      "loss": 0.9217,
+      "step": 321800
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 0.0005,
+      "loss": 0.9109,
+      "step": 321900
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 0.0005,
+      "loss": 0.9193,
+      "step": 322000
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 0.0005,
+      "loss": 0.9195,
+      "step": 322100
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 0.0005,
+      "loss": 0.9034,
+      "step": 322200
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 0.0005,
+      "loss": 0.9053,
+      "step": 322300
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 0.0005,
+      "loss": 0.9163,
+      "step": 322400
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 0.0005,
+      "loss": 0.9113,
+      "step": 322500
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 0.0005,
+      "loss": 0.9067,
+      "step": 322600
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9168,
+      "step": 322700
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9271,
+      "step": 322800
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 0.0005,
+      "loss": 0.9024,
+      "step": 322900
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 0.0005,
+      "loss": 0.9126,
+      "step": 323000
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 0.0005,
+      "loss": 0.9137,
+      "step": 323100
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 0.0005,
+      "loss": 0.922,
+      "step": 323200
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 0.0005,
+      "loss": 0.9191,
+      "step": 323300
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 0.0005,
+      "loss": 0.9099,
+      "step": 323400
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 0.0005,
+      "loss": 0.9366,
+      "step": 323500
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 0.0005,
+      "loss": 0.9236,
+      "step": 323600
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 0.0005,
+      "loss": 0.9264,
+      "step": 323700
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 0.0005,
+      "loss": 0.9089,
+      "step": 323800
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9362,
+      "step": 323900
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 0.0005,
+      "loss": 0.9119,
+      "step": 324000
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 0.0005,
+      "loss": 0.9206,
+      "step": 324100
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 0.0005,
+      "loss": 0.9215,
+      "step": 324200
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 0.0005,
+      "loss": 0.9125,
+      "step": 324300
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 0.0005,
+      "loss": 0.9413,
+      "step": 324400
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 0.0005,
+      "loss": 0.906,
+      "step": 324500
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 0.0005,
+      "loss": 0.9407,
+      "step": 324600
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 0.0005,
+      "loss": 0.9256,
+      "step": 324700
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 0.0005,
+      "loss": 0.9124,
+      "step": 324800
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 0.0005,
+      "loss": 0.9072,
+      "step": 324900
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 0.0005,
+      "loss": 0.9083,
+      "step": 325000
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 0.0005,
+      "loss": 0.9406,
+      "step": 325100
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 0.0005,
+      "loss": 0.9231,
+      "step": 325200
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9372,
+      "step": 325300
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 0.0005,
+      "loss": 0.932,
+      "step": 325400
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 0.0005,
+      "loss": 0.9148,
+      "step": 325500
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 0.0005,
+      "loss": 0.9375,
+      "step": 325600
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 0.0005,
+      "loss": 0.9312,
+      "step": 325700
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 0.0005,
+      "loss": 0.9183,
+      "step": 325800
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 0.0005,
+      "loss": 0.9026,
+      "step": 325900
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 0.0005,
+      "loss": 0.9128,
+      "step": 326000
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 0.0005,
+      "loss": 0.9456,
+      "step": 326100
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9392,
+      "step": 326200
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 0.0005,
+      "loss": 0.9533,
+      "step": 326300
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 0.0005,
+      "loss": 0.9302,
+      "step": 326400
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 0.0005,
+      "loss": 0.9388,
+      "step": 326500
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 0.0005,
+      "loss": 0.9214,
+      "step": 326600
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 0.0005,
+      "loss": 0.9304,
+      "step": 326700
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 0.0005,
+      "loss": 0.9424,
+      "step": 326800
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 0.0005,
+      "loss": 0.9408,
+      "step": 326900
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 0.0005,
+      "loss": 0.9365,
+      "step": 327000
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9276,
+      "step": 327100
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 0.0005,
+      "loss": 0.9325,
+      "step": 327200
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 0.0005,
+      "loss": 0.9497,
+      "step": 327300
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 0.0005,
+      "loss": 0.9383,
+      "step": 327400
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 0.0005,
+      "loss": 0.9388,
+      "step": 327500
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9443,
+      "step": 327600
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 0.0005,
+      "loss": 0.926,
+      "step": 327700
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 0.0005,
+      "loss": 0.9586,
+      "step": 327800
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 0.0005,
+      "loss": 0.9442,
+      "step": 327900
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 0.0005,
+      "loss": 0.9529,
+      "step": 328000
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 0.0005,
+      "loss": 0.9246,
+      "step": 328100
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 0.0005,
+      "loss": 0.9344,
+      "step": 328200
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 0.0005,
+      "loss": 0.9399,
+      "step": 328300
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 0.0005,
+      "loss": 0.9506,
+      "step": 328400
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9207,
+      "step": 328500
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 0.0005,
+      "loss": 0.9567,
+      "step": 328600
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9376,
+      "step": 328700
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9391,
+      "step": 328800
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 0.0005,
+      "loss": 0.9409,
+      "step": 328900
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 0.0005,
+      "loss": 0.9482,
+      "step": 329000
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 0.0005,
+      "loss": 0.9488,
+      "step": 329100
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 0.0005,
+      "loss": 0.9563,
+      "step": 329200
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 0.0005,
+      "loss": 0.939,
+      "step": 329300
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 0.0005,
+      "loss": 0.9355,
+      "step": 329400
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 0.0005,
+      "loss": 0.9403,
+      "step": 329500
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 0.0005,
+      "loss": 0.9377,
+      "step": 329600
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 0.0005,
+      "loss": 0.9481,
+      "step": 329700
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 0.0005,
+      "loss": 0.9491,
+      "step": 329800
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 0.0005,
+      "loss": 0.9412,
+      "step": 329900
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 0.0005,
+      "loss": 0.9335,
+      "step": 330000
+    },
+    {
+      "epoch": 25.88,
+      "eval_gen_len": 18.692769488831996,
+      "eval_loss": 2.503575086593628,
+      "eval_rouge1": 35.8974,
+      "eval_rouge2": 14.958,
+      "eval_rougeL": 29.7435,
+      "eval_rougeLsum": 29.7515,
+      "eval_runtime": 343.7355,
+      "eval_samples_per_second": 32.953,
+      "eval_steps_per_second": 2.06,
+      "step": 330000
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 0.0005,
+      "loss": 0.9622,
+      "step": 330100
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 0.0005,
+      "loss": 0.9253,
+      "step": 330200
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 0.0005,
+      "loss": 0.9583,
+      "step": 330300
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 0.0005,
+      "loss": 0.949,
+      "step": 330400
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 0.0005,
+      "loss": 0.9608,
+      "step": 330500
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 0.0005,
+      "loss": 0.9474,
+      "step": 330600
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 0.0005,
+      "loss": 0.9446,
+      "step": 330700
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 0.0005,
+      "loss": 0.9424,
+      "step": 330800
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 0.0005,
+      "loss": 0.9293,
+      "step": 330900
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9392,
+      "step": 331000
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9519,
+      "step": 331100
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 0.0005,
+      "loss": 0.9674,
+      "step": 331200
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 0.0005,
+      "loss": 0.947,
+      "step": 331300
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 0.0005,
+      "loss": 0.9739,
+      "step": 331400
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9587,
+      "step": 331500
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9201,
+      "step": 331600
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8644,
+      "step": 331700
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8581,
+      "step": 331800
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8868,
+      "step": 331900
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 0.0005,
+      "loss": 0.8868,
+      "step": 332000
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 0.0005,
+      "loss": 0.8634,
+      "step": 332100
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8793,
+      "step": 332200
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 0.0005,
+      "loss": 0.8779,
+      "step": 332300
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8793,
+      "step": 332400
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8923,
+      "step": 332500
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 0.0005,
+      "loss": 0.8774,
+      "step": 332600
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8699,
+      "step": 332700
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 0.0005,
+      "loss": 0.8708,
+      "step": 332800
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8899,
+      "step": 332900
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8779,
+      "step": 333000
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8926,
+      "step": 333100
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8917,
+      "step": 333200
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 0.0005,
+      "loss": 0.9002,
+      "step": 333300
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8864,
+      "step": 333400
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 0.0005,
+      "loss": 0.8914,
+      "step": 333500
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8801,
+      "step": 333600
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 0.0005,
+      "loss": 0.8886,
+      "step": 333700
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8956,
+      "step": 333800
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8889,
+      "step": 333900
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8937,
+      "step": 334000
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8842,
+      "step": 334100
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 0.0005,
+      "loss": 0.9001,
+      "step": 334200
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 0.0005,
+      "loss": 0.8991,
+      "step": 334300
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 0.0005,
+      "loss": 0.899,
+      "step": 334400
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 0.0005,
+      "loss": 0.9057,
+      "step": 334500
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 0.0005,
+      "loss": 0.8923,
+      "step": 334600
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 0.0005,
+      "loss": 0.9065,
+      "step": 334700
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 0.0005,
+      "loss": 0.889,
+      "step": 334800
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 0.0005,
+      "loss": 0.8918,
+      "step": 334900
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 0.0005,
+      "loss": 0.9046,
+      "step": 335000
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8922,
+      "step": 335100
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 0.0005,
+      "loss": 0.9002,
+      "step": 335200
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8995,
+      "step": 335300
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8953,
+      "step": 335400
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 0.0005,
+      "loss": 0.9064,
+      "step": 335500
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 0.0005,
+      "loss": 0.9207,
+      "step": 335600
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8872,
+      "step": 335700
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8966,
+      "step": 335800
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 0.0005,
+      "loss": 0.9181,
+      "step": 335900
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 0.0005,
+      "loss": 0.9238,
+      "step": 336000
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 0.0005,
+      "loss": 0.9176,
+      "step": 336100
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8847,
+      "step": 336200
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 0.0005,
+      "loss": 0.9041,
+      "step": 336300
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 0.0005,
+      "loss": 0.901,
+      "step": 336400
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8913,
+      "step": 336500
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 0.0005,
+      "loss": 0.916,
+      "step": 336600
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9105,
+      "step": 336700
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 0.0005,
+      "loss": 0.911,
+      "step": 336800
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 0.0005,
+      "loss": 0.919,
+      "step": 336900
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8979,
+      "step": 337000
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 0.0005,
+      "loss": 0.9193,
+      "step": 337100
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 0.0005,
+      "loss": 0.8913,
+      "step": 337200
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 0.0005,
+      "loss": 0.9286,
+      "step": 337300
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 0.0005,
+      "loss": 0.9046,
+      "step": 337400
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 0.0005,
+      "loss": 0.9207,
+      "step": 337500
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 0.0005,
+      "loss": 0.9098,
+      "step": 337600
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 0.0005,
+      "loss": 0.9177,
+      "step": 337700
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 0.0005,
+      "loss": 0.9051,
+      "step": 337800
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 0.0005,
+      "loss": 0.9074,
+      "step": 337900
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9157,
+      "step": 338000
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9032,
+      "step": 338100
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 0.0005,
+      "loss": 0.9159,
+      "step": 338200
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 0.0005,
+      "loss": 0.9227,
+      "step": 338300
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 0.0005,
+      "loss": 0.935,
+      "step": 338400
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 0.0005,
+      "loss": 0.9078,
+      "step": 338500
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 0.0005,
+      "loss": 0.9201,
+      "step": 338600
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 0.0005,
+      "loss": 0.9367,
+      "step": 338700
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 0.0005,
+      "loss": 0.9325,
+      "step": 338800
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 0.0005,
+      "loss": 0.8986,
+      "step": 338900
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9171,
+      "step": 339000
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 0.0005,
+      "loss": 0.9199,
+      "step": 339100
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 0.0005,
+      "loss": 0.8902,
+      "step": 339200
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 0.0005,
+      "loss": 0.9426,
+      "step": 339300
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 0.0005,
+      "loss": 0.9074,
+      "step": 339400
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 0.0005,
+      "loss": 0.9046,
+      "step": 339500
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 0.0005,
+      "loss": 0.9325,
+      "step": 339600
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 0.0005,
+      "loss": 0.934,
+      "step": 339700
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 0.0005,
+      "loss": 0.929,
+      "step": 339800
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9233,
+      "step": 339900
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 0.0005,
+      "loss": 0.9324,
+      "step": 340000
+    },
+    {
+      "epoch": 26.66,
+      "eval_gen_len": 18.74794738236073,
+      "eval_loss": 2.5247809886932373,
+      "eval_rouge1": 35.8077,
+      "eval_rouge2": 14.9114,
+      "eval_rougeL": 29.627,
+      "eval_rougeLsum": 29.6285,
+      "eval_runtime": 344.6219,
+      "eval_samples_per_second": 32.868,
+      "eval_steps_per_second": 2.054,
+      "step": 340000
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 0.0005,
+      "loss": 0.923,
+      "step": 340100
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 0.0005,
+      "loss": 0.9162,
+      "step": 340200
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9267,
+      "step": 340300
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9172,
+      "step": 340400
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 0.0005,
+      "loss": 0.9393,
+      "step": 340500
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 0.0005,
+      "loss": 0.9338,
+      "step": 340600
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 0.0005,
+      "loss": 0.944,
+      "step": 340700
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 0.0005,
+      "loss": 0.9187,
+      "step": 340800
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 0.0005,
+      "loss": 0.9293,
+      "step": 340900
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 0.0005,
+      "loss": 0.9361,
+      "step": 341000
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 0.0005,
+      "loss": 0.9298,
+      "step": 341100
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9349,
+      "step": 341200
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9216,
+      "step": 341300
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 0.0005,
+      "loss": 0.9445,
+      "step": 341400
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9376,
+      "step": 341500
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 0.0005,
+      "loss": 0.9266,
+      "step": 341600
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 0.0005,
+      "loss": 0.9311,
+      "step": 341700
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 0.0005,
+      "loss": 0.9399,
+      "step": 341800
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 0.0005,
+      "loss": 0.9059,
+      "step": 341900
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 0.0005,
+      "loss": 0.9241,
+      "step": 342000
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 0.0005,
+      "loss": 0.9276,
+      "step": 342100
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 0.0005,
+      "loss": 0.9259,
+      "step": 342200
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 0.0005,
+      "loss": 0.9262,
+      "step": 342300
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 0.0005,
+      "loss": 0.9346,
+      "step": 342400
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 0.0005,
+      "loss": 0.9406,
+      "step": 342500
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 0.0005,
+      "loss": 0.9522,
+      "step": 342600
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 0.0005,
+      "loss": 0.9023,
+      "step": 342700
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 0.0005,
+      "loss": 0.9389,
+      "step": 342800
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 0.0005,
+      "loss": 0.9364,
+      "step": 342900
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 0.0005,
+      "loss": 0.9338,
+      "step": 343000
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9258,
+      "step": 343100
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9315,
+      "step": 343200
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 0.0005,
+      "loss": 0.9448,
+      "step": 343300
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 0.0005,
+      "loss": 0.9283,
+      "step": 343400
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 0.0005,
+      "loss": 0.919,
+      "step": 343500
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 0.0005,
+      "loss": 0.9423,
+      "step": 343600
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 0.0005,
+      "loss": 0.9215,
+      "step": 343700
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9399,
+      "step": 343800
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 0.0005,
+      "loss": 0.9367,
+      "step": 343900
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 0.0005,
+      "loss": 0.956,
+      "step": 344000
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 0.0005,
+      "loss": 0.9347,
+      "step": 344100
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 0.0005,
+      "loss": 0.9514,
+      "step": 344200
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9499,
+      "step": 344300
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8721,
+      "step": 344400
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8669,
+      "step": 344500
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8606,
+      "step": 344600
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8757,
+      "step": 344700
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 0.0005,
+      "loss": 0.8588,
+      "step": 344800
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8642,
+      "step": 344900
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8582,
+      "step": 345000
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 0.0005,
+      "loss": 0.8578,
+      "step": 345100
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 0.0005,
+      "loss": 0.871,
+      "step": 345200
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 0.0005,
+      "loss": 0.8786,
+      "step": 345300
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8583,
+      "step": 345400
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8721,
+      "step": 345500
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 0.0005,
+      "loss": 0.8601,
+      "step": 345600
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8734,
+      "step": 345700
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8737,
+      "step": 345800
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8626,
+      "step": 345900
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8671,
+      "step": 346000
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8733,
+      "step": 346100
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 0.0005,
+      "loss": 0.8733,
+      "step": 346200
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8874,
+      "step": 346300
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8659,
+      "step": 346400
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 0.0005,
+      "loss": 0.8888,
+      "step": 346500
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8842,
+      "step": 346600
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8901,
+      "step": 346700
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8829,
+      "step": 346800
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8869,
+      "step": 346900
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8792,
+      "step": 347000
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 0.0005,
+      "loss": 0.8895,
+      "step": 347100
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 0.0005,
+      "loss": 0.8756,
+      "step": 347200
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 0.0005,
+      "loss": 0.8804,
+      "step": 347300
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 0.0005,
+      "loss": 0.8754,
+      "step": 347400
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8823,
+      "step": 347500
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 0.0005,
+      "loss": 0.8957,
+      "step": 347600
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8939,
+      "step": 347700
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8829,
+      "step": 347800
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8767,
+      "step": 347900
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8877,
+      "step": 348000
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8929,
+      "step": 348100
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 0.0005,
+      "loss": 0.884,
+      "step": 348200
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 0.0005,
+      "loss": 0.8813,
+      "step": 348300
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8865,
+      "step": 348400
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8928,
+      "step": 348500
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8821,
+      "step": 348600
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8857,
+      "step": 348700
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 0.0005,
+      "loss": 0.8991,
+      "step": 348800
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8767,
+      "step": 348900
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 0.0005,
+      "loss": 0.892,
+      "step": 349000
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 0.0005,
+      "loss": 0.9086,
+      "step": 349100
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8922,
+      "step": 349200
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8966,
+      "step": 349300
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 0.0005,
+      "loss": 0.9145,
+      "step": 349400
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8957,
+      "step": 349500
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 0.0005,
+      "loss": 0.9082,
+      "step": 349600
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 0.0005,
+      "loss": 0.8994,
+      "step": 349700
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8818,
+      "step": 349800
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 0.0005,
+      "loss": 0.8782,
+      "step": 349900
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 0.0005,
+      "loss": 0.9131,
+      "step": 350000
+    },
+    {
+      "epoch": 27.45,
+      "eval_gen_len": 18.75324446013949,
+      "eval_loss": 2.5338451862335205,
+      "eval_rouge1": 35.4764,
+      "eval_rouge2": 14.6233,
+      "eval_rougeL": 29.3295,
+      "eval_rougeLsum": 29.3271,
+      "eval_runtime": 342.5159,
+      "eval_samples_per_second": 33.07,
+      "eval_steps_per_second": 2.067,
+      "step": 350000
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8981,
+      "step": 350100
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 0.0005,
+      "loss": 0.8919,
+      "step": 350200
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8962,
+      "step": 350300
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 0.0005,
+      "loss": 0.9065,
+      "step": 350400
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8839,
+      "step": 350500
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 0.0005,
+      "loss": 0.9226,
+      "step": 350600
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8949,
+      "step": 350700
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 0.0005,
+      "loss": 0.9005,
+      "step": 350800
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8977,
+      "step": 350900
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 0.0005,
+      "loss": 0.908,
+      "step": 351000
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8956,
+      "step": 351100
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8906,
+      "step": 351200
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 0.0005,
+      "loss": 0.8817,
+      "step": 351300
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 0.0005,
+      "loss": 0.9132,
+      "step": 351400
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 0.0005,
+      "loss": 0.9114,
+      "step": 351500
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 0.0005,
+      "loss": 0.9161,
+      "step": 351600
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9128,
+      "step": 351700
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 0.0005,
+      "loss": 0.895,
+      "step": 351800
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 0.0005,
+      "loss": 0.914,
+      "step": 351900
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 0.0005,
+      "loss": 0.9082,
+      "step": 352000
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 0.0005,
+      "loss": 0.9267,
+      "step": 352100
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 0.0005,
+      "loss": 0.8968,
+      "step": 352200
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 0.0005,
+      "loss": 0.9162,
+      "step": 352300
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 0.0005,
+      "loss": 0.9174,
+      "step": 352400
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 0.0005,
+      "loss": 0.9075,
+      "step": 352500
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9275,
+      "step": 352600
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 0.0005,
+      "loss": 0.9007,
+      "step": 352700
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 0.0005,
+      "loss": 0.9094,
+      "step": 352800
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8958,
+      "step": 352900
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 0.0005,
+      "loss": 0.9158,
+      "step": 353000
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9148,
+      "step": 353100
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 0.0005,
+      "loss": 0.9048,
+      "step": 353200
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 0.0005,
+      "loss": 0.9168,
+      "step": 353300
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 0.0005,
+      "loss": 0.9027,
+      "step": 353400
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8975,
+      "step": 353500
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 0.0005,
+      "loss": 0.907,
+      "step": 353600
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 0.0005,
+      "loss": 0.929,
+      "step": 353700
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 0.0005,
+      "loss": 0.9171,
+      "step": 353800
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 0.0005,
+      "loss": 0.9269,
+      "step": 353900
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9198,
+      "step": 354000
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 0.0005,
+      "loss": 0.9323,
+      "step": 354100
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9231,
+      "step": 354200
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9159,
+      "step": 354300
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 0.0005,
+      "loss": 0.9269,
+      "step": 354400
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 0.0005,
+      "loss": 0.906,
+      "step": 354500
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 0.0005,
+      "loss": 0.9268,
+      "step": 354600
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 0.0005,
+      "loss": 0.9293,
+      "step": 354700
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 0.0005,
+      "loss": 0.9302,
+      "step": 354800
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 0.0005,
+      "loss": 0.9257,
+      "step": 354900
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 0.0005,
+      "loss": 0.926,
+      "step": 355000
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 0.0005,
+      "loss": 0.9234,
+      "step": 355100
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 0.0005,
+      "loss": 0.9299,
+      "step": 355200
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 0.0005,
+      "loss": 0.9336,
+      "step": 355300
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 0.0005,
+      "loss": 0.9407,
+      "step": 355400
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 0.0005,
+      "loss": 0.9227,
+      "step": 355500
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 0.0005,
+      "loss": 0.9281,
+      "step": 355600
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 0.0005,
+      "loss": 0.902,
+      "step": 355700
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 0.0005,
+      "loss": 0.909,
+      "step": 355800
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9284,
+      "step": 355900
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 0.0005,
+      "loss": 0.9104,
+      "step": 356000
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 0.0005,
+      "loss": 0.9239,
+      "step": 356100
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 0.0005,
+      "loss": 0.9364,
+      "step": 356200
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 0.0005,
+      "loss": 0.9207,
+      "step": 356300
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 0.0005,
+      "loss": 0.9208,
+      "step": 356400
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9331,
+      "step": 356500
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9394,
+      "step": 356600
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 0.0005,
+      "loss": 0.9175,
+      "step": 356700
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 0.0005,
+      "loss": 0.9379,
+      "step": 356800
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 0.0005,
+      "loss": 0.9293,
+      "step": 356900
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9164,
+      "step": 357000
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 0.0005,
+      "loss": 0.8842,
+      "step": 357100
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8249,
+      "step": 357200
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8539,
+      "step": 357300
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8682,
+      "step": 357400
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8495,
+      "step": 357500
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 0.0005,
+      "loss": 0.8562,
+      "step": 357600
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8425,
+      "step": 357700
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 0.0005,
+      "loss": 0.8658,
+      "step": 357800
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8537,
+      "step": 357900
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8539,
+      "step": 358000
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 0.0005,
+      "loss": 0.8655,
+      "step": 358100
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8737,
+      "step": 358200
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 0.0005,
+      "loss": 0.8596,
+      "step": 358300
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8669,
+      "step": 358400
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8513,
+      "step": 358500
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8544,
+      "step": 358600
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8656,
+      "step": 358700
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8528,
+      "step": 358800
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 0.0005,
+      "loss": 0.856,
+      "step": 358900
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 0.0005,
+      "loss": 0.8561,
+      "step": 359000
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8437,
+      "step": 359100
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 0.0005,
+      "loss": 0.871,
+      "step": 359200
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8716,
+      "step": 359300
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8601,
+      "step": 359400
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8735,
+      "step": 359500
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8689,
+      "step": 359600
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8721,
+      "step": 359700
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 0.0005,
+      "loss": 0.8717,
+      "step": 359800
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 0.0005,
+      "loss": 0.8497,
+      "step": 359900
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 0.0005,
+      "loss": 0.8805,
+      "step": 360000
+    },
+    {
+      "epoch": 28.23,
+      "eval_gen_len": 18.76789970866072,
+      "eval_loss": 2.5372776985168457,
+      "eval_rouge1": 35.7063,
+      "eval_rouge2": 14.7166,
+      "eval_rougeL": 29.4219,
+      "eval_rougeLsum": 29.4206,
+      "eval_runtime": 342.7529,
+      "eval_samples_per_second": 33.047,
+      "eval_steps_per_second": 2.066,
+      "step": 360000
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 0.0005,
+      "loss": 0.8709,
+      "step": 360100
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8777,
+      "step": 360200
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8792,
+      "step": 360300
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 0.0005,
+      "loss": 0.8701,
+      "step": 360400
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8977,
+      "step": 360500
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8732,
+      "step": 360600
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8741,
+      "step": 360700
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8822,
+      "step": 360800
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8917,
+      "step": 360900
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 0.0005,
+      "loss": 0.8797,
+      "step": 361000
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8814,
+      "step": 361100
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8797,
+      "step": 361200
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8891,
+      "step": 361300
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8845,
+      "step": 361400
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 0.0005,
+      "loss": 0.8806,
+      "step": 361500
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 0.0005,
+      "loss": 0.88,
+      "step": 361600
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8906,
+      "step": 361700
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 0.0005,
+      "loss": 0.8814,
+      "step": 361800
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8719,
+      "step": 361900
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8853,
+      "step": 362000
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 0.0005,
+      "loss": 0.8691,
+      "step": 362100
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 0.0005,
+      "loss": 0.8678,
+      "step": 362200
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8867,
+      "step": 362300
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 0.0005,
+      "loss": 0.898,
+      "step": 362400
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 0.0005,
+      "loss": 0.9137,
+      "step": 362500
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8931,
+      "step": 362600
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 0.0005,
+      "loss": 0.8859,
+      "step": 362700
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 0.0005,
+      "loss": 0.9005,
+      "step": 362800
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 0.0005,
+      "loss": 0.8763,
+      "step": 362900
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8683,
+      "step": 363000
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8866,
+      "step": 363100
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 0.0005,
+      "loss": 0.8963,
+      "step": 363200
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8846,
+      "step": 363300
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 0.0005,
+      "loss": 0.886,
+      "step": 363400
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 0.0005,
+      "loss": 0.8921,
+      "step": 363500
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 0.0005,
+      "loss": 0.898,
+      "step": 363600
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8784,
+      "step": 363700
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8979,
+      "step": 363800
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8878,
+      "step": 363900
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8999,
+      "step": 364000
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 0.0005,
+      "loss": 0.8778,
+      "step": 364100
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 0.0005,
+      "loss": 0.8856,
+      "step": 364200
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8938,
+      "step": 364300
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 0.0005,
+      "loss": 0.9011,
+      "step": 364400
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 0.0005,
+      "loss": 0.8987,
+      "step": 364500
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 0.0005,
+      "loss": 0.9002,
+      "step": 364600
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 0.0005,
+      "loss": 0.9035,
+      "step": 364700
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8931,
+      "step": 364800
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 0.0005,
+      "loss": 0.8884,
+      "step": 364900
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 0.0005,
+      "loss": 0.8985,
+      "step": 365000
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8982,
+      "step": 365100
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 0.0005,
+      "loss": 0.8899,
+      "step": 365200
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8943,
+      "step": 365300
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9133,
+      "step": 365400
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 0.0005,
+      "loss": 0.8853,
+      "step": 365500
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 0.0005,
+      "loss": 0.9038,
+      "step": 365600
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8863,
+      "step": 365700
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 0.0005,
+      "loss": 0.9016,
+      "step": 365800
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 0.0005,
+      "loss": 0.8923,
+      "step": 365900
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 0.0005,
+      "loss": 0.9003,
+      "step": 366000
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 0.0005,
+      "loss": 0.886,
+      "step": 366100
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 0.0005,
+      "loss": 0.9096,
+      "step": 366200
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 0.0005,
+      "loss": 0.9223,
+      "step": 366300
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 0.0005,
+      "loss": 0.9097,
+      "step": 366400
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8826,
+      "step": 366500
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 0.0005,
+      "loss": 0.9088,
+      "step": 366600
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9049,
+      "step": 366700
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 0.0005,
+      "loss": 0.9141,
+      "step": 366800
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 0.0005,
+      "loss": 0.9033,
+      "step": 366900
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9207,
+      "step": 367000
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 0.0005,
+      "loss": 0.9334,
+      "step": 367100
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 0.0005,
+      "loss": 0.9036,
+      "step": 367200
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 0.0005,
+      "loss": 0.9075,
+      "step": 367300
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 0.0005,
+      "loss": 0.9105,
+      "step": 367400
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 0.0005,
+      "loss": 0.8969,
+      "step": 367500
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 0.0005,
+      "loss": 0.883,
+      "step": 367600
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 0.0005,
+      "loss": 0.9088,
+      "step": 367700
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 0.0005,
+      "loss": 0.9039,
+      "step": 367800
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8916,
+      "step": 367900
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 0.0005,
+      "loss": 0.9001,
+      "step": 368000
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 0.0005,
+      "loss": 0.921,
+      "step": 368100
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 0.0005,
+      "loss": 0.9076,
+      "step": 368200
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8996,
+      "step": 368300
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8978,
+      "step": 368400
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 0.0005,
+      "loss": 0.9024,
+      "step": 368500
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9298,
+      "step": 368600
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 0.0005,
+      "loss": 0.9093,
+      "step": 368700
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 0.0005,
+      "loss": 0.9188,
+      "step": 368800
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8863,
+      "step": 368900
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 0.0005,
+      "loss": 0.9089,
+      "step": 369000
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 0.0005,
+      "loss": 0.9015,
+      "step": 369100
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 0.0005,
+      "loss": 0.9233,
+      "step": 369200
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8923,
+      "step": 369300
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 0.0005,
+      "loss": 0.9033,
+      "step": 369400
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 0.0005,
+      "loss": 0.9084,
+      "step": 369500
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 0.0005,
+      "loss": 0.9086,
+      "step": 369600
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 0.0005,
+      "loss": 0.9156,
+      "step": 369700
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9118,
+      "step": 369800
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8237,
+      "step": 369900
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8535,
+      "step": 370000
+    },
+    {
+      "epoch": 29.02,
+      "eval_gen_len": 18.747417674582856,
+      "eval_loss": 2.6049208641052246,
+      "eval_rouge1": 35.5556,
+      "eval_rouge2": 14.7079,
+      "eval_rougeL": 29.4102,
+      "eval_rougeLsum": 29.4123,
+      "eval_runtime": 342.2048,
+      "eval_samples_per_second": 33.1,
+      "eval_steps_per_second": 2.069,
+      "step": 370000
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8266,
+      "step": 370100
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8482,
+      "step": 370200
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 0.0005,
+      "loss": 0.8397,
+      "step": 370300
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8545,
+      "step": 370400
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8445,
+      "step": 370500
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 0.0005,
+      "loss": 0.8375,
+      "step": 370600
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8407,
+      "step": 370700
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 0.0005,
+      "loss": 0.8441,
+      "step": 370800
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8493,
+      "step": 370900
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8368,
+      "step": 371000
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 0.0005,
+      "loss": 0.8442,
+      "step": 371100
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8446,
+      "step": 371200
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8462,
+      "step": 371300
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8461,
+      "step": 371400
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8486,
+      "step": 371500
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8603,
+      "step": 371600
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 0.0005,
+      "loss": 0.8696,
+      "step": 371700
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8638,
+      "step": 371800
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8647,
+      "step": 371900
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 0.0005,
+      "loss": 0.8609,
+      "step": 372000
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 0.0005,
+      "loss": 0.852,
+      "step": 372100
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8582,
+      "step": 372200
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8579,
+      "step": 372300
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 0.0005,
+      "loss": 0.834,
+      "step": 372400
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8731,
+      "step": 372500
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 0.0005,
+      "loss": 0.8474,
+      "step": 372600
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 0.0005,
+      "loss": 0.8539,
+      "step": 372700
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 0.0005,
+      "loss": 0.8584,
+      "step": 372800
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 0.0005,
+      "loss": 0.8567,
+      "step": 372900
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8579,
+      "step": 373000
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 0.0005,
+      "loss": 0.8576,
+      "step": 373100
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8628,
+      "step": 373200
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8639,
+      "step": 373300
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8677,
+      "step": 373400
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8661,
+      "step": 373500
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8609,
+      "step": 373600
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 0.0005,
+      "loss": 0.8804,
+      "step": 373700
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 0.0005,
+      "loss": 0.8659,
+      "step": 373800
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8627,
+      "step": 373900
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8622,
+      "step": 374000
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 0.0005,
+      "loss": 0.873,
+      "step": 374100
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8749,
+      "step": 374200
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 0.0005,
+      "loss": 0.8654,
+      "step": 374300
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8854,
+      "step": 374400
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 0.0005,
+      "loss": 0.861,
+      "step": 374500
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8615,
+      "step": 374600
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8702,
+      "step": 374700
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8605,
+      "step": 374800
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 0.0005,
+      "loss": 0.8555,
+      "step": 374900
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8733,
+      "step": 375000
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8929,
+      "step": 375100
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 0.0005,
+      "loss": 0.8697,
+      "step": 375200
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8687,
+      "step": 375300
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 0.0005,
+      "loss": 0.8573,
+      "step": 375400
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8935,
+      "step": 375500
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 0.0005,
+      "loss": 0.861,
+      "step": 375600
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 0.0005,
+      "loss": 0.8742,
+      "step": 375700
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8649,
+      "step": 375800
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 0.0005,
+      "loss": 0.8776,
+      "step": 375900
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8897,
+      "step": 376000
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8972,
+      "step": 376100
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8943,
+      "step": 376200
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 0.0005,
+      "loss": 0.8796,
+      "step": 376300
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8884,
+      "step": 376400
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8724,
+      "step": 376500
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8796,
+      "step": 376600
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 0.0005,
+      "loss": 0.873,
+      "step": 376700
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 0.0005,
+      "loss": 0.885,
+      "step": 376800
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 0.0005,
+      "loss": 0.864,
+      "step": 376900
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 0.0005,
+      "loss": 0.8937,
+      "step": 377000
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8861,
+      "step": 377100
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 0.0005,
+      "loss": 0.882,
+      "step": 377200
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 0.0005,
+      "loss": 0.8777,
+      "step": 377300
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 0.0005,
+      "loss": 0.8981,
+      "step": 377400
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 0.0005,
+      "loss": 0.8643,
+      "step": 377500
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8898,
+      "step": 377600
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 0.0005,
+      "loss": 0.8839,
+      "step": 377700
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 0.0005,
+      "loss": 0.9064,
+      "step": 377800
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8864,
+      "step": 377900
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 0.0005,
+      "loss": 0.8682,
+      "step": 378000
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 0.0005,
+      "loss": 0.9041,
+      "step": 378100
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 0.0005,
+      "loss": 0.8807,
+      "step": 378200
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8757,
+      "step": 378300
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8837,
+      "step": 378400
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8701,
+      "step": 378500
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 0.0005,
+      "loss": 0.8868,
+      "step": 378600
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8875,
+      "step": 378700
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 0.0005,
+      "loss": 0.8934,
+      "step": 378800
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 0.0005,
+      "loss": 0.8924,
+      "step": 378900
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8787,
+      "step": 379000
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 0.0005,
+      "loss": 0.878,
+      "step": 379100
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 0.0005,
+      "loss": 0.9067,
+      "step": 379200
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8977,
+      "step": 379300
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 0.0005,
+      "loss": 0.8806,
+      "step": 379400
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8896,
+      "step": 379500
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 0.0005,
+      "loss": 0.8944,
+      "step": 379600
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 0.0005,
+      "loss": 0.9,
+      "step": 379700
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 0.0005,
+      "loss": 0.8928,
+      "step": 379800
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 0.0005,
+      "loss": 0.8904,
+      "step": 379900
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 0.0005,
+      "loss": 0.8929,
+      "step": 380000
+    },
+    {
+      "epoch": 29.8,
+      "eval_gen_len": 18.703451929019156,
+      "eval_loss": 2.553347110748291,
+      "eval_rouge1": 35.8775,
+      "eval_rouge2": 14.8452,
+      "eval_rougeL": 29.6381,
+      "eval_rougeLsum": 29.6405,
+      "eval_runtime": 344.0885,
+      "eval_samples_per_second": 32.919,
+      "eval_steps_per_second": 2.058,
+      "step": 380000
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 0.0005,
+      "loss": 0.9025,
+      "step": 380100
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8897,
+      "step": 380200
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 0.0005,
+      "loss": 0.8934,
+      "step": 380300
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8792,
+      "step": 380400
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 0.0005,
+      "loss": 0.9071,
+      "step": 380500
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8875,
+      "step": 380600
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8598,
+      "step": 380700
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 0.0005,
+      "loss": 0.9036,
+      "step": 380800
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 0.0005,
+      "loss": 0.8997,
+      "step": 380900
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 0.0005,
+      "loss": 0.9014,
+      "step": 381000
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8798,
+      "step": 381100
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8978,
+      "step": 381200
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8987,
+      "step": 381300
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 0.0005,
+      "loss": 0.877,
+      "step": 381400
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8892,
+      "step": 381500
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 0.0005,
+      "loss": 0.9068,
+      "step": 381600
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8889,
+      "step": 381700
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8967,
+      "step": 381800
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 0.0005,
+      "loss": 0.8883,
+      "step": 381900
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9073,
+      "step": 382000
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 0.0005,
+      "loss": 0.9061,
+      "step": 382100
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8932,
+      "step": 382200
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 0.0005,
+      "loss": 0.9205,
+      "step": 382300
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 0.0005,
+      "loss": 0.9188,
+      "step": 382400
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 0.0005,
+      "loss": 0.9018,
+      "step": 382500
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 0.0005,
+      "loss": 0.8755,
+      "step": 382600
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8238,
+      "step": 382700
+    },
+    {
+      "epoch": 30.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8308,
+      "step": 382800
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8384,
+      "step": 382900
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8284,
+      "step": 383000
+    },
+    {
+      "epoch": 30.04,
+      "learning_rate": 0.0005,
+      "loss": 0.8354,
+      "step": 383100
+    },
+    {
+      "epoch": 30.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8276,
+      "step": 383200
+    },
+    {
+      "epoch": 30.06,
+      "learning_rate": 0.0005,
+      "loss": 0.8413,
+      "step": 383300
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8349,
+      "step": 383400
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8465,
+      "step": 383500
+    },
+    {
+      "epoch": 30.08,
+      "learning_rate": 0.0005,
+      "loss": 0.8471,
+      "step": 383600
+    },
+    {
+      "epoch": 30.09,
+      "learning_rate": 0.0005,
+      "loss": 0.851,
+      "step": 383700
+    },
+    {
+      "epoch": 30.1,
+      "learning_rate": 0.0005,
+      "loss": 0.8236,
+      "step": 383800
+    },
+    {
+      "epoch": 30.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8438,
+      "step": 383900
+    },
+    {
+      "epoch": 30.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8308,
+      "step": 384000
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8391,
+      "step": 384100
+    },
+    {
+      "epoch": 30.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8476,
+      "step": 384200
+    },
+    {
+      "epoch": 30.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8446,
+      "step": 384300
+    },
+    {
+      "epoch": 30.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8403,
+      "step": 384400
+    },
+    {
+      "epoch": 30.15,
+      "learning_rate": 0.0005,
+      "loss": 0.8481,
+      "step": 384500
+    },
+    {
+      "epoch": 30.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8366,
+      "step": 384600
+    },
+    {
+      "epoch": 30.17,
+      "learning_rate": 0.0005,
+      "loss": 0.8312,
+      "step": 384700
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8391,
+      "step": 384800
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 0.0005,
+      "loss": 0.825,
+      "step": 384900
+    },
+    {
+      "epoch": 30.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8344,
+      "step": 385000
+    },
+    {
+      "epoch": 30.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8362,
+      "step": 385100
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8422,
+      "step": 385200
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8838,
+      "step": 385300
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 0.0005,
+      "loss": 0.838,
+      "step": 385400
+    },
+    {
+      "epoch": 30.23,
+      "learning_rate": 0.0005,
+      "loss": 0.8577,
+      "step": 385500
+    },
+    {
+      "epoch": 30.24,
+      "learning_rate": 0.0005,
+      "loss": 0.8481,
+      "step": 385600
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8423,
+      "step": 385700
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8441,
+      "step": 385800
+    },
+    {
+      "epoch": 30.26,
+      "learning_rate": 0.0005,
+      "loss": 0.8661,
+      "step": 385900
+    },
+    {
+      "epoch": 30.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8555,
+      "step": 386000
+    },
+    {
+      "epoch": 30.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8512,
+      "step": 386100
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 0.0005,
+      "loss": 0.853,
+      "step": 386200
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8591,
+      "step": 386300
+    },
+    {
+      "epoch": 30.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8608,
+      "step": 386400
+    },
+    {
+      "epoch": 30.31,
+      "learning_rate": 0.0005,
+      "loss": 0.8438,
+      "step": 386500
+    },
+    {
+      "epoch": 30.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8514,
+      "step": 386600
+    },
+    {
+      "epoch": 30.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8619,
+      "step": 386700
+    },
+    {
+      "epoch": 30.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8399,
+      "step": 386800
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8481,
+      "step": 386900
+    },
+    {
+      "epoch": 30.35,
+      "learning_rate": 0.0005,
+      "loss": 0.8487,
+      "step": 387000
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8656,
+      "step": 387100
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8567,
+      "step": 387200
+    },
+    {
+      "epoch": 30.37,
+      "learning_rate": 0.0005,
+      "loss": 0.8613,
+      "step": 387300
+    },
+    {
+      "epoch": 30.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8593,
+      "step": 387400
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8511,
+      "step": 387500
+    },
+    {
+      "epoch": 30.4,
+      "learning_rate": 0.0005,
+      "loss": 0.8573,
+      "step": 387600
+    },
+    {
+      "epoch": 30.4,
+      "learning_rate": 0.0005,
+      "loss": 0.8588,
+      "step": 387700
+    },
+    {
+      "epoch": 30.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8709,
+      "step": 387800
+    },
+    {
+      "epoch": 30.42,
+      "learning_rate": 0.0005,
+      "loss": 0.874,
+      "step": 387900
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8742,
+      "step": 388000
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8474,
+      "step": 388100
+    },
+    {
+      "epoch": 30.44,
+      "learning_rate": 0.0005,
+      "loss": 0.8629,
+      "step": 388200
+    },
+    {
+      "epoch": 30.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8897,
+      "step": 388300
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 0.0005,
+      "loss": 0.8466,
+      "step": 388400
+    },
+    {
+      "epoch": 30.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8659,
+      "step": 388500
+    },
+    {
+      "epoch": 30.47,
+      "learning_rate": 0.0005,
+      "loss": 0.871,
+      "step": 388600
+    },
+    {
+      "epoch": 30.48,
+      "learning_rate": 0.0005,
+      "loss": 0.871,
+      "step": 388700
+    },
+    {
+      "epoch": 30.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8802,
+      "step": 388800
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8554,
+      "step": 388900
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 0.0005,
+      "loss": 0.8594,
+      "step": 389000
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 0.0005,
+      "loss": 0.8702,
+      "step": 389100
+    },
+    {
+      "epoch": 30.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8748,
+      "step": 389200
+    },
+    {
+      "epoch": 30.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8588,
+      "step": 389300
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8666,
+      "step": 389400
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8618,
+      "step": 389500
+    },
+    {
+      "epoch": 30.55,
+      "learning_rate": 0.0005,
+      "loss": 0.8643,
+      "step": 389600
+    },
+    {
+      "epoch": 30.56,
+      "learning_rate": 0.0005,
+      "loss": 0.8677,
+      "step": 389700
+    },
+    {
+      "epoch": 30.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8715,
+      "step": 389800
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 0.0005,
+      "loss": 0.866,
+      "step": 389900
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 0.0005,
+      "loss": 0.857,
+      "step": 390000
+    },
+    {
+      "epoch": 30.58,
+      "eval_gen_len": 18.76269091551161,
+      "eval_loss": 2.564743757247925,
+      "eval_rouge1": 35.752,
+      "eval_rouge2": 14.8207,
+      "eval_rougeL": 29.5313,
+      "eval_rougeLsum": 29.5352,
+      "eval_runtime": 343.9829,
+      "eval_samples_per_second": 32.929,
+      "eval_steps_per_second": 2.058,
+      "step": 390000
+    },
+    {
+      "epoch": 30.59,
+      "learning_rate": 0.0005,
+      "loss": 0.8713,
+      "step": 390100
+    },
+    {
+      "epoch": 30.6,
+      "learning_rate": 0.0005,
+      "loss": 0.8539,
+      "step": 390200
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8625,
+      "step": 390300
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8884,
+      "step": 390400
+    },
+    {
+      "epoch": 30.62,
+      "learning_rate": 0.0005,
+      "loss": 0.8713,
+      "step": 390500
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8708,
+      "step": 390600
+    },
+    {
+      "epoch": 30.64,
+      "learning_rate": 0.0005,
+      "loss": 0.8752,
+      "step": 390700
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8657,
+      "step": 390800
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8835,
+      "step": 390900
+    },
+    {
+      "epoch": 30.66,
+      "learning_rate": 0.0005,
+      "loss": 0.8784,
+      "step": 391000
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8621,
+      "step": 391100
+    },
+    {
+      "epoch": 30.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8807,
+      "step": 391200
+    },
+    {
+      "epoch": 30.69,
+      "learning_rate": 0.0005,
+      "loss": 0.8958,
+      "step": 391300
+    },
+    {
+      "epoch": 30.69,
+      "learning_rate": 0.0005,
+      "loss": 0.8668,
+      "step": 391400
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8647,
+      "step": 391500
+    },
+    {
+      "epoch": 30.71,
+      "learning_rate": 0.0005,
+      "loss": 0.8656,
+      "step": 391600
+    },
+    {
+      "epoch": 30.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8991,
+      "step": 391700
+    },
+    {
+      "epoch": 30.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8732,
+      "step": 391800
+    },
+    {
+      "epoch": 30.73,
+      "learning_rate": 0.0005,
+      "loss": 0.8672,
+      "step": 391900
+    },
+    {
+      "epoch": 30.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8937,
+      "step": 392000
+    },
+    {
+      "epoch": 30.75,
+      "learning_rate": 0.0005,
+      "loss": 0.8924,
+      "step": 392100
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8955,
+      "step": 392200
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8815,
+      "step": 392300
+    },
+    {
+      "epoch": 30.77,
+      "learning_rate": 0.0005,
+      "loss": 0.8734,
+      "step": 392400
+    },
+    {
+      "epoch": 30.78,
+      "learning_rate": 0.0005,
+      "loss": 0.8741,
+      "step": 392500
+    },
+    {
+      "epoch": 30.79,
+      "learning_rate": 0.0005,
+      "loss": 0.8778,
+      "step": 392600
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 0.0005,
+      "loss": 0.8865,
+      "step": 392700
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 0.0005,
+      "loss": 0.8772,
+      "step": 392800
+    },
+    {
+      "epoch": 30.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8612,
+      "step": 392900
+    },
+    {
+      "epoch": 30.82,
+      "learning_rate": 0.0005,
+      "loss": 0.8835,
+      "step": 393000
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8639,
+      "step": 393100
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 0.0005,
+      "loss": 0.871,
+      "step": 393200
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 0.0005,
+      "loss": 0.8982,
+      "step": 393300
+    },
+    {
+      "epoch": 30.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8853,
+      "step": 393400
+    },
+    {
+      "epoch": 30.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8718,
+      "step": 393500
+    },
+    {
+      "epoch": 30.87,
+      "learning_rate": 0.0005,
+      "loss": 0.9022,
+      "step": 393600
+    },
+    {
+      "epoch": 30.87,
+      "learning_rate": 0.0005,
+      "loss": 0.8702,
+      "step": 393700
+    },
+    {
+      "epoch": 30.88,
+      "learning_rate": 0.0005,
+      "loss": 0.9148,
+      "step": 393800
+    },
+    {
+      "epoch": 30.89,
+      "learning_rate": 0.0005,
+      "loss": 0.9045,
+      "step": 393900
+    },
+    {
+      "epoch": 30.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8866,
+      "step": 394000
+    },
+    {
+      "epoch": 30.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8754,
+      "step": 394100
+    },
+    {
+      "epoch": 30.91,
+      "learning_rate": 0.0005,
+      "loss": 0.8849,
+      "step": 394200
+    },
+    {
+      "epoch": 30.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8729,
+      "step": 394300
+    },
+    {
+      "epoch": 30.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8926,
+      "step": 394400
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8843,
+      "step": 394500
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8856,
+      "step": 394600
+    },
+    {
+      "epoch": 30.95,
+      "learning_rate": 0.0005,
+      "loss": 0.9011,
+      "step": 394700
+    },
+    {
+      "epoch": 30.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8963,
+      "step": 394800
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8861,
+      "step": 394900
+    },
+    {
+      "epoch": 30.98,
+      "learning_rate": 0.0005,
+      "loss": 0.875,
+      "step": 395000
+    },
+    {
+      "epoch": 30.98,
+      "learning_rate": 0.0005,
+      "loss": 0.8904,
+      "step": 395100
+    },
+    {
+      "epoch": 30.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8788,
+      "step": 395200
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 0.0005,
+      "loss": 0.901,
+      "step": 395300
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8241,
+      "step": 395400
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8258,
+      "step": 395500
+    },
+    {
+      "epoch": 31.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8048,
+      "step": 395600
+    },
+    {
+      "epoch": 31.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8037,
+      "step": 395700
+    },
+    {
+      "epoch": 31.04,
+      "learning_rate": 0.0005,
+      "loss": 0.8181,
+      "step": 395800
+    },
+    {
+      "epoch": 31.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8227,
+      "step": 395900
+    },
+    {
+      "epoch": 31.05,
+      "learning_rate": 0.0005,
+      "loss": 0.826,
+      "step": 396000
+    },
+    {
+      "epoch": 31.06,
+      "learning_rate": 0.0005,
+      "loss": 0.8066,
+      "step": 396100
+    },
+    {
+      "epoch": 31.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8147,
+      "step": 396200
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 0.0005,
+      "loss": 0.8222,
+      "step": 396300
+    },
+    {
+      "epoch": 31.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8213,
+      "step": 396400
+    },
+    {
+      "epoch": 31.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8357,
+      "step": 396500
+    },
+    {
+      "epoch": 31.1,
+      "learning_rate": 0.0005,
+      "loss": 0.8089,
+      "step": 396600
+    },
+    {
+      "epoch": 31.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8465,
+      "step": 396700
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8177,
+      "step": 396800
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8205,
+      "step": 396900
+    },
+    {
+      "epoch": 31.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8268,
+      "step": 397000
+    },
+    {
+      "epoch": 31.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8385,
+      "step": 397100
+    },
+    {
+      "epoch": 31.15,
+      "learning_rate": 0.0005,
+      "loss": 0.8185,
+      "step": 397200
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8325,
+      "step": 397300
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8368,
+      "step": 397400
+    },
+    {
+      "epoch": 31.17,
+      "learning_rate": 0.0005,
+      "loss": 0.824,
+      "step": 397500
+    },
+    {
+      "epoch": 31.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8224,
+      "step": 397600
+    },
+    {
+      "epoch": 31.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8478,
+      "step": 397700
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8418,
+      "step": 397800
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 0.0005,
+      "loss": 0.831,
+      "step": 397900
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8311,
+      "step": 398000
+    },
+    {
+      "epoch": 31.22,
+      "learning_rate": 0.0005,
+      "loss": 0.8241,
+      "step": 398100
+    },
+    {
+      "epoch": 31.23,
+      "learning_rate": 0.0005,
+      "loss": 0.8347,
+      "step": 398200
+    },
+    {
+      "epoch": 31.23,
+      "learning_rate": 0.0005,
+      "loss": 0.8421,
+      "step": 398300
+    },
+    {
+      "epoch": 31.24,
+      "learning_rate": 0.0005,
+      "loss": 0.8371,
+      "step": 398400
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8403,
+      "step": 398500
+    },
+    {
+      "epoch": 31.26,
+      "learning_rate": 0.0005,
+      "loss": 0.8554,
+      "step": 398600
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8366,
+      "step": 398700
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 0.0005,
+      "loss": 0.835,
+      "step": 398800
+    },
+    {
+      "epoch": 31.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8565,
+      "step": 398900
+    },
+    {
+      "epoch": 31.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8365,
+      "step": 399000
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8549,
+      "step": 399100
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8254,
+      "step": 399200
+    },
+    {
+      "epoch": 31.31,
+      "learning_rate": 0.0005,
+      "loss": 0.8566,
+      "step": 399300
+    },
+    {
+      "epoch": 31.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8628,
+      "step": 399400
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8504,
+      "step": 399500
+    },
+    {
+      "epoch": 31.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8688,
+      "step": 399600
+    },
+    {
+      "epoch": 31.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8524,
+      "step": 399700
+    },
+    {
+      "epoch": 31.35,
+      "learning_rate": 0.0005,
+      "loss": 0.8486,
+      "step": 399800
+    },
+    {
+      "epoch": 31.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8542,
+      "step": 399900
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 0.0005,
+      "loss": 0.8505,
+      "step": 400000
+    },
+    {
+      "epoch": 31.37,
+      "eval_gen_len": 18.719872870133308,
+      "eval_loss": 2.595099925994873,
+      "eval_rouge1": 35.7844,
+      "eval_rouge2": 14.8342,
+      "eval_rougeL": 29.5181,
+      "eval_rougeLsum": 29.5218,
+      "eval_runtime": 342.3534,
+      "eval_samples_per_second": 33.086,
+      "eval_steps_per_second": 2.068,
+      "step": 400000
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8618,
+      "step": 400100
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8535,
+      "step": 400200
+    },
+    {
+      "epoch": 31.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8547,
+      "step": 400300
+    },
+    {
+      "epoch": 31.4,
+      "learning_rate": 0.0005,
+      "loss": 0.8452,
+      "step": 400400
+    },
+    {
+      "epoch": 31.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8375,
+      "step": 400500
+    },
+    {
+      "epoch": 31.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8443,
+      "step": 400600
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 0.0005,
+      "loss": 0.8542,
+      "step": 400700
+    },
+    {
+      "epoch": 31.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8519,
+      "step": 400800
+    },
+    {
+      "epoch": 31.44,
+      "learning_rate": 0.0005,
+      "loss": 0.8681,
+      "step": 400900
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8433,
+      "step": 401000
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8371,
+      "step": 401100
+    },
+    {
+      "epoch": 31.46,
+      "learning_rate": 0.0005,
+      "loss": 0.8632,
+      "step": 401200
+    },
+    {
+      "epoch": 31.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8532,
+      "step": 401300
+    },
+    {
+      "epoch": 31.48,
+      "learning_rate": 0.0005,
+      "loss": 0.8427,
+      "step": 401400
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8553,
+      "step": 401500
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8598,
+      "step": 401600
+    },
+    {
+      "epoch": 31.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8512,
+      "step": 401700
+    },
+    {
+      "epoch": 31.51,
+      "learning_rate": 0.0005,
+      "loss": 0.8387,
+      "step": 401800
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 0.0005,
+      "loss": 0.855,
+      "step": 401900
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8589,
+      "step": 402000
+    },
+    {
+      "epoch": 31.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8537,
+      "step": 402100
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8422,
+      "step": 402200
+    },
+    {
+      "epoch": 31.55,
+      "learning_rate": 0.0005,
+      "loss": 0.8759,
+      "step": 402300
+    },
+    {
+      "epoch": 31.56,
+      "learning_rate": 0.0005,
+      "loss": 0.8618,
+      "step": 402400
+    },
+    {
+      "epoch": 31.56,
+      "learning_rate": 0.0005,
+      "loss": 0.859,
+      "step": 402500
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8693,
+      "step": 402600
+    },
+    {
+      "epoch": 31.58,
+      "learning_rate": 0.0005,
+      "loss": 0.863,
+      "step": 402700
+    },
+    {
+      "epoch": 31.59,
+      "learning_rate": 0.0005,
+      "loss": 0.8679,
+      "step": 402800
+    },
+    {
+      "epoch": 31.6,
+      "learning_rate": 0.0005,
+      "loss": 0.866,
+      "step": 402900
+    },
+    {
+      "epoch": 31.6,
+      "learning_rate": 0.0005,
+      "loss": 0.8508,
+      "step": 403000
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 0.0005,
+      "loss": 0.851,
+      "step": 403100
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 0.0005,
+      "loss": 0.8433,
+      "step": 403200
+    },
+    {
+      "epoch": 31.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8641,
+      "step": 403300
+    },
+    {
+      "epoch": 31.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8781,
+      "step": 403400
+    },
+    {
+      "epoch": 31.64,
+      "learning_rate": 0.0005,
+      "loss": 0.8403,
+      "step": 403500
+    },
+    {
+      "epoch": 31.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8734,
+      "step": 403600
+    },
+    {
+      "epoch": 31.66,
+      "learning_rate": 0.0005,
+      "loss": 0.8626,
+      "step": 403700
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8661,
+      "step": 403800
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8631,
+      "step": 403900
+    },
+    {
+      "epoch": 31.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8629,
+      "step": 404000
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 0.0005,
+      "loss": 0.8507,
+      "step": 404100
+    },
+    {
+      "epoch": 31.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8746,
+      "step": 404200
+    },
+    {
+      "epoch": 31.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8681,
+      "step": 404300
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 0.0005,
+      "loss": 0.8592,
+      "step": 404400
+    },
+    {
+      "epoch": 31.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8675,
+      "step": 404500
+    },
+    {
+      "epoch": 31.73,
+      "learning_rate": 0.0005,
+      "loss": 0.8671,
+      "step": 404600
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8652,
+      "step": 404700
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8773,
+      "step": 404800
+    },
+    {
+      "epoch": 31.75,
+      "learning_rate": 0.0005,
+      "loss": 0.8654,
+      "step": 404900
+    },
+    {
+      "epoch": 31.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8605,
+      "step": 405000
+    },
+    {
+      "epoch": 31.77,
+      "learning_rate": 0.0005,
+      "loss": 0.8804,
+      "step": 405100
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 0.0005,
+      "loss": 0.8659,
+      "step": 405200
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 0.0005,
+      "loss": 0.8667,
+      "step": 405300
+    },
+    {
+      "epoch": 31.79,
+      "learning_rate": 0.0005,
+      "loss": 0.8781,
+      "step": 405400
+    },
+    {
+      "epoch": 31.8,
+      "learning_rate": 0.0005,
+      "loss": 0.8708,
+      "step": 405500
+    },
+    {
+      "epoch": 31.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8592,
+      "step": 405600
+    },
+    {
+      "epoch": 31.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8699,
+      "step": 405700
+    },
+    {
+      "epoch": 31.82,
+      "learning_rate": 0.0005,
+      "loss": 0.8521,
+      "step": 405800
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8488,
+      "step": 405900
+    },
+    {
+      "epoch": 31.84,
+      "learning_rate": 0.0005,
+      "loss": 0.8702,
+      "step": 406000
+    },
+    {
+      "epoch": 31.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8754,
+      "step": 406100
+    },
+    {
+      "epoch": 31.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8887,
+      "step": 406200
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8962,
+      "step": 406300
+    },
+    {
+      "epoch": 31.87,
+      "learning_rate": 0.0005,
+      "loss": 0.8809,
+      "step": 406400
+    },
+    {
+      "epoch": 31.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8716,
+      "step": 406500
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8661,
+      "step": 406600
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8727,
+      "step": 406700
+    },
+    {
+      "epoch": 31.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8766,
+      "step": 406800
+    },
+    {
+      "epoch": 31.91,
+      "learning_rate": 0.0005,
+      "loss": 0.8715,
+      "step": 406900
+    },
+    {
+      "epoch": 31.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8755,
+      "step": 407000
+    },
+    {
+      "epoch": 31.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8814,
+      "step": 407100
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8662,
+      "step": 407200
+    },
+    {
+      "epoch": 31.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8679,
+      "step": 407300
+    },
+    {
+      "epoch": 31.95,
+      "learning_rate": 0.0005,
+      "loss": 0.884,
+      "step": 407400
+    },
+    {
+      "epoch": 31.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8806,
+      "step": 407500
+    },
+    {
+      "epoch": 31.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8633,
+      "step": 407600
+    },
+    {
+      "epoch": 31.97,
+      "learning_rate": 0.0005,
+      "loss": 0.876,
+      "step": 407700
+    },
+    {
+      "epoch": 31.98,
+      "learning_rate": 0.0005,
+      "loss": 0.8903,
+      "step": 407800
+    },
+    {
+      "epoch": 31.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8641,
+      "step": 407900
+    },
+    {
+      "epoch": 31.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8814,
+      "step": 408000
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 0.0005,
+      "loss": 0.8659,
+      "step": 408100
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8103,
+      "step": 408200
+    },
+    {
+      "epoch": 32.02,
+      "learning_rate": 0.0005,
+      "loss": 0.8034,
+      "step": 408300
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8117,
+      "step": 408400
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 0.0005,
+      "loss": 0.8234,
+      "step": 408500
+    },
+    {
+      "epoch": 32.04,
+      "learning_rate": 0.0005,
+      "loss": 0.8076,
+      "step": 408600
+    },
+    {
+      "epoch": 32.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8231,
+      "step": 408700
+    },
+    {
+      "epoch": 32.06,
+      "learning_rate": 0.0005,
+      "loss": 0.8245,
+      "step": 408800
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8092,
+      "step": 408900
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 0.0005,
+      "loss": 0.8033,
+      "step": 409000
+    },
+    {
+      "epoch": 32.08,
+      "learning_rate": 0.0005,
+      "loss": 0.802,
+      "step": 409100
+    },
+    {
+      "epoch": 32.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8198,
+      "step": 409200
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 0.0005,
+      "loss": 0.8133,
+      "step": 409300
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 0.0005,
+      "loss": 0.8177,
+      "step": 409400
+    },
+    {
+      "epoch": 32.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8127,
+      "step": 409500
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8241,
+      "step": 409600
+    },
+    {
+      "epoch": 32.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8276,
+      "step": 409700
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 0.0005,
+      "loss": 0.815,
+      "step": 409800
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8279,
+      "step": 409900
+    },
+    {
+      "epoch": 32.15,
+      "learning_rate": 0.0005,
+      "loss": 0.8233,
+      "step": 410000
+    },
+    {
+      "epoch": 32.15,
+      "eval_gen_len": 18.735410964951,
+      "eval_loss": 2.6317381858825684,
+      "eval_rouge1": 35.8198,
+      "eval_rouge2": 14.9264,
+      "eval_rougeL": 29.5674,
+      "eval_rougeLsum": 29.564,
+      "eval_runtime": 343.4071,
+      "eval_samples_per_second": 32.984,
+      "eval_steps_per_second": 2.062,
+      "step": 410000
+    },
+    {
+      "epoch": 32.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8139,
+      "step": 410100
+    },
+    {
+      "epoch": 32.17,
+      "learning_rate": 0.0005,
+      "loss": 0.823,
+      "step": 410200
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 0.0005,
+      "loss": 0.819,
+      "step": 410300
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8439,
+      "step": 410400
+    },
+    {
+      "epoch": 32.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8349,
+      "step": 410500
+    },
+    {
+      "epoch": 32.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8028,
+      "step": 410600
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8309,
+      "step": 410700
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8243,
+      "step": 410800
+    },
+    {
+      "epoch": 32.22,
+      "learning_rate": 0.0005,
+      "loss": 0.8322,
+      "step": 410900
+    },
+    {
+      "epoch": 32.23,
+      "learning_rate": 0.0005,
+      "loss": 0.816,
+      "step": 411000
+    },
+    {
+      "epoch": 32.24,
+      "learning_rate": 0.0005,
+      "loss": 0.8437,
+      "step": 411100
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8385,
+      "step": 411200
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8055,
+      "step": 411300
+    },
+    {
+      "epoch": 32.26,
+      "learning_rate": 0.0005,
+      "loss": 0.8231,
+      "step": 411400
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8337,
+      "step": 411500
+    },
+    {
+      "epoch": 32.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8319,
+      "step": 411600
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8245,
+      "step": 411700
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8285,
+      "step": 411800
+    },
+    {
+      "epoch": 32.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8207,
+      "step": 411900
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 0.0005,
+      "loss": 0.8266,
+      "step": 412000
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 0.0005,
+      "loss": 0.832,
+      "step": 412100
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 0.0005,
+      "loss": 0.84,
+      "step": 412200
+    },
+    {
+      "epoch": 32.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8291,
+      "step": 412300
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8189,
+      "step": 412400
+    },
+    {
+      "epoch": 32.35,
+      "learning_rate": 0.0005,
+      "loss": 0.8504,
+      "step": 412500
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 0.0005,
+      "loss": 0.843,
+      "step": 412600
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8379,
+      "step": 412700
+    },
+    {
+      "epoch": 32.37,
+      "learning_rate": 0.0005,
+      "loss": 0.8225,
+      "step": 412800
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8354,
+      "step": 412900
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8359,
+      "step": 413000
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8427,
+      "step": 413100
+    },
+    {
+      "epoch": 32.4,
+      "learning_rate": 0.0005,
+      "loss": 0.8146,
+      "step": 413200
+    },
+    {
+      "epoch": 32.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8387,
+      "step": 413300
+    },
+    {
+      "epoch": 32.42,
+      "learning_rate": 0.0005,
+      "loss": 0.8524,
+      "step": 413400
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8341,
+      "step": 413500
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8378,
+      "step": 413600
+    },
+    {
+      "epoch": 32.44,
+      "learning_rate": 0.0005,
+      "loss": 0.8534,
+      "step": 413700
+    },
+    {
+      "epoch": 32.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8507,
+      "step": 413800
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 0.0005,
+      "loss": 0.8479,
+      "step": 413900
+    },
+    {
+      "epoch": 32.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8138,
+      "step": 414000
+    },
+    {
+      "epoch": 32.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8471,
+      "step": 414100
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 0.0005,
+      "loss": 0.8454,
+      "step": 414200
+    },
+    {
+      "epoch": 32.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8327,
+      "step": 414300
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8529,
+      "step": 414400
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8436,
+      "step": 414500
+    },
+    {
+      "epoch": 32.51,
+      "learning_rate": 0.0005,
+      "loss": 0.8353,
+      "step": 414600
+    },
+    {
+      "epoch": 32.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8485,
+      "step": 414700
+    },
+    {
+      "epoch": 32.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8327,
+      "step": 414800
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 0.0005,
+      "loss": 0.832,
+      "step": 414900
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8455,
+      "step": 415000
+    },
+    {
+      "epoch": 32.55,
+      "learning_rate": 0.0005,
+      "loss": 0.8402,
+      "step": 415100
+    },
+    {
+      "epoch": 32.56,
+      "learning_rate": 0.0005,
+      "loss": 0.8399,
+      "step": 415200
+    },
+    {
+      "epoch": 32.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8634,
+      "step": 415300
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 0.0005,
+      "loss": 0.8414,
+      "step": 415400
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 0.0005,
+      "loss": 0.848,
+      "step": 415500
+    },
+    {
+      "epoch": 32.59,
+      "learning_rate": 0.0005,
+      "loss": 0.8433,
+      "step": 415600
+    },
+    {
+      "epoch": 32.6,
+      "learning_rate": 0.0005,
+      "loss": 0.85,
+      "step": 415700
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8562,
+      "step": 415800
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8374,
+      "step": 415900
+    },
+    {
+      "epoch": 32.62,
+      "learning_rate": 0.0005,
+      "loss": 0.8537,
+      "step": 416000
+    },
+    {
+      "epoch": 32.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8536,
+      "step": 416100
+    },
+    {
+      "epoch": 32.64,
+      "learning_rate": 0.0005,
+      "loss": 0.8637,
+      "step": 416200
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8491,
+      "step": 416300
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 0.0005,
+      "loss": 0.851,
+      "step": 416400
+    },
+    {
+      "epoch": 32.66,
+      "learning_rate": 0.0005,
+      "loss": 0.8657,
+      "step": 416500
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8447,
+      "step": 416600
+    },
+    {
+      "epoch": 32.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8502,
+      "step": 416700
+    },
+    {
+      "epoch": 32.69,
+      "learning_rate": 0.0005,
+      "loss": 0.8609,
+      "step": 416800
+    },
+    {
+      "epoch": 32.69,
+      "learning_rate": 0.0005,
+      "loss": 0.8596,
+      "step": 416900
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8572,
+      "step": 417000
+    },
+    {
+      "epoch": 32.71,
+      "learning_rate": 0.0005,
+      "loss": 0.8593,
+      "step": 417100
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8561,
+      "step": 417200
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8424,
+      "step": 417300
+    },
+    {
+      "epoch": 32.73,
+      "learning_rate": 0.0005,
+      "loss": 0.8594,
+      "step": 417400
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8809,
+      "step": 417500
+    },
+    {
+      "epoch": 32.75,
+      "learning_rate": 0.0005,
+      "loss": 0.8634,
+      "step": 417600
+    },
+    {
+      "epoch": 32.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8524,
+      "step": 417700
+    },
+    {
+      "epoch": 32.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8551,
+      "step": 417800
+    },
+    {
+      "epoch": 32.77,
+      "learning_rate": 0.0005,
+      "loss": 0.8659,
+      "step": 417900
+    },
+    {
+      "epoch": 32.78,
+      "learning_rate": 0.0005,
+      "loss": 0.8441,
+      "step": 418000
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 0.0005,
+      "loss": 0.8558,
+      "step": 418100
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 0.0005,
+      "loss": 0.8747,
+      "step": 418200
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 0.0005,
+      "loss": 0.8671,
+      "step": 418300
+    },
+    {
+      "epoch": 32.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8614,
+      "step": 418400
+    },
+    {
+      "epoch": 32.82,
+      "learning_rate": 0.0005,
+      "loss": 0.8528,
+      "step": 418500
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8658,
+      "step": 418600
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8742,
+      "step": 418700
+    },
+    {
+      "epoch": 32.84,
+      "learning_rate": 0.0005,
+      "loss": 0.868,
+      "step": 418800
+    },
+    {
+      "epoch": 32.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8659,
+      "step": 418900
+    },
+    {
+      "epoch": 32.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8639,
+      "step": 419000
+    },
+    {
+      "epoch": 32.87,
+      "learning_rate": 0.0005,
+      "loss": 0.8603,
+      "step": 419100
+    },
+    {
+      "epoch": 32.87,
+      "learning_rate": 0.0005,
+      "loss": 0.8698,
+      "step": 419200
+    },
+    {
+      "epoch": 32.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8699,
+      "step": 419300
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8617,
+      "step": 419400
+    },
+    {
+      "epoch": 32.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8572,
+      "step": 419500
+    },
+    {
+      "epoch": 32.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8419,
+      "step": 419600
+    },
+    {
+      "epoch": 32.91,
+      "learning_rate": 0.0005,
+      "loss": 0.8746,
+      "step": 419700
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8786,
+      "step": 419800
+    },
+    {
+      "epoch": 32.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8527,
+      "step": 419900
+    },
+    {
+      "epoch": 32.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8825,
+      "step": 420000
+    },
+    {
+      "epoch": 32.94,
+      "eval_gen_len": 18.7486536593979,
+      "eval_loss": 2.5763282775878906,
+      "eval_rouge1": 35.6833,
+      "eval_rouge2": 14.7874,
+      "eval_rougeL": 29.531,
+      "eval_rougeLsum": 29.5185,
+      "eval_runtime": 344.2647,
+      "eval_samples_per_second": 32.902,
+      "eval_steps_per_second": 2.057,
+      "step": 420000
+    },
+    {
+      "epoch": 32.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8753,
+      "step": 420100
+    },
+    {
+      "epoch": 32.95,
+      "learning_rate": 0.0005,
+      "loss": 0.8629,
+      "step": 420200
+    },
+    {
+      "epoch": 32.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8628,
+      "step": 420300
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8622,
+      "step": 420400
+    },
+    {
+      "epoch": 32.98,
+      "learning_rate": 0.0005,
+      "loss": 0.8761,
+      "step": 420500
+    },
+    {
+      "epoch": 32.98,
+      "learning_rate": 0.0005,
+      "loss": 0.8817,
+      "step": 420600
+    },
+    {
+      "epoch": 32.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8628,
+      "step": 420700
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 0.0005,
+      "loss": 0.8663,
+      "step": 420800
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 0.0005,
+      "loss": 0.8209,
+      "step": 420900
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 0.0005,
+      "loss": 0.7882,
+      "step": 421000
+    },
+    {
+      "epoch": 33.02,
+      "learning_rate": 0.0005,
+      "loss": 0.7939,
+      "step": 421100
+    },
+    {
+      "epoch": 33.03,
+      "learning_rate": 0.0005,
+      "loss": 0.7939,
+      "step": 421200
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 0.0005,
+      "loss": 0.7872,
+      "step": 421300
+    },
+    {
+      "epoch": 33.05,
+      "learning_rate": 0.0005,
+      "loss": 0.7913,
+      "step": 421400
+    },
+    {
+      "epoch": 33.05,
+      "learning_rate": 0.0005,
+      "loss": 0.8159,
+      "step": 421500
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 0.0005,
+      "loss": 0.7871,
+      "step": 421600
+    },
+    {
+      "epoch": 33.07,
+      "learning_rate": 0.0005,
+      "loss": 0.7965,
+      "step": 421700
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 0.0005,
+      "loss": 0.7984,
+      "step": 421800
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8078,
+      "step": 421900
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 0.0005,
+      "loss": 0.8111,
+      "step": 422000
+    },
+    {
+      "epoch": 33.1,
+      "learning_rate": 0.0005,
+      "loss": 0.7984,
+      "step": 422100
+    },
+    {
+      "epoch": 33.11,
+      "learning_rate": 0.0005,
+      "loss": 0.8112,
+      "step": 422200
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8259,
+      "step": 422300
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8179,
+      "step": 422400
+    },
+    {
+      "epoch": 33.13,
+      "learning_rate": 0.0005,
+      "loss": 0.8379,
+      "step": 422500
+    },
+    {
+      "epoch": 33.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8086,
+      "step": 422600
+    },
+    {
+      "epoch": 33.15,
+      "learning_rate": 0.0005,
+      "loss": 0.8182,
+      "step": 422700
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8001,
+      "step": 422800
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8129,
+      "step": 422900
+    },
+    {
+      "epoch": 33.17,
+      "learning_rate": 0.0005,
+      "loss": 0.8063,
+      "step": 423000
+    },
+    {
+      "epoch": 33.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8167,
+      "step": 423100
+    },
+    {
+      "epoch": 33.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8182,
+      "step": 423200
+    },
+    {
+      "epoch": 33.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8283,
+      "step": 423300
+    },
+    {
+      "epoch": 33.2,
+      "learning_rate": 0.0005,
+      "loss": 0.807,
+      "step": 423400
+    },
+    {
+      "epoch": 33.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8323,
+      "step": 423500
+    },
+    {
+      "epoch": 33.22,
+      "learning_rate": 0.0005,
+      "loss": 0.8145,
+      "step": 423600
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 0.0005,
+      "loss": 0.7967,
+      "step": 423700
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 0.0005,
+      "loss": 0.8177,
+      "step": 423800
+    },
+    {
+      "epoch": 33.24,
+      "learning_rate": 0.0005,
+      "loss": 0.8056,
+      "step": 423900
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8275,
+      "step": 424000
+    },
+    {
+      "epoch": 33.26,
+      "learning_rate": 0.0005,
+      "loss": 0.8216,
+      "step": 424100
+    },
+    {
+      "epoch": 33.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8307,
+      "step": 424200
+    },
+    {
+      "epoch": 33.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8285,
+      "step": 424300
+    },
+    {
+      "epoch": 33.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8256,
+      "step": 424400
+    },
+    {
+      "epoch": 33.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8234,
+      "step": 424500
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8097,
+      "step": 424600
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8142,
+      "step": 424700
+    },
+    {
+      "epoch": 33.31,
+      "learning_rate": 0.0005,
+      "loss": 0.8241,
+      "step": 424800
+    },
+    {
+      "epoch": 33.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8156,
+      "step": 424900
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8056,
+      "step": 425000
+    },
+    {
+      "epoch": 33.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8278,
+      "step": 425100
+    },
+    {
+      "epoch": 33.34,
+      "learning_rate": 0.0005,
+      "loss": 0.817,
+      "step": 425200
+    },
+    {
+      "epoch": 33.35,
+      "learning_rate": 0.0005,
+      "loss": 0.8169,
+      "step": 425300
+    },
+    {
+      "epoch": 33.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8139,
+      "step": 425400
+    },
+    {
+      "epoch": 33.37,
+      "learning_rate": 0.0005,
+      "loss": 0.8209,
+      "step": 425500
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8239,
+      "step": 425600
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8167,
+      "step": 425700
+    },
+    {
+      "epoch": 33.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8227,
+      "step": 425800
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 0.0005,
+      "loss": 0.834,
+      "step": 425900
+    },
+    {
+      "epoch": 33.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8476,
+      "step": 426000
+    },
+    {
+      "epoch": 33.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8246,
+      "step": 426100
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 0.0005,
+      "loss": 0.8288,
+      "step": 426200
+    },
+    {
+      "epoch": 33.43,
+      "learning_rate": 0.0005,
+      "loss": 0.823,
+      "step": 426300
+    },
+    {
+      "epoch": 33.44,
+      "learning_rate": 0.0005,
+      "loss": 0.8368,
+      "step": 426400
+    },
+    {
+      "epoch": 33.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8374,
+      "step": 426500
+    },
+    {
+      "epoch": 33.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8432,
+      "step": 426600
+    },
+    {
+      "epoch": 33.46,
+      "learning_rate": 0.0005,
+      "loss": 0.8436,
+      "step": 426700
+    },
+    {
+      "epoch": 33.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8309,
+      "step": 426800
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 0.0005,
+      "loss": 0.8322,
+      "step": 426900
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 0.0005,
+      "loss": 0.8413,
+      "step": 427000
+    },
+    {
+      "epoch": 33.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8403,
+      "step": 427100
+    },
+    {
+      "epoch": 33.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8394,
+      "step": 427200
+    },
+    {
+      "epoch": 33.51,
+      "learning_rate": 0.0005,
+      "loss": 0.843,
+      "step": 427300
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8253,
+      "step": 427400
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8426,
+      "step": 427500
+    },
+    {
+      "epoch": 33.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8413,
+      "step": 427600
+    },
+    {
+      "epoch": 33.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8456,
+      "step": 427700
+    },
+    {
+      "epoch": 33.55,
+      "learning_rate": 0.0005,
+      "loss": 0.8459,
+      "step": 427800
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 0.0005,
+      "loss": 0.8189,
+      "step": 427900
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 0.0005,
+      "loss": 0.8424,
+      "step": 428000
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8268,
+      "step": 428100
+    },
+    {
+      "epoch": 33.58,
+      "learning_rate": 0.0005,
+      "loss": 0.8405,
+      "step": 428200
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 0.0005,
+      "loss": 0.8277,
+      "step": 428300
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 0.0005,
+      "loss": 0.8327,
+      "step": 428400
+    },
+    {
+      "epoch": 33.6,
+      "learning_rate": 0.0005,
+      "loss": 0.8342,
+      "step": 428500
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8367,
+      "step": 428600
+    },
+    {
+      "epoch": 33.62,
+      "learning_rate": 0.0005,
+      "loss": 0.829,
+      "step": 428700
+    },
+    {
+      "epoch": 33.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8265,
+      "step": 428800
+    },
+    {
+      "epoch": 33.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8416,
+      "step": 428900
+    },
+    {
+      "epoch": 33.64,
+      "learning_rate": 0.0005,
+      "loss": 0.8204,
+      "step": 429000
+    },
+    {
+      "epoch": 33.65,
+      "learning_rate": 0.0005,
+      "loss": 0.834,
+      "step": 429100
+    },
+    {
+      "epoch": 33.66,
+      "learning_rate": 0.0005,
+      "loss": 0.8527,
+      "step": 429200
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8515,
+      "step": 429300
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8335,
+      "step": 429400
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8428,
+      "step": 429500
+    },
+    {
+      "epoch": 33.69,
+      "learning_rate": 0.0005,
+      "loss": 0.8431,
+      "step": 429600
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8593,
+      "step": 429700
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8496,
+      "step": 429800
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 0.0005,
+      "loss": 0.8791,
+      "step": 429900
+    },
+    {
+      "epoch": 33.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8553,
+      "step": 430000
+    },
+    {
+      "epoch": 33.72,
+      "eval_gen_len": 18.7672817162532,
+      "eval_loss": 2.5715715885162354,
+      "eval_rouge1": 35.804,
+      "eval_rouge2": 14.8203,
+      "eval_rougeL": 29.528,
+      "eval_rougeLsum": 29.5216,
+      "eval_runtime": 342.741,
+      "eval_samples_per_second": 33.048,
+      "eval_steps_per_second": 2.066,
+      "step": 430000
+    },
+    {
+      "epoch": 33.73,
+      "learning_rate": 0.0005,
+      "loss": 0.852,
+      "step": 430100
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8455,
+      "step": 430200
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 0.0005,
+      "loss": 0.86,
+      "step": 430300
+    },
+    {
+      "epoch": 33.75,
+      "learning_rate": 0.0005,
+      "loss": 0.8403,
+      "step": 430400
+    },
+    {
+      "epoch": 33.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8532,
+      "step": 430500
+    },
+    {
+      "epoch": 33.77,
+      "learning_rate": 0.0005,
+      "loss": 0.8397,
+      "step": 430600
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 0.0005,
+      "loss": 0.8421,
+      "step": 430700
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 0.0005,
+      "loss": 0.8572,
+      "step": 430800
+    },
+    {
+      "epoch": 33.79,
+      "learning_rate": 0.0005,
+      "loss": 0.8428,
+      "step": 430900
+    },
+    {
+      "epoch": 33.8,
+      "learning_rate": 0.0005,
+      "loss": 0.8125,
+      "step": 431000
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8541,
+      "step": 431100
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8338,
+      "step": 431200
+    },
+    {
+      "epoch": 33.82,
+      "learning_rate": 0.0005,
+      "loss": 0.8576,
+      "step": 431300
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8411,
+      "step": 431400
+    },
+    {
+      "epoch": 33.84,
+      "learning_rate": 0.0005,
+      "loss": 0.8495,
+      "step": 431500
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8603,
+      "step": 431600
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8696,
+      "step": 431700
+    },
+    {
+      "epoch": 33.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8594,
+      "step": 431800
+    },
+    {
+      "epoch": 33.87,
+      "learning_rate": 0.0005,
+      "loss": 0.8386,
+      "step": 431900
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8496,
+      "step": 432000
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8465,
+      "step": 432100
+    },
+    {
+      "epoch": 33.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8594,
+      "step": 432200
+    },
+    {
+      "epoch": 33.9,
+      "learning_rate": 0.0005,
+      "loss": 0.854,
+      "step": 432300
+    },
+    {
+      "epoch": 33.91,
+      "learning_rate": 0.0005,
+      "loss": 0.8447,
+      "step": 432400
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8608,
+      "step": 432500
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8572,
+      "step": 432600
+    },
+    {
+      "epoch": 33.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8624,
+      "step": 432700
+    },
+    {
+      "epoch": 33.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8635,
+      "step": 432800
+    },
+    {
+      "epoch": 33.95,
+      "learning_rate": 0.0005,
+      "loss": 0.8521,
+      "step": 432900
+    },
+    {
+      "epoch": 33.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8615,
+      "step": 433000
+    },
+    {
+      "epoch": 33.96,
+      "learning_rate": 0.0005,
+      "loss": 0.835,
+      "step": 433100
+    },
+    {
+      "epoch": 33.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8391,
+      "step": 433200
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 0.0005,
+      "loss": 0.844,
+      "step": 433300
+    },
+    {
+      "epoch": 33.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8657,
+      "step": 433400
+    },
+    {
+      "epoch": 33.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8686,
+      "step": 433500
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 0.0005,
+      "loss": 0.8344,
+      "step": 433600
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 0.0005,
+      "loss": 0.7766,
+      "step": 433700
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 0.0005,
+      "loss": 0.7651,
+      "step": 433800
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 0.0005,
+      "loss": 0.7894,
+      "step": 433900
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 0.0005,
+      "loss": 0.783,
+      "step": 434000
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 0.0005,
+      "loss": 0.7997,
+      "step": 434100
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 0.0005,
+      "loss": 0.7916,
+      "step": 434200
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 0.0005,
+      "loss": 0.806,
+      "step": 434300
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 0.0005,
+      "loss": 0.7905,
+      "step": 434400
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 0.0005,
+      "loss": 0.7827,
+      "step": 434500
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 0.0005,
+      "loss": 0.7867,
+      "step": 434600
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 0.0005,
+      "loss": 0.7875,
+      "step": 434700
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 0.0005,
+      "loss": 0.798,
+      "step": 434800
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 0.0005,
+      "loss": 0.8086,
+      "step": 434900
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 0.0005,
+      "loss": 0.7837,
+      "step": 435000
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 0.0005,
+      "loss": 0.8006,
+      "step": 435100
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 0.0005,
+      "loss": 0.7904,
+      "step": 435200
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 0.0005,
+      "loss": 0.7889,
+      "step": 435300
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8047,
+      "step": 435400
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 0.0005,
+      "loss": 0.8064,
+      "step": 435500
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 0.0005,
+      "loss": 0.7976,
+      "step": 435600
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 0.0005,
+      "loss": 0.8184,
+      "step": 435700
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 0.0005,
+      "loss": 0.8029,
+      "step": 435800
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 0.0005,
+      "loss": 0.814,
+      "step": 435900
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8016,
+      "step": 436000
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 0.0005,
+      "loss": 0.8048,
+      "step": 436100
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8168,
+      "step": 436200
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 0.0005,
+      "loss": 0.8007,
+      "step": 436300
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 0.0005,
+      "loss": 0.7969,
+      "step": 436400
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 0.0005,
+      "loss": 0.8047,
+      "step": 436500
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 0.0005,
+      "loss": 0.81,
+      "step": 436600
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8079,
+      "step": 436700
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 0.0005,
+      "loss": 0.7928,
+      "step": 436800
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 0.0005,
+      "loss": 0.8157,
+      "step": 436900
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8159,
+      "step": 437000
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8248,
+      "step": 437100
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8182,
+      "step": 437200
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8099,
+      "step": 437300
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8223,
+      "step": 437400
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 0.0005,
+      "loss": 0.8014,
+      "step": 437500
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8313,
+      "step": 437600
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8054,
+      "step": 437700
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8245,
+      "step": 437800
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8153,
+      "step": 437900
+    },
+    {
+      "epoch": 34.35,
+      "learning_rate": 0.0005,
+      "loss": 0.8125,
+      "step": 438000
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8074,
+      "step": 438100
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8155,
+      "step": 438200
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 0.0005,
+      "loss": 0.8164,
+      "step": 438300
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8073,
+      "step": 438400
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 0.0005,
+      "loss": 0.83,
+      "step": 438500
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8174,
+      "step": 438600
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 0.0005,
+      "loss": 0.8199,
+      "step": 438700
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8117,
+      "step": 438800
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 0.0005,
+      "loss": 0.8318,
+      "step": 438900
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8325,
+      "step": 439000
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 0.0005,
+      "loss": 0.815,
+      "step": 439100
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 0.0005,
+      "loss": 0.8214,
+      "step": 439200
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8253,
+      "step": 439300
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 0.0005,
+      "loss": 0.841,
+      "step": 439400
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8074,
+      "step": 439500
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8385,
+      "step": 439600
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 0.0005,
+      "loss": 0.822,
+      "step": 439700
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8159,
+      "step": 439800
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8264,
+      "step": 439900
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8183,
+      "step": 440000
+    },
+    {
+      "epoch": 34.5,
+      "eval_gen_len": 18.763838615697008,
+      "eval_loss": 2.6066930294036865,
+      "eval_rouge1": 35.8093,
+      "eval_rouge2": 14.8115,
+      "eval_rougeL": 29.5759,
+      "eval_rougeLsum": 29.5784,
+      "eval_runtime": 344.3419,
+      "eval_samples_per_second": 32.895,
+      "eval_steps_per_second": 2.056,
+      "step": 440000
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 0.0005,
+      "loss": 0.812,
+      "step": 440100
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 0.0005,
+      "loss": 0.84,
+      "step": 440200
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8214,
+      "step": 440300
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8131,
+      "step": 440400
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8257,
+      "step": 440500
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 0.0005,
+      "loss": 0.831,
+      "step": 440600
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 0.0005,
+      "loss": 0.8385,
+      "step": 440700
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8258,
+      "step": 440800
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8128,
+      "step": 440900
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 0.0005,
+      "loss": 0.8396,
+      "step": 441000
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 0.0005,
+      "loss": 0.797,
+      "step": 441100
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 0.0005,
+      "loss": 0.8389,
+      "step": 441200
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8328,
+      "step": 441300
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8357,
+      "step": 441400
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 0.0005,
+      "loss": 0.8416,
+      "step": 441500
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8182,
+      "step": 441600
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 0.0005,
+      "loss": 0.827,
+      "step": 441700
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8377,
+      "step": 441800
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8376,
+      "step": 441900
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 0.0005,
+      "loss": 0.8373,
+      "step": 442000
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8355,
+      "step": 442100
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 0.0005,
+      "loss": 0.831,
+      "step": 442200
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8319,
+      "step": 442300
+    },
+    {
+      "epoch": 34.69,
+      "learning_rate": 0.0005,
+      "loss": 0.838,
+      "step": 442400
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8341,
+      "step": 442500
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 0.0005,
+      "loss": 0.8517,
+      "step": 442600
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8445,
+      "step": 442700
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8481,
+      "step": 442800
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 0.0005,
+      "loss": 0.8378,
+      "step": 442900
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8282,
+      "step": 443000
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 0.0005,
+      "loss": 0.8313,
+      "step": 443100
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8331,
+      "step": 443200
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8289,
+      "step": 443300
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 0.0005,
+      "loss": 0.8486,
+      "step": 443400
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 0.0005,
+      "loss": 0.8296,
+      "step": 443500
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 0.0005,
+      "loss": 0.825,
+      "step": 443600
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 0.0005,
+      "loss": 0.8518,
+      "step": 443700
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 0.0005,
+      "loss": 0.841,
+      "step": 443800
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8487,
+      "step": 443900
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 0.0005,
+      "loss": 0.84,
+      "step": 444000
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8461,
+      "step": 444100
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8311,
+      "step": 444200
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 0.0005,
+      "loss": 0.839,
+      "step": 444300
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8496,
+      "step": 444400
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8523,
+      "step": 444500
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 0.0005,
+      "loss": 0.8429,
+      "step": 444600
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 0.0005,
+      "loss": 0.8418,
+      "step": 444700
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8368,
+      "step": 444800
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8457,
+      "step": 444900
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 0.0005,
+      "loss": 0.853,
+      "step": 445000
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8323,
+      "step": 445100
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 0.0005,
+      "loss": 0.8477,
+      "step": 445200
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8549,
+      "step": 445300
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8326,
+      "step": 445400
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8582,
+      "step": 445500
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8332,
+      "step": 445600
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 0.0005,
+      "loss": 0.8394,
+      "step": 445700
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8373,
+      "step": 445800
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8423,
+      "step": 445900
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8352,
+      "step": 446000
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 0.0005,
+      "loss": 0.8462,
+      "step": 446100
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8499,
+      "step": 446200
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 0.0005,
+      "loss": 0.8428,
+      "step": 446300
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 0.0005,
+      "loss": 0.7978,
+      "step": 446400
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 0.0005,
+      "loss": 0.7763,
+      "step": 446500
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 0.0005,
+      "loss": 0.7864,
+      "step": 446600
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 0.0005,
+      "loss": 0.7784,
+      "step": 446700
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 0.0005,
+      "loss": 0.7811,
+      "step": 446800
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 0.0005,
+      "loss": 0.7874,
+      "step": 446900
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 0.0005,
+      "loss": 0.7826,
+      "step": 447000
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 0.0005,
+      "loss": 0.7766,
+      "step": 447100
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 0.0005,
+      "loss": 0.7829,
+      "step": 447200
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 0.0005,
+      "loss": 0.7731,
+      "step": 447300
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 0.0005,
+      "loss": 0.7956,
+      "step": 447400
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 0.0005,
+      "loss": 0.7856,
+      "step": 447500
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 0.0005,
+      "loss": 0.7957,
+      "step": 447600
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 0.0005,
+      "loss": 0.7681,
+      "step": 447700
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 0.0005,
+      "loss": 0.7797,
+      "step": 447800
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 0.0005,
+      "loss": 0.7957,
+      "step": 447900
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 0.0005,
+      "loss": 0.7806,
+      "step": 448000
+    },
+    {
+      "epoch": 35.14,
+      "learning_rate": 0.0005,
+      "loss": 0.8021,
+      "step": 448100
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 0.0005,
+      "loss": 0.7923,
+      "step": 448200
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 0.0005,
+      "loss": 0.8019,
+      "step": 448300
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 0.0005,
+      "loss": 0.7935,
+      "step": 448400
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 0.0005,
+      "loss": 0.7987,
+      "step": 448500
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 0.0005,
+      "loss": 0.798,
+      "step": 448600
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 0.0005,
+      "loss": 0.7904,
+      "step": 448700
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 0.0005,
+      "loss": 0.8029,
+      "step": 448800
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 0.0005,
+      "loss": 0.7862,
+      "step": 448900
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 0.0005,
+      "loss": 0.7884,
+      "step": 449000
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 0.0005,
+      "loss": 0.7895,
+      "step": 449100
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 0.0005,
+      "loss": 0.7948,
+      "step": 449200
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 0.0005,
+      "loss": 0.7953,
+      "step": 449300
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 0.0005,
+      "loss": 0.807,
+      "step": 449400
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8105,
+      "step": 449500
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 0.0005,
+      "loss": 0.7959,
+      "step": 449600
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 0.0005,
+      "loss": 0.8048,
+      "step": 449700
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 0.0005,
+      "loss": 0.7986,
+      "step": 449800
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8037,
+      "step": 449900
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8095,
+      "step": 450000
+    },
+    {
+      "epoch": 35.29,
+      "eval_gen_len": 18.77716959477355,
+      "eval_loss": 2.638524293899536,
+      "eval_rouge1": 35.734,
+      "eval_rouge2": 14.7352,
+      "eval_rougeL": 29.5245,
+      "eval_rougeLsum": 29.5196,
+      "eval_runtime": 343.351,
+      "eval_samples_per_second": 32.99,
+      "eval_steps_per_second": 2.062,
+      "step": 450000
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8028,
+      "step": 450100
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 0.0005,
+      "loss": 0.8031,
+      "step": 450200
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 0.0005,
+      "loss": 0.8257,
+      "step": 450300
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 0.0005,
+      "loss": 0.8024,
+      "step": 450400
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8068,
+      "step": 450500
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 0.0005,
+      "loss": 0.803,
+      "step": 450600
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 0.0005,
+      "loss": 0.8003,
+      "step": 450700
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 0.0005,
+      "loss": 0.8013,
+      "step": 450800
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 0.0005,
+      "loss": 0.8163,
+      "step": 450900
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 0.0005,
+      "loss": 0.7973,
+      "step": 451000
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 0.0005,
+      "loss": 0.7942,
+      "step": 451100
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 0.0005,
+      "loss": 0.803,
+      "step": 451200
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 0.0005,
+      "loss": 0.7992,
+      "step": 451300
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 0.0005,
+      "loss": 0.8124,
+      "step": 451400
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 0.0005,
+      "loss": 0.7982,
+      "step": 451500
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 0.0005,
+      "loss": 0.8111,
+      "step": 451600
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 0.0005,
+      "loss": 0.7959,
+      "step": 451700
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8066,
+      "step": 451800
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 0.0005,
+      "loss": 0.8037,
+      "step": 451900
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8163,
+      "step": 452000
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8144,
+      "step": 452100
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 0.0005,
+      "loss": 0.7978,
+      "step": 452200
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8063,
+      "step": 452300
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 0.0005,
+      "loss": 0.8142,
+      "step": 452400
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 0.0005,
+      "loss": 0.8319,
+      "step": 452500
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 0.0005,
+      "loss": 0.8081,
+      "step": 452600
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8118,
+      "step": 452700
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 0.0005,
+      "loss": 0.822,
+      "step": 452800
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8152,
+      "step": 452900
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8046,
+      "step": 453000
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8165,
+      "step": 453100
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 0.0005,
+      "loss": 0.81,
+      "step": 453200
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 0.0005,
+      "loss": 0.7984,
+      "step": 453300
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 0.0005,
+      "loss": 0.806,
+      "step": 453400
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 0.0005,
+      "loss": 0.8164,
+      "step": 453500
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 0.0005,
+      "loss": 0.813,
+      "step": 453600
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 0.0005,
+      "loss": 0.8072,
+      "step": 453700
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 0.0005,
+      "loss": 0.8135,
+      "step": 453800
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 0.0005,
+      "loss": 0.8073,
+      "step": 453900
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 0.0005,
+      "loss": 0.815,
+      "step": 454000
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8239,
+      "step": 454100
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 0.0005,
+      "loss": 0.8274,
+      "step": 454200
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8192,
+      "step": 454300
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8422,
+      "step": 454400
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 0.0005,
+      "loss": 0.8196,
+      "step": 454500
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8139,
+      "step": 454600
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 0.0005,
+      "loss": 0.8382,
+      "step": 454700
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 0.0005,
+      "loss": 0.8197,
+      "step": 454800
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8385,
+      "step": 454900
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8177,
+      "step": 455000
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 0.0005,
+      "loss": 0.814,
+      "step": 455100
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8306,
+      "step": 455200
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8461,
+      "step": 455300
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 0.0005,
+      "loss": 0.8416,
+      "step": 455400
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8048,
+      "step": 455500
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 0.0005,
+      "loss": 0.8334,
+      "step": 455600
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8359,
+      "step": 455700
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8078,
+      "step": 455800
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 0.0005,
+      "loss": 0.8337,
+      "step": 455900
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 0.0005,
+      "loss": 0.827,
+      "step": 456000
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 0.0005,
+      "loss": 0.8268,
+      "step": 456100
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 0.0005,
+      "loss": 0.811,
+      "step": 456200
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 0.0005,
+      "loss": 0.833,
+      "step": 456300
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 0.0005,
+      "loss": 0.839,
+      "step": 456400
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 0.0005,
+      "loss": 0.837,
+      "step": 456500
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8271,
+      "step": 456600
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8288,
+      "step": 456700
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 0.0005,
+      "loss": 0.8316,
+      "step": 456800
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8124,
+      "step": 456900
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 0.0005,
+      "loss": 0.8154,
+      "step": 457000
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8378,
+      "step": 457100
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8521,
+      "step": 457200
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8235,
+      "step": 457300
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 0.0005,
+      "loss": 0.84,
+      "step": 457400
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8393,
+      "step": 457500
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8429,
+      "step": 457600
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8259,
+      "step": 457700
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8353,
+      "step": 457800
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 0.0005,
+      "loss": 0.8472,
+      "step": 457900
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8356,
+      "step": 458000
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8406,
+      "step": 458100
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8403,
+      "step": 458200
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8295,
+      "step": 458300
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 0.0005,
+      "loss": 0.8491,
+      "step": 458400
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8041,
+      "step": 458500
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8446,
+      "step": 458600
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8517,
+      "step": 458700
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 0.0005,
+      "loss": 0.8476,
+      "step": 458800
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8332,
+      "step": 458900
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8464,
+      "step": 459000
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 0.0005,
+      "loss": 0.8185,
+      "step": 459100
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 0.0005,
+      "loss": 0.7558,
+      "step": 459200
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 0.0005,
+      "loss": 0.7795,
+      "step": 459300
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 0.0005,
+      "loss": 0.7633,
+      "step": 459400
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 0.0005,
+      "loss": 0.7774,
+      "step": 459500
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 0.0005,
+      "loss": 0.786,
+      "step": 459600
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 0.0005,
+      "loss": 0.7673,
+      "step": 459700
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 0.0005,
+      "loss": 0.7816,
+      "step": 459800
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 0.0005,
+      "loss": 0.7744,
+      "step": 459900
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 0.0005,
+      "loss": 0.7855,
+      "step": 460000
+    },
+    {
+      "epoch": 36.07,
+      "eval_gen_len": 18.767546570142137,
+      "eval_loss": 2.670743465423584,
+      "eval_rouge1": 35.881,
+      "eval_rouge2": 14.9235,
+      "eval_rougeL": 29.6308,
+      "eval_rougeLsum": 29.6326,
+      "eval_runtime": 357.5591,
+      "eval_samples_per_second": 31.679,
+      "eval_steps_per_second": 1.98,
+      "step": 460000
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 0.0005,
+      "loss": 0.7765,
+      "step": 460100
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 0.0005,
+      "loss": 0.7711,
+      "step": 460200
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 0.0005,
+      "loss": 0.7955,
+      "step": 460300
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 0.0005,
+      "loss": 0.7766,
+      "step": 460400
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 0.0005,
+      "loss": 0.7848,
+      "step": 460500
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 0.0005,
+      "loss": 0.7788,
+      "step": 460600
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 0.0005,
+      "loss": 0.7853,
+      "step": 460700
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 0.0005,
+      "loss": 0.7926,
+      "step": 460800
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 0.0005,
+      "loss": 0.7713,
+      "step": 460900
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 0.0005,
+      "loss": 0.7747,
+      "step": 461000
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 0.0005,
+      "loss": 0.7857,
+      "step": 461100
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 0.0005,
+      "loss": 0.7767,
+      "step": 461200
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 0.0005,
+      "loss": 0.7687,
+      "step": 461300
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 0.0005,
+      "loss": 0.7738,
+      "step": 461400
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 0.0005,
+      "loss": 0.7698,
+      "step": 461500
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 0.0005,
+      "loss": 0.7822,
+      "step": 461600
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 0.0005,
+      "loss": 0.796,
+      "step": 461700
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 0.0005,
+      "loss": 0.7873,
+      "step": 461800
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 0.0005,
+      "loss": 0.7893,
+      "step": 461900
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 0.0005,
+      "loss": 0.7887,
+      "step": 462000
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 0.0005,
+      "loss": 0.7934,
+      "step": 462100
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 0.0005,
+      "loss": 0.8007,
+      "step": 462200
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 0.0005,
+      "loss": 0.7955,
+      "step": 462300
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 0.0005,
+      "loss": 0.7928,
+      "step": 462400
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 0.0005,
+      "loss": 0.7973,
+      "step": 462500
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 0.0005,
+      "loss": 0.8038,
+      "step": 462600
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 0.0005,
+      "loss": 0.7919,
+      "step": 462700
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 0.0005,
+      "loss": 0.8013,
+      "step": 462800
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 0.0005,
+      "loss": 0.7972,
+      "step": 462900
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 0.0005,
+      "loss": 0.7906,
+      "step": 463000
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 0.0005,
+      "loss": 0.7877,
+      "step": 463100
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 0.0005,
+      "loss": 0.7763,
+      "step": 463200
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 0.0005,
+      "loss": 0.8125,
+      "step": 463300
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 0.0005,
+      "loss": 0.7908,
+      "step": 463400
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 0.0005,
+      "loss": 0.7881,
+      "step": 463500
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 0.0005,
+      "loss": 0.7967,
+      "step": 463600
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 0.0005,
+      "loss": 0.813,
+      "step": 463700
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 0.0005,
+      "loss": 0.7929,
+      "step": 463800
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 0.0005,
+      "loss": 0.8054,
+      "step": 463900
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 0.0005,
+      "loss": 0.8029,
+      "step": 464000
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 0.0005,
+      "loss": 0.7978,
+      "step": 464100
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 0.0005,
+      "loss": 0.8142,
+      "step": 464200
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 0.0005,
+      "loss": 0.7955,
+      "step": 464300
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 0.0005,
+      "loss": 0.7899,
+      "step": 464400
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8003,
+      "step": 464500
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8066,
+      "step": 464600
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 0.0005,
+      "loss": 0.7888,
+      "step": 464700
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 0.0005,
+      "loss": 0.794,
+      "step": 464800
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 0.0005,
+      "loss": 0.7944,
+      "step": 464900
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 0.0005,
+      "loss": 0.8059,
+      "step": 465000
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 0.0005,
+      "loss": 0.8035,
+      "step": 465100
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 0.0005,
+      "loss": 0.7903,
+      "step": 465200
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 0.0005,
+      "loss": 0.7972,
+      "step": 465300
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 0.0005,
+      "loss": 0.8111,
+      "step": 465400
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 0.0005,
+      "loss": 0.7994,
+      "step": 465500
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 0.0005,
+      "loss": 0.8135,
+      "step": 465600
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 0.0005,
+      "loss": 0.8025,
+      "step": 465700
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 0.0005,
+      "loss": 0.802,
+      "step": 465800
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 0.0005,
+      "loss": 0.7956,
+      "step": 465900
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8108,
+      "step": 466000
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 0.0005,
+      "loss": 0.8147,
+      "step": 466100
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 0.0005,
+      "loss": 0.8047,
+      "step": 466200
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8173,
+      "step": 466300
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8066,
+      "step": 466400
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 0.0005,
+      "loss": 0.8079,
+      "step": 466500
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 0.0005,
+      "loss": 0.7877,
+      "step": 466600
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 0.0005,
+      "loss": 0.8162,
+      "step": 466700
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8069,
+      "step": 466800
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8191,
+      "step": 466900
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 0.0005,
+      "loss": 0.807,
+      "step": 467000
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8132,
+      "step": 467100
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 0.0005,
+      "loss": 0.7967,
+      "step": 467200
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8126,
+      "step": 467300
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8263,
+      "step": 467400
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 0.0005,
+      "loss": 0.822,
+      "step": 467500
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8118,
+      "step": 467600
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8003,
+      "step": 467700
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8103,
+      "step": 467800
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 0.0005,
+      "loss": 0.8028,
+      "step": 467900
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8207,
+      "step": 468000
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 0.0005,
+      "loss": 0.8168,
+      "step": 468100
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8461,
+      "step": 468200
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 0.0005,
+      "loss": 0.8257,
+      "step": 468300
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 0.0005,
+      "loss": 0.8007,
+      "step": 468400
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8206,
+      "step": 468500
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 0.0005,
+      "loss": 0.8275,
+      "step": 468600
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8306,
+      "step": 468700
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 0.0005,
+      "loss": 0.7945,
+      "step": 468800
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 0.0005,
+      "loss": 0.8237,
+      "step": 468900
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 0.0005,
+      "loss": 0.8024,
+      "step": 469000
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 0.0005,
+      "loss": 0.8356,
+      "step": 469100
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 0.0005,
+      "loss": 0.8196,
+      "step": 469200
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 0.0005,
+      "loss": 0.8132,
+      "step": 469300
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8178,
+      "step": 469400
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 0.0005,
+      "loss": 0.817,
+      "step": 469500
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8178,
+      "step": 469600
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8326,
+      "step": 469700
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 0.0005,
+      "loss": 0.8334,
+      "step": 469800
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8169,
+      "step": 469900
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8477,
+      "step": 470000
+    },
+    {
+      "epoch": 36.86,
+      "eval_gen_len": 18.735852388099232,
+      "eval_loss": 2.60902738571167,
+      "eval_rouge1": 35.7078,
+      "eval_rouge2": 14.7556,
+      "eval_rougeL": 29.5286,
+      "eval_rougeLsum": 29.5256,
+      "eval_runtime": 378.8936,
+      "eval_samples_per_second": 29.895,
+      "eval_steps_per_second": 1.869,
+      "step": 470000
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8233,
+      "step": 470100
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 0.0005,
+      "loss": 0.8172,
+      "step": 470200
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8357,
+      "step": 470300
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8409,
+      "step": 470400
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8167,
+      "step": 470500
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8222,
+      "step": 470600
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 0.0005,
+      "loss": 0.8258,
+      "step": 470700
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8166,
+      "step": 470800
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8391,
+      "step": 470900
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8468,
+      "step": 471000
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8225,
+      "step": 471100
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 0.0005,
+      "loss": 0.8308,
+      "step": 471200
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8239,
+      "step": 471300
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8429,
+      "step": 471400
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8231,
+      "step": 471500
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 0.0005,
+      "loss": 0.8362,
+      "step": 471600
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8313,
+      "step": 471700
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 0.0005,
+      "loss": 0.8286,
+      "step": 471800
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 0.0005,
+      "loss": 0.7754,
+      "step": 471900
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 0.0005,
+      "loss": 0.7617,
+      "step": 472000
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 0.0005,
+      "loss": 0.7606,
+      "step": 472100
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 0.0005,
+      "loss": 0.7469,
+      "step": 472200
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 0.0005,
+      "loss": 0.7591,
+      "step": 472300
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 0.0005,
+      "loss": 0.7543,
+      "step": 472400
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 0.0005,
+      "loss": 0.7544,
+      "step": 472500
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 0.0005,
+      "loss": 0.7685,
+      "step": 472600
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 0.0005,
+      "loss": 0.7577,
+      "step": 472700
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 0.0005,
+      "loss": 0.7495,
+      "step": 472800
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 0.0005,
+      "loss": 0.7498,
+      "step": 472900
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 0.0005,
+      "loss": 0.7712,
+      "step": 473000
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 0.0005,
+      "loss": 0.7464,
+      "step": 473100
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 0.0005,
+      "loss": 0.7627,
+      "step": 473200
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 0.0005,
+      "loss": 0.7591,
+      "step": 473300
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 0.0005,
+      "loss": 0.7803,
+      "step": 473400
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 0.0005,
+      "loss": 0.7655,
+      "step": 473500
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 0.0005,
+      "loss": 0.7818,
+      "step": 473600
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 0.0005,
+      "loss": 0.7823,
+      "step": 473700
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 0.0005,
+      "loss": 0.7796,
+      "step": 473800
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 0.0005,
+      "loss": 0.7766,
+      "step": 473900
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 0.0005,
+      "loss": 0.7712,
+      "step": 474000
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 0.0005,
+      "loss": 0.7847,
+      "step": 474100
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 0.0005,
+      "loss": 0.7692,
+      "step": 474200
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 0.0005,
+      "loss": 0.7817,
+      "step": 474300
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 0.0005,
+      "loss": 0.7833,
+      "step": 474400
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 0.0005,
+      "loss": 0.7743,
+      "step": 474500
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 0.0005,
+      "loss": 0.7668,
+      "step": 474600
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 0.0005,
+      "loss": 0.7677,
+      "step": 474700
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 0.0005,
+      "loss": 0.7812,
+      "step": 474800
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 0.0005,
+      "loss": 0.7863,
+      "step": 474900
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 0.0005,
+      "loss": 0.781,
+      "step": 475000
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 0.0005,
+      "loss": 0.779,
+      "step": 475100
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 0.0005,
+      "loss": 0.7852,
+      "step": 475200
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 0.0005,
+      "loss": 0.7828,
+      "step": 475300
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 0.0005,
+      "loss": 0.7738,
+      "step": 475400
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 0.0005,
+      "loss": 0.7747,
+      "step": 475500
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 0.0005,
+      "loss": 0.7884,
+      "step": 475600
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 0.0005,
+      "loss": 0.7846,
+      "step": 475700
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 0.0005,
+      "loss": 0.7871,
+      "step": 475800
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 0.0005,
+      "loss": 0.78,
+      "step": 475900
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 0.0005,
+      "loss": 0.7907,
+      "step": 476000
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 0.0005,
+      "loss": 0.7768,
+      "step": 476100
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 0.0005,
+      "loss": 0.7854,
+      "step": 476200
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 0.0005,
+      "loss": 0.7853,
+      "step": 476300
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 0.0005,
+      "loss": 0.7806,
+      "step": 476400
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 0.0005,
+      "loss": 0.7993,
+      "step": 476500
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 0.0005,
+      "loss": 0.7925,
+      "step": 476600
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 0.0005,
+      "loss": 0.7759,
+      "step": 476700
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 0.0005,
+      "loss": 0.7989,
+      "step": 476800
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 0.0005,
+      "loss": 0.7949,
+      "step": 476900
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 0.0005,
+      "loss": 0.7947,
+      "step": 477000
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 0.0005,
+      "loss": 0.7873,
+      "step": 477100
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 0.0005,
+      "loss": 0.8054,
+      "step": 477200
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 0.0005,
+      "loss": 0.8011,
+      "step": 477300
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 0.0005,
+      "loss": 0.7734,
+      "step": 477400
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 0.0005,
+      "loss": 0.7926,
+      "step": 477500
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 0.0005,
+      "loss": 0.8041,
+      "step": 477600
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 0.0005,
+      "loss": 0.7928,
+      "step": 477700
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 0.0005,
+      "loss": 0.7862,
+      "step": 477800
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 0.0005,
+      "loss": 0.8035,
+      "step": 477900
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 0.0005,
+      "loss": 0.791,
+      "step": 478000
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 0.0005,
+      "loss": 0.7836,
+      "step": 478100
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 0.0005,
+      "loss": 0.803,
+      "step": 478200
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 0.0005,
+      "loss": 0.7891,
+      "step": 478300
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 0.0005,
+      "loss": 0.7887,
+      "step": 478400
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 0.0005,
+      "loss": 0.7961,
+      "step": 478500
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8055,
+      "step": 478600
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 0.0005,
+      "loss": 0.8102,
+      "step": 478700
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 0.0005,
+      "loss": 0.8058,
+      "step": 478800
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 0.0005,
+      "loss": 0.8053,
+      "step": 478900
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 0.0005,
+      "loss": 0.7839,
+      "step": 479000
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8064,
+      "step": 479100
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 0.0005,
+      "loss": 0.7746,
+      "step": 479200
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 0.0005,
+      "loss": 0.8084,
+      "step": 479300
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 0.0005,
+      "loss": 0.8039,
+      "step": 479400
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 0.0005,
+      "loss": 0.8147,
+      "step": 479500
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 0.0005,
+      "loss": 0.7917,
+      "step": 479600
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 0.0005,
+      "loss": 0.7955,
+      "step": 479700
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8174,
+      "step": 479800
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8039,
+      "step": 479900
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 0.0005,
+      "loss": 0.819,
+      "step": 480000
+    },
+    {
+      "epoch": 37.64,
+      "eval_gen_len": 18.777787587181074,
+      "eval_loss": 2.638998508453369,
+      "eval_rouge1": 35.7961,
+      "eval_rouge2": 14.8984,
+      "eval_rougeL": 29.577,
+      "eval_rougeLsum": 29.574,
+      "eval_runtime": 381.3863,
+      "eval_samples_per_second": 29.7,
+      "eval_steps_per_second": 1.856,
+      "step": 480000
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 0.0005,
+      "loss": 0.8115,
+      "step": 480100
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 0.0005,
+      "loss": 0.7937,
+      "step": 480200
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 0.0005,
+      "loss": 0.8177,
+      "step": 480300
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 0.0005,
+      "loss": 0.8204,
+      "step": 480400
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8079,
+      "step": 480500
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 0.0005,
+      "loss": 0.7922,
+      "step": 480600
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8047,
+      "step": 480700
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 0.0005,
+      "loss": 0.7963,
+      "step": 480800
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 0.0005,
+      "loss": 0.7966,
+      "step": 480900
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 0.0005,
+      "loss": 0.817,
+      "step": 481000
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 0.0005,
+      "loss": 0.8118,
+      "step": 481100
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 0.0005,
+      "loss": 0.7964,
+      "step": 481200
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 0.0005,
+      "loss": 0.8205,
+      "step": 481300
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 0.0005,
+      "loss": 0.8132,
+      "step": 481400
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8093,
+      "step": 481500
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 0.0005,
+      "loss": 0.8237,
+      "step": 481600
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 0.0005,
+      "loss": 0.8174,
+      "step": 481700
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 0.0005,
+      "loss": 0.7873,
+      "step": 481800
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 0.0005,
+      "loss": 0.8104,
+      "step": 481900
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 0.0005,
+      "loss": 0.8336,
+      "step": 482000
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8169,
+      "step": 482100
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8128,
+      "step": 482200
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 0.0005,
+      "loss": 0.8081,
+      "step": 482300
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 0.0005,
+      "loss": 0.7982,
+      "step": 482400
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 0.0005,
+      "loss": 0.8139,
+      "step": 482500
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 0.0005,
+      "loss": 0.82,
+      "step": 482600
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 0.0005,
+      "loss": 0.8118,
+      "step": 482700
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8155,
+      "step": 482800
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 0.0005,
+      "loss": 0.8242,
+      "step": 482900
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8231,
+      "step": 483000
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8049,
+      "step": 483100
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 0.0005,
+      "loss": 0.8155,
+      "step": 483200
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8248,
+      "step": 483300
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 0.0005,
+      "loss": 0.8062,
+      "step": 483400
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8294,
+      "step": 483500
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8376,
+      "step": 483600
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8136,
+      "step": 483700
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8185,
+      "step": 483800
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 0.0005,
+      "loss": 0.8062,
+      "step": 483900
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 0.0005,
+      "loss": 0.8199,
+      "step": 484000
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8214,
+      "step": 484100
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8285,
+      "step": 484200
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 0.0005,
+      "loss": 0.8259,
+      "step": 484300
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8283,
+      "step": 484400
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8267,
+      "step": 484500
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 0.0005,
+      "loss": 0.7963,
+      "step": 484600
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 0.0005,
+      "loss": 0.7441,
+      "step": 484700
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 0.0005,
+      "loss": 0.764,
+      "step": 484800
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 0.0005,
+      "loss": 0.7441,
+      "step": 484900
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 0.0005,
+      "loss": 0.7467,
+      "step": 485000
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 0.0005,
+      "loss": 0.7483,
+      "step": 485100
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 0.0005,
+      "loss": 0.7541,
+      "step": 485200
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 0.0005,
+      "loss": 0.7637,
+      "step": 485300
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 0.0005,
+      "loss": 0.7627,
+      "step": 485400
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 0.0005,
+      "loss": 0.7428,
+      "step": 485500
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 0.0005,
+      "loss": 0.7634,
+      "step": 485600
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 0.0005,
+      "loss": 0.754,
+      "step": 485700
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 0.0005,
+      "loss": 0.764,
+      "step": 485800
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 0.0005,
+      "loss": 0.7703,
+      "step": 485900
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 0.0005,
+      "loss": 0.7607,
+      "step": 486000
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 0.0005,
+      "loss": 0.7593,
+      "step": 486100
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 0.0005,
+      "loss": 0.7632,
+      "step": 486200
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 0.0005,
+      "loss": 0.7546,
+      "step": 486300
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 0.0005,
+      "loss": 0.781,
+      "step": 486400
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 0.0005,
+      "loss": 0.7796,
+      "step": 486500
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 0.0005,
+      "loss": 0.7885,
+      "step": 486600
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 0.0005,
+      "loss": 0.7649,
+      "step": 486700
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 0.0005,
+      "loss": 0.7591,
+      "step": 486800
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 0.0005,
+      "loss": 0.7552,
+      "step": 486900
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 0.0005,
+      "loss": 0.7642,
+      "step": 487000
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 0.0005,
+      "loss": 0.7598,
+      "step": 487100
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 0.0005,
+      "loss": 0.7733,
+      "step": 487200
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 0.0005,
+      "loss": 0.7665,
+      "step": 487300
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 0.0005,
+      "loss": 0.7725,
+      "step": 487400
+    },
+    {
+      "epoch": 38.23,
+      "learning_rate": 0.0005,
+      "loss": 0.765,
+      "step": 487500
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 0.0005,
+      "loss": 0.7849,
+      "step": 487600
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 0.0005,
+      "loss": 0.774,
+      "step": 487700
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 0.0005,
+      "loss": 0.7727,
+      "step": 487800
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 0.0005,
+      "loss": 0.7863,
+      "step": 487900
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 0.0005,
+      "loss": 0.7672,
+      "step": 488000
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 0.0005,
+      "loss": 0.7788,
+      "step": 488100
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 0.0005,
+      "loss": 0.7728,
+      "step": 488200
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 0.0005,
+      "loss": 0.7769,
+      "step": 488300
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 0.0005,
+      "loss": 0.7859,
+      "step": 488400
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 0.0005,
+      "loss": 0.7622,
+      "step": 488500
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 0.0005,
+      "loss": 0.7606,
+      "step": 488600
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 0.0005,
+      "loss": 0.7834,
+      "step": 488700
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 0.0005,
+      "loss": 0.7724,
+      "step": 488800
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 0.0005,
+      "loss": 0.7873,
+      "step": 488900
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 0.0005,
+      "loss": 0.7833,
+      "step": 489000
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 0.0005,
+      "loss": 0.7713,
+      "step": 489100
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 0.0005,
+      "loss": 0.771,
+      "step": 489200
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 0.0005,
+      "loss": 0.7899,
+      "step": 489300
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 0.0005,
+      "loss": 0.7779,
+      "step": 489400
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 0.0005,
+      "loss": 0.7835,
+      "step": 489500
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 0.0005,
+      "loss": 0.7883,
+      "step": 489600
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 0.0005,
+      "loss": 0.773,
+      "step": 489700
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 0.0005,
+      "loss": 0.7867,
+      "step": 489800
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 0.0005,
+      "loss": 0.7703,
+      "step": 489900
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 0.0005,
+      "loss": 0.7805,
+      "step": 490000
+    },
+    {
+      "epoch": 38.43,
+      "eval_gen_len": 18.730908448839056,
+      "eval_loss": 2.6787099838256836,
+      "eval_rouge1": 35.6705,
+      "eval_rouge2": 14.71,
+      "eval_rougeL": 29.4423,
+      "eval_rougeLsum": 29.4417,
+      "eval_runtime": 382.2597,
+      "eval_samples_per_second": 29.632,
+      "eval_steps_per_second": 1.852,
+      "step": 490000
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 0.0005,
+      "loss": 0.7859,
+      "step": 490100
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 0.0005,
+      "loss": 0.7903,
+      "step": 490200
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 0.0005,
+      "loss": 0.7715,
+      "step": 490300
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 0.0005,
+      "loss": 0.7713,
+      "step": 490400
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 0.0005,
+      "loss": 0.7675,
+      "step": 490500
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 0.0005,
+      "loss": 0.7855,
+      "step": 490600
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 0.0005,
+      "loss": 0.7847,
+      "step": 490700
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 0.0005,
+      "loss": 0.7773,
+      "step": 490800
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 0.0005,
+      "loss": 0.7874,
+      "step": 490900
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 0.0005,
+      "loss": 0.7904,
+      "step": 491000
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 0.0005,
+      "loss": 0.7847,
+      "step": 491100
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 0.0005,
+      "loss": 0.7626,
+      "step": 491200
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 0.0005,
+      "loss": 0.8014,
+      "step": 491300
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 0.0005,
+      "loss": 0.7699,
+      "step": 491400
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 0.0005,
+      "loss": 0.7917,
+      "step": 491500
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 0.0005,
+      "loss": 0.7804,
+      "step": 491600
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 0.0005,
+      "loss": 0.7782,
+      "step": 491700
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 0.0005,
+      "loss": 0.7905,
+      "step": 491800
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 0.0005,
+      "loss": 0.8008,
+      "step": 491900
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 0.0005,
+      "loss": 0.7952,
+      "step": 492000
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 0.0005,
+      "loss": 0.7836,
+      "step": 492100
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 0.0005,
+      "loss": 0.7916,
+      "step": 492200
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 0.0005,
+      "loss": 0.8032,
+      "step": 492300
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 0.0005,
+      "loss": 0.7988,
+      "step": 492400
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 0.0005,
+      "loss": 0.8059,
+      "step": 492500
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 0.0005,
+      "loss": 0.8002,
+      "step": 492600
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 0.0005,
+      "loss": 0.8033,
+      "step": 492700
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 0.0005,
+      "loss": 0.8018,
+      "step": 492800
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 0.0005,
+      "loss": 0.7887,
+      "step": 492900
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 0.0005,
+      "loss": 0.7899,
+      "step": 493000
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 0.0005,
+      "loss": 0.787,
+      "step": 493100
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 0.0005,
+      "loss": 0.7851,
+      "step": 493200
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 0.0005,
+      "loss": 0.8015,
+      "step": 493300
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 0.0005,
+      "loss": 0.8058,
+      "step": 493400
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 0.0005,
+      "loss": 0.8148,
+      "step": 493500
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 0.0005,
+      "loss": 0.7937,
+      "step": 493600
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 0.0005,
+      "loss": 0.7883,
+      "step": 493700
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 0.0005,
+      "loss": 0.7989,
+      "step": 493800
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 0.0005,
+      "loss": 0.8148,
+      "step": 493900
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 0.0005,
+      "loss": 0.793,
+      "step": 494000
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 0.0005,
+      "loss": 0.7904,
+      "step": 494100
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 0.0005,
+      "loss": 0.8014,
+      "step": 494200
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 0.0005,
+      "loss": 0.8086,
+      "step": 494300
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 0.0005,
+      "loss": 0.7984,
+      "step": 494400
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 0.0005,
+      "loss": 0.7979,
+      "step": 494500
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 0.0005,
+      "loss": 0.7977,
+      "step": 494600
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 0.0005,
+      "loss": 0.7853,
+      "step": 494700
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 0.0005,
+      "loss": 0.7845,
+      "step": 494800
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 0.0005,
+      "loss": 0.8026,
+      "step": 494900
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 0.0005,
+      "loss": 0.8048,
+      "step": 495000
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 0.0005,
+      "loss": 0.8033,
+      "step": 495100
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 0.0005,
+      "loss": 0.7907,
+      "step": 495200
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 0.0005,
+      "loss": 0.809,
+      "step": 495300
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 0.0005,
+      "loss": 0.7964,
+      "step": 495400
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8103,
+      "step": 495500
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 0.0005,
+      "loss": 0.8152,
+      "step": 495600
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 0.0005,
+      "loss": 0.797,
+      "step": 495700
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 0.0005,
+      "loss": 0.8186,
+      "step": 495800
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 0.0005,
+      "loss": 0.7971,
+      "step": 495900
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8047,
+      "step": 496000
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 0.0005,
+      "loss": 0.8059,
+      "step": 496100
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 0.0005,
+      "loss": 0.825,
+      "step": 496200
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 0.0005,
+      "loss": 0.8032,
+      "step": 496300
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 0.0005,
+      "loss": 0.8058,
+      "step": 496400
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8043,
+      "step": 496500
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 0.0005,
+      "loss": 0.8195,
+      "step": 496600
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 0.0005,
+      "loss": 0.7943,
+      "step": 496700
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 0.0005,
+      "loss": 0.8087,
+      "step": 496800
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8194,
+      "step": 496900
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 0.0005,
+      "loss": 0.8167,
+      "step": 497000
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 0.0005,
+      "loss": 0.7995,
+      "step": 497100
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 0.0005,
+      "loss": 0.8111,
+      "step": 497200
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 0.0005,
+      "loss": 0.8107,
+      "step": 497300
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 0.0005,
+      "loss": 0.7537,
+      "step": 497400
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 0.0005,
+      "loss": 0.735,
+      "step": 497500
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 0.0005,
+      "loss": 0.7464,
+      "step": 497600
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 0.0005,
+      "loss": 0.7368,
+      "step": 497700
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 0.0005,
+      "loss": 0.7398,
+      "step": 497800
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 0.0005,
+      "loss": 0.7409,
+      "step": 497900
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 0.0005,
+      "loss": 0.735,
+      "step": 498000
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 0.0005,
+      "loss": 0.7382,
+      "step": 498100
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 0.0005,
+      "loss": 0.7348,
+      "step": 498200
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 0.0005,
+      "loss": 0.7364,
+      "step": 498300
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 0.0005,
+      "loss": 0.7536,
+      "step": 498400
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 0.0005,
+      "loss": 0.7668,
+      "step": 498500
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 0.0005,
+      "loss": 0.7428,
+      "step": 498600
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 0.0005,
+      "loss": 0.7446,
+      "step": 498700
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 0.0005,
+      "loss": 0.7583,
+      "step": 498800
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 0.0005,
+      "loss": 0.7429,
+      "step": 498900
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 0.0005,
+      "loss": 0.7537,
+      "step": 499000
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 0.0005,
+      "loss": 0.7436,
+      "step": 499100
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 0.0005,
+      "loss": 0.7541,
+      "step": 499200
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 0.0005,
+      "loss": 0.7643,
+      "step": 499300
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 0.0005,
+      "loss": 0.745,
+      "step": 499400
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 0.0005,
+      "loss": 0.7421,
+      "step": 499500
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 0.0005,
+      "loss": 0.7588,
+      "step": 499600
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 0.0005,
+      "loss": 0.7629,
+      "step": 499700
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 0.0005,
+      "loss": 0.7602,
+      "step": 499800
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 0.0005,
+      "loss": 0.7571,
+      "step": 499900
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 0.0005,
+      "loss": 0.7541,
+      "step": 500000
+    },
+    {
+      "epoch": 39.21,
+      "eval_gen_len": 18.742297166063388,
+      "eval_loss": 2.7122867107391357,
+      "eval_rouge1": 35.952,
+      "eval_rouge2": 15.0205,
+      "eval_rougeL": 29.7389,
+      "eval_rougeLsum": 29.7331,
+      "eval_runtime": 382.2157,
+      "eval_samples_per_second": 29.635,
+      "eval_steps_per_second": 1.852,
+      "step": 500000
+    }
+  ],
+  "max_steps": 637600,
+  "num_train_epochs": 50,
+  "total_flos": 2.1183059506360812e+18,
+  "trial_name": null,
+  "trial_params": null
+}