diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31279 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 39.993726474278546,
+  "global_step": 510000,
+  "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": 2.9557,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.8658,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.8174,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005,
+      "loss": 2.7738,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005,
+      "loss": 2.8055,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005,
+      "loss": 2.7804,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005,
+      "loss": 2.7898,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005,
+      "loss": 2.7531,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005,
+      "loss": 2.7081,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005,
+      "loss": 2.7108,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005,
+      "loss": 2.7372,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005,
+      "loss": 2.7245,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005,
+      "loss": 2.718,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005,
+      "loss": 2.71,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005,
+      "loss": 2.6816,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005,
+      "loss": 2.7044,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005,
+      "loss": 2.6867,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005,
+      "loss": 2.7035,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005,
+      "loss": 2.6853,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005,
+      "loss": 2.6772,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005,
+      "loss": 2.6647,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005,
+      "loss": 2.6923,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005,
+      "loss": 2.6514,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005,
+      "loss": 2.6542,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 2.6391,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 2.6527,
+      "step": 2600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 2.6658,
+      "step": 2700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 2.6537,
+      "step": 2800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 2.67,
+      "step": 2900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 2.6605,
+      "step": 3000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 2.6516,
+      "step": 3100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 2.6358,
+      "step": 3200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 2.6302,
+      "step": 3300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 2.6007,
+      "step": 3400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 2.6063,
+      "step": 3500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 2.6016,
+      "step": 3600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 2.6013,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 2.5946,
+      "step": 3800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 2.6054,
+      "step": 3900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 2.6073,
+      "step": 4000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 2.6101,
+      "step": 4100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 2.6027,
+      "step": 4200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 2.5753,
+      "step": 4300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 2.5974,
+      "step": 4400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 2.6356,
+      "step": 4500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 2.5751,
+      "step": 4600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 2.5813,
+      "step": 4700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 2.5925,
+      "step": 4800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 2.5726,
+      "step": 4900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 2.593,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 2.5702,
+      "step": 5100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 2.5746,
+      "step": 5200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 2.5799,
+      "step": 5300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 2.5502,
+      "step": 5400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 2.5812,
+      "step": 5500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 2.5866,
+      "step": 5600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 2.5586,
+      "step": 5700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 2.568,
+      "step": 5800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 2.5671,
+      "step": 5900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 2.5502,
+      "step": 6000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 2.5526,
+      "step": 6100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 2.5421,
+      "step": 6200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 2.5462,
+      "step": 6300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 2.5394,
+      "step": 6400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 2.5563,
+      "step": 6500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 2.5762,
+      "step": 6600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 2.5232,
+      "step": 6700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 2.5681,
+      "step": 6800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 2.565,
+      "step": 6900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 2.516,
+      "step": 7000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 2.5506,
+      "step": 7100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 2.5251,
+      "step": 7200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 2.4942,
+      "step": 7300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 2.5034,
+      "step": 7400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 2.5232,
+      "step": 7500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 2.523,
+      "step": 7600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 2.5113,
+      "step": 7700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 2.5103,
+      "step": 7800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 2.512,
+      "step": 7900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 2.5053,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 2.5327,
+      "step": 8100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 2.5199,
+      "step": 8200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 2.5373,
+      "step": 8300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 2.484,
+      "step": 8400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 2.5555,
+      "step": 8500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 2.4938,
+      "step": 8600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 2.5342,
+      "step": 8700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 2.4963,
+      "step": 8800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 2.5022,
+      "step": 8900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 2.5201,
+      "step": 9000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 2.4868,
+      "step": 9100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 2.4804,
+      "step": 9200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 2.4936,
+      "step": 9300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 2.5081,
+      "step": 9400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 2.4851,
+      "step": 9500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 2.498,
+      "step": 9600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 2.4868,
+      "step": 9700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 2.5126,
+      "step": 9800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 2.5031,
+      "step": 9900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 2.4989,
+      "step": 10000
+    },
+    {
+      "epoch": 0.78,
+      "eval_gen_len": 18.83164121126512,
+      "eval_loss": 2.2748022079467773,
+      "eval_rouge1": 31.1136,
+      "eval_rouge2": 10.0478,
+      "eval_rougeL": 24.9492,
+      "eval_rougeLsum": 24.9433,
+      "eval_runtime": 361.4284,
+      "eval_samples_per_second": 31.34,
+      "eval_steps_per_second": 1.959,
+      "step": 10000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 2.4972,
+      "step": 10100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 2.4781,
+      "step": 10200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 2.4716,
+      "step": 10300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 2.47,
+      "step": 10400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 2.5001,
+      "step": 10500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 2.5084,
+      "step": 10600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 2.4845,
+      "step": 10700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 2.469,
+      "step": 10800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 2.4909,
+      "step": 10900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 2.4785,
+      "step": 11000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 2.4664,
+      "step": 11100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 2.4913,
+      "step": 11200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 2.4588,
+      "step": 11300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 2.4712,
+      "step": 11400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 2.4857,
+      "step": 11500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 2.4542,
+      "step": 11600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 2.4724,
+      "step": 11700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 2.48,
+      "step": 11800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 2.4975,
+      "step": 11900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 2.4908,
+      "step": 12000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 2.4845,
+      "step": 12100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 2.474,
+      "step": 12200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 2.4674,
+      "step": 12300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 2.4571,
+      "step": 12400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 2.4618,
+      "step": 12500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 2.5003,
+      "step": 12600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 2.4793,
+      "step": 12700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 2.4271,
+      "step": 12800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005,
+      "loss": 2.372,
+      "step": 12900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005,
+      "loss": 2.3714,
+      "step": 13000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005,
+      "loss": 2.3498,
+      "step": 13100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005,
+      "loss": 2.3581,
+      "step": 13200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005,
+      "loss": 2.3774,
+      "step": 13300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0005,
+      "loss": 2.3658,
+      "step": 13400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0005,
+      "loss": 2.4074,
+      "step": 13500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005,
+      "loss": 2.4005,
+      "step": 13600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005,
+      "loss": 2.3915,
+      "step": 13700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0005,
+      "loss": 2.3894,
+      "step": 13800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0005,
+      "loss": 2.3981,
+      "step": 13900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0005,
+      "loss": 2.3875,
+      "step": 14000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0005,
+      "loss": 2.3823,
+      "step": 14100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0005,
+      "loss": 2.3678,
+      "step": 14200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0005,
+      "loss": 2.3891,
+      "step": 14300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0005,
+      "loss": 2.3688,
+      "step": 14400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005,
+      "loss": 2.3524,
+      "step": 14500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005,
+      "loss": 2.3719,
+      "step": 14600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0005,
+      "loss": 2.3646,
+      "step": 14700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0005,
+      "loss": 2.3875,
+      "step": 14800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0005,
+      "loss": 2.3717,
+      "step": 14900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0005,
+      "loss": 2.3585,
+      "step": 15000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0005,
+      "loss": 2.348,
+      "step": 15100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0005,
+      "loss": 2.3713,
+      "step": 15200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0005,
+      "loss": 2.392,
+      "step": 15300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0005,
+      "loss": 2.3611,
+      "step": 15400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0005,
+      "loss": 2.3781,
+      "step": 15500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0005,
+      "loss": 2.3851,
+      "step": 15600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0005,
+      "loss": 2.3868,
+      "step": 15700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0005,
+      "loss": 2.3603,
+      "step": 15800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005,
+      "loss": 2.3947,
+      "step": 15900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005,
+      "loss": 2.3965,
+      "step": 16000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0005,
+      "loss": 2.3836,
+      "step": 16100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0005,
+      "loss": 2.3423,
+      "step": 16200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0005,
+      "loss": 2.3669,
+      "step": 16300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0005,
+      "loss": 2.3748,
+      "step": 16400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0005,
+      "loss": 2.3593,
+      "step": 16500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0005,
+      "loss": 2.3559,
+      "step": 16600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0005,
+      "loss": 2.3652,
+      "step": 16700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0005,
+      "loss": 2.3841,
+      "step": 16800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0005,
+      "loss": 2.3874,
+      "step": 16900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0005,
+      "loss": 2.3607,
+      "step": 17000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0005,
+      "loss": 2.3849,
+      "step": 17100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0005,
+      "loss": 2.3809,
+      "step": 17200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0005,
+      "loss": 2.359,
+      "step": 17300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0005,
+      "loss": 2.3618,
+      "step": 17400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005,
+      "loss": 2.3758,
+      "step": 17500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005,
+      "loss": 2.3538,
+      "step": 17600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005,
+      "loss": 2.3733,
+      "step": 17700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005,
+      "loss": 2.3735,
+      "step": 17800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005,
+      "loss": 2.3859,
+      "step": 17900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005,
+      "loss": 2.3508,
+      "step": 18000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005,
+      "loss": 2.3859,
+      "step": 18100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005,
+      "loss": 2.3567,
+      "step": 18200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005,
+      "loss": 2.3292,
+      "step": 18300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005,
+      "loss": 2.3882,
+      "step": 18400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005,
+      "loss": 2.3592,
+      "step": 18500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005,
+      "loss": 2.3594,
+      "step": 18600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005,
+      "loss": 2.359,
+      "step": 18700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005,
+      "loss": 2.3549,
+      "step": 18800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005,
+      "loss": 2.3962,
+      "step": 18900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005,
+      "loss": 2.3413,
+      "step": 19000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005,
+      "loss": 2.3636,
+      "step": 19100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005,
+      "loss": 2.3381,
+      "step": 19200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005,
+      "loss": 2.363,
+      "step": 19300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005,
+      "loss": 2.3687,
+      "step": 19400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005,
+      "loss": 2.3359,
+      "step": 19500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005,
+      "loss": 2.361,
+      "step": 19600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005,
+      "loss": 2.3808,
+      "step": 19700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005,
+      "loss": 2.347,
+      "step": 19800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005,
+      "loss": 2.3695,
+      "step": 19900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005,
+      "loss": 2.3456,
+      "step": 20000
+    },
+    {
+      "epoch": 1.57,
+      "eval_gen_len": 18.73682351902534,
+      "eval_loss": 2.202566385269165,
+      "eval_rouge1": 31.9461,
+      "eval_rouge2": 10.8723,
+      "eval_rougeL": 25.7822,
+      "eval_rougeLsum": 25.7746,
+      "eval_runtime": 358.9411,
+      "eval_samples_per_second": 31.557,
+      "eval_steps_per_second": 1.972,
+      "step": 20000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005,
+      "loss": 2.3445,
+      "step": 20100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005,
+      "loss": 2.36,
+      "step": 20200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005,
+      "loss": 2.3749,
+      "step": 20300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005,
+      "loss": 2.3606,
+      "step": 20400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005,
+      "loss": 2.3375,
+      "step": 20500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005,
+      "loss": 2.3448,
+      "step": 20600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005,
+      "loss": 2.3698,
+      "step": 20700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005,
+      "loss": 2.3403,
+      "step": 20800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005,
+      "loss": 2.354,
+      "step": 20900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005,
+      "loss": 2.3578,
+      "step": 21000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005,
+      "loss": 2.3598,
+      "step": 21100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0005,
+      "loss": 2.3262,
+      "step": 21200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0005,
+      "loss": 2.3667,
+      "step": 21300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0005,
+      "loss": 2.3801,
+      "step": 21400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005,
+      "loss": 2.3638,
+      "step": 21500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005,
+      "loss": 2.3559,
+      "step": 21600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0005,
+      "loss": 2.3595,
+      "step": 21700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0005,
+      "loss": 2.3472,
+      "step": 21800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005,
+      "loss": 2.3415,
+      "step": 21900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0005,
+      "loss": 2.3427,
+      "step": 22000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0005,
+      "loss": 2.3306,
+      "step": 22100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0005,
+      "loss": 2.3147,
+      "step": 22200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0005,
+      "loss": 2.3402,
+      "step": 22300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0005,
+      "loss": 2.3294,
+      "step": 22400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0005,
+      "loss": 2.383,
+      "step": 22500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0005,
+      "loss": 2.3562,
+      "step": 22600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0005,
+      "loss": 2.3583,
+      "step": 22700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0005,
+      "loss": 2.353,
+      "step": 22800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0005,
+      "loss": 2.3428,
+      "step": 22900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0005,
+      "loss": 2.3461,
+      "step": 23000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0005,
+      "loss": 2.3351,
+      "step": 23100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0005,
+      "loss": 2.3735,
+      "step": 23200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0005,
+      "loss": 2.3722,
+      "step": 23300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0005,
+      "loss": 2.3445,
+      "step": 23400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0005,
+      "loss": 2.2993,
+      "step": 23500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0005,
+      "loss": 2.3267,
+      "step": 23600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0005,
+      "loss": 2.3075,
+      "step": 23700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0005,
+      "loss": 2.3993,
+      "step": 23800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0005,
+      "loss": 2.3256,
+      "step": 23900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0005,
+      "loss": 2.3395,
+      "step": 24000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0005,
+      "loss": 2.3395,
+      "step": 24100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0005,
+      "loss": 2.3619,
+      "step": 24200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0005,
+      "loss": 2.3502,
+      "step": 24300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0005,
+      "loss": 2.3278,
+      "step": 24400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0005,
+      "loss": 2.3603,
+      "step": 24500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0005,
+      "loss": 2.3323,
+      "step": 24600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0005,
+      "loss": 2.374,
+      "step": 24700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0005,
+      "loss": 2.3298,
+      "step": 24800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0005,
+      "loss": 2.3274,
+      "step": 24900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0005,
+      "loss": 2.3451,
+      "step": 25000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0005,
+      "loss": 2.2976,
+      "step": 25100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0005,
+      "loss": 2.3222,
+      "step": 25200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0005,
+      "loss": 2.3013,
+      "step": 25300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0005,
+      "loss": 2.3352,
+      "step": 25400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0005,
+      "loss": 2.3124,
+      "step": 25500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0005,
+      "loss": 2.2476,
+      "step": 25600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005,
+      "loss": 2.2515,
+      "step": 25700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005,
+      "loss": 2.2521,
+      "step": 25800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005,
+      "loss": 2.2177,
+      "step": 25900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005,
+      "loss": 2.2554,
+      "step": 26000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005,
+      "loss": 2.2345,
+      "step": 26100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005,
+      "loss": 2.2548,
+      "step": 26200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005,
+      "loss": 2.2591,
+      "step": 26300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005,
+      "loss": 2.2414,
+      "step": 26400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005,
+      "loss": 2.2562,
+      "step": 26500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005,
+      "loss": 2.2722,
+      "step": 26600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005,
+      "loss": 2.2945,
+      "step": 26700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005,
+      "loss": 2.2473,
+      "step": 26800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005,
+      "loss": 2.2406,
+      "step": 26900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005,
+      "loss": 2.2683,
+      "step": 27000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005,
+      "loss": 2.2277,
+      "step": 27100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005,
+      "loss": 2.2773,
+      "step": 27200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005,
+      "loss": 2.2379,
+      "step": 27300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005,
+      "loss": 2.2632,
+      "step": 27400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005,
+      "loss": 2.2494,
+      "step": 27500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005,
+      "loss": 2.2149,
+      "step": 27600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005,
+      "loss": 2.2357,
+      "step": 27700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005,
+      "loss": 2.2573,
+      "step": 27800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005,
+      "loss": 2.2577,
+      "step": 27900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005,
+      "loss": 2.2526,
+      "step": 28000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005,
+      "loss": 2.2573,
+      "step": 28100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005,
+      "loss": 2.2741,
+      "step": 28200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005,
+      "loss": 2.2442,
+      "step": 28300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005,
+      "loss": 2.2483,
+      "step": 28400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005,
+      "loss": 2.2582,
+      "step": 28500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005,
+      "loss": 2.275,
+      "step": 28600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005,
+      "loss": 2.2972,
+      "step": 28700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005,
+      "loss": 2.2356,
+      "step": 28800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005,
+      "loss": 2.2517,
+      "step": 28900
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005,
+      "loss": 2.3137,
+      "step": 29000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005,
+      "loss": 2.2788,
+      "step": 29100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005,
+      "loss": 2.2562,
+      "step": 29200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005,
+      "loss": 2.2825,
+      "step": 29300
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005,
+      "loss": 2.2082,
+      "step": 29400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005,
+      "loss": 2.2491,
+      "step": 29500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005,
+      "loss": 2.2299,
+      "step": 29600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005,
+      "loss": 2.2678,
+      "step": 29700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005,
+      "loss": 2.2961,
+      "step": 29800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005,
+      "loss": 2.2751,
+      "step": 29900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005,
+      "loss": 2.2444,
+      "step": 30000
+    },
+    {
+      "epoch": 2.35,
+      "eval_gen_len": 18.769223978105412,
+      "eval_loss": 2.1571497917175293,
+      "eval_rouge1": 32.5123,
+      "eval_rouge2": 11.4387,
+      "eval_rougeL": 26.3083,
+      "eval_rougeLsum": 26.2953,
+      "eval_runtime": 360.0544,
+      "eval_samples_per_second": 31.459,
+      "eval_steps_per_second": 1.966,
+      "step": 30000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005,
+      "loss": 2.2231,
+      "step": 30100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005,
+      "loss": 2.2138,
+      "step": 30200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005,
+      "loss": 2.2555,
+      "step": 30300
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005,
+      "loss": 2.255,
+      "step": 30400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005,
+      "loss": 2.2482,
+      "step": 30500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005,
+      "loss": 2.2546,
+      "step": 30600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005,
+      "loss": 2.2404,
+      "step": 30700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005,
+      "loss": 2.2568,
+      "step": 30800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005,
+      "loss": 2.2331,
+      "step": 30900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005,
+      "loss": 2.2175,
+      "step": 31000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005,
+      "loss": 2.2343,
+      "step": 31100
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005,
+      "loss": 2.2838,
+      "step": 31200
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005,
+      "loss": 2.2531,
+      "step": 31300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005,
+      "loss": 2.2971,
+      "step": 31400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005,
+      "loss": 2.2527,
+      "step": 31500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005,
+      "loss": 2.2508,
+      "step": 31600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005,
+      "loss": 2.2392,
+      "step": 31700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005,
+      "loss": 2.2263,
+      "step": 31800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005,
+      "loss": 2.2048,
+      "step": 31900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005,
+      "loss": 2.2687,
+      "step": 32000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005,
+      "loss": 2.2202,
+      "step": 32100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005,
+      "loss": 2.2439,
+      "step": 32200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005,
+      "loss": 2.2705,
+      "step": 32300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005,
+      "loss": 2.2384,
+      "step": 32400
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005,
+      "loss": 2.2517,
+      "step": 32500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005,
+      "loss": 2.2336,
+      "step": 32600
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005,
+      "loss": 2.2587,
+      "step": 32700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005,
+      "loss": 2.2716,
+      "step": 32800
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005,
+      "loss": 2.2294,
+      "step": 32900
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005,
+      "loss": 2.2819,
+      "step": 33000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005,
+      "loss": 2.2461,
+      "step": 33100
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005,
+      "loss": 2.2399,
+      "step": 33200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005,
+      "loss": 2.2518,
+      "step": 33300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005,
+      "loss": 2.2797,
+      "step": 33400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005,
+      "loss": 2.2575,
+      "step": 33500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005,
+      "loss": 2.2288,
+      "step": 33600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005,
+      "loss": 2.2504,
+      "step": 33700
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005,
+      "loss": 2.229,
+      "step": 33800
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005,
+      "loss": 2.2625,
+      "step": 33900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005,
+      "loss": 2.2648,
+      "step": 34000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005,
+      "loss": 2.2677,
+      "step": 34100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005,
+      "loss": 2.2825,
+      "step": 34200
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005,
+      "loss": 2.3043,
+      "step": 34300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005,
+      "loss": 2.2503,
+      "step": 34400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005,
+      "loss": 2.2582,
+      "step": 34500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005,
+      "loss": 2.2516,
+      "step": 34600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005,
+      "loss": 2.2583,
+      "step": 34700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005,
+      "loss": 2.2899,
+      "step": 34800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005,
+      "loss": 2.2794,
+      "step": 34900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005,
+      "loss": 2.2469,
+      "step": 35000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005,
+      "loss": 2.2274,
+      "step": 35100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005,
+      "loss": 2.2566,
+      "step": 35200
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005,
+      "loss": 2.2657,
+      "step": 35300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005,
+      "loss": 2.2718,
+      "step": 35400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005,
+      "loss": 2.2226,
+      "step": 35500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005,
+      "loss": 2.2495,
+      "step": 35600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005,
+      "loss": 2.2368,
+      "step": 35700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005,
+      "loss": 2.2553,
+      "step": 35800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005,
+      "loss": 2.2305,
+      "step": 35900
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005,
+      "loss": 2.2716,
+      "step": 36000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005,
+      "loss": 2.2758,
+      "step": 36100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005,
+      "loss": 2.2743,
+      "step": 36200
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005,
+      "loss": 2.2621,
+      "step": 36300
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005,
+      "loss": 2.2489,
+      "step": 36400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005,
+      "loss": 2.2363,
+      "step": 36500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005,
+      "loss": 2.257,
+      "step": 36600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005,
+      "loss": 2.2738,
+      "step": 36700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005,
+      "loss": 2.2153,
+      "step": 36800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005,
+      "loss": 2.2587,
+      "step": 36900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005,
+      "loss": 2.2536,
+      "step": 37000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005,
+      "loss": 2.2509,
+      "step": 37100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005,
+      "loss": 2.2362,
+      "step": 37200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005,
+      "loss": 2.2259,
+      "step": 37300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005,
+      "loss": 2.2235,
+      "step": 37400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005,
+      "loss": 2.2439,
+      "step": 37500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005,
+      "loss": 2.2646,
+      "step": 37600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005,
+      "loss": 2.2301,
+      "step": 37700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005,
+      "loss": 2.2284,
+      "step": 37800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005,
+      "loss": 2.2703,
+      "step": 37900
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005,
+      "loss": 2.2647,
+      "step": 38000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005,
+      "loss": 2.2456,
+      "step": 38100
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005,
+      "loss": 2.2619,
+      "step": 38200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005,
+      "loss": 2.21,
+      "step": 38300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005,
+      "loss": 2.1604,
+      "step": 38400
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005,
+      "loss": 2.165,
+      "step": 38500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0005,
+      "loss": 2.1502,
+      "step": 38600
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0005,
+      "loss": 2.1564,
+      "step": 38700
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0005,
+      "loss": 2.147,
+      "step": 38800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0005,
+      "loss": 2.171,
+      "step": 38900
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0005,
+      "loss": 2.1522,
+      "step": 39000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0005,
+      "loss": 2.19,
+      "step": 39100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0005,
+      "loss": 2.1632,
+      "step": 39200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0005,
+      "loss": 2.1739,
+      "step": 39300
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0005,
+      "loss": 2.1466,
+      "step": 39400
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0005,
+      "loss": 2.1726,
+      "step": 39500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0005,
+      "loss": 2.1659,
+      "step": 39600
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0005,
+      "loss": 2.1573,
+      "step": 39700
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0005,
+      "loss": 2.1853,
+      "step": 39800
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0005,
+      "loss": 2.1446,
+      "step": 39900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0005,
+      "loss": 2.1901,
+      "step": 40000
+    },
+    {
+      "epoch": 3.14,
+      "eval_gen_len": 18.773461640328417,
+      "eval_loss": 2.1257381439208984,
+      "eval_rouge1": 32.8553,
+      "eval_rouge2": 11.7404,
+      "eval_rougeL": 26.6114,
+      "eval_rougeLsum": 26.6102,
+      "eval_runtime": 359.175,
+      "eval_samples_per_second": 31.536,
+      "eval_steps_per_second": 1.971,
+      "step": 40000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0005,
+      "loss": 2.1934,
+      "step": 40100
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0005,
+      "loss": 2.1582,
+      "step": 40200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0005,
+      "loss": 2.1633,
+      "step": 40300
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0005,
+      "loss": 2.1623,
+      "step": 40400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0005,
+      "loss": 2.1895,
+      "step": 40500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0005,
+      "loss": 2.1656,
+      "step": 40600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0005,
+      "loss": 2.1944,
+      "step": 40700
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0005,
+      "loss": 2.1575,
+      "step": 40800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005,
+      "loss": 2.1717,
+      "step": 40900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005,
+      "loss": 2.1541,
+      "step": 41000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005,
+      "loss": 2.1976,
+      "step": 41100
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005,
+      "loss": 2.1578,
+      "step": 41200
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005,
+      "loss": 2.1661,
+      "step": 41300
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005,
+      "loss": 2.2012,
+      "step": 41400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005,
+      "loss": 2.1878,
+      "step": 41500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005,
+      "loss": 2.144,
+      "step": 41600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005,
+      "loss": 2.1595,
+      "step": 41700
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005,
+      "loss": 2.1741,
+      "step": 41800
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005,
+      "loss": 2.1908,
+      "step": 41900
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005,
+      "loss": 2.1943,
+      "step": 42000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005,
+      "loss": 2.1714,
+      "step": 42100
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005,
+      "loss": 2.1638,
+      "step": 42200
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005,
+      "loss": 2.1751,
+      "step": 42300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005,
+      "loss": 2.1649,
+      "step": 42400
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005,
+      "loss": 2.2036,
+      "step": 42500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005,
+      "loss": 2.1772,
+      "step": 42600
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005,
+      "loss": 2.16,
+      "step": 42700
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005,
+      "loss": 2.1918,
+      "step": 42800
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005,
+      "loss": 2.1737,
+      "step": 42900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005,
+      "loss": 2.1684,
+      "step": 43000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005,
+      "loss": 2.1722,
+      "step": 43100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0005,
+      "loss": 2.1881,
+      "step": 43200
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0005,
+      "loss": 2.1944,
+      "step": 43300
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0005,
+      "loss": 2.192,
+      "step": 43400
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0005,
+      "loss": 2.1617,
+      "step": 43500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0005,
+      "loss": 2.2029,
+      "step": 43600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0005,
+      "loss": 2.1596,
+      "step": 43700
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0005,
+      "loss": 2.1793,
+      "step": 43800
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0005,
+      "loss": 2.1792,
+      "step": 43900
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0005,
+      "loss": 2.1892,
+      "step": 44000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0005,
+      "loss": 2.1759,
+      "step": 44100
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0005,
+      "loss": 2.1724,
+      "step": 44200
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0005,
+      "loss": 2.1689,
+      "step": 44300
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0005,
+      "loss": 2.1707,
+      "step": 44400
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0005,
+      "loss": 2.1845,
+      "step": 44500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0005,
+      "loss": 2.1765,
+      "step": 44600
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0005,
+      "loss": 2.1788,
+      "step": 44700
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0005,
+      "loss": 2.1824,
+      "step": 44800
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0005,
+      "loss": 2.1826,
+      "step": 44900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0005,
+      "loss": 2.1299,
+      "step": 45000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0005,
+      "loss": 2.1819,
+      "step": 45100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0005,
+      "loss": 2.1741,
+      "step": 45200
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0005,
+      "loss": 2.1896,
+      "step": 45300
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0005,
+      "loss": 2.1772,
+      "step": 45400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0005,
+      "loss": 2.2326,
+      "step": 45500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0005,
+      "loss": 2.1733,
+      "step": 45600
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0005,
+      "loss": 2.2147,
+      "step": 45700
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0005,
+      "loss": 2.1753,
+      "step": 45800
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0005,
+      "loss": 2.187,
+      "step": 45900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0005,
+      "loss": 2.1863,
+      "step": 46000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0005,
+      "loss": 2.1605,
+      "step": 46100
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0005,
+      "loss": 2.1484,
+      "step": 46200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0005,
+      "loss": 2.2143,
+      "step": 46300
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0005,
+      "loss": 2.1812,
+      "step": 46400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0005,
+      "loss": 2.1725,
+      "step": 46500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0005,
+      "loss": 2.2017,
+      "step": 46600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0005,
+      "loss": 2.225,
+      "step": 46700
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0005,
+      "loss": 2.1981,
+      "step": 46800
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0005,
+      "loss": 2.1845,
+      "step": 46900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0005,
+      "loss": 2.1595,
+      "step": 47000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0005,
+      "loss": 2.1557,
+      "step": 47100
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0005,
+      "loss": 2.1729,
+      "step": 47200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0005,
+      "loss": 2.1649,
+      "step": 47300
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0005,
+      "loss": 2.1793,
+      "step": 47400
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0005,
+      "loss": 2.1388,
+      "step": 47500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0005,
+      "loss": 2.1773,
+      "step": 47600
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0005,
+      "loss": 2.1915,
+      "step": 47700
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0005,
+      "loss": 2.1809,
+      "step": 47800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0005,
+      "loss": 2.1909,
+      "step": 47900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0005,
+      "loss": 2.153,
+      "step": 48000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0005,
+      "loss": 2.1878,
+      "step": 48100
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0005,
+      "loss": 2.1892,
+      "step": 48200
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0005,
+      "loss": 2.1804,
+      "step": 48300
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0005,
+      "loss": 2.1694,
+      "step": 48400
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0005,
+      "loss": 2.208,
+      "step": 48500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0005,
+      "loss": 2.153,
+      "step": 48600
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0005,
+      "loss": 2.215,
+      "step": 48700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0005,
+      "loss": 2.1499,
+      "step": 48800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0005,
+      "loss": 2.1766,
+      "step": 48900
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0005,
+      "loss": 2.1973,
+      "step": 49000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0005,
+      "loss": 2.2039,
+      "step": 49100
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0005,
+      "loss": 2.1866,
+      "step": 49200
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0005,
+      "loss": 2.1763,
+      "step": 49300
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0005,
+      "loss": 2.1737,
+      "step": 49400
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0005,
+      "loss": 2.2036,
+      "step": 49500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0005,
+      "loss": 2.21,
+      "step": 49600
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0005,
+      "loss": 2.137,
+      "step": 49700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0005,
+      "loss": 2.1908,
+      "step": 49800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0005,
+      "loss": 2.1764,
+      "step": 49900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0005,
+      "loss": 2.2058,
+      "step": 50000
+    },
+    {
+      "epoch": 3.92,
+      "eval_gen_len": 18.717577469762514,
+      "eval_loss": 2.0990633964538574,
+      "eval_rouge1": 33.2448,
+      "eval_rouge2": 12.1995,
+      "eval_rougeL": 27.0821,
+      "eval_rougeLsum": 27.086,
+      "eval_runtime": 360.4822,
+      "eval_samples_per_second": 31.422,
+      "eval_steps_per_second": 1.964,
+      "step": 50000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0005,
+      "loss": 2.1585,
+      "step": 50100
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0005,
+      "loss": 2.1888,
+      "step": 50200
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0005,
+      "loss": 2.2,
+      "step": 50300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0005,
+      "loss": 2.1772,
+      "step": 50400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0005,
+      "loss": 2.2016,
+      "step": 50500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0005,
+      "loss": 2.1783,
+      "step": 50600
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0005,
+      "loss": 2.1639,
+      "step": 50700
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0005,
+      "loss": 2.1585,
+      "step": 50800
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0005,
+      "loss": 2.1983,
+      "step": 50900
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0005,
+      "loss": 2.1695,
+      "step": 51000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0005,
+      "loss": 2.0872,
+      "step": 51100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0005,
+      "loss": 2.1085,
+      "step": 51200
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0005,
+      "loss": 2.1441,
+      "step": 51300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0005,
+      "loss": 2.0954,
+      "step": 51400
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0005,
+      "loss": 2.1041,
+      "step": 51500
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0005,
+      "loss": 2.0745,
+      "step": 51600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0005,
+      "loss": 2.097,
+      "step": 51700
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0005,
+      "loss": 2.119,
+      "step": 51800
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0005,
+      "loss": 2.0559,
+      "step": 51900
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0005,
+      "loss": 2.1543,
+      "step": 52000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0005,
+      "loss": 2.1138,
+      "step": 52100
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0005,
+      "loss": 2.0916,
+      "step": 52200
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0005,
+      "loss": 2.1135,
+      "step": 52300
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0005,
+      "loss": 2.0917,
+      "step": 52400
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0005,
+      "loss": 2.1246,
+      "step": 52500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0005,
+      "loss": 2.1137,
+      "step": 52600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0005,
+      "loss": 2.1029,
+      "step": 52700
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0005,
+      "loss": 2.103,
+      "step": 52800
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0005,
+      "loss": 2.1063,
+      "step": 52900
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0005,
+      "loss": 2.1103,
+      "step": 53000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0005,
+      "loss": 2.1083,
+      "step": 53100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0005,
+      "loss": 2.1142,
+      "step": 53200
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0005,
+      "loss": 2.1066,
+      "step": 53300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0005,
+      "loss": 2.1003,
+      "step": 53400
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0005,
+      "loss": 2.0934,
+      "step": 53500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0005,
+      "loss": 2.0904,
+      "step": 53600
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0005,
+      "loss": 2.141,
+      "step": 53700
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0005,
+      "loss": 2.0869,
+      "step": 53800
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0005,
+      "loss": 2.1202,
+      "step": 53900
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0005,
+      "loss": 2.1131,
+      "step": 54000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0005,
+      "loss": 2.1181,
+      "step": 54100
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0005,
+      "loss": 2.1484,
+      "step": 54200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0005,
+      "loss": 2.1127,
+      "step": 54300
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0005,
+      "loss": 2.1079,
+      "step": 54400
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0957,
+      "step": 54500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0005,
+      "loss": 2.1231,
+      "step": 54600
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0005,
+      "loss": 2.1139,
+      "step": 54700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0005,
+      "loss": 2.096,
+      "step": 54800
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0005,
+      "loss": 2.1174,
+      "step": 54900
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0005,
+      "loss": 2.1045,
+      "step": 55000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0005,
+      "loss": 2.0988,
+      "step": 55100
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0005,
+      "loss": 2.0954,
+      "step": 55200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0005,
+      "loss": 2.0964,
+      "step": 55300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0005,
+      "loss": 2.1128,
+      "step": 55400
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0005,
+      "loss": 2.0874,
+      "step": 55500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0005,
+      "loss": 2.1303,
+      "step": 55600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0005,
+      "loss": 2.1261,
+      "step": 55700
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0005,
+      "loss": 2.0916,
+      "step": 55800
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0005,
+      "loss": 2.0894,
+      "step": 55900
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0005,
+      "loss": 2.1365,
+      "step": 56000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0005,
+      "loss": 2.1081,
+      "step": 56100
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0005,
+      "loss": 2.143,
+      "step": 56200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0005,
+      "loss": 2.1147,
+      "step": 56300
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0005,
+      "loss": 2.1072,
+      "step": 56400
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0005,
+      "loss": 2.1,
+      "step": 56500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0005,
+      "loss": 2.1373,
+      "step": 56600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0005,
+      "loss": 2.1385,
+      "step": 56700
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0005,
+      "loss": 2.1063,
+      "step": 56800
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0005,
+      "loss": 2.1563,
+      "step": 56900
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0005,
+      "loss": 2.1238,
+      "step": 57000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0005,
+      "loss": 2.1087,
+      "step": 57100
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0005,
+      "loss": 2.1064,
+      "step": 57200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0005,
+      "loss": 2.0979,
+      "step": 57300
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0005,
+      "loss": 2.0988,
+      "step": 57400
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0005,
+      "loss": 2.115,
+      "step": 57500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0005,
+      "loss": 2.1191,
+      "step": 57600
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0005,
+      "loss": 2.098,
+      "step": 57700
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0005,
+      "loss": 2.1081,
+      "step": 57800
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0005,
+      "loss": 2.134,
+      "step": 57900
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0005,
+      "loss": 2.1103,
+      "step": 58000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0005,
+      "loss": 2.1114,
+      "step": 58100
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0005,
+      "loss": 2.1201,
+      "step": 58200
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0005,
+      "loss": 2.1435,
+      "step": 58300
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0005,
+      "loss": 2.1254,
+      "step": 58400
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0005,
+      "loss": 2.1204,
+      "step": 58500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0005,
+      "loss": 2.1461,
+      "step": 58600
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0005,
+      "loss": 2.1181,
+      "step": 58700
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0005,
+      "loss": 2.1405,
+      "step": 58800
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0005,
+      "loss": 2.1213,
+      "step": 58900
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0005,
+      "loss": 2.1087,
+      "step": 59000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0005,
+      "loss": 2.1094,
+      "step": 59100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0005,
+      "loss": 2.12,
+      "step": 59200
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0005,
+      "loss": 2.0994,
+      "step": 59300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0005,
+      "loss": 2.1193,
+      "step": 59400
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0005,
+      "loss": 2.1288,
+      "step": 59500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0005,
+      "loss": 2.1091,
+      "step": 59600
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0005,
+      "loss": 2.1191,
+      "step": 59700
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0005,
+      "loss": 2.1305,
+      "step": 59800
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0005,
+      "loss": 2.1271,
+      "step": 59900
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0005,
+      "loss": 2.1531,
+      "step": 60000
+    },
+    {
+      "epoch": 4.71,
+      "eval_gen_len": 18.730025602542597,
+      "eval_loss": 2.0838534832000732,
+      "eval_rouge1": 33.6123,
+      "eval_rouge2": 12.46,
+      "eval_rougeL": 27.3966,
+      "eval_rougeLsum": 27.3902,
+      "eval_runtime": 366.4271,
+      "eval_samples_per_second": 30.912,
+      "eval_steps_per_second": 1.932,
+      "step": 60000
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0005,
+      "loss": 2.1047,
+      "step": 60100
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0005,
+      "loss": 2.1382,
+      "step": 60200
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0005,
+      "loss": 2.1807,
+      "step": 60300
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0005,
+      "loss": 2.1061,
+      "step": 60400
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0005,
+      "loss": 2.1272,
+      "step": 60500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0005,
+      "loss": 2.1286,
+      "step": 60600
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0005,
+      "loss": 2.1149,
+      "step": 60700
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0005,
+      "loss": 2.1097,
+      "step": 60800
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0005,
+      "loss": 2.1103,
+      "step": 60900
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0005,
+      "loss": 2.121,
+      "step": 61000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0005,
+      "loss": 2.1304,
+      "step": 61100
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0005,
+      "loss": 2.1219,
+      "step": 61200
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0005,
+      "loss": 2.1391,
+      "step": 61300
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0005,
+      "loss": 2.1386,
+      "step": 61400
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0005,
+      "loss": 2.1213,
+      "step": 61500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0005,
+      "loss": 2.0743,
+      "step": 61600
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0005,
+      "loss": 2.101,
+      "step": 61700
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0005,
+      "loss": 2.1325,
+      "step": 61800
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0005,
+      "loss": 2.128,
+      "step": 61900
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0005,
+      "loss": 2.1176,
+      "step": 62000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0005,
+      "loss": 2.1379,
+      "step": 62100
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0005,
+      "loss": 2.1102,
+      "step": 62200
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0005,
+      "loss": 2.149,
+      "step": 62300
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0005,
+      "loss": 2.1386,
+      "step": 62400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0005,
+      "loss": 2.1165,
+      "step": 62500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0005,
+      "loss": 2.1297,
+      "step": 62600
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0005,
+      "loss": 2.1164,
+      "step": 62700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0005,
+      "loss": 2.1092,
+      "step": 62800
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0005,
+      "loss": 2.1188,
+      "step": 62900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0005,
+      "loss": 2.128,
+      "step": 63000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0005,
+      "loss": 2.1438,
+      "step": 63100
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0005,
+      "loss": 2.119,
+      "step": 63200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0005,
+      "loss": 2.1456,
+      "step": 63300
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0005,
+      "loss": 2.1314,
+      "step": 63400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0005,
+      "loss": 2.1293,
+      "step": 63500
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0005,
+      "loss": 2.1111,
+      "step": 63600
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0005,
+      "loss": 2.1195,
+      "step": 63700
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0005,
+      "loss": 2.1054,
+      "step": 63800
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0005,
+      "loss": 2.0384,
+      "step": 63900
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0005,
+      "loss": 2.034,
+      "step": 64000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0005,
+      "loss": 2.0256,
+      "step": 64100
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0005,
+      "loss": 2.0358,
+      "step": 64200
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0005,
+      "loss": 2.0021,
+      "step": 64300
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0005,
+      "loss": 2.0582,
+      "step": 64400
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0005,
+      "loss": 2.0526,
+      "step": 64500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0005,
+      "loss": 2.0456,
+      "step": 64600
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0005,
+      "loss": 2.0602,
+      "step": 64700
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0005,
+      "loss": 2.0504,
+      "step": 64800
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0005,
+      "loss": 2.0517,
+      "step": 64900
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0005,
+      "loss": 2.0499,
+      "step": 65000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0005,
+      "loss": 2.0296,
+      "step": 65100
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0005,
+      "loss": 2.054,
+      "step": 65200
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0005,
+      "loss": 2.0454,
+      "step": 65300
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0005,
+      "loss": 2.0635,
+      "step": 65400
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0005,
+      "loss": 2.0474,
+      "step": 65500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0005,
+      "loss": 2.0691,
+      "step": 65600
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0005,
+      "loss": 2.0333,
+      "step": 65700
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0005,
+      "loss": 2.0033,
+      "step": 65800
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0005,
+      "loss": 2.0756,
+      "step": 65900
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0005,
+      "loss": 2.0386,
+      "step": 66000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0005,
+      "loss": 2.0326,
+      "step": 66100
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0005,
+      "loss": 2.0672,
+      "step": 66200
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0005,
+      "loss": 2.0822,
+      "step": 66300
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0005,
+      "loss": 2.0693,
+      "step": 66400
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0005,
+      "loss": 2.0563,
+      "step": 66500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0005,
+      "loss": 2.0756,
+      "step": 66600
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0005,
+      "loss": 2.0132,
+      "step": 66700
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0005,
+      "loss": 2.0786,
+      "step": 66800
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0005,
+      "loss": 2.067,
+      "step": 66900
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0005,
+      "loss": 2.059,
+      "step": 67000
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0005,
+      "loss": 2.0199,
+      "step": 67100
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0458,
+      "step": 67200
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0005,
+      "loss": 2.0783,
+      "step": 67300
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0005,
+      "loss": 2.076,
+      "step": 67400
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0005,
+      "loss": 2.0714,
+      "step": 67500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0005,
+      "loss": 2.0723,
+      "step": 67600
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0005,
+      "loss": 2.0727,
+      "step": 67700
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0005,
+      "loss": 2.0356,
+      "step": 67800
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0005,
+      "loss": 2.0634,
+      "step": 67900
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0005,
+      "loss": 2.0605,
+      "step": 68000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0005,
+      "loss": 2.0342,
+      "step": 68100
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0005,
+      "loss": 2.0354,
+      "step": 68200
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0005,
+      "loss": 2.0479,
+      "step": 68300
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0005,
+      "loss": 2.0752,
+      "step": 68400
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0005,
+      "loss": 2.0633,
+      "step": 68500
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0005,
+      "loss": 2.0621,
+      "step": 68600
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0005,
+      "loss": 2.0963,
+      "step": 68700
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0005,
+      "loss": 2.0484,
+      "step": 68800
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0005,
+      "loss": 2.0551,
+      "step": 68900
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0005,
+      "loss": 2.0968,
+      "step": 69000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0005,
+      "loss": 2.0809,
+      "step": 69100
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0005,
+      "loss": 2.0783,
+      "step": 69200
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0005,
+      "loss": 2.0593,
+      "step": 69300
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0005,
+      "loss": 2.0443,
+      "step": 69400
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0005,
+      "loss": 2.0586,
+      "step": 69500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0005,
+      "loss": 2.0708,
+      "step": 69600
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0005,
+      "loss": 2.079,
+      "step": 69700
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0005,
+      "loss": 2.0888,
+      "step": 69800
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0005,
+      "loss": 2.0972,
+      "step": 69900
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0005,
+      "loss": 2.063,
+      "step": 70000
+    },
+    {
+      "epoch": 5.49,
+      "eval_gen_len": 18.784497219034165,
+      "eval_loss": 2.068556785583496,
+      "eval_rouge1": 33.6877,
+      "eval_rouge2": 12.6196,
+      "eval_rougeL": 27.5291,
+      "eval_rougeLsum": 27.5307,
+      "eval_runtime": 359.5783,
+      "eval_samples_per_second": 31.501,
+      "eval_steps_per_second": 1.969,
+      "step": 70000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0005,
+      "loss": 2.0793,
+      "step": 70100
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0005,
+      "loss": 2.0636,
+      "step": 70200
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0005,
+      "loss": 2.0758,
+      "step": 70300
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0005,
+      "loss": 2.0628,
+      "step": 70400
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0005,
+      "loss": 2.0622,
+      "step": 70500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0005,
+      "loss": 2.0957,
+      "step": 70600
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0005,
+      "loss": 2.096,
+      "step": 70700
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0005,
+      "loss": 2.0606,
+      "step": 70800
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0005,
+      "loss": 2.1034,
+      "step": 70900
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0005,
+      "loss": 2.0591,
+      "step": 71000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0005,
+      "loss": 2.0884,
+      "step": 71100
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0005,
+      "loss": 2.0633,
+      "step": 71200
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0005,
+      "loss": 2.0562,
+      "step": 71300
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0005,
+      "loss": 2.0772,
+      "step": 71400
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0005,
+      "loss": 2.126,
+      "step": 71500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0005,
+      "loss": 2.0717,
+      "step": 71600
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0005,
+      "loss": 2.081,
+      "step": 71700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0005,
+      "loss": 2.0639,
+      "step": 71800
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0005,
+      "loss": 2.1004,
+      "step": 71900
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0005,
+      "loss": 2.0877,
+      "step": 72000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0005,
+      "loss": 2.0836,
+      "step": 72100
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0005,
+      "loss": 2.0649,
+      "step": 72200
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0005,
+      "loss": 2.0571,
+      "step": 72300
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0005,
+      "loss": 2.0496,
+      "step": 72400
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0005,
+      "loss": 2.0761,
+      "step": 72500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0005,
+      "loss": 2.0765,
+      "step": 72600
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0005,
+      "loss": 2.0764,
+      "step": 72700
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0005,
+      "loss": 2.0524,
+      "step": 72800
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0005,
+      "loss": 2.0802,
+      "step": 72900
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0005,
+      "loss": 2.0551,
+      "step": 73000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0005,
+      "loss": 2.0552,
+      "step": 73100
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0005,
+      "loss": 2.0579,
+      "step": 73200
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0005,
+      "loss": 2.0506,
+      "step": 73300
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0005,
+      "loss": 2.0771,
+      "step": 73400
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0005,
+      "loss": 2.0851,
+      "step": 73500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0005,
+      "loss": 2.0828,
+      "step": 73600
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0005,
+      "loss": 2.0987,
+      "step": 73700
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0005,
+      "loss": 2.1023,
+      "step": 73800
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0005,
+      "loss": 2.0703,
+      "step": 73900
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0005,
+      "loss": 2.0784,
+      "step": 74000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0005,
+      "loss": 2.0518,
+      "step": 74100
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0005,
+      "loss": 2.0901,
+      "step": 74200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0005,
+      "loss": 2.0442,
+      "step": 74300
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0005,
+      "loss": 2.0926,
+      "step": 74400
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0005,
+      "loss": 2.0789,
+      "step": 74500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0005,
+      "loss": 2.0929,
+      "step": 74600
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0005,
+      "loss": 2.0689,
+      "step": 74700
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0005,
+      "loss": 2.1078,
+      "step": 74800
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0005,
+      "loss": 2.0821,
+      "step": 74900
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0005,
+      "loss": 2.0605,
+      "step": 75000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0005,
+      "loss": 2.0762,
+      "step": 75100
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0005,
+      "loss": 2.0827,
+      "step": 75200
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0005,
+      "loss": 2.0828,
+      "step": 75300
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0005,
+      "loss": 2.0767,
+      "step": 75400
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0005,
+      "loss": 2.0732,
+      "step": 75500
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0005,
+      "loss": 2.0692,
+      "step": 75600
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0005,
+      "loss": 2.0544,
+      "step": 75700
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0005,
+      "loss": 2.0736,
+      "step": 75800
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0005,
+      "loss": 2.0803,
+      "step": 75900
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0005,
+      "loss": 2.0889,
+      "step": 76000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0005,
+      "loss": 2.0841,
+      "step": 76100
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0005,
+      "loss": 2.064,
+      "step": 76200
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0005,
+      "loss": 2.0575,
+      "step": 76300
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0005,
+      "loss": 2.0763,
+      "step": 76400
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0005,
+      "loss": 2.0605,
+      "step": 76500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0005,
+      "loss": 2.0113,
+      "step": 76600
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0005,
+      "loss": 1.9809,
+      "step": 76700
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0005,
+      "loss": 2.0006,
+      "step": 76800
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0005,
+      "loss": 1.9569,
+      "step": 76900
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0005,
+      "loss": 1.985,
+      "step": 77000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0005,
+      "loss": 1.9736,
+      "step": 77100
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0005,
+      "loss": 1.9773,
+      "step": 77200
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0005,
+      "loss": 2.0067,
+      "step": 77300
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0005,
+      "loss": 2.0069,
+      "step": 77400
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0005,
+      "loss": 2.0018,
+      "step": 77500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0005,
+      "loss": 1.9895,
+      "step": 77600
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0005,
+      "loss": 1.9927,
+      "step": 77700
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0005,
+      "loss": 1.9861,
+      "step": 77800
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0005,
+      "loss": 1.9965,
+      "step": 77900
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0005,
+      "loss": 2.0095,
+      "step": 78000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0005,
+      "loss": 2.0137,
+      "step": 78100
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0005,
+      "loss": 1.9971,
+      "step": 78200
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0005,
+      "loss": 2.0021,
+      "step": 78300
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0005,
+      "loss": 2.0098,
+      "step": 78400
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0005,
+      "loss": 2.029,
+      "step": 78500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0005,
+      "loss": 2.0294,
+      "step": 78600
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0005,
+      "loss": 2.0184,
+      "step": 78700
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0005,
+      "loss": 2.0426,
+      "step": 78800
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0005,
+      "loss": 2.0107,
+      "step": 78900
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0005,
+      "loss": 2.0161,
+      "step": 79000
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0005,
+      "loss": 2.0044,
+      "step": 79100
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0005,
+      "loss": 2.0025,
+      "step": 79200
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0005,
+      "loss": 2.0023,
+      "step": 79300
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0005,
+      "loss": 2.0455,
+      "step": 79400
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0005,
+      "loss": 2.0443,
+      "step": 79500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0005,
+      "loss": 2.0365,
+      "step": 79600
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0005,
+      "loss": 2.0225,
+      "step": 79700
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0005,
+      "loss": 2.0021,
+      "step": 79800
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0005,
+      "loss": 2.0114,
+      "step": 79900
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0005,
+      "loss": 1.982,
+      "step": 80000
+    },
+    {
+      "epoch": 6.27,
+      "eval_gen_len": 18.763926900326652,
+      "eval_loss": 2.057493209838867,
+      "eval_rouge1": 33.9585,
+      "eval_rouge2": 12.853,
+      "eval_rougeL": 27.7139,
+      "eval_rougeLsum": 27.7071,
+      "eval_runtime": 362.3051,
+      "eval_samples_per_second": 31.264,
+      "eval_steps_per_second": 1.954,
+      "step": 80000
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0005,
+      "loss": 2.0384,
+      "step": 80100
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0005,
+      "loss": 2.0101,
+      "step": 80200
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0005,
+      "loss": 1.9886,
+      "step": 80300
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0005,
+      "loss": 2.0217,
+      "step": 80400
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0005,
+      "loss": 1.9968,
+      "step": 80500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0005,
+      "loss": 2.0091,
+      "step": 80600
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0005,
+      "loss": 2.0001,
+      "step": 80700
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0005,
+      "loss": 2.0355,
+      "step": 80800
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0005,
+      "loss": 2.0076,
+      "step": 80900
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0005,
+      "loss": 2.0297,
+      "step": 81000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0005,
+      "loss": 1.9883,
+      "step": 81100
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0005,
+      "loss": 1.9958,
+      "step": 81200
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0005,
+      "loss": 2.0411,
+      "step": 81300
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0005,
+      "loss": 1.9781,
+      "step": 81400
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0005,
+      "loss": 2.047,
+      "step": 81500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0005,
+      "loss": 2.0289,
+      "step": 81600
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0005,
+      "loss": 2.0189,
+      "step": 81700
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0005,
+      "loss": 2.086,
+      "step": 81800
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0005,
+      "loss": 2.0131,
+      "step": 81900
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0005,
+      "loss": 2.0036,
+      "step": 82000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0005,
+      "loss": 2.0146,
+      "step": 82100
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0005,
+      "loss": 2.0362,
+      "step": 82200
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0005,
+      "loss": 2.0268,
+      "step": 82300
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0005,
+      "loss": 2.0321,
+      "step": 82400
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0005,
+      "loss": 2.0105,
+      "step": 82500
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0005,
+      "loss": 2.0522,
+      "step": 82600
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0005,
+      "loss": 2.0456,
+      "step": 82700
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0005,
+      "loss": 1.9935,
+      "step": 82800
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0005,
+      "loss": 2.041,
+      "step": 82900
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0005,
+      "loss": 1.9851,
+      "step": 83000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0005,
+      "loss": 2.0574,
+      "step": 83100
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0005,
+      "loss": 2.0324,
+      "step": 83200
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0005,
+      "loss": 2.0367,
+      "step": 83300
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0005,
+      "loss": 2.0264,
+      "step": 83400
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0005,
+      "loss": 2.0314,
+      "step": 83500
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9554,
+      "step": 83600
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0005,
+      "loss": 2.0183,
+      "step": 83700
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0005,
+      "loss": 2.0161,
+      "step": 83800
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0005,
+      "loss": 2.0337,
+      "step": 83900
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0005,
+      "loss": 2.0075,
+      "step": 84000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0005,
+      "loss": 2.0191,
+      "step": 84100
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0005,
+      "loss": 2.0136,
+      "step": 84200
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0005,
+      "loss": 2.0535,
+      "step": 84300
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0005,
+      "loss": 2.0413,
+      "step": 84400
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0005,
+      "loss": 2.0472,
+      "step": 84500
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0005,
+      "loss": 2.0352,
+      "step": 84600
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0005,
+      "loss": 2.0219,
+      "step": 84700
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0005,
+      "loss": 2.0327,
+      "step": 84800
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0005,
+      "loss": 2.0169,
+      "step": 84900
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0005,
+      "loss": 2.0091,
+      "step": 85000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0005,
+      "loss": 2.0307,
+      "step": 85100
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0005,
+      "loss": 2.0415,
+      "step": 85200
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0005,
+      "loss": 1.9875,
+      "step": 85300
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0005,
+      "loss": 2.0242,
+      "step": 85400
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0005,
+      "loss": 2.0333,
+      "step": 85500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0005,
+      "loss": 2.0341,
+      "step": 85600
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0005,
+      "loss": 2.0467,
+      "step": 85700
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0005,
+      "loss": 2.0223,
+      "step": 85800
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0005,
+      "loss": 2.0483,
+      "step": 85900
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0005,
+      "loss": 2.0412,
+      "step": 86000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0005,
+      "loss": 2.0299,
+      "step": 86100
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0005,
+      "loss": 2.0148,
+      "step": 86200
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0005,
+      "loss": 2.0199,
+      "step": 86300
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0005,
+      "loss": 2.0394,
+      "step": 86400
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0005,
+      "loss": 2.0118,
+      "step": 86500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0005,
+      "loss": 2.0217,
+      "step": 86600
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0005,
+      "loss": 2.063,
+      "step": 86700
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0005,
+      "loss": 2.0358,
+      "step": 86800
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0005,
+      "loss": 2.0223,
+      "step": 86900
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0005,
+      "loss": 2.0308,
+      "step": 87000
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0005,
+      "loss": 2.0555,
+      "step": 87100
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0005,
+      "loss": 2.0664,
+      "step": 87200
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0005,
+      "loss": 2.0429,
+      "step": 87300
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0005,
+      "loss": 2.0329,
+      "step": 87400
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0005,
+      "loss": 2.0086,
+      "step": 87500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0005,
+      "loss": 2.0284,
+      "step": 87600
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0005,
+      "loss": 2.0535,
+      "step": 87700
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0005,
+      "loss": 2.0376,
+      "step": 87800
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0005,
+      "loss": 2.0191,
+      "step": 87900
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0005,
+      "loss": 2.0417,
+      "step": 88000
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0005,
+      "loss": 2.0254,
+      "step": 88100
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0005,
+      "loss": 2.0469,
+      "step": 88200
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0005,
+      "loss": 2.0422,
+      "step": 88300
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0005,
+      "loss": 2.0291,
+      "step": 88400
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0005,
+      "loss": 2.0549,
+      "step": 88500
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0005,
+      "loss": 2.0494,
+      "step": 88600
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0005,
+      "loss": 2.0522,
+      "step": 88700
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0005,
+      "loss": 2.0315,
+      "step": 88800
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0005,
+      "loss": 2.0284,
+      "step": 88900
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0005,
+      "loss": 2.0619,
+      "step": 89000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0005,
+      "loss": 2.0335,
+      "step": 89100
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0005,
+      "loss": 2.0259,
+      "step": 89200
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0005,
+      "loss": 2.0142,
+      "step": 89300
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0005,
+      "loss": 1.9777,
+      "step": 89400
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0005,
+      "loss": 1.9655,
+      "step": 89500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0005,
+      "loss": 1.9457,
+      "step": 89600
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0005,
+      "loss": 1.9775,
+      "step": 89700
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0005,
+      "loss": 1.9603,
+      "step": 89800
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.0005,
+      "loss": 1.9705,
+      "step": 89900
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0005,
+      "loss": 1.9568,
+      "step": 90000
+    },
+    {
+      "epoch": 7.06,
+      "eval_gen_len": 18.745916835878873,
+      "eval_loss": 2.060124635696411,
+      "eval_rouge1": 34.3703,
+      "eval_rouge2": 13.056,
+      "eval_rougeL": 28.0273,
+      "eval_rougeLsum": 28.0164,
+      "eval_runtime": 360.3535,
+      "eval_samples_per_second": 31.433,
+      "eval_steps_per_second": 1.965,
+      "step": 90000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0005,
+      "loss": 1.95,
+      "step": 90100
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0005,
+      "loss": 1.9641,
+      "step": 90200
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.0005,
+      "loss": 1.9718,
+      "step": 90300
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.0005,
+      "loss": 1.9298,
+      "step": 90400
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.0005,
+      "loss": 1.926,
+      "step": 90500
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.0005,
+      "loss": 1.9711,
+      "step": 90600
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.0005,
+      "loss": 1.955,
+      "step": 90700
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0005,
+      "loss": 1.9411,
+      "step": 90800
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0005,
+      "loss": 1.9471,
+      "step": 90900
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0005,
+      "loss": 1.9949,
+      "step": 91000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0005,
+      "loss": 1.9662,
+      "step": 91100
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0005,
+      "loss": 1.9512,
+      "step": 91200
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0005,
+      "loss": 1.9485,
+      "step": 91300
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0005,
+      "loss": 1.9587,
+      "step": 91400
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0005,
+      "loss": 2.0031,
+      "step": 91500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0005,
+      "loss": 1.9903,
+      "step": 91600
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.0005,
+      "loss": 1.9852,
+      "step": 91700
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0005,
+      "loss": 1.9856,
+      "step": 91800
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0005,
+      "loss": 1.9691,
+      "step": 91900
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0005,
+      "loss": 1.9728,
+      "step": 92000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.0005,
+      "loss": 1.9831,
+      "step": 92100
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0005,
+      "loss": 1.9617,
+      "step": 92200
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.0005,
+      "loss": 1.9783,
+      "step": 92300
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0005,
+      "loss": 1.9817,
+      "step": 92400
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0005,
+      "loss": 1.9759,
+      "step": 92500
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0005,
+      "loss": 1.9912,
+      "step": 92600
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0005,
+      "loss": 1.9836,
+      "step": 92700
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0005,
+      "loss": 1.9792,
+      "step": 92800
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0005,
+      "loss": 1.9728,
+      "step": 92900
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0005,
+      "loss": 2.0051,
+      "step": 93000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0005,
+      "loss": 1.9884,
+      "step": 93100
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0005,
+      "loss": 1.9679,
+      "step": 93200
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0005,
+      "loss": 1.9732,
+      "step": 93300
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0005,
+      "loss": 1.9627,
+      "step": 93400
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.0005,
+      "loss": 1.9745,
+      "step": 93500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0005,
+      "loss": 1.9982,
+      "step": 93600
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0005,
+      "loss": 1.9901,
+      "step": 93700
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0005,
+      "loss": 1.9544,
+      "step": 93800
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0005,
+      "loss": 1.9727,
+      "step": 93900
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0005,
+      "loss": 1.9721,
+      "step": 94000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0005,
+      "loss": 1.9807,
+      "step": 94100
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0005,
+      "loss": 1.937,
+      "step": 94200
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0005,
+      "loss": 1.9661,
+      "step": 94300
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0005,
+      "loss": 1.9868,
+      "step": 94400
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0005,
+      "loss": 1.9727,
+      "step": 94500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.0005,
+      "loss": 1.9726,
+      "step": 94600
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0005,
+      "loss": 1.9914,
+      "step": 94700
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0005,
+      "loss": 1.9804,
+      "step": 94800
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.0005,
+      "loss": 1.9533,
+      "step": 94900
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.0005,
+      "loss": 1.9625,
+      "step": 95000
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.0005,
+      "loss": 1.986,
+      "step": 95100
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.0005,
+      "loss": 1.9745,
+      "step": 95200
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.0005,
+      "loss": 2.0023,
+      "step": 95300
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0005,
+      "loss": 1.9857,
+      "step": 95400
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0005,
+      "loss": 1.9837,
+      "step": 95500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0005,
+      "loss": 1.9924,
+      "step": 95600
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0005,
+      "loss": 1.9984,
+      "step": 95700
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.0005,
+      "loss": 1.9832,
+      "step": 95800
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9741,
+      "step": 95900
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9719,
+      "step": 96000
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9789,
+      "step": 96100
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9968,
+      "step": 96200
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.0005,
+      "loss": 1.9551,
+      "step": 96300
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0005,
+      "loss": 2.0159,
+      "step": 96400
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0005,
+      "loss": 1.9721,
+      "step": 96500
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0005,
+      "loss": 1.9896,
+      "step": 96600
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0005,
+      "loss": 2.004,
+      "step": 96700
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.0005,
+      "loss": 1.9564,
+      "step": 96800
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9491,
+      "step": 96900
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9866,
+      "step": 97000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9746,
+      "step": 97100
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0005,
+      "loss": 1.9724,
+      "step": 97200
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9936,
+      "step": 97300
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.0005,
+      "loss": 1.9812,
+      "step": 97400
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.0005,
+      "loss": 2.0067,
+      "step": 97500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.0005,
+      "loss": 1.9797,
+      "step": 97600
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.0005,
+      "loss": 2.0005,
+      "step": 97700
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0005,
+      "loss": 1.9963,
+      "step": 97800
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.0005,
+      "loss": 2.0044,
+      "step": 97900
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0005,
+      "loss": 1.9896,
+      "step": 98000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0005,
+      "loss": 1.9853,
+      "step": 98100
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9957,
+      "step": 98200
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.0005,
+      "loss": 2.0128,
+      "step": 98300
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0005,
+      "loss": 1.9685,
+      "step": 98400
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0005,
+      "loss": 2.0051,
+      "step": 98500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.0005,
+      "loss": 1.9703,
+      "step": 98600
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.0005,
+      "loss": 2.0022,
+      "step": 98700
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.0005,
+      "loss": 2.0205,
+      "step": 98800
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0005,
+      "loss": 2.0008,
+      "step": 98900
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0005,
+      "loss": 2.0121,
+      "step": 99000
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.0005,
+      "loss": 2.0063,
+      "step": 99100
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.0005,
+      "loss": 1.9981,
+      "step": 99200
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.0005,
+      "loss": 1.9838,
+      "step": 99300
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.0005,
+      "loss": 1.9923,
+      "step": 99400
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.0005,
+      "loss": 1.9959,
+      "step": 99500
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.0005,
+      "loss": 1.9924,
+      "step": 99600
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.0005,
+      "loss": 2.0133,
+      "step": 99700
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.0005,
+      "loss": 1.9995,
+      "step": 99800
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.0005,
+      "loss": 2.0026,
+      "step": 99900
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.0005,
+      "loss": 1.9736,
+      "step": 100000
+    },
+    {
+      "epoch": 7.84,
+      "eval_gen_len": 18.74485742032312,
+      "eval_loss": 2.0384180545806885,
+      "eval_rouge1": 34.1724,
+      "eval_rouge2": 13.072,
+      "eval_rougeL": 27.9429,
+      "eval_rougeLsum": 27.9294,
+      "eval_runtime": 358.3007,
+      "eval_samples_per_second": 31.613,
+      "eval_steps_per_second": 1.976,
+      "step": 100000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.0005,
+      "loss": 2.0139,
+      "step": 100100
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0005,
+      "loss": 2.0064,
+      "step": 100200
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0005,
+      "loss": 1.9845,
+      "step": 100300
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0005,
+      "loss": 1.9921,
+      "step": 100400
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.0005,
+      "loss": 1.9755,
+      "step": 100500
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.0005,
+      "loss": 2.0093,
+      "step": 100600
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0005,
+      "loss": 2.0127,
+      "step": 100700
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0005,
+      "loss": 1.9878,
+      "step": 100800
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0005,
+      "loss": 1.9925,
+      "step": 100900
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.0005,
+      "loss": 1.9798,
+      "step": 101000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.0005,
+      "loss": 2.0125,
+      "step": 101100
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0005,
+      "loss": 2.012,
+      "step": 101200
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0005,
+      "loss": 1.9951,
+      "step": 101300
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.0005,
+      "loss": 2.0181,
+      "step": 101400
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.0005,
+      "loss": 2.0121,
+      "step": 101500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.0005,
+      "loss": 2.0009,
+      "step": 101600
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.0005,
+      "loss": 1.9855,
+      "step": 101700
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.0005,
+      "loss": 1.9993,
+      "step": 101800
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0005,
+      "loss": 1.9784,
+      "step": 101900
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.0005,
+      "loss": 1.9794,
+      "step": 102000
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.0005,
+      "loss": 1.9454,
+      "step": 102100
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.0005,
+      "loss": 1.9057,
+      "step": 102200
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.0005,
+      "loss": 1.912,
+      "step": 102300
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.0005,
+      "loss": 1.9005,
+      "step": 102400
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.0005,
+      "loss": 1.9037,
+      "step": 102500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.0005,
+      "loss": 1.9229,
+      "step": 102600
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.0005,
+      "loss": 1.9096,
+      "step": 102700
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.0005,
+      "loss": 1.9353,
+      "step": 102800
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.0005,
+      "loss": 1.9332,
+      "step": 102900
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.0005,
+      "loss": 1.9093,
+      "step": 103000
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.0005,
+      "loss": 1.9085,
+      "step": 103100
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.0005,
+      "loss": 1.9351,
+      "step": 103200
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.0005,
+      "loss": 1.9295,
+      "step": 103300
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.0005,
+      "loss": 1.919,
+      "step": 103400
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.0005,
+      "loss": 1.9265,
+      "step": 103500
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.0005,
+      "loss": 1.923,
+      "step": 103600
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.0005,
+      "loss": 1.9284,
+      "step": 103700
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.0005,
+      "loss": 1.9155,
+      "step": 103800
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.0005,
+      "loss": 1.9245,
+      "step": 103900
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.0005,
+      "loss": 1.9388,
+      "step": 104000
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.0005,
+      "loss": 1.9247,
+      "step": 104100
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.0005,
+      "loss": 1.9535,
+      "step": 104200
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.0005,
+      "loss": 1.9312,
+      "step": 104300
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0005,
+      "loss": 1.924,
+      "step": 104400
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0005,
+      "loss": 1.944,
+      "step": 104500
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 0.0005,
+      "loss": 1.9328,
+      "step": 104600
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.0005,
+      "loss": 1.926,
+      "step": 104700
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.0005,
+      "loss": 1.9264,
+      "step": 104800
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.0005,
+      "loss": 1.92,
+      "step": 104900
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.0005,
+      "loss": 1.94,
+      "step": 105000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.0005,
+      "loss": 1.9447,
+      "step": 105100
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 0.0005,
+      "loss": 1.9307,
+      "step": 105200
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.0005,
+      "loss": 1.9461,
+      "step": 105300
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.0005,
+      "loss": 1.9426,
+      "step": 105400
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.0005,
+      "loss": 1.9638,
+      "step": 105500
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.0005,
+      "loss": 1.9381,
+      "step": 105600
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.0005,
+      "loss": 1.9362,
+      "step": 105700
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.0005,
+      "loss": 1.9646,
+      "step": 105800
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.0005,
+      "loss": 1.9434,
+      "step": 105900
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.0005,
+      "loss": 1.9305,
+      "step": 106000
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.0005,
+      "loss": 1.9263,
+      "step": 106100
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.0005,
+      "loss": 1.9568,
+      "step": 106200
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.0005,
+      "loss": 1.941,
+      "step": 106300
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.0005,
+      "loss": 1.9541,
+      "step": 106400
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 0.0005,
+      "loss": 1.9446,
+      "step": 106500
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.0005,
+      "loss": 1.9513,
+      "step": 106600
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 0.0005,
+      "loss": 1.9394,
+      "step": 106700
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0005,
+      "loss": 1.9623,
+      "step": 106800
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0005,
+      "loss": 1.9638,
+      "step": 106900
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.0005,
+      "loss": 1.9246,
+      "step": 107000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.0005,
+      "loss": 1.9464,
+      "step": 107100
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.0005,
+      "loss": 1.9515,
+      "step": 107200
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.0005,
+      "loss": 1.9604,
+      "step": 107300
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.0005,
+      "loss": 1.9294,
+      "step": 107400
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.0005,
+      "loss": 1.9376,
+      "step": 107500
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 0.0005,
+      "loss": 1.9506,
+      "step": 107600
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.0005,
+      "loss": 1.9304,
+      "step": 107700
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.0005,
+      "loss": 1.958,
+      "step": 107800
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.0005,
+      "loss": 1.9809,
+      "step": 107900
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.0005,
+      "loss": 1.9446,
+      "step": 108000
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.0005,
+      "loss": 1.9503,
+      "step": 108100
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.0005,
+      "loss": 1.9452,
+      "step": 108200
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 0.0005,
+      "loss": 1.9523,
+      "step": 108300
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.0005,
+      "loss": 1.9206,
+      "step": 108400
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 0.0005,
+      "loss": 1.958,
+      "step": 108500
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9445,
+      "step": 108600
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9558,
+      "step": 108700
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9497,
+      "step": 108800
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9523,
+      "step": 108900
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.0005,
+      "loss": 1.9334,
+      "step": 109000
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9342,
+      "step": 109100
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9514,
+      "step": 109200
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.0005,
+      "loss": 1.979,
+      "step": 109300
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 0.0005,
+      "loss": 1.9424,
+      "step": 109400
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.0005,
+      "loss": 1.9586,
+      "step": 109500
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.0005,
+      "loss": 1.9227,
+      "step": 109600
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9335,
+      "step": 109700
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9582,
+      "step": 109800
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 0.0005,
+      "loss": 1.9475,
+      "step": 109900
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9571,
+      "step": 110000
+    },
+    {
+      "epoch": 8.63,
+      "eval_gen_len": 18.777257879403194,
+      "eval_loss": 2.0414655208587646,
+      "eval_rouge1": 34.5053,
+      "eval_rouge2": 13.283,
+      "eval_rougeL": 28.19,
+      "eval_rougeLsum": 28.1836,
+      "eval_runtime": 360.6557,
+      "eval_samples_per_second": 31.407,
+      "eval_steps_per_second": 1.963,
+      "step": 110000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9556,
+      "step": 110100
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 0.0005,
+      "loss": 1.9607,
+      "step": 110200
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 0.0005,
+      "loss": 1.9676,
+      "step": 110300
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 0.0005,
+      "loss": 1.9648,
+      "step": 110400
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.0005,
+      "loss": 1.9451,
+      "step": 110500
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.0005,
+      "loss": 1.9554,
+      "step": 110600
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.0005,
+      "loss": 1.9645,
+      "step": 110700
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 0.0005,
+      "loss": 1.9936,
+      "step": 110800
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9629,
+      "step": 110900
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9463,
+      "step": 111000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 0.0005,
+      "loss": 1.9464,
+      "step": 111100
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 0.0005,
+      "loss": 1.9193,
+      "step": 111200
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 0.0005,
+      "loss": 1.9795,
+      "step": 111300
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.0005,
+      "loss": 1.9997,
+      "step": 111400
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.0005,
+      "loss": 1.9712,
+      "step": 111500
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 0.0005,
+      "loss": 1.9425,
+      "step": 111600
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 0.0005,
+      "loss": 1.9544,
+      "step": 111700
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.0005,
+      "loss": 1.9692,
+      "step": 111800
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.0005,
+      "loss": 1.9595,
+      "step": 111900
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.0005,
+      "loss": 1.974,
+      "step": 112000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 0.0005,
+      "loss": 1.9502,
+      "step": 112100
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.0005,
+      "loss": 1.972,
+      "step": 112200
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.0005,
+      "loss": 1.9803,
+      "step": 112300
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.0005,
+      "loss": 1.9553,
+      "step": 112400
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 0.0005,
+      "loss": 1.9793,
+      "step": 112500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.0005,
+      "loss": 1.9593,
+      "step": 112600
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 0.0005,
+      "loss": 1.9687,
+      "step": 112700
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.0005,
+      "loss": 1.949,
+      "step": 112800
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.0005,
+      "loss": 1.9785,
+      "step": 112900
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 0.0005,
+      "loss": 1.9625,
+      "step": 113000
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 0.0005,
+      "loss": 1.9763,
+      "step": 113100
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.0005,
+      "loss": 1.9741,
+      "step": 113200
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.0005,
+      "loss": 1.9307,
+      "step": 113300
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 0.0005,
+      "loss": 1.9471,
+      "step": 113400
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 0.0005,
+      "loss": 1.9682,
+      "step": 113500
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 0.0005,
+      "loss": 2.0082,
+      "step": 113600
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.0005,
+      "loss": 1.9933,
+      "step": 113700
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.0005,
+      "loss": 1.9674,
+      "step": 113800
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 0.0005,
+      "loss": 1.9422,
+      "step": 113900
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 0.0005,
+      "loss": 1.9523,
+      "step": 114000
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.0005,
+      "loss": 1.9825,
+      "step": 114100
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 0.0005,
+      "loss": 1.9663,
+      "step": 114200
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 0.0005,
+      "loss": 1.9994,
+      "step": 114300
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 0.0005,
+      "loss": 1.9455,
+      "step": 114400
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.0005,
+      "loss": 1.9715,
+      "step": 114500
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.0005,
+      "loss": 1.9657,
+      "step": 114600
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.0005,
+      "loss": 1.9587,
+      "step": 114700
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 0.0005,
+      "loss": 1.9472,
+      "step": 114800
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 0.0005,
+      "loss": 1.8601,
+      "step": 114900
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 0.0005,
+      "loss": 1.8969,
+      "step": 115000
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 0.0005,
+      "loss": 1.8682,
+      "step": 115100
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 0.0005,
+      "loss": 1.9005,
+      "step": 115200
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.0005,
+      "loss": 1.8729,
+      "step": 115300
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 0.0005,
+      "loss": 1.8889,
+      "step": 115400
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 0.0005,
+      "loss": 1.8802,
+      "step": 115500
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.0005,
+      "loss": 1.8617,
+      "step": 115600
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.0005,
+      "loss": 1.8912,
+      "step": 115700
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 0.0005,
+      "loss": 1.8954,
+      "step": 115800
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 0.0005,
+      "loss": 1.9125,
+      "step": 115900
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 0.0005,
+      "loss": 1.8967,
+      "step": 116000
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 0.0005,
+      "loss": 1.8914,
+      "step": 116100
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 0.0005,
+      "loss": 1.8867,
+      "step": 116200
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 0.0005,
+      "loss": 1.9153,
+      "step": 116300
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.0005,
+      "loss": 1.9004,
+      "step": 116400
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8644,
+      "step": 116500
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8856,
+      "step": 116600
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 0.0005,
+      "loss": 1.8949,
+      "step": 116700
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.0005,
+      "loss": 1.8782,
+      "step": 116800
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8724,
+      "step": 116900
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 0.0005,
+      "loss": 1.9133,
+      "step": 117000
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 0.0005,
+      "loss": 1.8869,
+      "step": 117100
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.0005,
+      "loss": 1.9042,
+      "step": 117200
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 0.0005,
+      "loss": 1.8915,
+      "step": 117300
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 0.0005,
+      "loss": 1.9083,
+      "step": 117400
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 0.0005,
+      "loss": 1.9009,
+      "step": 117500
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.0005,
+      "loss": 1.8974,
+      "step": 117600
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.0005,
+      "loss": 1.9142,
+      "step": 117700
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 0.0005,
+      "loss": 1.9136,
+      "step": 117800
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.0005,
+      "loss": 1.9299,
+      "step": 117900
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.0005,
+      "loss": 1.8972,
+      "step": 118000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 0.0005,
+      "loss": 1.9068,
+      "step": 118100
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 0.0005,
+      "loss": 1.8926,
+      "step": 118200
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.0005,
+      "loss": 1.9263,
+      "step": 118300
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.0005,
+      "loss": 1.9052,
+      "step": 118400
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 0.0005,
+      "loss": 1.9119,
+      "step": 118500
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 0.0005,
+      "loss": 1.8979,
+      "step": 118600
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 0.0005,
+      "loss": 1.8753,
+      "step": 118700
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.0005,
+      "loss": 1.8867,
+      "step": 118800
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.0005,
+      "loss": 1.8759,
+      "step": 118900
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 0.0005,
+      "loss": 1.932,
+      "step": 119000
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.0005,
+      "loss": 1.9213,
+      "step": 119100
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 0.0005,
+      "loss": 1.9241,
+      "step": 119200
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 0.0005,
+      "loss": 1.9039,
+      "step": 119300
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 0.0005,
+      "loss": 1.9411,
+      "step": 119400
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 0.0005,
+      "loss": 1.9071,
+      "step": 119500
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 0.0005,
+      "loss": 1.9214,
+      "step": 119600
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.0005,
+      "loss": 1.8785,
+      "step": 119700
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.0005,
+      "loss": 1.9119,
+      "step": 119800
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 0.0005,
+      "loss": 1.9278,
+      "step": 119900
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 0.0005,
+      "loss": 1.9196,
+      "step": 120000
+    },
+    {
+      "epoch": 9.41,
+      "eval_gen_len": 18.812042023483713,
+      "eval_loss": 2.018998384475708,
+      "eval_rouge1": 34.5454,
+      "eval_rouge2": 13.3829,
+      "eval_rougeL": 28.2239,
+      "eval_rougeLsum": 28.2114,
+      "eval_runtime": 359.993,
+      "eval_samples_per_second": 31.465,
+      "eval_steps_per_second": 1.967,
+      "step": 120000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 0.0005,
+      "loss": 1.9188,
+      "step": 120100
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.0005,
+      "loss": 1.9056,
+      "step": 120200
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.0005,
+      "loss": 1.9468,
+      "step": 120300
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 0.0005,
+      "loss": 1.9084,
+      "step": 120400
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 0.0005,
+      "loss": 1.9327,
+      "step": 120500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 0.0005,
+      "loss": 1.9426,
+      "step": 120600
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.0005,
+      "loss": 1.9135,
+      "step": 120700
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.0005,
+      "loss": 1.8908,
+      "step": 120800
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 0.0005,
+      "loss": 1.9273,
+      "step": 120900
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 0.0005,
+      "loss": 1.9364,
+      "step": 121000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.0005,
+      "loss": 1.906,
+      "step": 121100
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.0005,
+      "loss": 1.9405,
+      "step": 121200
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 0.0005,
+      "loss": 1.9018,
+      "step": 121300
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 0.0005,
+      "loss": 1.9183,
+      "step": 121400
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9258,
+      "step": 121500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 0.0005,
+      "loss": 1.8952,
+      "step": 121600
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9412,
+      "step": 121700
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 0.0005,
+      "loss": 1.9284,
+      "step": 121800
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9197,
+      "step": 121900
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.0005,
+      "loss": 1.9171,
+      "step": 122000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.0005,
+      "loss": 1.9278,
+      "step": 122100
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 0.0005,
+      "loss": 1.9395,
+      "step": 122200
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 0.0005,
+      "loss": 1.936,
+      "step": 122300
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 0.0005,
+      "loss": 1.9,
+      "step": 122400
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9328,
+      "step": 122500
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9329,
+      "step": 122600
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 0.0005,
+      "loss": 1.9407,
+      "step": 122700
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9219,
+      "step": 122800
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 0.0005,
+      "loss": 1.9634,
+      "step": 122900
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.0005,
+      "loss": 1.9186,
+      "step": 123000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.0005,
+      "loss": 1.9361,
+      "step": 123100
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 0.0005,
+      "loss": 1.9279,
+      "step": 123200
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 0.0005,
+      "loss": 1.9251,
+      "step": 123300
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.0005,
+      "loss": 1.9143,
+      "step": 123400
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.0005,
+      "loss": 1.9565,
+      "step": 123500
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 0.0005,
+      "loss": 1.9419,
+      "step": 123600
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9147,
+      "step": 123700
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.0005,
+      "loss": 1.8996,
+      "step": 123800
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 0.0005,
+      "loss": 1.9299,
+      "step": 123900
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 0.0005,
+      "loss": 1.9303,
+      "step": 124000
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 0.0005,
+      "loss": 1.9447,
+      "step": 124100
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.0005,
+      "loss": 1.9251,
+      "step": 124200
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 0.0005,
+      "loss": 1.9279,
+      "step": 124300
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 0.0005,
+      "loss": 1.8948,
+      "step": 124400
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 0.0005,
+      "loss": 1.9331,
+      "step": 124500
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 0.0005,
+      "loss": 1.9378,
+      "step": 124600
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 0.0005,
+      "loss": 1.9366,
+      "step": 124700
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 0.0005,
+      "loss": 1.9875,
+      "step": 124800
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 0.0005,
+      "loss": 1.9354,
+      "step": 124900
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 0.0005,
+      "loss": 1.9318,
+      "step": 125000
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 0.0005,
+      "loss": 1.9058,
+      "step": 125100
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 0.0005,
+      "loss": 1.9133,
+      "step": 125200
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.0005,
+      "loss": 1.9353,
+      "step": 125300
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.0005,
+      "loss": 1.9616,
+      "step": 125400
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 0.0005,
+      "loss": 1.9395,
+      "step": 125500
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 0.0005,
+      "loss": 1.9405,
+      "step": 125600
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8969,
+      "step": 125700
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 0.0005,
+      "loss": 1.9206,
+      "step": 125800
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8971,
+      "step": 125900
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 0.0005,
+      "loss": 1.912,
+      "step": 126000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 0.0005,
+      "loss": 1.9517,
+      "step": 126100
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.0005,
+      "loss": 1.9252,
+      "step": 126200
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.0005,
+      "loss": 1.9225,
+      "step": 126300
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 0.0005,
+      "loss": 1.943,
+      "step": 126400
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.0005,
+      "loss": 1.9287,
+      "step": 126500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 0.0005,
+      "loss": 1.9797,
+      "step": 126600
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 0.0005,
+      "loss": 1.9319,
+      "step": 126700
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 0.0005,
+      "loss": 1.9392,
+      "step": 126800
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 0.0005,
+      "loss": 1.9354,
+      "step": 126900
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 0.0005,
+      "loss": 1.9127,
+      "step": 127000
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 0.0005,
+      "loss": 1.9682,
+      "step": 127100
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 0.0005,
+      "loss": 1.9153,
+      "step": 127200
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 0.0005,
+      "loss": 1.9246,
+      "step": 127300
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.0005,
+      "loss": 1.9302,
+      "step": 127400
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0005,
+      "loss": 1.9188,
+      "step": 127500
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 0.0005,
+      "loss": 1.9,
+      "step": 127600
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 0.0005,
+      "loss": 1.847,
+      "step": 127700
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 0.0005,
+      "loss": 1.8283,
+      "step": 127800
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 0.0005,
+      "loss": 1.8143,
+      "step": 127900
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 0.0005,
+      "loss": 1.8273,
+      "step": 128000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.0005,
+      "loss": 1.8709,
+      "step": 128100
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.0005,
+      "loss": 1.8487,
+      "step": 128200
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 0.0005,
+      "loss": 1.8281,
+      "step": 128300
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 0.0005,
+      "loss": 1.8641,
+      "step": 128400
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.0005,
+      "loss": 1.862,
+      "step": 128500
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.0005,
+      "loss": 1.8973,
+      "step": 128600
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 0.0005,
+      "loss": 1.8294,
+      "step": 128700
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 0.0005,
+      "loss": 1.8656,
+      "step": 128800
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 0.0005,
+      "loss": 1.8619,
+      "step": 128900
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 0.0005,
+      "loss": 1.8493,
+      "step": 129000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 0.0005,
+      "loss": 1.8471,
+      "step": 129100
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.0005,
+      "loss": 1.8663,
+      "step": 129200
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8781,
+      "step": 129300
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 0.0005,
+      "loss": 1.8959,
+      "step": 129400
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.0005,
+      "loss": 1.8703,
+      "step": 129500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.0005,
+      "loss": 1.8461,
+      "step": 129600
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8471,
+      "step": 129700
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 0.0005,
+      "loss": 1.8793,
+      "step": 129800
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.0005,
+      "loss": 1.9014,
+      "step": 129900
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.0005,
+      "loss": 1.8524,
+      "step": 130000
+    },
+    {
+      "epoch": 10.19,
+      "eval_gen_len": 18.745651981989937,
+      "eval_loss": 2.035221815109253,
+      "eval_rouge1": 34.7913,
+      "eval_rouge2": 13.5607,
+      "eval_rougeL": 28.4207,
+      "eval_rougeLsum": 28.4117,
+      "eval_runtime": 358.3534,
+      "eval_samples_per_second": 31.608,
+      "eval_steps_per_second": 1.976,
+      "step": 130000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 0.0005,
+      "loss": 1.8683,
+      "step": 130100
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 0.0005,
+      "loss": 1.8492,
+      "step": 130200
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 0.0005,
+      "loss": 1.8901,
+      "step": 130300
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.0005,
+      "loss": 1.8592,
+      "step": 130400
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.0005,
+      "loss": 1.8946,
+      "step": 130500
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 0.0005,
+      "loss": 1.8495,
+      "step": 130600
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 0.0005,
+      "loss": 1.8598,
+      "step": 130700
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 0.0005,
+      "loss": 1.8755,
+      "step": 130800
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 0.0005,
+      "loss": 1.8473,
+      "step": 130900
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 0.0005,
+      "loss": 1.8724,
+      "step": 131000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 0.0005,
+      "loss": 1.8806,
+      "step": 131100
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 0.0005,
+      "loss": 1.8776,
+      "step": 131200
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 0.0005,
+      "loss": 1.8739,
+      "step": 131300
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 0.0005,
+      "loss": 1.8831,
+      "step": 131400
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 0.0005,
+      "loss": 1.8764,
+      "step": 131500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 0.0005,
+      "loss": 1.8965,
+      "step": 131600
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 0.0005,
+      "loss": 1.8842,
+      "step": 131700
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.0005,
+      "loss": 1.8578,
+      "step": 131800
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.0005,
+      "loss": 1.8977,
+      "step": 131900
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 0.0005,
+      "loss": 1.8717,
+      "step": 132000
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 0.0005,
+      "loss": 1.8595,
+      "step": 132100
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.0005,
+      "loss": 1.8855,
+      "step": 132200
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.0005,
+      "loss": 1.9059,
+      "step": 132300
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 0.0005,
+      "loss": 1.8721,
+      "step": 132400
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 0.0005,
+      "loss": 1.9069,
+      "step": 132500
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 0.0005,
+      "loss": 1.8871,
+      "step": 132600
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 0.0005,
+      "loss": 1.888,
+      "step": 132700
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 0.0005,
+      "loss": 1.8623,
+      "step": 132800
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 0.0005,
+      "loss": 1.8942,
+      "step": 132900
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 0.0005,
+      "loss": 1.9247,
+      "step": 133000
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 0.0005,
+      "loss": 1.871,
+      "step": 133100
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.0005,
+      "loss": 1.8971,
+      "step": 133200
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.0005,
+      "loss": 1.8707,
+      "step": 133300
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 0.0005,
+      "loss": 1.8797,
+      "step": 133400
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 0.0005,
+      "loss": 1.8896,
+      "step": 133500
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.0005,
+      "loss": 1.878,
+      "step": 133600
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.0005,
+      "loss": 1.8719,
+      "step": 133700
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 0.0005,
+      "loss": 1.8828,
+      "step": 133800
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 0.0005,
+      "loss": 1.91,
+      "step": 133900
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 0.0005,
+      "loss": 1.8776,
+      "step": 134000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.0005,
+      "loss": 1.8828,
+      "step": 134100
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.0005,
+      "loss": 1.8739,
+      "step": 134200
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 0.0005,
+      "loss": 1.9038,
+      "step": 134300
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 0.0005,
+      "loss": 1.9058,
+      "step": 134400
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 0.0005,
+      "loss": 1.8967,
+      "step": 134500
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 0.0005,
+      "loss": 1.9057,
+      "step": 134600
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 0.0005,
+      "loss": 1.8868,
+      "step": 134700
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 0.0005,
+      "loss": 1.8957,
+      "step": 134800
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 0.0005,
+      "loss": 1.8816,
+      "step": 134900
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.0005,
+      "loss": 1.8901,
+      "step": 135000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.0005,
+      "loss": 1.8964,
+      "step": 135100
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 0.0005,
+      "loss": 1.8975,
+      "step": 135200
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 0.0005,
+      "loss": 1.9138,
+      "step": 135300
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 0.0005,
+      "loss": 1.9164,
+      "step": 135400
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 0.0005,
+      "loss": 1.8969,
+      "step": 135500
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9051,
+      "step": 135600
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 0.0005,
+      "loss": 1.8947,
+      "step": 135700
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.0005,
+      "loss": 1.9156,
+      "step": 135800
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.0005,
+      "loss": 1.906,
+      "step": 135900
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.0005,
+      "loss": 1.8948,
+      "step": 136000
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 0.0005,
+      "loss": 1.8997,
+      "step": 136100
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 0.0005,
+      "loss": 1.9112,
+      "step": 136200
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 0.0005,
+      "loss": 1.9091,
+      "step": 136300
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9083,
+      "step": 136400
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 0.0005,
+      "loss": 1.9142,
+      "step": 136500
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 0.0005,
+      "loss": 1.9074,
+      "step": 136600
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 0.0005,
+      "loss": 1.8614,
+      "step": 136700
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 0.0005,
+      "loss": 1.8908,
+      "step": 136800
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 0.0005,
+      "loss": 1.9128,
+      "step": 136900
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 0.0005,
+      "loss": 1.9006,
+      "step": 137000
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 0.0005,
+      "loss": 1.8909,
+      "step": 137100
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 0.0005,
+      "loss": 1.8812,
+      "step": 137200
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.0005,
+      "loss": 1.9118,
+      "step": 137300
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8966,
+      "step": 137400
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 0.0005,
+      "loss": 1.9206,
+      "step": 137500
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 0.0005,
+      "loss": 1.9065,
+      "step": 137600
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 0.0005,
+      "loss": 1.8795,
+      "step": 137700
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 0.0005,
+      "loss": 1.9157,
+      "step": 137800
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 0.0005,
+      "loss": 1.9352,
+      "step": 137900
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8882,
+      "step": 138000
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8767,
+      "step": 138100
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8922,
+      "step": 138200
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.0005,
+      "loss": 1.902,
+      "step": 138300
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.0005,
+      "loss": 1.9059,
+      "step": 138400
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 0.0005,
+      "loss": 1.9211,
+      "step": 138500
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8916,
+      "step": 138600
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.0005,
+      "loss": 1.936,
+      "step": 138700
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.0005,
+      "loss": 1.9028,
+      "step": 138800
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 0.0005,
+      "loss": 1.9066,
+      "step": 138900
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8575,
+      "step": 139000
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 0.0005,
+      "loss": 1.9116,
+      "step": 139100
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.0005,
+      "loss": 1.9183,
+      "step": 139200
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8794,
+      "step": 139300
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 0.0005,
+      "loss": 1.9203,
+      "step": 139400
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 0.0005,
+      "loss": 1.9218,
+      "step": 139500
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 0.0005,
+      "loss": 1.9005,
+      "step": 139600
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 0.0005,
+      "loss": 1.9146,
+      "step": 139700
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 0.0005,
+      "loss": 1.9101,
+      "step": 139800
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 0.0005,
+      "loss": 1.9226,
+      "step": 139900
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8968,
+      "step": 140000
+    },
+    {
+      "epoch": 10.98,
+      "eval_gen_len": 18.780524410700096,
+      "eval_loss": 2.0128087997436523,
+      "eval_rouge1": 34.8079,
+      "eval_rouge2": 13.6337,
+      "eval_rougeL": 28.4248,
+      "eval_rougeLsum": 28.4272,
+      "eval_runtime": 363.0976,
+      "eval_samples_per_second": 31.195,
+      "eval_steps_per_second": 1.95,
+      "step": 140000
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 0.0005,
+      "loss": 1.908,
+      "step": 140100
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 0.0005,
+      "loss": 1.9198,
+      "step": 140200
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.0005,
+      "loss": 1.8756,
+      "step": 140300
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 0.0005,
+      "loss": 1.842,
+      "step": 140400
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 0.0005,
+      "loss": 1.8369,
+      "step": 140500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7975,
+      "step": 140600
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.0005,
+      "loss": 1.8204,
+      "step": 140700
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.0005,
+      "loss": 1.8019,
+      "step": 140800
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 0.0005,
+      "loss": 1.8289,
+      "step": 140900
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.0005,
+      "loss": 1.8119,
+      "step": 141000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.0005,
+      "loss": 1.8118,
+      "step": 141100
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 0.0005,
+      "loss": 1.8111,
+      "step": 141200
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 0.0005,
+      "loss": 1.8147,
+      "step": 141300
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 0.0005,
+      "loss": 1.8222,
+      "step": 141400
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.0005,
+      "loss": 1.8397,
+      "step": 141500
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.0005,
+      "loss": 1.8353,
+      "step": 141600
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 0.0005,
+      "loss": 1.854,
+      "step": 141700
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 0.0005,
+      "loss": 1.8107,
+      "step": 141800
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 0.0005,
+      "loss": 1.8227,
+      "step": 141900
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8332,
+      "step": 142000
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8161,
+      "step": 142100
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 0.0005,
+      "loss": 1.8332,
+      "step": 142200
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 0.0005,
+      "loss": 1.8283,
+      "step": 142300
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8493,
+      "step": 142400
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8284,
+      "step": 142500
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.0005,
+      "loss": 1.8085,
+      "step": 142600
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 0.0005,
+      "loss": 1.8523,
+      "step": 142700
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 0.0005,
+      "loss": 1.8412,
+      "step": 142800
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.0005,
+      "loss": 1.8628,
+      "step": 142900
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.0005,
+      "loss": 1.8296,
+      "step": 143000
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 0.0005,
+      "loss": 1.8591,
+      "step": 143100
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 0.0005,
+      "loss": 1.8804,
+      "step": 143200
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 0.0005,
+      "loss": 1.8345,
+      "step": 143300
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.0005,
+      "loss": 1.8262,
+      "step": 143400
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.0005,
+      "loss": 1.8762,
+      "step": 143500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 0.0005,
+      "loss": 1.8287,
+      "step": 143600
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.0005,
+      "loss": 1.8452,
+      "step": 143700
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.0005,
+      "loss": 1.8575,
+      "step": 143800
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.0005,
+      "loss": 1.8701,
+      "step": 143900
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 0.0005,
+      "loss": 1.8291,
+      "step": 144000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.0005,
+      "loss": 1.8768,
+      "step": 144100
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 0.0005,
+      "loss": 1.8742,
+      "step": 144200
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 0.0005,
+      "loss": 1.8481,
+      "step": 144300
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 0.0005,
+      "loss": 1.8406,
+      "step": 144400
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 0.0005,
+      "loss": 1.8386,
+      "step": 144500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.0005,
+      "loss": 1.8815,
+      "step": 144600
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 0.0005,
+      "loss": 1.8318,
+      "step": 144700
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 0.0005,
+      "loss": 1.8546,
+      "step": 144800
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 0.0005,
+      "loss": 1.889,
+      "step": 144900
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.0005,
+      "loss": 1.8426,
+      "step": 145000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 0.0005,
+      "loss": 1.8647,
+      "step": 145100
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.0005,
+      "loss": 1.8627,
+      "step": 145200
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.0005,
+      "loss": 1.8513,
+      "step": 145300
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 0.0005,
+      "loss": 1.8939,
+      "step": 145400
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 0.0005,
+      "loss": 1.8071,
+      "step": 145500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 0.0005,
+      "loss": 1.871,
+      "step": 145600
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 0.0005,
+      "loss": 1.8685,
+      "step": 145700
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 0.0005,
+      "loss": 1.8702,
+      "step": 145800
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 0.0005,
+      "loss": 1.8751,
+      "step": 145900
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 0.0005,
+      "loss": 1.8614,
+      "step": 146000
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.0005,
+      "loss": 1.8781,
+      "step": 146100
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.0005,
+      "loss": 1.8487,
+      "step": 146200
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 0.0005,
+      "loss": 1.829,
+      "step": 146300
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 0.0005,
+      "loss": 1.8417,
+      "step": 146400
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 0.0005,
+      "loss": 1.8634,
+      "step": 146500
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 0.0005,
+      "loss": 1.8713,
+      "step": 146600
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 0.0005,
+      "loss": 1.8597,
+      "step": 146700
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 0.0005,
+      "loss": 1.8772,
+      "step": 146800
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 0.0005,
+      "loss": 1.8325,
+      "step": 146900
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 0.0005,
+      "loss": 1.8753,
+      "step": 147000
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 0.0005,
+      "loss": 1.8868,
+      "step": 147100
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 0.0005,
+      "loss": 1.8634,
+      "step": 147200
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.0005,
+      "loss": 1.8759,
+      "step": 147300
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 0.0005,
+      "loss": 1.8571,
+      "step": 147400
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 0.0005,
+      "loss": 1.8848,
+      "step": 147500
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 0.0005,
+      "loss": 1.8947,
+      "step": 147600
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.0005,
+      "loss": 1.8594,
+      "step": 147700
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 0.0005,
+      "loss": 1.8686,
+      "step": 147800
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 0.0005,
+      "loss": 1.8718,
+      "step": 147900
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 0.0005,
+      "loss": 1.8632,
+      "step": 148000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 0.0005,
+      "loss": 1.8723,
+      "step": 148100
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 0.0005,
+      "loss": 1.845,
+      "step": 148200
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 0.0005,
+      "loss": 1.9104,
+      "step": 148300
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 0.0005,
+      "loss": 1.8627,
+      "step": 148400
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.0005,
+      "loss": 1.86,
+      "step": 148500
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.0005,
+      "loss": 1.8749,
+      "step": 148600
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 0.0005,
+      "loss": 1.8809,
+      "step": 148700
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 0.0005,
+      "loss": 1.8597,
+      "step": 148800
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 0.0005,
+      "loss": 1.8427,
+      "step": 148900
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 0.0005,
+      "loss": 1.8585,
+      "step": 149000
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 0.0005,
+      "loss": 1.8673,
+      "step": 149100
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 0.0005,
+      "loss": 1.871,
+      "step": 149200
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 0.0005,
+      "loss": 1.8847,
+      "step": 149300
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 0.0005,
+      "loss": 1.8439,
+      "step": 149400
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 0.0005,
+      "loss": 1.8545,
+      "step": 149500
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 0.0005,
+      "loss": 1.873,
+      "step": 149600
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.0005,
+      "loss": 1.8647,
+      "step": 149700
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 0.0005,
+      "loss": 1.8725,
+      "step": 149800
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.0005,
+      "loss": 1.8905,
+      "step": 149900
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.0005,
+      "loss": 1.8758,
+      "step": 150000
+    },
+    {
+      "epoch": 11.76,
+      "eval_gen_len": 18.80771607663106,
+      "eval_loss": 2.0203864574432373,
+      "eval_rouge1": 34.6581,
+      "eval_rouge2": 13.5851,
+      "eval_rougeL": 28.3861,
+      "eval_rougeLsum": 28.3839,
+      "eval_runtime": 370.8862,
+      "eval_samples_per_second": 30.54,
+      "eval_steps_per_second": 1.909,
+      "step": 150000
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8741,
+      "step": 150100
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 0.0005,
+      "loss": 1.8762,
+      "step": 150200
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8504,
+      "step": 150300
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8605,
+      "step": 150400
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 0.0005,
+      "loss": 1.8651,
+      "step": 150500
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8913,
+      "step": 150600
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 0.0005,
+      "loss": 1.9204,
+      "step": 150700
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8737,
+      "step": 150800
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8794,
+      "step": 150900
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 0.0005,
+      "loss": 1.9059,
+      "step": 151000
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 0.0005,
+      "loss": 1.9089,
+      "step": 151100
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8743,
+      "step": 151200
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8677,
+      "step": 151300
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8821,
+      "step": 151400
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8667,
+      "step": 151500
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 0.0005,
+      "loss": 1.8949,
+      "step": 151600
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8714,
+      "step": 151700
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8637,
+      "step": 151800
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 0.0005,
+      "loss": 1.8878,
+      "step": 151900
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8625,
+      "step": 152000
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8858,
+      "step": 152100
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8811,
+      "step": 152200
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8735,
+      "step": 152300
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8728,
+      "step": 152400
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8751,
+      "step": 152500
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8996,
+      "step": 152600
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.0005,
+      "loss": 1.9027,
+      "step": 152700
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8875,
+      "step": 152800
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8858,
+      "step": 152900
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 0.0005,
+      "loss": 1.8822,
+      "step": 153000
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.0005,
+      "loss": 1.8282,
+      "step": 153100
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.0005,
+      "loss": 1.7881,
+      "step": 153200
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 0.0005,
+      "loss": 1.7834,
+      "step": 153300
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7981,
+      "step": 153400
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.0005,
+      "loss": 1.796,
+      "step": 153500
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 0.0005,
+      "loss": 1.7744,
+      "step": 153600
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 0.0005,
+      "loss": 1.7785,
+      "step": 153700
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 0.0005,
+      "loss": 1.7937,
+      "step": 153800
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 0.0005,
+      "loss": 1.7746,
+      "step": 153900
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 0.0005,
+      "loss": 1.8114,
+      "step": 154000
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 0.0005,
+      "loss": 1.814,
+      "step": 154100
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 0.0005,
+      "loss": 1.7709,
+      "step": 154200
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 0.0005,
+      "loss": 1.7994,
+      "step": 154300
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 0.0005,
+      "loss": 1.8088,
+      "step": 154400
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.0005,
+      "loss": 1.8016,
+      "step": 154500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.0005,
+      "loss": 1.8205,
+      "step": 154600
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 0.0005,
+      "loss": 1.826,
+      "step": 154700
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8543,
+      "step": 154800
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.0005,
+      "loss": 1.8301,
+      "step": 154900
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.0005,
+      "loss": 1.797,
+      "step": 155000
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 0.0005,
+      "loss": 1.8202,
+      "step": 155100
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8305,
+      "step": 155200
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 0.0005,
+      "loss": 1.8421,
+      "step": 155300
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.0005,
+      "loss": 1.8219,
+      "step": 155400
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.0005,
+      "loss": 1.848,
+      "step": 155500
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 0.0005,
+      "loss": 1.8103,
+      "step": 155600
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 0.0005,
+      "loss": 1.8199,
+      "step": 155700
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.0005,
+      "loss": 1.8423,
+      "step": 155800
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 0.0005,
+      "loss": 1.8153,
+      "step": 155900
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 0.0005,
+      "loss": 1.8072,
+      "step": 156000
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 0.0005,
+      "loss": 1.8351,
+      "step": 156100
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 0.0005,
+      "loss": 1.8175,
+      "step": 156200
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 0.0005,
+      "loss": 1.8411,
+      "step": 156300
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 0.0005,
+      "loss": 1.8558,
+      "step": 156400
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 0.0005,
+      "loss": 1.83,
+      "step": 156500
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 0.0005,
+      "loss": 1.8168,
+      "step": 156600
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 0.0005,
+      "loss": 1.8293,
+      "step": 156700
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 0.0005,
+      "loss": 1.8293,
+      "step": 156800
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 0.0005,
+      "loss": 1.8264,
+      "step": 156900
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.0005,
+      "loss": 1.832,
+      "step": 157000
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 0.0005,
+      "loss": 1.8167,
+      "step": 157100
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 0.0005,
+      "loss": 1.8483,
+      "step": 157200
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 0.0005,
+      "loss": 1.836,
+      "step": 157300
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 0.0005,
+      "loss": 1.8409,
+      "step": 157400
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 0.0005,
+      "loss": 1.8207,
+      "step": 157500
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 0.0005,
+      "loss": 1.8434,
+      "step": 157600
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 0.0005,
+      "loss": 1.8394,
+      "step": 157700
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 0.0005,
+      "loss": 1.8232,
+      "step": 157800
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 0.0005,
+      "loss": 1.8328,
+      "step": 157900
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 0.0005,
+      "loss": 1.8358,
+      "step": 158000
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 0.0005,
+      "loss": 1.8278,
+      "step": 158100
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 0.0005,
+      "loss": 1.7989,
+      "step": 158200
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 0.0005,
+      "loss": 1.8207,
+      "step": 158300
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 0.0005,
+      "loss": 1.8147,
+      "step": 158400
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 0.0005,
+      "loss": 1.8319,
+      "step": 158500
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 0.0005,
+      "loss": 1.823,
+      "step": 158600
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.0005,
+      "loss": 1.8673,
+      "step": 158700
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.0005,
+      "loss": 1.8437,
+      "step": 158800
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 0.0005,
+      "loss": 1.8654,
+      "step": 158900
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 0.0005,
+      "loss": 1.8534,
+      "step": 159000
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 0.0005,
+      "loss": 1.8327,
+      "step": 159100
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 0.0005,
+      "loss": 1.8365,
+      "step": 159200
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 0.0005,
+      "loss": 1.8231,
+      "step": 159300
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 0.0005,
+      "loss": 1.8104,
+      "step": 159400
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 0.0005,
+      "loss": 1.8582,
+      "step": 159500
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.0005,
+      "loss": 1.8159,
+      "step": 159600
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.0005,
+      "loss": 1.8491,
+      "step": 159700
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 0.0005,
+      "loss": 1.8316,
+      "step": 159800
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 0.0005,
+      "loss": 1.8465,
+      "step": 159900
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 0.0005,
+      "loss": 1.831,
+      "step": 160000
+    },
+    {
+      "epoch": 12.55,
+      "eval_gen_len": 18.794914805332393,
+      "eval_loss": 2.022197961807251,
+      "eval_rouge1": 35.0141,
+      "eval_rouge2": 13.742,
+      "eval_rougeL": 28.6326,
+      "eval_rougeLsum": 28.6266,
+      "eval_runtime": 366.073,
+      "eval_samples_per_second": 30.942,
+      "eval_steps_per_second": 1.934,
+      "step": 160000
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 0.0005,
+      "loss": 1.8515,
+      "step": 160100
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 0.0005,
+      "loss": 1.8349,
+      "step": 160200
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.0005,
+      "loss": 1.8328,
+      "step": 160300
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 0.0005,
+      "loss": 1.8438,
+      "step": 160400
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 0.0005,
+      "loss": 1.8385,
+      "step": 160500
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 0.0005,
+      "loss": 1.8492,
+      "step": 160600
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 0.0005,
+      "loss": 1.8458,
+      "step": 160700
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 0.0005,
+      "loss": 1.8523,
+      "step": 160800
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 0.0005,
+      "loss": 1.8413,
+      "step": 160900
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.0005,
+      "loss": 1.8455,
+      "step": 161000
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.0005,
+      "loss": 1.8402,
+      "step": 161100
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 0.0005,
+      "loss": 1.836,
+      "step": 161200
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 0.0005,
+      "loss": 1.826,
+      "step": 161300
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.0005,
+      "loss": 1.8346,
+      "step": 161400
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.0005,
+      "loss": 1.8614,
+      "step": 161500
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 0.0005,
+      "loss": 1.8132,
+      "step": 161600
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 0.0005,
+      "loss": 1.8829,
+      "step": 161700
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 0.0005,
+      "loss": 1.8424,
+      "step": 161800
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.0005,
+      "loss": 1.8492,
+      "step": 161900
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.0005,
+      "loss": 1.8379,
+      "step": 162000
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 0.0005,
+      "loss": 1.8563,
+      "step": 162100
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 0.0005,
+      "loss": 1.8041,
+      "step": 162200
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 0.0005,
+      "loss": 1.8593,
+      "step": 162300
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 0.0005,
+      "loss": 1.8275,
+      "step": 162400
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 0.0005,
+      "loss": 1.8515,
+      "step": 162500
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 0.0005,
+      "loss": 1.8346,
+      "step": 162600
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 0.0005,
+      "loss": 1.856,
+      "step": 162700
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8382,
+      "step": 162800
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8589,
+      "step": 162900
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 0.0005,
+      "loss": 1.8555,
+      "step": 163000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8464,
+      "step": 163100
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 0.0005,
+      "loss": 1.8498,
+      "step": 163200
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8542,
+      "step": 163300
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8414,
+      "step": 163400
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8498,
+      "step": 163500
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8572,
+      "step": 163600
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8524,
+      "step": 163700
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.0005,
+      "loss": 1.8547,
+      "step": 163800
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.0005,
+      "loss": 1.8326,
+      "step": 163900
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8452,
+      "step": 164000
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8259,
+      "step": 164100
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8649,
+      "step": 164200
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8535,
+      "step": 164300
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 0.0005,
+      "loss": 1.8384,
+      "step": 164400
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8549,
+      "step": 164500
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.0005,
+      "loss": 1.888,
+      "step": 164600
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8618,
+      "step": 164700
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8834,
+      "step": 164800
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8608,
+      "step": 164900
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8636,
+      "step": 165000
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8552,
+      "step": 165100
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8288,
+      "step": 165200
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8426,
+      "step": 165300
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8586,
+      "step": 165400
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8693,
+      "step": 165500
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8556,
+      "step": 165600
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8689,
+      "step": 165700
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 0.0005,
+      "loss": 1.8191,
+      "step": 165800
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 0.0005,
+      "loss": 1.7918,
+      "step": 165900
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 0.0005,
+      "loss": 1.8066,
+      "step": 166000
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7596,
+      "step": 166100
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7748,
+      "step": 166200
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 0.0005,
+      "loss": 1.7756,
+      "step": 166300
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 0.0005,
+      "loss": 1.7872,
+      "step": 166400
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.0005,
+      "loss": 1.78,
+      "step": 166500
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.0005,
+      "loss": 1.7967,
+      "step": 166600
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 0.0005,
+      "loss": 1.7719,
+      "step": 166700
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 0.0005,
+      "loss": 1.7773,
+      "step": 166800
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 0.0005,
+      "loss": 1.7931,
+      "step": 166900
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 0.0005,
+      "loss": 1.7814,
+      "step": 167000
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 0.0005,
+      "loss": 1.7707,
+      "step": 167100
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 0.0005,
+      "loss": 1.7711,
+      "step": 167200
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7758,
+      "step": 167300
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 0.0005,
+      "loss": 1.772,
+      "step": 167400
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 0.0005,
+      "loss": 1.7794,
+      "step": 167500
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 0.0005,
+      "loss": 1.8027,
+      "step": 167600
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 0.0005,
+      "loss": 1.7688,
+      "step": 167700
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 0.0005,
+      "loss": 1.7958,
+      "step": 167800
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 0.0005,
+      "loss": 1.8164,
+      "step": 167900
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 0.0005,
+      "loss": 1.7801,
+      "step": 168000
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 0.0005,
+      "loss": 1.7964,
+      "step": 168100
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 0.0005,
+      "loss": 1.784,
+      "step": 168200
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 0.0005,
+      "loss": 1.8051,
+      "step": 168300
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7861,
+      "step": 168400
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7915,
+      "step": 168500
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 0.0005,
+      "loss": 1.8157,
+      "step": 168600
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 0.0005,
+      "loss": 1.7867,
+      "step": 168700
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7561,
+      "step": 168800
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7943,
+      "step": 168900
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7808,
+      "step": 169000
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 0.0005,
+      "loss": 1.7856,
+      "step": 169100
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 0.0005,
+      "loss": 1.8041,
+      "step": 169200
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 0.0005,
+      "loss": 1.8003,
+      "step": 169300
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 0.0005,
+      "loss": 1.8013,
+      "step": 169400
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 0.0005,
+      "loss": 1.7939,
+      "step": 169500
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.0005,
+      "loss": 1.8059,
+      "step": 169600
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 0.0005,
+      "loss": 1.8036,
+      "step": 169700
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.0005,
+      "loss": 1.7763,
+      "step": 169800
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.0005,
+      "loss": 1.7864,
+      "step": 169900
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 0.0005,
+      "loss": 1.8027,
+      "step": 170000
+    },
+    {
+      "epoch": 13.33,
+      "eval_gen_len": 18.75721726847356,
+      "eval_loss": 2.0271575450897217,
+      "eval_rouge1": 35.2864,
+      "eval_rouge2": 13.9499,
+      "eval_rougeL": 28.9284,
+      "eval_rougeLsum": 28.9215,
+      "eval_runtime": 360.6463,
+      "eval_samples_per_second": 31.408,
+      "eval_steps_per_second": 1.963,
+      "step": 170000
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 0.0005,
+      "loss": 1.8209,
+      "step": 170100
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7839,
+      "step": 170200
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 0.0005,
+      "loss": 1.8168,
+      "step": 170300
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 0.0005,
+      "loss": 1.8084,
+      "step": 170400
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7727,
+      "step": 170500
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 0.0005,
+      "loss": 1.7981,
+      "step": 170600
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.0005,
+      "loss": 1.806,
+      "step": 170700
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7888,
+      "step": 170800
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 0.0005,
+      "loss": 1.8181,
+      "step": 170900
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 0.0005,
+      "loss": 1.8123,
+      "step": 171000
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 0.0005,
+      "loss": 1.8251,
+      "step": 171100
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.0005,
+      "loss": 1.8066,
+      "step": 171200
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7705,
+      "step": 171300
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7931,
+      "step": 171400
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 0.0005,
+      "loss": 1.8041,
+      "step": 171500
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.0005,
+      "loss": 1.8164,
+      "step": 171600
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.0005,
+      "loss": 1.8147,
+      "step": 171700
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 0.0005,
+      "loss": 1.8222,
+      "step": 171800
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.0005,
+      "loss": 1.8062,
+      "step": 171900
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 0.0005,
+      "loss": 1.8085,
+      "step": 172000
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 0.0005,
+      "loss": 1.8116,
+      "step": 172100
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 0.0005,
+      "loss": 1.8293,
+      "step": 172200
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 0.0005,
+      "loss": 1.8549,
+      "step": 172300
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 0.0005,
+      "loss": 1.8357,
+      "step": 172400
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 0.0005,
+      "loss": 1.8385,
+      "step": 172500
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.0005,
+      "loss": 1.8104,
+      "step": 172600
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.0005,
+      "loss": 1.8192,
+      "step": 172700
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7848,
+      "step": 172800
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 0.0005,
+      "loss": 1.826,
+      "step": 172900
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.0005,
+      "loss": 1.812,
+      "step": 173000
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.0005,
+      "loss": 1.8255,
+      "step": 173100
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 0.0005,
+      "loss": 1.8037,
+      "step": 173200
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 0.0005,
+      "loss": 1.8181,
+      "step": 173300
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 0.0005,
+      "loss": 1.8365,
+      "step": 173400
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.0005,
+      "loss": 1.8409,
+      "step": 173500
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.0005,
+      "loss": 1.8047,
+      "step": 173600
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 0.0005,
+      "loss": 1.8418,
+      "step": 173700
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 0.0005,
+      "loss": 1.8107,
+      "step": 173800
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.0005,
+      "loss": 1.8037,
+      "step": 173900
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.0005,
+      "loss": 1.8453,
+      "step": 174000
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 0.0005,
+      "loss": 1.8125,
+      "step": 174100
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 0.0005,
+      "loss": 1.8392,
+      "step": 174200
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 0.0005,
+      "loss": 1.7939,
+      "step": 174300
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 0.0005,
+      "loss": 1.8239,
+      "step": 174400
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 0.0005,
+      "loss": 1.8236,
+      "step": 174500
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 0.0005,
+      "loss": 1.8072,
+      "step": 174600
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 0.0005,
+      "loss": 1.8417,
+      "step": 174700
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 0.0005,
+      "loss": 1.8382,
+      "step": 174800
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7897,
+      "step": 174900
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 0.0005,
+      "loss": 1.8221,
+      "step": 175000
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 0.0005,
+      "loss": 1.8289,
+      "step": 175100
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 0.0005,
+      "loss": 1.8416,
+      "step": 175200
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 0.0005,
+      "loss": 1.8429,
+      "step": 175300
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 0.0005,
+      "loss": 1.8408,
+      "step": 175400
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 0.0005,
+      "loss": 1.836,
+      "step": 175500
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8222,
+      "step": 175600
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 0.0005,
+      "loss": 1.8146,
+      "step": 175700
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8134,
+      "step": 175800
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.0005,
+      "loss": 1.7945,
+      "step": 175900
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 0.0005,
+      "loss": 1.8222,
+      "step": 176000
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8414,
+      "step": 176100
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8085,
+      "step": 176200
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8338,
+      "step": 176300
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8468,
+      "step": 176400
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8403,
+      "step": 176500
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.0005,
+      "loss": 1.8421,
+      "step": 176600
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8111,
+      "step": 176700
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8125,
+      "step": 176800
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8113,
+      "step": 176900
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8023,
+      "step": 177000
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 0.0005,
+      "loss": 1.8249,
+      "step": 177100
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8316,
+      "step": 177200
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8266,
+      "step": 177300
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 0.0005,
+      "loss": 1.8324,
+      "step": 177400
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8145,
+      "step": 177500
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8484,
+      "step": 177600
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8215,
+      "step": 177700
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 0.0005,
+      "loss": 1.83,
+      "step": 177800
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8315,
+      "step": 177900
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 0.0005,
+      "loss": 1.882,
+      "step": 178000
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8308,
+      "step": 178100
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8354,
+      "step": 178200
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8254,
+      "step": 178300
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8696,
+      "step": 178400
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 0.0005,
+      "loss": 1.8469,
+      "step": 178500
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 0.0005,
+      "loss": 1.7778,
+      "step": 178600
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 0.0005,
+      "loss": 1.7202,
+      "step": 178700
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 0.0005,
+      "loss": 1.7704,
+      "step": 178800
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7825,
+      "step": 178900
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 0.0005,
+      "loss": 1.7526,
+      "step": 179000
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 0.0005,
+      "loss": 1.777,
+      "step": 179100
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 0.0005,
+      "loss": 1.7551,
+      "step": 179200
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 0.0005,
+      "loss": 1.7501,
+      "step": 179300
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 0.0005,
+      "loss": 1.7711,
+      "step": 179400
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 0.0005,
+      "loss": 1.756,
+      "step": 179500
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 0.0005,
+      "loss": 1.76,
+      "step": 179600
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 0.0005,
+      "loss": 1.7467,
+      "step": 179700
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 0.0005,
+      "loss": 1.7563,
+      "step": 179800
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 0.0005,
+      "loss": 1.7338,
+      "step": 179900
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7544,
+      "step": 180000
+    },
+    {
+      "epoch": 14.12,
+      "eval_gen_len": 18.749006797916483,
+      "eval_loss": 2.0176799297332764,
+      "eval_rouge1": 35.1221,
+      "eval_rouge2": 13.8252,
+      "eval_rougeL": 28.7503,
+      "eval_rougeLsum": 28.7366,
+      "eval_runtime": 359.2813,
+      "eval_samples_per_second": 31.527,
+      "eval_steps_per_second": 1.971,
+      "step": 180000
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7795,
+      "step": 180100
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 0.0005,
+      "loss": 1.7548,
+      "step": 180200
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 0.0005,
+      "loss": 1.7473,
+      "step": 180300
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.0005,
+      "loss": 1.7341,
+      "step": 180400
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.0005,
+      "loss": 1.7936,
+      "step": 180500
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 0.0005,
+      "loss": 1.7506,
+      "step": 180600
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 0.0005,
+      "loss": 1.7988,
+      "step": 180700
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 0.0005,
+      "loss": 1.7907,
+      "step": 180800
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 0.0005,
+      "loss": 1.7804,
+      "step": 180900
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 0.0005,
+      "loss": 1.7554,
+      "step": 181000
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 0.0005,
+      "loss": 1.7807,
+      "step": 181100
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7703,
+      "step": 181200
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 0.0005,
+      "loss": 1.7901,
+      "step": 181300
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 0.0005,
+      "loss": 1.796,
+      "step": 181400
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 0.0005,
+      "loss": 1.789,
+      "step": 181500
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7764,
+      "step": 181600
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7761,
+      "step": 181700
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.0005,
+      "loss": 1.7824,
+      "step": 181800
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.0005,
+      "loss": 1.7522,
+      "step": 181900
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 0.0005,
+      "loss": 1.7723,
+      "step": 182000
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 0.0005,
+      "loss": 1.7678,
+      "step": 182100
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 0.0005,
+      "loss": 1.7886,
+      "step": 182200
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7694,
+      "step": 182300
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7767,
+      "step": 182400
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 0.0005,
+      "loss": 1.7684,
+      "step": 182500
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 0.0005,
+      "loss": 1.7488,
+      "step": 182600
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7981,
+      "step": 182700
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7871,
+      "step": 182800
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7587,
+      "step": 182900
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 0.0005,
+      "loss": 1.8053,
+      "step": 183000
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 0.0005,
+      "loss": 1.7758,
+      "step": 183100
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7966,
+      "step": 183200
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7478,
+      "step": 183300
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 0.0005,
+      "loss": 1.7651,
+      "step": 183400
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7924,
+      "step": 183500
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 0.0005,
+      "loss": 1.7729,
+      "step": 183600
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 0.0005,
+      "loss": 1.7818,
+      "step": 183700
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 0.0005,
+      "loss": 1.7805,
+      "step": 183800
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 0.0005,
+      "loss": 1.8025,
+      "step": 183900
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7769,
+      "step": 184000
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 0.0005,
+      "loss": 1.798,
+      "step": 184100
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7751,
+      "step": 184200
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 0.0005,
+      "loss": 1.7869,
+      "step": 184300
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 0.0005,
+      "loss": 1.7882,
+      "step": 184400
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 0.0005,
+      "loss": 1.7908,
+      "step": 184500
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.0005,
+      "loss": 1.819,
+      "step": 184600
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.0005,
+      "loss": 1.794,
+      "step": 184700
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 0.0005,
+      "loss": 1.7894,
+      "step": 184800
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 0.0005,
+      "loss": 1.8124,
+      "step": 184900
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7684,
+      "step": 185000
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7929,
+      "step": 185100
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7906,
+      "step": 185200
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7852,
+      "step": 185300
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7982,
+      "step": 185400
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 0.0005,
+      "loss": 1.8103,
+      "step": 185500
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 0.0005,
+      "loss": 1.8011,
+      "step": 185600
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 0.0005,
+      "loss": 1.77,
+      "step": 185700
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7581,
+      "step": 185800
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7856,
+      "step": 185900
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7908,
+      "step": 186000
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7839,
+      "step": 186100
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 0.0005,
+      "loss": 1.8084,
+      "step": 186200
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7967,
+      "step": 186300
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7827,
+      "step": 186400
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.0005,
+      "loss": 1.8011,
+      "step": 186500
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.0005,
+      "loss": 1.8161,
+      "step": 186600
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 0.0005,
+      "loss": 1.8012,
+      "step": 186700
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.0005,
+      "loss": 1.8069,
+      "step": 186800
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.0005,
+      "loss": 1.8105,
+      "step": 186900
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7905,
+      "step": 187000
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 0.0005,
+      "loss": 1.803,
+      "step": 187100
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7951,
+      "step": 187200
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 0.0005,
+      "loss": 1.8003,
+      "step": 187300
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7834,
+      "step": 187400
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 0.0005,
+      "loss": 1.827,
+      "step": 187500
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7933,
+      "step": 187600
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 0.0005,
+      "loss": 1.8315,
+      "step": 187700
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7726,
+      "step": 187800
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7941,
+      "step": 187900
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 0.0005,
+      "loss": 1.7815,
+      "step": 188000
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 0.0005,
+      "loss": 1.7911,
+      "step": 188100
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 0.0005,
+      "loss": 1.8138,
+      "step": 188200
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8124,
+      "step": 188300
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 0.0005,
+      "loss": 1.7925,
+      "step": 188400
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 0.0005,
+      "loss": 1.7887,
+      "step": 188500
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 0.0005,
+      "loss": 1.8255,
+      "step": 188600
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7837,
+      "step": 188700
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8091,
+      "step": 188800
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 0.0005,
+      "loss": 1.8004,
+      "step": 188900
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 0.0005,
+      "loss": 1.8177,
+      "step": 189000
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 0.0005,
+      "loss": 1.8072,
+      "step": 189100
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 0.0005,
+      "loss": 1.8119,
+      "step": 189200
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 0.0005,
+      "loss": 1.7918,
+      "step": 189300
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 0.0005,
+      "loss": 1.797,
+      "step": 189400
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8045,
+      "step": 189500
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 0.0005,
+      "loss": 1.815,
+      "step": 189600
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 0.0005,
+      "loss": 1.8034,
+      "step": 189700
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 0.0005,
+      "loss": 1.7908,
+      "step": 189800
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 0.0005,
+      "loss": 1.8085,
+      "step": 189900
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8096,
+      "step": 190000
+    },
+    {
+      "epoch": 14.9,
+      "eval_gen_len": 18.80727465348283,
+      "eval_loss": 2.011542558670044,
+      "eval_rouge1": 35.1261,
+      "eval_rouge2": 13.9669,
+      "eval_rougeL": 28.8032,
+      "eval_rougeLsum": 28.7986,
+      "eval_runtime": 359.3121,
+      "eval_samples_per_second": 31.524,
+      "eval_steps_per_second": 1.97,
+      "step": 190000
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 0.0005,
+      "loss": 1.7785,
+      "step": 190100
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8173,
+      "step": 190200
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8061,
+      "step": 190300
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 0.0005,
+      "loss": 1.8585,
+      "step": 190400
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 0.0005,
+      "loss": 1.8302,
+      "step": 190500
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8016,
+      "step": 190600
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 0.0005,
+      "loss": 1.8222,
+      "step": 190700
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8069,
+      "step": 190800
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 0.0005,
+      "loss": 1.8366,
+      "step": 190900
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8089,
+      "step": 191000
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8057,
+      "step": 191100
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 0.0005,
+      "loss": 1.8068,
+      "step": 191200
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 0.0005,
+      "loss": 1.7916,
+      "step": 191300
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 0.0005,
+      "loss": 1.7352,
+      "step": 191400
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 0.0005,
+      "loss": 1.7374,
+      "step": 191500
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7113,
+      "step": 191600
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7363,
+      "step": 191700
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.0005,
+      "loss": 1.759,
+      "step": 191800
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 0.0005,
+      "loss": 1.7333,
+      "step": 191900
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 0.0005,
+      "loss": 1.7388,
+      "step": 192000
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 0.0005,
+      "loss": 1.7501,
+      "step": 192100
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 0.0005,
+      "loss": 1.7265,
+      "step": 192200
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 0.0005,
+      "loss": 1.7163,
+      "step": 192300
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 0.0005,
+      "loss": 1.7372,
+      "step": 192400
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.0005,
+      "loss": 1.7249,
+      "step": 192500
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.0005,
+      "loss": 1.7426,
+      "step": 192600
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 0.0005,
+      "loss": 1.7574,
+      "step": 192700
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7542,
+      "step": 192800
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.0005,
+      "loss": 1.7373,
+      "step": 192900
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.0005,
+      "loss": 1.7739,
+      "step": 193000
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 0.0005,
+      "loss": 1.7292,
+      "step": 193100
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 0.0005,
+      "loss": 1.7559,
+      "step": 193200
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 0.0005,
+      "loss": 1.7664,
+      "step": 193300
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 0.0005,
+      "loss": 1.775,
+      "step": 193400
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 0.0005,
+      "loss": 1.7353,
+      "step": 193500
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 0.0005,
+      "loss": 1.7059,
+      "step": 193600
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 0.0005,
+      "loss": 1.7496,
+      "step": 193700
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 0.0005,
+      "loss": 1.7722,
+      "step": 193800
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7518,
+      "step": 193900
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7517,
+      "step": 194000
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 0.0005,
+      "loss": 1.7901,
+      "step": 194100
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 0.0005,
+      "loss": 1.7402,
+      "step": 194200
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7294,
+      "step": 194300
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7142,
+      "step": 194400
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7213,
+      "step": 194500
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 0.0005,
+      "loss": 1.7418,
+      "step": 194600
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 0.0005,
+      "loss": 1.7736,
+      "step": 194700
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.0005,
+      "loss": 1.7709,
+      "step": 194800
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.0005,
+      "loss": 1.7365,
+      "step": 194900
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 0.0005,
+      "loss": 1.7571,
+      "step": 195000
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7207,
+      "step": 195100
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 0.0005,
+      "loss": 1.7616,
+      "step": 195200
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.0005,
+      "loss": 1.752,
+      "step": 195300
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.0005,
+      "loss": 1.737,
+      "step": 195400
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7397,
+      "step": 195500
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7586,
+      "step": 195600
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7357,
+      "step": 195700
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.0005,
+      "loss": 1.742,
+      "step": 195800
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 0.0005,
+      "loss": 1.7802,
+      "step": 195900
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7528,
+      "step": 196000
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 0.0005,
+      "loss": 1.7203,
+      "step": 196100
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7815,
+      "step": 196200
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7547,
+      "step": 196300
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 0.0005,
+      "loss": 1.761,
+      "step": 196400
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 0.0005,
+      "loss": 1.7246,
+      "step": 196500
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 0.0005,
+      "loss": 1.7984,
+      "step": 196600
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7696,
+      "step": 196700
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7424,
+      "step": 196800
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7836,
+      "step": 196900
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 0.0005,
+      "loss": 1.763,
+      "step": 197000
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 0.0005,
+      "loss": 1.7935,
+      "step": 197100
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 0.0005,
+      "loss": 1.757,
+      "step": 197200
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 0.0005,
+      "loss": 1.7406,
+      "step": 197300
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.0005,
+      "loss": 1.7726,
+      "step": 197400
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 0.0005,
+      "loss": 1.7588,
+      "step": 197500
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.0005,
+      "loss": 1.775,
+      "step": 197600
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.0005,
+      "loss": 1.7454,
+      "step": 197700
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7705,
+      "step": 197800
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 0.0005,
+      "loss": 1.8015,
+      "step": 197900
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7881,
+      "step": 198000
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7773,
+      "step": 198100
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7813,
+      "step": 198200
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7729,
+      "step": 198300
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 0.0005,
+      "loss": 1.77,
+      "step": 198400
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7687,
+      "step": 198500
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.0005,
+      "loss": 1.786,
+      "step": 198600
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 0.0005,
+      "loss": 1.765,
+      "step": 198700
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7615,
+      "step": 198800
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 0.0005,
+      "loss": 1.7887,
+      "step": 198900
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7714,
+      "step": 199000
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 0.0005,
+      "loss": 1.8128,
+      "step": 199100
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7684,
+      "step": 199200
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 0.0005,
+      "loss": 1.769,
+      "step": 199300
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 0.0005,
+      "loss": 1.7688,
+      "step": 199400
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 0.0005,
+      "loss": 1.7897,
+      "step": 199500
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 0.0005,
+      "loss": 1.7827,
+      "step": 199600
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 0.0005,
+      "loss": 1.775,
+      "step": 199700
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 0.0005,
+      "loss": 1.7955,
+      "step": 199800
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7774,
+      "step": 199900
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 0.0005,
+      "loss": 1.8084,
+      "step": 200000
+    },
+    {
+      "epoch": 15.68,
+      "eval_gen_len": 18.788293458108942,
+      "eval_loss": 2.014420747756958,
+      "eval_rouge1": 35.1009,
+      "eval_rouge2": 13.9441,
+      "eval_rougeL": 28.7814,
+      "eval_rougeLsum": 28.7778,
+      "eval_runtime": 359.2501,
+      "eval_samples_per_second": 31.53,
+      "eval_steps_per_second": 1.971,
+      "step": 200000
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7672,
+      "step": 200100
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7826,
+      "step": 200200
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7923,
+      "step": 200300
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7803,
+      "step": 200400
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7878,
+      "step": 200500
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7987,
+      "step": 200600
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.0005,
+      "loss": 1.775,
+      "step": 200700
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 0.0005,
+      "loss": 1.7686,
+      "step": 200800
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 0.0005,
+      "loss": 1.7859,
+      "step": 200900
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 0.0005,
+      "loss": 1.7605,
+      "step": 201000
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 0.0005,
+      "loss": 1.8035,
+      "step": 201100
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 0.0005,
+      "loss": 1.7632,
+      "step": 201200
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 0.0005,
+      "loss": 1.7976,
+      "step": 201300
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 0.0005,
+      "loss": 1.7718,
+      "step": 201400
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7786,
+      "step": 201500
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 0.0005,
+      "loss": 1.7525,
+      "step": 201600
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 0.0005,
+      "loss": 1.7773,
+      "step": 201700
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 0.0005,
+      "loss": 1.7534,
+      "step": 201800
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.0005,
+      "loss": 1.7707,
+      "step": 201900
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 0.0005,
+      "loss": 1.7756,
+      "step": 202000
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 0.0005,
+      "loss": 1.797,
+      "step": 202100
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.0005,
+      "loss": 1.7775,
+      "step": 202200
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.0005,
+      "loss": 1.8021,
+      "step": 202300
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 0.0005,
+      "loss": 1.8092,
+      "step": 202400
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 0.0005,
+      "loss": 1.802,
+      "step": 202500
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 0.0005,
+      "loss": 1.7979,
+      "step": 202600
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.0005,
+      "loss": 1.8046,
+      "step": 202700
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.0005,
+      "loss": 1.7676,
+      "step": 202800
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 0.0005,
+      "loss": 1.7956,
+      "step": 202900
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 0.0005,
+      "loss": 1.7877,
+      "step": 203000
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.0005,
+      "loss": 1.7923,
+      "step": 203100
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.0005,
+      "loss": 1.7798,
+      "step": 203200
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 0.0005,
+      "loss": 1.805,
+      "step": 203300
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 0.0005,
+      "loss": 1.7862,
+      "step": 203400
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 0.0005,
+      "loss": 1.8017,
+      "step": 203500
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 0.0005,
+      "loss": 1.7967,
+      "step": 203600
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 0.0005,
+      "loss": 1.7794,
+      "step": 203700
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 0.0005,
+      "loss": 1.8175,
+      "step": 203800
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 0.0005,
+      "loss": 1.7917,
+      "step": 203900
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 0.0005,
+      "loss": 1.7841,
+      "step": 204000
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.0005,
+      "loss": 1.7394,
+      "step": 204100
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.0005,
+      "loss": 1.6738,
+      "step": 204200
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 0.0005,
+      "loss": 1.6943,
+      "step": 204300
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 0.0005,
+      "loss": 1.7132,
+      "step": 204400
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.0005,
+      "loss": 1.7137,
+      "step": 204500
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.0005,
+      "loss": 1.7141,
+      "step": 204600
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 0.0005,
+      "loss": 1.6959,
+      "step": 204700
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 0.0005,
+      "loss": 1.7451,
+      "step": 204800
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 0.0005,
+      "loss": 1.713,
+      "step": 204900
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.0005,
+      "loss": 1.7364,
+      "step": 205000
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.0005,
+      "loss": 1.7192,
+      "step": 205100
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 0.0005,
+      "loss": 1.7199,
+      "step": 205200
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 0.0005,
+      "loss": 1.734,
+      "step": 205300
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 0.0005,
+      "loss": 1.6913,
+      "step": 205400
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7102,
+      "step": 205500
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7346,
+      "step": 205600
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 0.0005,
+      "loss": 1.7258,
+      "step": 205700
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 0.0005,
+      "loss": 1.6992,
+      "step": 205800
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.0005,
+      "loss": 1.7197,
+      "step": 205900
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.0005,
+      "loss": 1.7702,
+      "step": 206000
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 0.0005,
+      "loss": 1.7227,
+      "step": 206100
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 0.0005,
+      "loss": 1.7175,
+      "step": 206200
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 0.0005,
+      "loss": 1.7367,
+      "step": 206300
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.0005,
+      "loss": 1.7115,
+      "step": 206400
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.0005,
+      "loss": 1.751,
+      "step": 206500
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 0.0005,
+      "loss": 1.7377,
+      "step": 206600
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7243,
+      "step": 206700
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.0005,
+      "loss": 1.7267,
+      "step": 206800
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.0005,
+      "loss": 1.7448,
+      "step": 206900
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 0.0005,
+      "loss": 1.7357,
+      "step": 207000
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7057,
+      "step": 207100
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7444,
+      "step": 207200
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.0005,
+      "loss": 1.7404,
+      "step": 207300
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.0005,
+      "loss": 1.7089,
+      "step": 207400
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 0.0005,
+      "loss": 1.7147,
+      "step": 207500
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 0.0005,
+      "loss": 1.745,
+      "step": 207600
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 0.0005,
+      "loss": 1.7442,
+      "step": 207700
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7184,
+      "step": 207800
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 0.0005,
+      "loss": 1.724,
+      "step": 207900
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 0.0005,
+      "loss": 1.7249,
+      "step": 208000
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 0.0005,
+      "loss": 1.7072,
+      "step": 208100
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7377,
+      "step": 208200
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7466,
+      "step": 208300
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7426,
+      "step": 208400
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7699,
+      "step": 208500
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 0.0005,
+      "loss": 1.7459,
+      "step": 208600
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7269,
+      "step": 208700
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7108,
+      "step": 208800
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 0.0005,
+      "loss": 1.7786,
+      "step": 208900
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7564,
+      "step": 209000
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 0.0005,
+      "loss": 1.7284,
+      "step": 209100
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 0.0005,
+      "loss": 1.728,
+      "step": 209200
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 0.0005,
+      "loss": 1.7521,
+      "step": 209300
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 0.0005,
+      "loss": 1.7642,
+      "step": 209400
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7576,
+      "step": 209500
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7292,
+      "step": 209600
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7253,
+      "step": 209700
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 0.0005,
+      "loss": 1.7542,
+      "step": 209800
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 0.0005,
+      "loss": 1.7427,
+      "step": 209900
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 0.0005,
+      "loss": 1.7272,
+      "step": 210000
+    },
+    {
+      "epoch": 16.47,
+      "eval_gen_len": 18.780789264589036,
+      "eval_loss": 2.018299102783203,
+      "eval_rouge1": 35.0675,
+      "eval_rouge2": 13.9534,
+      "eval_rougeL": 28.8007,
+      "eval_rougeLsum": 28.7924,
+      "eval_runtime": 359.03,
+      "eval_samples_per_second": 31.549,
+      "eval_steps_per_second": 1.972,
+      "step": 210000
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 0.0005,
+      "loss": 1.7419,
+      "step": 210100
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 0.0005,
+      "loss": 1.7509,
+      "step": 210200
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 0.0005,
+      "loss": 1.7605,
+      "step": 210300
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 0.0005,
+      "loss": 1.7442,
+      "step": 210400
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7492,
+      "step": 210500
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7504,
+      "step": 210600
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7649,
+      "step": 210700
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7518,
+      "step": 210800
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7234,
+      "step": 210900
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7515,
+      "step": 211000
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7684,
+      "step": 211100
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 0.0005,
+      "loss": 1.7516,
+      "step": 211200
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7365,
+      "step": 211300
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7448,
+      "step": 211400
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7608,
+      "step": 211500
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7207,
+      "step": 211600
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 0.0005,
+      "loss": 1.7543,
+      "step": 211700
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7662,
+      "step": 211800
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7219,
+      "step": 211900
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7558,
+      "step": 212000
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 0.0005,
+      "loss": 1.7425,
+      "step": 212100
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 0.0005,
+      "loss": 1.7497,
+      "step": 212200
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 0.0005,
+      "loss": 1.7257,
+      "step": 212300
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7546,
+      "step": 212400
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7781,
+      "step": 212500
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 0.0005,
+      "loss": 1.7755,
+      "step": 212600
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7529,
+      "step": 212700
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7525,
+      "step": 212800
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7715,
+      "step": 212900
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7378,
+      "step": 213000
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7646,
+      "step": 213100
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7489,
+      "step": 213200
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7538,
+      "step": 213300
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7877,
+      "step": 213400
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 0.0005,
+      "loss": 1.7733,
+      "step": 213500
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 0.0005,
+      "loss": 1.7807,
+      "step": 213600
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 0.0005,
+      "loss": 1.7642,
+      "step": 213700
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.0005,
+      "loss": 1.7609,
+      "step": 213800
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.0005,
+      "loss": 1.7679,
+      "step": 213900
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 0.0005,
+      "loss": 1.7719,
+      "step": 214000
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 0.0005,
+      "loss": 1.7707,
+      "step": 214100
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7702,
+      "step": 214200
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 0.0005,
+      "loss": 1.7301,
+      "step": 214300
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 0.0005,
+      "loss": 1.7522,
+      "step": 214400
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 0.0005,
+      "loss": 1.7738,
+      "step": 214500
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 0.0005,
+      "loss": 1.7551,
+      "step": 214600
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 0.0005,
+      "loss": 1.742,
+      "step": 214700
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 0.0005,
+      "loss": 1.7641,
+      "step": 214800
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 0.0005,
+      "loss": 1.7828,
+      "step": 214900
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 0.0005,
+      "loss": 1.7799,
+      "step": 215000
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 0.0005,
+      "loss": 1.7476,
+      "step": 215100
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.0005,
+      "loss": 1.7662,
+      "step": 215200
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.0005,
+      "loss": 1.7764,
+      "step": 215300
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 0.0005,
+      "loss": 1.79,
+      "step": 215400
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 0.0005,
+      "loss": 1.7862,
+      "step": 215500
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0005,
+      "loss": 1.7792,
+      "step": 215600
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0005,
+      "loss": 1.7772,
+      "step": 215700
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.0005,
+      "loss": 1.8195,
+      "step": 215800
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 0.0005,
+      "loss": 1.7426,
+      "step": 215900
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 0.0005,
+      "loss": 1.781,
+      "step": 216000
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.0005,
+      "loss": 1.7399,
+      "step": 216100
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.0005,
+      "loss": 1.774,
+      "step": 216200
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 0.0005,
+      "loss": 1.7855,
+      "step": 216300
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 0.0005,
+      "loss": 1.785,
+      "step": 216400
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 0.0005,
+      "loss": 1.7603,
+      "step": 216500
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 0.0005,
+      "loss": 1.7576,
+      "step": 216600
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 0.0005,
+      "loss": 1.7952,
+      "step": 216700
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 0.0005,
+      "loss": 1.7494,
+      "step": 216800
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 0.0005,
+      "loss": 1.7225,
+      "step": 216900
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 0.0005,
+      "loss": 1.6896,
+      "step": 217000
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 0.0005,
+      "loss": 1.7069,
+      "step": 217100
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 0.0005,
+      "loss": 1.6877,
+      "step": 217200
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 0.0005,
+      "loss": 1.6936,
+      "step": 217300
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 0.0005,
+      "loss": 1.6868,
+      "step": 217400
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.0005,
+      "loss": 1.6801,
+      "step": 217500
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.0005,
+      "loss": 1.7087,
+      "step": 217600
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 0.0005,
+      "loss": 1.677,
+      "step": 217700
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 0.0005,
+      "loss": 1.7113,
+      "step": 217800
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 0.0005,
+      "loss": 1.7145,
+      "step": 217900
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.0005,
+      "loss": 1.6894,
+      "step": 218000
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.0005,
+      "loss": 1.7059,
+      "step": 218100
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 0.0005,
+      "loss": 1.6714,
+      "step": 218200
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7143,
+      "step": 218300
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.0005,
+      "loss": 1.7091,
+      "step": 218400
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.0005,
+      "loss": 1.6903,
+      "step": 218500
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 0.0005,
+      "loss": 1.7064,
+      "step": 218600
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 0.0005,
+      "loss": 1.6881,
+      "step": 218700
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 0.0005,
+      "loss": 1.6891,
+      "step": 218800
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 0.0005,
+      "loss": 1.7152,
+      "step": 218900
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 0.0005,
+      "loss": 1.7019,
+      "step": 219000
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 0.0005,
+      "loss": 1.7281,
+      "step": 219100
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 0.0005,
+      "loss": 1.7069,
+      "step": 219200
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 0.0005,
+      "loss": 1.7337,
+      "step": 219300
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7059,
+      "step": 219400
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7184,
+      "step": 219500
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 0.0005,
+      "loss": 1.7176,
+      "step": 219600
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 0.0005,
+      "loss": 1.6903,
+      "step": 219700
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7106,
+      "step": 219800
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 0.0005,
+      "loss": 1.7179,
+      "step": 219900
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 0.0005,
+      "loss": 1.7223,
+      "step": 220000
+    },
+    {
+      "epoch": 17.25,
+      "eval_gen_len": 18.785556634589916,
+      "eval_loss": 2.0205140113830566,
+      "eval_rouge1": 35.2932,
+      "eval_rouge2": 14.0579,
+      "eval_rougeL": 28.9651,
+      "eval_rougeLsum": 28.9628,
+      "eval_runtime": 358.8009,
+      "eval_samples_per_second": 31.569,
+      "eval_steps_per_second": 1.973,
+      "step": 220000
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 0.0005,
+      "loss": 1.7245,
+      "step": 220100
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 0.0005,
+      "loss": 1.6959,
+      "step": 220200
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 0.0005,
+      "loss": 1.7083,
+      "step": 220300
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 0.0005,
+      "loss": 1.7132,
+      "step": 220400
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 0.0005,
+      "loss": 1.7206,
+      "step": 220500
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7128,
+      "step": 220600
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.0005,
+      "loss": 1.711,
+      "step": 220700
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.0005,
+      "loss": 1.7169,
+      "step": 220800
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 0.0005,
+      "loss": 1.6995,
+      "step": 220900
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7092,
+      "step": 221000
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7284,
+      "step": 221100
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7167,
+      "step": 221200
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7499,
+      "step": 221300
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 0.0005,
+      "loss": 1.693,
+      "step": 221400
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 0.0005,
+      "loss": 1.7267,
+      "step": 221500
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 0.0005,
+      "loss": 1.6923,
+      "step": 221600
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7253,
+      "step": 221700
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 0.0005,
+      "loss": 1.7074,
+      "step": 221800
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 0.0005,
+      "loss": 1.6936,
+      "step": 221900
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 0.0005,
+      "loss": 1.7314,
+      "step": 222000
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 0.0005,
+      "loss": 1.7109,
+      "step": 222100
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 0.0005,
+      "loss": 1.7574,
+      "step": 222200
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 0.0005,
+      "loss": 1.6917,
+      "step": 222300
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7473,
+      "step": 222400
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 0.0005,
+      "loss": 1.7132,
+      "step": 222500
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 0.0005,
+      "loss": 1.7103,
+      "step": 222600
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 0.0005,
+      "loss": 1.717,
+      "step": 222700
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 0.0005,
+      "loss": 1.7215,
+      "step": 222800
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 0.0005,
+      "loss": 1.7374,
+      "step": 222900
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 0.0005,
+      "loss": 1.6961,
+      "step": 223000
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.0005,
+      "loss": 1.7261,
+      "step": 223100
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.0005,
+      "loss": 1.7277,
+      "step": 223200
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7473,
+      "step": 223300
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7192,
+      "step": 223400
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7496,
+      "step": 223500
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7392,
+      "step": 223600
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7387,
+      "step": 223700
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6937,
+      "step": 223800
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 0.0005,
+      "loss": 1.7106,
+      "step": 223900
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7596,
+      "step": 224000
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.0005,
+      "loss": 1.743,
+      "step": 224100
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7321,
+      "step": 224200
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7312,
+      "step": 224300
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 0.0005,
+      "loss": 1.737,
+      "step": 224400
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7505,
+      "step": 224500
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7366,
+      "step": 224600
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7694,
+      "step": 224700
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 0.0005,
+      "loss": 1.7462,
+      "step": 224800
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.0005,
+      "loss": 1.7142,
+      "step": 224900
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.0005,
+      "loss": 1.7402,
+      "step": 225000
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.0005,
+      "loss": 1.7443,
+      "step": 225100
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7299,
+      "step": 225200
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 0.0005,
+      "loss": 1.7341,
+      "step": 225300
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7406,
+      "step": 225400
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7466,
+      "step": 225500
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7195,
+      "step": 225600
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7175,
+      "step": 225700
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7269,
+      "step": 225800
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7371,
+      "step": 225900
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7362,
+      "step": 226000
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7418,
+      "step": 226100
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.0005,
+      "loss": 1.7583,
+      "step": 226200
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 0.0005,
+      "loss": 1.7261,
+      "step": 226300
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 0.0005,
+      "loss": 1.7558,
+      "step": 226400
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 0.0005,
+      "loss": 1.7613,
+      "step": 226500
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 0.0005,
+      "loss": 1.7171,
+      "step": 226600
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 0.0005,
+      "loss": 1.7267,
+      "step": 226700
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.0005,
+      "loss": 1.7391,
+      "step": 226800
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.0005,
+      "loss": 1.7509,
+      "step": 226900
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7602,
+      "step": 227000
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 0.0005,
+      "loss": 1.7495,
+      "step": 227100
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.0005,
+      "loss": 1.7496,
+      "step": 227200
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.0005,
+      "loss": 1.7514,
+      "step": 227300
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.0005,
+      "loss": 1.7236,
+      "step": 227400
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 0.0005,
+      "loss": 1.7744,
+      "step": 227500
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 0.0005,
+      "loss": 1.7376,
+      "step": 227600
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.0005,
+      "loss": 1.7838,
+      "step": 227700
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.0005,
+      "loss": 1.7483,
+      "step": 227800
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 0.0005,
+      "loss": 1.7373,
+      "step": 227900
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.0005,
+      "loss": 1.763,
+      "step": 228000
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 0.0005,
+      "loss": 1.7521,
+      "step": 228100
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 0.0005,
+      "loss": 1.7713,
+      "step": 228200
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 0.0005,
+      "loss": 1.7472,
+      "step": 228300
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 0.0005,
+      "loss": 1.7377,
+      "step": 228400
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 0.0005,
+      "loss": 1.7492,
+      "step": 228500
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 0.0005,
+      "loss": 1.7903,
+      "step": 228600
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 0.0005,
+      "loss": 1.7734,
+      "step": 228700
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 0.0005,
+      "loss": 1.743,
+      "step": 228800
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 0.0005,
+      "loss": 1.7376,
+      "step": 228900
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 0.0005,
+      "loss": 1.7402,
+      "step": 229000
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.0005,
+      "loss": 1.7615,
+      "step": 229100
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.0005,
+      "loss": 1.7213,
+      "step": 229200
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 0.0005,
+      "loss": 1.7485,
+      "step": 229300
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 0.0005,
+      "loss": 1.7361,
+      "step": 229400
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 0.0005,
+      "loss": 1.7568,
+      "step": 229500
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 0.0005,
+      "loss": 1.6941,
+      "step": 229600
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 0.0005,
+      "loss": 1.6589,
+      "step": 229700
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 0.0005,
+      "loss": 1.6802,
+      "step": 229800
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 0.0005,
+      "loss": 1.6694,
+      "step": 229900
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.0005,
+      "loss": 1.6742,
+      "step": 230000
+    },
+    {
+      "epoch": 18.04,
+      "eval_gen_len": 18.784497219034165,
+      "eval_loss": 2.0137345790863037,
+      "eval_rouge1": 35.3791,
+      "eval_rouge2": 14.1491,
+      "eval_rougeL": 29.0175,
+      "eval_rougeLsum": 29.0086,
+      "eval_runtime": 361.5305,
+      "eval_samples_per_second": 31.331,
+      "eval_steps_per_second": 1.958,
+      "step": 230000
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.0005,
+      "loss": 1.6665,
+      "step": 230100
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 0.0005,
+      "loss": 1.6606,
+      "step": 230200
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 0.0005,
+      "loss": 1.677,
+      "step": 230300
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 0.0005,
+      "loss": 1.6896,
+      "step": 230400
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 0.0005,
+      "loss": 1.6786,
+      "step": 230500
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 0.0005,
+      "loss": 1.653,
+      "step": 230600
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 0.0005,
+      "loss": 1.7007,
+      "step": 230700
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 0.0005,
+      "loss": 1.6996,
+      "step": 230800
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 0.0005,
+      "loss": 1.6833,
+      "step": 230900
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 0.0005,
+      "loss": 1.6877,
+      "step": 231000
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 0.0005,
+      "loss": 1.7055,
+      "step": 231100
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.0005,
+      "loss": 1.6772,
+      "step": 231200
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 0.0005,
+      "loss": 1.6763,
+      "step": 231300
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 0.0005,
+      "loss": 1.69,
+      "step": 231400
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 0.0005,
+      "loss": 1.6732,
+      "step": 231500
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 0.0005,
+      "loss": 1.6955,
+      "step": 231600
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 0.0005,
+      "loss": 1.7267,
+      "step": 231700
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 0.0005,
+      "loss": 1.6848,
+      "step": 231800
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 0.0005,
+      "loss": 1.6835,
+      "step": 231900
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 0.0005,
+      "loss": 1.689,
+      "step": 232000
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 0.0005,
+      "loss": 1.6716,
+      "step": 232100
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 0.0005,
+      "loss": 1.6968,
+      "step": 232200
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.0005,
+      "loss": 1.6927,
+      "step": 232300
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.0005,
+      "loss": 1.678,
+      "step": 232400
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 0.0005,
+      "loss": 1.6848,
+      "step": 232500
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 0.0005,
+      "loss": 1.6858,
+      "step": 232600
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 0.0005,
+      "loss": 1.6891,
+      "step": 232700
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 0.0005,
+      "loss": 1.7047,
+      "step": 232800
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 0.0005,
+      "loss": 1.6695,
+      "step": 232900
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 0.0005,
+      "loss": 1.7147,
+      "step": 233000
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 0.0005,
+      "loss": 1.681,
+      "step": 233100
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 0.0005,
+      "loss": 1.703,
+      "step": 233200
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7181,
+      "step": 233300
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.0005,
+      "loss": 1.6858,
+      "step": 233400
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.0005,
+      "loss": 1.695,
+      "step": 233500
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 0.0005,
+      "loss": 1.6958,
+      "step": 233600
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6873,
+      "step": 233700
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 0.0005,
+      "loss": 1.7194,
+      "step": 233800
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7067,
+      "step": 233900
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 0.0005,
+      "loss": 1.7099,
+      "step": 234000
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 0.0005,
+      "loss": 1.6824,
+      "step": 234100
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.0005,
+      "loss": 1.6951,
+      "step": 234200
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.0005,
+      "loss": 1.6935,
+      "step": 234300
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 0.0005,
+      "loss": 1.7125,
+      "step": 234400
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 0.0005,
+      "loss": 1.729,
+      "step": 234500
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.0005,
+      "loss": 1.7548,
+      "step": 234600
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.0005,
+      "loss": 1.6915,
+      "step": 234700
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 0.0005,
+      "loss": 1.677,
+      "step": 234800
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6872,
+      "step": 234900
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 0.0005,
+      "loss": 1.6984,
+      "step": 235000
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6906,
+      "step": 235100
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 0.0005,
+      "loss": 1.7143,
+      "step": 235200
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 0.0005,
+      "loss": 1.7065,
+      "step": 235300
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 0.0005,
+      "loss": 1.7155,
+      "step": 235400
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 0.0005,
+      "loss": 1.7136,
+      "step": 235500
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.0005,
+      "loss": 1.703,
+      "step": 235600
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.0005,
+      "loss": 1.6929,
+      "step": 235700
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 0.0005,
+      "loss": 1.7074,
+      "step": 235800
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 0.0005,
+      "loss": 1.6988,
+      "step": 235900
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7208,
+      "step": 236000
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.0005,
+      "loss": 1.7038,
+      "step": 236100
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 0.0005,
+      "loss": 1.7127,
+      "step": 236200
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7226,
+      "step": 236300
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7001,
+      "step": 236400
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6968,
+      "step": 236500
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6991,
+      "step": 236600
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 0.0005,
+      "loss": 1.7272,
+      "step": 236700
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7066,
+      "step": 236800
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 0.0005,
+      "loss": 1.7438,
+      "step": 236900
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7206,
+      "step": 237000
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.0005,
+      "loss": 1.751,
+      "step": 237100
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 0.0005,
+      "loss": 1.7057,
+      "step": 237200
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 0.0005,
+      "loss": 1.7001,
+      "step": 237300
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7321,
+      "step": 237400
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 0.0005,
+      "loss": 1.733,
+      "step": 237500
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 0.0005,
+      "loss": 1.704,
+      "step": 237600
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 0.0005,
+      "loss": 1.7253,
+      "step": 237700
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 0.0005,
+      "loss": 1.7128,
+      "step": 237800
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7277,
+      "step": 237900
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7428,
+      "step": 238000
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 0.0005,
+      "loss": 1.7269,
+      "step": 238100
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7137,
+      "step": 238200
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7148,
+      "step": 238300
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7229,
+      "step": 238400
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.0005,
+      "loss": 1.7405,
+      "step": 238500
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7477,
+      "step": 238600
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7204,
+      "step": 238700
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7565,
+      "step": 238800
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.0005,
+      "loss": 1.7104,
+      "step": 238900
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 0.0005,
+      "loss": 1.748,
+      "step": 239000
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 0.0005,
+      "loss": 1.7363,
+      "step": 239100
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 0.0005,
+      "loss": 1.7393,
+      "step": 239200
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 0.0005,
+      "loss": 1.7151,
+      "step": 239300
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 0.0005,
+      "loss": 1.7496,
+      "step": 239400
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 0.0005,
+      "loss": 1.7605,
+      "step": 239500
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 0.0005,
+      "loss": 1.7371,
+      "step": 239600
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7388,
+      "step": 239700
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7166,
+      "step": 239800
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 0.0005,
+      "loss": 1.7391,
+      "step": 239900
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 0.0005,
+      "loss": 1.7056,
+      "step": 240000
+    },
+    {
+      "epoch": 18.82,
+      "eval_gen_len": 18.775668756069567,
+      "eval_loss": 2.0135035514831543,
+      "eval_rouge1": 35.5094,
+      "eval_rouge2": 14.2676,
+      "eval_rougeL": 29.0733,
+      "eval_rougeLsum": 29.0642,
+      "eval_runtime": 364.5866,
+      "eval_samples_per_second": 31.068,
+      "eval_steps_per_second": 1.942,
+      "step": 240000
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 0.0005,
+      "loss": 1.7024,
+      "step": 240100
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6971,
+      "step": 240200
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6981,
+      "step": 240300
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 0.0005,
+      "loss": 1.7311,
+      "step": 240400
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 0.0005,
+      "loss": 1.7257,
+      "step": 240500
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6868,
+      "step": 240600
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 0.0005,
+      "loss": 1.7249,
+      "step": 240700
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 0.0005,
+      "loss": 1.7212,
+      "step": 240800
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 0.0005,
+      "loss": 1.7294,
+      "step": 240900
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 0.0005,
+      "loss": 1.7355,
+      "step": 241000
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.0005,
+      "loss": 1.7353,
+      "step": 241100
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.0005,
+      "loss": 1.7472,
+      "step": 241200
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 0.0005,
+      "loss": 1.7534,
+      "step": 241300
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 0.0005,
+      "loss": 1.7067,
+      "step": 241400
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 0.0005,
+      "loss": 1.7399,
+      "step": 241500
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 0.0005,
+      "loss": 1.7203,
+      "step": 241600
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 0.0005,
+      "loss": 1.7496,
+      "step": 241700
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 0.0005,
+      "loss": 1.7307,
+      "step": 241800
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 0.0005,
+      "loss": 1.7298,
+      "step": 241900
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 0.0005,
+      "loss": 1.7253,
+      "step": 242000
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 0.0005,
+      "loss": 1.7364,
+      "step": 242100
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 0.0005,
+      "loss": 1.7221,
+      "step": 242200
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 0.0005,
+      "loss": 1.7532,
+      "step": 242300
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 0.0005,
+      "loss": 1.6476,
+      "step": 242400
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 0.0005,
+      "loss": 1.6458,
+      "step": 242500
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 0.0005,
+      "loss": 1.627,
+      "step": 242600
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 0.0005,
+      "loss": 1.6363,
+      "step": 242700
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 0.0005,
+      "loss": 1.6743,
+      "step": 242800
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 0.0005,
+      "loss": 1.666,
+      "step": 242900
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 0.0005,
+      "loss": 1.6507,
+      "step": 243000
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 0.0005,
+      "loss": 1.6605,
+      "step": 243100
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 0.0005,
+      "loss": 1.6324,
+      "step": 243200
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 0.0005,
+      "loss": 1.6725,
+      "step": 243300
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 0.0005,
+      "loss": 1.6617,
+      "step": 243400
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 0.0005,
+      "loss": 1.6396,
+      "step": 243500
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 0.0005,
+      "loss": 1.6576,
+      "step": 243600
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 0.0005,
+      "loss": 1.6778,
+      "step": 243700
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 0.0005,
+      "loss": 1.6664,
+      "step": 243800
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 0.0005,
+      "loss": 1.7057,
+      "step": 243900
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 0.0005,
+      "loss": 1.6805,
+      "step": 244000
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 0.0005,
+      "loss": 1.6807,
+      "step": 244100
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 0.0005,
+      "loss": 1.6879,
+      "step": 244200
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 0.0005,
+      "loss": 1.6696,
+      "step": 244300
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 0.0005,
+      "loss": 1.6631,
+      "step": 244400
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 0.0005,
+      "loss": 1.672,
+      "step": 244500
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 0.0005,
+      "loss": 1.6705,
+      "step": 244600
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 0.0005,
+      "loss": 1.6861,
+      "step": 244700
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 0.0005,
+      "loss": 1.6579,
+      "step": 244800
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 0.0005,
+      "loss": 1.6765,
+      "step": 244900
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 0.0005,
+      "loss": 1.7092,
+      "step": 245000
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 0.0005,
+      "loss": 1.6675,
+      "step": 245100
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 0.0005,
+      "loss": 1.6635,
+      "step": 245200
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 0.0005,
+      "loss": 1.678,
+      "step": 245300
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 0.0005,
+      "loss": 1.664,
+      "step": 245400
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 0.0005,
+      "loss": 1.6954,
+      "step": 245500
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 0.0005,
+      "loss": 1.6661,
+      "step": 245600
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 0.0005,
+      "loss": 1.6942,
+      "step": 245700
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 0.0005,
+      "loss": 1.6785,
+      "step": 245800
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 0.0005,
+      "loss": 1.6845,
+      "step": 245900
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 0.0005,
+      "loss": 1.6905,
+      "step": 246000
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 0.0005,
+      "loss": 1.7119,
+      "step": 246100
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6895,
+      "step": 246200
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6958,
+      "step": 246300
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 0.0005,
+      "loss": 1.6489,
+      "step": 246400
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 0.0005,
+      "loss": 1.672,
+      "step": 246500
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 0.0005,
+      "loss": 1.7014,
+      "step": 246600
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 0.0005,
+      "loss": 1.6828,
+      "step": 246700
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 0.0005,
+      "loss": 1.6354,
+      "step": 246800
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 0.0005,
+      "loss": 1.6998,
+      "step": 246900
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 0.0005,
+      "loss": 1.679,
+      "step": 247000
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 0.0005,
+      "loss": 1.6901,
+      "step": 247100
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.0005,
+      "loss": 1.6976,
+      "step": 247200
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.0005,
+      "loss": 1.6883,
+      "step": 247300
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 0.0005,
+      "loss": 1.6643,
+      "step": 247400
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 0.0005,
+      "loss": 1.6772,
+      "step": 247500
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6931,
+      "step": 247600
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6521,
+      "step": 247700
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 0.0005,
+      "loss": 1.7068,
+      "step": 247800
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6937,
+      "step": 247900
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 0.0005,
+      "loss": 1.72,
+      "step": 248000
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 0.0005,
+      "loss": 1.7136,
+      "step": 248100
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 0.0005,
+      "loss": 1.6937,
+      "step": 248200
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 0.0005,
+      "loss": 1.6706,
+      "step": 248300
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 0.0005,
+      "loss": 1.6952,
+      "step": 248400
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 0.0005,
+      "loss": 1.698,
+      "step": 248500
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 0.0005,
+      "loss": 1.699,
+      "step": 248600
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 0.0005,
+      "loss": 1.6925,
+      "step": 248700
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6807,
+      "step": 248800
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 0.0005,
+      "loss": 1.6993,
+      "step": 248900
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 0.0005,
+      "loss": 1.6907,
+      "step": 249000
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7015,
+      "step": 249100
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 0.0005,
+      "loss": 1.7157,
+      "step": 249200
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6905,
+      "step": 249300
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 0.0005,
+      "loss": 1.6841,
+      "step": 249400
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.0005,
+      "loss": 1.6621,
+      "step": 249500
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7077,
+      "step": 249600
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 0.0005,
+      "loss": 1.6981,
+      "step": 249700
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 0.0005,
+      "loss": 1.6905,
+      "step": 249800
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 0.0005,
+      "loss": 1.7028,
+      "step": 249900
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6849,
+      "step": 250000
+    },
+    {
+      "epoch": 19.6,
+      "eval_gen_len": 18.80047673700009,
+      "eval_loss": 2.020770311355591,
+      "eval_rouge1": 35.1918,
+      "eval_rouge2": 14.0743,
+      "eval_rougeL": 28.8604,
+      "eval_rougeLsum": 28.846,
+      "eval_runtime": 358.4886,
+      "eval_samples_per_second": 31.597,
+      "eval_steps_per_second": 1.975,
+      "step": 250000
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 0.0005,
+      "loss": 1.6831,
+      "step": 250100
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 0.0005,
+      "loss": 1.7192,
+      "step": 250200
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 0.0005,
+      "loss": 1.7143,
+      "step": 250300
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 0.0005,
+      "loss": 1.708,
+      "step": 250400
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 0.0005,
+      "loss": 1.714,
+      "step": 250500
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 0.0005,
+      "loss": 1.7072,
+      "step": 250600
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7227,
+      "step": 250700
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 0.0005,
+      "loss": 1.7089,
+      "step": 250800
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7213,
+      "step": 250900
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 0.0005,
+      "loss": 1.7167,
+      "step": 251000
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 0.0005,
+      "loss": 1.7012,
+      "step": 251100
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 0.0005,
+      "loss": 1.6903,
+      "step": 251200
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6836,
+      "step": 251300
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6714,
+      "step": 251400
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 0.0005,
+      "loss": 1.7137,
+      "step": 251500
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6813,
+      "step": 251600
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 0.0005,
+      "loss": 1.7059,
+      "step": 251700
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 0.0005,
+      "loss": 1.6909,
+      "step": 251800
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 0.0005,
+      "loss": 1.7011,
+      "step": 251900
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 0.0005,
+      "loss": 1.7112,
+      "step": 252000
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 0.0005,
+      "loss": 1.7428,
+      "step": 252100
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 0.0005,
+      "loss": 1.7332,
+      "step": 252200
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 0.0005,
+      "loss": 1.7068,
+      "step": 252300
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 0.0005,
+      "loss": 1.7146,
+      "step": 252400
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7073,
+      "step": 252500
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 0.0005,
+      "loss": 1.7071,
+      "step": 252600
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 0.0005,
+      "loss": 1.7196,
+      "step": 252700
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 0.0005,
+      "loss": 1.7188,
+      "step": 252800
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 0.0005,
+      "loss": 1.691,
+      "step": 252900
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 0.0005,
+      "loss": 1.7275,
+      "step": 253000
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 0.0005,
+      "loss": 1.6993,
+      "step": 253100
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 0.0005,
+      "loss": 1.7168,
+      "step": 253200
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 0.0005,
+      "loss": 1.6927,
+      "step": 253300
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 0.0005,
+      "loss": 1.7066,
+      "step": 253400
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 0.0005,
+      "loss": 1.7022,
+      "step": 253500
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6908,
+      "step": 253600
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 0.0005,
+      "loss": 1.7124,
+      "step": 253700
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 0.0005,
+      "loss": 1.7099,
+      "step": 253800
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 0.0005,
+      "loss": 1.7335,
+      "step": 253900
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 0.0005,
+      "loss": 1.7446,
+      "step": 254000
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 0.0005,
+      "loss": 1.7187,
+      "step": 254100
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6978,
+      "step": 254200
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 0.0005,
+      "loss": 1.7339,
+      "step": 254300
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 0.0005,
+      "loss": 1.698,
+      "step": 254400
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 0.0005,
+      "loss": 1.7041,
+      "step": 254500
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 0.0005,
+      "loss": 1.7173,
+      "step": 254600
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 0.0005,
+      "loss": 1.7025,
+      "step": 254700
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6967,
+      "step": 254800
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 0.0005,
+      "loss": 1.6963,
+      "step": 254900
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.0005,
+      "loss": 1.7125,
+      "step": 255000
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6839,
+      "step": 255100
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 0.0005,
+      "loss": 1.6403,
+      "step": 255200
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 0.0005,
+      "loss": 1.6344,
+      "step": 255300
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 0.0005,
+      "loss": 1.643,
+      "step": 255400
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 0.0005,
+      "loss": 1.6433,
+      "step": 255500
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 0.0005,
+      "loss": 1.631,
+      "step": 255600
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 0.0005,
+      "loss": 1.662,
+      "step": 255700
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 0.0005,
+      "loss": 1.6108,
+      "step": 255800
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 0.0005,
+      "loss": 1.6505,
+      "step": 255900
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 0.0005,
+      "loss": 1.6687,
+      "step": 256000
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 0.0005,
+      "loss": 1.6718,
+      "step": 256100
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 0.0005,
+      "loss": 1.6497,
+      "step": 256200
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 0.0005,
+      "loss": 1.6455,
+      "step": 256300
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 0.0005,
+      "loss": 1.6545,
+      "step": 256400
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 0.0005,
+      "loss": 1.6349,
+      "step": 256500
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 0.0005,
+      "loss": 1.6368,
+      "step": 256600
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 0.0005,
+      "loss": 1.624,
+      "step": 256700
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 0.0005,
+      "loss": 1.6588,
+      "step": 256800
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 0.0005,
+      "loss": 1.6254,
+      "step": 256900
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 0.0005,
+      "loss": 1.6433,
+      "step": 257000
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 0.0005,
+      "loss": 1.6505,
+      "step": 257100
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 0.0005,
+      "loss": 1.6722,
+      "step": 257200
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 0.0005,
+      "loss": 1.6373,
+      "step": 257300
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 0.0005,
+      "loss": 1.6586,
+      "step": 257400
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 0.0005,
+      "loss": 1.6745,
+      "step": 257500
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 0.0005,
+      "loss": 1.665,
+      "step": 257600
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 0.0005,
+      "loss": 1.656,
+      "step": 257700
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 0.0005,
+      "loss": 1.6748,
+      "step": 257800
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 0.0005,
+      "loss": 1.6635,
+      "step": 257900
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 0.0005,
+      "loss": 1.6579,
+      "step": 258000
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 0.0005,
+      "loss": 1.6716,
+      "step": 258100
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 0.0005,
+      "loss": 1.6748,
+      "step": 258200
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 0.0005,
+      "loss": 1.688,
+      "step": 258300
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 0.0005,
+      "loss": 1.6389,
+      "step": 258400
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 0.0005,
+      "loss": 1.6642,
+      "step": 258500
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 0.0005,
+      "loss": 1.6493,
+      "step": 258600
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 0.0005,
+      "loss": 1.6601,
+      "step": 258700
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 0.0005,
+      "loss": 1.6424,
+      "step": 258800
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 0.0005,
+      "loss": 1.6607,
+      "step": 258900
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6403,
+      "step": 259000
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 0.0005,
+      "loss": 1.693,
+      "step": 259100
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6657,
+      "step": 259200
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6836,
+      "step": 259300
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 0.0005,
+      "loss": 1.6652,
+      "step": 259400
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 0.0005,
+      "loss": 1.6512,
+      "step": 259500
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 0.0005,
+      "loss": 1.671,
+      "step": 259600
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 0.0005,
+      "loss": 1.6558,
+      "step": 259700
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 0.0005,
+      "loss": 1.6636,
+      "step": 259800
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 0.0005,
+      "loss": 1.6533,
+      "step": 259900
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 0.0005,
+      "loss": 1.6784,
+      "step": 260000
+    },
+    {
+      "epoch": 20.39,
+      "eval_gen_len": 18.790235719961156,
+      "eval_loss": 2.0271031856536865,
+      "eval_rouge1": 35.165,
+      "eval_rouge2": 14.0225,
+      "eval_rougeL": 28.8773,
+      "eval_rougeLsum": 28.8644,
+      "eval_runtime": 359.9035,
+      "eval_samples_per_second": 31.472,
+      "eval_steps_per_second": 1.967,
+      "step": 260000
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 0.0005,
+      "loss": 1.6699,
+      "step": 260100
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 0.0005,
+      "loss": 1.6597,
+      "step": 260200
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 0.0005,
+      "loss": 1.6813,
+      "step": 260300
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6641,
+      "step": 260400
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 0.0005,
+      "loss": 1.6587,
+      "step": 260500
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6786,
+      "step": 260600
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6915,
+      "step": 260700
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 0.0005,
+      "loss": 1.659,
+      "step": 260800
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 0.0005,
+      "loss": 1.6866,
+      "step": 260900
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 0.0005,
+      "loss": 1.6817,
+      "step": 261000
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 0.0005,
+      "loss": 1.6575,
+      "step": 261100
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 0.0005,
+      "loss": 1.6908,
+      "step": 261200
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 0.0005,
+      "loss": 1.6865,
+      "step": 261300
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 0.0005,
+      "loss": 1.6798,
+      "step": 261400
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6394,
+      "step": 261500
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6804,
+      "step": 261600
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 0.0005,
+      "loss": 1.679,
+      "step": 261700
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 0.0005,
+      "loss": 1.7161,
+      "step": 261800
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 0.0005,
+      "loss": 1.6768,
+      "step": 261900
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 0.0005,
+      "loss": 1.7062,
+      "step": 262000
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6279,
+      "step": 262100
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 0.0005,
+      "loss": 1.6862,
+      "step": 262200
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 0.0005,
+      "loss": 1.7075,
+      "step": 262300
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 0.0005,
+      "loss": 1.6938,
+      "step": 262400
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 0.0005,
+      "loss": 1.7088,
+      "step": 262500
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 0.0005,
+      "loss": 1.6599,
+      "step": 262600
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6652,
+      "step": 262700
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 0.0005,
+      "loss": 1.6543,
+      "step": 262800
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 0.0005,
+      "loss": 1.6803,
+      "step": 262900
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 0.0005,
+      "loss": 1.6728,
+      "step": 263000
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 0.0005,
+      "loss": 1.6599,
+      "step": 263100
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 0.0005,
+      "loss": 1.6822,
+      "step": 263200
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 0.0005,
+      "loss": 1.6958,
+      "step": 263300
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7065,
+      "step": 263400
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 0.0005,
+      "loss": 1.7043,
+      "step": 263500
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 0.0005,
+      "loss": 1.6914,
+      "step": 263600
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 0.0005,
+      "loss": 1.701,
+      "step": 263700
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6959,
+      "step": 263800
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6852,
+      "step": 263900
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 0.0005,
+      "loss": 1.6723,
+      "step": 264000
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6832,
+      "step": 264100
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 0.0005,
+      "loss": 1.65,
+      "step": 264200
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6954,
+      "step": 264300
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6687,
+      "step": 264400
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 0.0005,
+      "loss": 1.693,
+      "step": 264500
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 0.0005,
+      "loss": 1.6815,
+      "step": 264600
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 0.0005,
+      "loss": 1.6688,
+      "step": 264700
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 0.0005,
+      "loss": 1.6602,
+      "step": 264800
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 0.0005,
+      "loss": 1.6585,
+      "step": 264900
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6895,
+      "step": 265000
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 0.0005,
+      "loss": 1.6925,
+      "step": 265100
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7074,
+      "step": 265200
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7013,
+      "step": 265300
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 0.0005,
+      "loss": 1.6727,
+      "step": 265400
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 0.0005,
+      "loss": 1.6998,
+      "step": 265500
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 0.0005,
+      "loss": 1.687,
+      "step": 265600
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6983,
+      "step": 265700
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 0.0005,
+      "loss": 1.7113,
+      "step": 265800
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 0.0005,
+      "loss": 1.6944,
+      "step": 265900
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 0.0005,
+      "loss": 1.7157,
+      "step": 266000
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 0.0005,
+      "loss": 1.7116,
+      "step": 266100
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 0.0005,
+      "loss": 1.7307,
+      "step": 266200
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 0.0005,
+      "loss": 1.6907,
+      "step": 266300
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6861,
+      "step": 266400
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 0.0005,
+      "loss": 1.6892,
+      "step": 266500
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 0.0005,
+      "loss": 1.6904,
+      "step": 266600
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 0.0005,
+      "loss": 1.7172,
+      "step": 266700
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 0.0005,
+      "loss": 1.7,
+      "step": 266800
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6893,
+      "step": 266900
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 0.0005,
+      "loss": 1.6898,
+      "step": 267000
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 0.0005,
+      "loss": 1.7073,
+      "step": 267100
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 0.0005,
+      "loss": 1.703,
+      "step": 267200
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6952,
+      "step": 267300
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 0.0005,
+      "loss": 1.6893,
+      "step": 267400
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 0.0005,
+      "loss": 1.7366,
+      "step": 267500
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 0.0005,
+      "loss": 1.7323,
+      "step": 267600
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 0.0005,
+      "loss": 1.7095,
+      "step": 267700
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6746,
+      "step": 267800
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5848,
+      "step": 267900
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 0.0005,
+      "loss": 1.64,
+      "step": 268000
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 0.0005,
+      "loss": 1.6432,
+      "step": 268100
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 0.0005,
+      "loss": 1.6222,
+      "step": 268200
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 0.0005,
+      "loss": 1.6085,
+      "step": 268300
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 0.0005,
+      "loss": 1.6059,
+      "step": 268400
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 0.0005,
+      "loss": 1.6361,
+      "step": 268500
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 0.0005,
+      "loss": 1.6125,
+      "step": 268600
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 0.0005,
+      "loss": 1.6033,
+      "step": 268700
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 0.0005,
+      "loss": 1.6163,
+      "step": 268800
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 0.0005,
+      "loss": 1.6157,
+      "step": 268900
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 0.0005,
+      "loss": 1.6505,
+      "step": 269000
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 0.0005,
+      "loss": 1.6267,
+      "step": 269100
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 0.0005,
+      "loss": 1.6007,
+      "step": 269200
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 0.0005,
+      "loss": 1.6224,
+      "step": 269300
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 0.0005,
+      "loss": 1.6151,
+      "step": 269400
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 0.0005,
+      "loss": 1.6311,
+      "step": 269500
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 0.0005,
+      "loss": 1.6143,
+      "step": 269600
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 0.0005,
+      "loss": 1.6499,
+      "step": 269700
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 0.0005,
+      "loss": 1.6286,
+      "step": 269800
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 0.0005,
+      "loss": 1.6327,
+      "step": 269900
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 0.0005,
+      "loss": 1.6273,
+      "step": 270000
+    },
+    {
+      "epoch": 21.17,
+      "eval_gen_len": 18.794120243665578,
+      "eval_loss": 2.0332844257354736,
+      "eval_rouge1": 35.267,
+      "eval_rouge2": 14.1753,
+      "eval_rougeL": 28.973,
+      "eval_rougeLsum": 28.9735,
+      "eval_runtime": 364.1194,
+      "eval_samples_per_second": 31.108,
+      "eval_steps_per_second": 1.944,
+      "step": 270000
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 0.0005,
+      "loss": 1.6709,
+      "step": 270100
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 0.0005,
+      "loss": 1.6506,
+      "step": 270200
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 0.0005,
+      "loss": 1.6591,
+      "step": 270300
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 0.0005,
+      "loss": 1.6446,
+      "step": 270400
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 0.0005,
+      "loss": 1.6299,
+      "step": 270500
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 0.0005,
+      "loss": 1.6684,
+      "step": 270600
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 0.0005,
+      "loss": 1.6636,
+      "step": 270700
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 0.0005,
+      "loss": 1.6326,
+      "step": 270800
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 0.0005,
+      "loss": 1.6458,
+      "step": 270900
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 0.0005,
+      "loss": 1.6513,
+      "step": 271000
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 0.0005,
+      "loss": 1.6618,
+      "step": 271100
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 0.0005,
+      "loss": 1.6441,
+      "step": 271200
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 0.0005,
+      "loss": 1.64,
+      "step": 271300
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 0.0005,
+      "loss": 1.6454,
+      "step": 271400
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 0.0005,
+      "loss": 1.6246,
+      "step": 271500
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 0.0005,
+      "loss": 1.6632,
+      "step": 271600
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6478,
+      "step": 271700
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6651,
+      "step": 271800
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 0.0005,
+      "loss": 1.626,
+      "step": 271900
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6519,
+      "step": 272000
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 0.0005,
+      "loss": 1.66,
+      "step": 272100
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 0.0005,
+      "loss": 1.6479,
+      "step": 272200
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 0.0005,
+      "loss": 1.6556,
+      "step": 272300
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 0.0005,
+      "loss": 1.6563,
+      "step": 272400
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 0.0005,
+      "loss": 1.6392,
+      "step": 272500
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 0.0005,
+      "loss": 1.6474,
+      "step": 272600
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 0.0005,
+      "loss": 1.6544,
+      "step": 272700
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 0.0005,
+      "loss": 1.6677,
+      "step": 272800
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 0.0005,
+      "loss": 1.6584,
+      "step": 272900
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 0.0005,
+      "loss": 1.656,
+      "step": 273000
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6487,
+      "step": 273100
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6614,
+      "step": 273200
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 0.0005,
+      "loss": 1.6839,
+      "step": 273300
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6518,
+      "step": 273400
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 0.0005,
+      "loss": 1.6572,
+      "step": 273500
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 0.0005,
+      "loss": 1.6604,
+      "step": 273600
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 0.0005,
+      "loss": 1.683,
+      "step": 273700
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 0.0005,
+      "loss": 1.6378,
+      "step": 273800
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 0.0005,
+      "loss": 1.6439,
+      "step": 273900
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 0.0005,
+      "loss": 1.6667,
+      "step": 274000
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 0.0005,
+      "loss": 1.6472,
+      "step": 274100
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 0.0005,
+      "loss": 1.639,
+      "step": 274200
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6757,
+      "step": 274300
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 0.0005,
+      "loss": 1.6633,
+      "step": 274400
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 0.0005,
+      "loss": 1.65,
+      "step": 274500
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 0.0005,
+      "loss": 1.67,
+      "step": 274600
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 0.0005,
+      "loss": 1.6726,
+      "step": 274700
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6634,
+      "step": 274800
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 0.0005,
+      "loss": 1.6844,
+      "step": 274900
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 0.0005,
+      "loss": 1.6708,
+      "step": 275000
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 0.0005,
+      "loss": 1.6485,
+      "step": 275100
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 0.0005,
+      "loss": 1.6648,
+      "step": 275200
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 0.0005,
+      "loss": 1.6691,
+      "step": 275300
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6751,
+      "step": 275400
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6635,
+      "step": 275500
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 0.0005,
+      "loss": 1.6693,
+      "step": 275600
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 0.0005,
+      "loss": 1.6764,
+      "step": 275700
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 0.0005,
+      "loss": 1.6629,
+      "step": 275800
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 0.0005,
+      "loss": 1.6837,
+      "step": 275900
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 0.0005,
+      "loss": 1.6856,
+      "step": 276000
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 0.0005,
+      "loss": 1.6608,
+      "step": 276100
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 0.0005,
+      "loss": 1.665,
+      "step": 276200
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 0.0005,
+      "loss": 1.6734,
+      "step": 276300
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 0.0005,
+      "loss": 1.6561,
+      "step": 276400
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 0.0005,
+      "loss": 1.6762,
+      "step": 276500
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6431,
+      "step": 276600
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 0.0005,
+      "loss": 1.6631,
+      "step": 276700
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 0.0005,
+      "loss": 1.7076,
+      "step": 276800
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6691,
+      "step": 276900
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 0.0005,
+      "loss": 1.6689,
+      "step": 277000
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6895,
+      "step": 277100
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 0.0005,
+      "loss": 1.701,
+      "step": 277200
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 0.0005,
+      "loss": 1.6691,
+      "step": 277300
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 0.0005,
+      "loss": 1.6708,
+      "step": 277400
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 0.0005,
+      "loss": 1.6708,
+      "step": 277500
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 0.0005,
+      "loss": 1.7038,
+      "step": 277600
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6813,
+      "step": 277700
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6738,
+      "step": 277800
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 0.0005,
+      "loss": 1.6991,
+      "step": 277900
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7033,
+      "step": 278000
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 0.0005,
+      "loss": 1.6716,
+      "step": 278100
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 0.0005,
+      "loss": 1.678,
+      "step": 278200
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 0.0005,
+      "loss": 1.6645,
+      "step": 278300
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 0.0005,
+      "loss": 1.6913,
+      "step": 278400
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6807,
+      "step": 278500
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 0.0005,
+      "loss": 1.6646,
+      "step": 278600
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 0.0005,
+      "loss": 1.6968,
+      "step": 278700
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 0.0005,
+      "loss": 1.6384,
+      "step": 278800
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6578,
+      "step": 278900
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 0.0005,
+      "loss": 1.6761,
+      "step": 279000
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6749,
+      "step": 279100
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6953,
+      "step": 279200
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 0.0005,
+      "loss": 1.6829,
+      "step": 279300
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 0.0005,
+      "loss": 1.6733,
+      "step": 279400
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 0.0005,
+      "loss": 1.6732,
+      "step": 279500
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6764,
+      "step": 279600
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6588,
+      "step": 279700
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 0.0005,
+      "loss": 1.6779,
+      "step": 279800
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 0.0005,
+      "loss": 1.6869,
+      "step": 279900
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 0.0005,
+      "loss": 1.7009,
+      "step": 280000
+    },
+    {
+      "epoch": 21.96,
+      "eval_gen_len": 18.80851063829787,
+      "eval_loss": 2.0156733989715576,
+      "eval_rouge1": 35.3219,
+      "eval_rouge2": 14.2115,
+      "eval_rougeL": 28.9797,
+      "eval_rougeLsum": 28.9688,
+      "eval_runtime": 361.1372,
+      "eval_samples_per_second": 31.365,
+      "eval_steps_per_second": 1.96,
+      "step": 280000
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 0.0005,
+      "loss": 1.6889,
+      "step": 280100
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 0.0005,
+      "loss": 1.6603,
+      "step": 280200
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6909,
+      "step": 280300
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 0.0005,
+      "loss": 1.7038,
+      "step": 280400
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6765,
+      "step": 280500
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6226,
+      "step": 280600
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 0.0005,
+      "loss": 1.6016,
+      "step": 280700
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 0.0005,
+      "loss": 1.6137,
+      "step": 280800
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 0.0005,
+      "loss": 1.6311,
+      "step": 280900
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 0.0005,
+      "loss": 1.6032,
+      "step": 281000
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 0.0005,
+      "loss": 1.586,
+      "step": 281100
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 0.0005,
+      "loss": 1.6228,
+      "step": 281200
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 0.0005,
+      "loss": 1.5936,
+      "step": 281300
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 0.0005,
+      "loss": 1.6262,
+      "step": 281400
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 0.0005,
+      "loss": 1.6091,
+      "step": 281500
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 0.0005,
+      "loss": 1.6138,
+      "step": 281600
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 0.0005,
+      "loss": 1.6125,
+      "step": 281700
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 0.0005,
+      "loss": 1.636,
+      "step": 281800
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5914,
+      "step": 281900
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 0.0005,
+      "loss": 1.6495,
+      "step": 282000
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 0.0005,
+      "loss": 1.6053,
+      "step": 282100
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 0.0005,
+      "loss": 1.6091,
+      "step": 282200
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 0.0005,
+      "loss": 1.6329,
+      "step": 282300
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 0.0005,
+      "loss": 1.6053,
+      "step": 282400
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 0.0005,
+      "loss": 1.5926,
+      "step": 282500
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5977,
+      "step": 282600
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 0.0005,
+      "loss": 1.59,
+      "step": 282700
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 0.0005,
+      "loss": 1.6346,
+      "step": 282800
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 0.0005,
+      "loss": 1.5948,
+      "step": 282900
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 0.0005,
+      "loss": 1.6221,
+      "step": 283000
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 0.0005,
+      "loss": 1.6059,
+      "step": 283100
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 0.0005,
+      "loss": 1.6255,
+      "step": 283200
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 0.0005,
+      "loss": 1.6016,
+      "step": 283300
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 0.0005,
+      "loss": 1.6287,
+      "step": 283400
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 0.0005,
+      "loss": 1.6254,
+      "step": 283500
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 0.0005,
+      "loss": 1.623,
+      "step": 283600
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 0.0005,
+      "loss": 1.6471,
+      "step": 283700
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 0.0005,
+      "loss": 1.6425,
+      "step": 283800
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 0.0005,
+      "loss": 1.6553,
+      "step": 283900
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 0.0005,
+      "loss": 1.6522,
+      "step": 284000
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 0.0005,
+      "loss": 1.6446,
+      "step": 284100
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 0.0005,
+      "loss": 1.6494,
+      "step": 284200
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 0.0005,
+      "loss": 1.6355,
+      "step": 284300
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 0.0005,
+      "loss": 1.612,
+      "step": 284400
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6425,
+      "step": 284500
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 0.0005,
+      "loss": 1.6426,
+      "step": 284600
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6647,
+      "step": 284700
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6259,
+      "step": 284800
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 0.0005,
+      "loss": 1.6348,
+      "step": 284900
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 0.0005,
+      "loss": 1.6271,
+      "step": 285000
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 0.0005,
+      "loss": 1.6255,
+      "step": 285100
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 0.0005,
+      "loss": 1.6273,
+      "step": 285200
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 0.0005,
+      "loss": 1.6208,
+      "step": 285300
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 0.0005,
+      "loss": 1.639,
+      "step": 285400
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 0.0005,
+      "loss": 1.6401,
+      "step": 285500
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 0.0005,
+      "loss": 1.631,
+      "step": 285600
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 0.0005,
+      "loss": 1.6275,
+      "step": 285700
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 0.0005,
+      "loss": 1.6289,
+      "step": 285800
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6153,
+      "step": 285900
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 0.0005,
+      "loss": 1.6571,
+      "step": 286000
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6038,
+      "step": 286100
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6535,
+      "step": 286200
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 0.0005,
+      "loss": 1.6414,
+      "step": 286300
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 0.0005,
+      "loss": 1.642,
+      "step": 286400
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 0.0005,
+      "loss": 1.6574,
+      "step": 286500
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 0.0005,
+      "loss": 1.612,
+      "step": 286600
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 0.0005,
+      "loss": 1.6388,
+      "step": 286700
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 0.0005,
+      "loss": 1.6577,
+      "step": 286800
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 0.0005,
+      "loss": 1.6306,
+      "step": 286900
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6493,
+      "step": 287000
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6496,
+      "step": 287100
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 0.0005,
+      "loss": 1.6655,
+      "step": 287200
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 0.0005,
+      "loss": 1.6347,
+      "step": 287300
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 0.0005,
+      "loss": 1.6628,
+      "step": 287400
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6369,
+      "step": 287500
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6818,
+      "step": 287600
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 0.0005,
+      "loss": 1.6497,
+      "step": 287700
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 0.0005,
+      "loss": 1.6406,
+      "step": 287800
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 0.0005,
+      "loss": 1.6353,
+      "step": 287900
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 0.0005,
+      "loss": 1.6637,
+      "step": 288000
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 0.0005,
+      "loss": 1.6416,
+      "step": 288100
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6283,
+      "step": 288200
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 0.0005,
+      "loss": 1.667,
+      "step": 288300
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 0.0005,
+      "loss": 1.6687,
+      "step": 288400
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 0.0005,
+      "loss": 1.6424,
+      "step": 288500
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 0.0005,
+      "loss": 1.661,
+      "step": 288600
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 0.0005,
+      "loss": 1.6544,
+      "step": 288700
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 0.0005,
+      "loss": 1.66,
+      "step": 288800
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 0.0005,
+      "loss": 1.661,
+      "step": 288900
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 0.0005,
+      "loss": 1.6509,
+      "step": 289000
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 0.0005,
+      "loss": 1.6761,
+      "step": 289100
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 0.0005,
+      "loss": 1.6578,
+      "step": 289200
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6454,
+      "step": 289300
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6606,
+      "step": 289400
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 0.0005,
+      "loss": 1.6326,
+      "step": 289500
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6758,
+      "step": 289600
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 0.0005,
+      "loss": 1.6564,
+      "step": 289700
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 0.0005,
+      "loss": 1.656,
+      "step": 289800
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6915,
+      "step": 289900
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 0.0005,
+      "loss": 1.6582,
+      "step": 290000
+    },
+    {
+      "epoch": 22.74,
+      "eval_gen_len": 18.805862099408493,
+      "eval_loss": 2.0187759399414062,
+      "eval_rouge1": 35.4028,
+      "eval_rouge2": 14.3182,
+      "eval_rougeL": 29.1245,
+      "eval_rougeLsum": 29.1239,
+      "eval_runtime": 358.1851,
+      "eval_samples_per_second": 31.623,
+      "eval_steps_per_second": 1.977,
+      "step": 290000
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 0.0005,
+      "loss": 1.6384,
+      "step": 290100
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 0.0005,
+      "loss": 1.6645,
+      "step": 290200
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 0.0005,
+      "loss": 1.6506,
+      "step": 290300
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 0.0005,
+      "loss": 1.6602,
+      "step": 290400
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6624,
+      "step": 290500
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 0.0005,
+      "loss": 1.661,
+      "step": 290600
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 0.0005,
+      "loss": 1.659,
+      "step": 290700
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 0.0005,
+      "loss": 1.7059,
+      "step": 290800
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 0.0005,
+      "loss": 1.6346,
+      "step": 290900
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 0.0005,
+      "loss": 1.6551,
+      "step": 291000
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 0.0005,
+      "loss": 1.6512,
+      "step": 291100
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6733,
+      "step": 291200
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6672,
+      "step": 291300
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 0.0005,
+      "loss": 1.6735,
+      "step": 291400
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 0.0005,
+      "loss": 1.6639,
+      "step": 291500
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6709,
+      "step": 291600
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6485,
+      "step": 291700
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 0.0005,
+      "loss": 1.6701,
+      "step": 291800
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6622,
+      "step": 291900
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 0.0005,
+      "loss": 1.6651,
+      "step": 292000
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 0.0005,
+      "loss": 1.6575,
+      "step": 292100
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 0.0005,
+      "loss": 1.661,
+      "step": 292200
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 0.0005,
+      "loss": 1.6728,
+      "step": 292300
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6599,
+      "step": 292400
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 0.0005,
+      "loss": 1.6628,
+      "step": 292500
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 0.0005,
+      "loss": 1.6632,
+      "step": 292600
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 0.0005,
+      "loss": 1.642,
+      "step": 292700
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6767,
+      "step": 292800
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 0.0005,
+      "loss": 1.6652,
+      "step": 292900
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6488,
+      "step": 293000
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6915,
+      "step": 293100
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 0.0005,
+      "loss": 1.6965,
+      "step": 293200
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6508,
+      "step": 293300
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5917,
+      "step": 293400
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5818,
+      "step": 293500
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 0.0005,
+      "loss": 1.6064,
+      "step": 293600
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 0.0005,
+      "loss": 1.6003,
+      "step": 293700
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 0.0005,
+      "loss": 1.5817,
+      "step": 293800
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5865,
+      "step": 293900
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 0.0005,
+      "loss": 1.5629,
+      "step": 294000
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 0.0005,
+      "loss": 1.5881,
+      "step": 294100
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5776,
+      "step": 294200
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 0.0005,
+      "loss": 1.59,
+      "step": 294300
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5971,
+      "step": 294400
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 0.0005,
+      "loss": 1.6124,
+      "step": 294500
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 0.0005,
+      "loss": 1.6203,
+      "step": 294600
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 0.0005,
+      "loss": 1.574,
+      "step": 294700
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 0.0005,
+      "loss": 1.6149,
+      "step": 294800
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 0.0005,
+      "loss": 1.6092,
+      "step": 294900
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5889,
+      "step": 295000
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5929,
+      "step": 295100
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 0.0005,
+      "loss": 1.6198,
+      "step": 295200
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5972,
+      "step": 295300
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5884,
+      "step": 295400
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 0.0005,
+      "loss": 1.5839,
+      "step": 295500
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 0.0005,
+      "loss": 1.6108,
+      "step": 295600
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 0.0005,
+      "loss": 1.5895,
+      "step": 295700
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5911,
+      "step": 295800
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5722,
+      "step": 295900
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 0.0005,
+      "loss": 1.6111,
+      "step": 296000
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 0.0005,
+      "loss": 1.6076,
+      "step": 296100
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 0.0005,
+      "loss": 1.6228,
+      "step": 296200
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 0.0005,
+      "loss": 1.5836,
+      "step": 296300
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 0.0005,
+      "loss": 1.616,
+      "step": 296400
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 0.0005,
+      "loss": 1.6038,
+      "step": 296500
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 0.0005,
+      "loss": 1.5998,
+      "step": 296600
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 0.0005,
+      "loss": 1.6057,
+      "step": 296700
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 0.0005,
+      "loss": 1.6069,
+      "step": 296800
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 0.0005,
+      "loss": 1.6165,
+      "step": 296900
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 0.0005,
+      "loss": 1.6075,
+      "step": 297000
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 0.0005,
+      "loss": 1.6274,
+      "step": 297100
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6295,
+      "step": 297200
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6176,
+      "step": 297300
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 0.0005,
+      "loss": 1.6193,
+      "step": 297400
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6231,
+      "step": 297500
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 0.0005,
+      "loss": 1.6155,
+      "step": 297600
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 0.0005,
+      "loss": 1.6239,
+      "step": 297700
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 0.0005,
+      "loss": 1.6217,
+      "step": 297800
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5893,
+      "step": 297900
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 0.0005,
+      "loss": 1.632,
+      "step": 298000
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 0.0005,
+      "loss": 1.614,
+      "step": 298100
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 0.0005,
+      "loss": 1.604,
+      "step": 298200
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 0.0005,
+      "loss": 1.6348,
+      "step": 298300
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 0.0005,
+      "loss": 1.6524,
+      "step": 298400
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 0.0005,
+      "loss": 1.6346,
+      "step": 298500
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6335,
+      "step": 298600
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6356,
+      "step": 298700
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 0.0005,
+      "loss": 1.6304,
+      "step": 298800
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6099,
+      "step": 298900
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 0.0005,
+      "loss": 1.6269,
+      "step": 299000
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 0.0005,
+      "loss": 1.6192,
+      "step": 299100
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 0.0005,
+      "loss": 1.6656,
+      "step": 299200
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 0.0005,
+      "loss": 1.6019,
+      "step": 299300
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 0.0005,
+      "loss": 1.6528,
+      "step": 299400
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 0.0005,
+      "loss": 1.6116,
+      "step": 299500
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 0.0005,
+      "loss": 1.6486,
+      "step": 299600
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 0.0005,
+      "loss": 1.6194,
+      "step": 299700
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6463,
+      "step": 299800
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 0.0005,
+      "loss": 1.628,
+      "step": 299900
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 0.0005,
+      "loss": 1.6391,
+      "step": 300000
+    },
+    {
+      "epoch": 23.53,
+      "eval_gen_len": 18.781583826255847,
+      "eval_loss": 2.024906635284424,
+      "eval_rouge1": 35.371,
+      "eval_rouge2": 14.234,
+      "eval_rougeL": 29.0754,
+      "eval_rougeLsum": 29.072,
+      "eval_runtime": 361.8269,
+      "eval_samples_per_second": 31.305,
+      "eval_steps_per_second": 1.957,
+      "step": 300000
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 0.0005,
+      "loss": 1.666,
+      "step": 300100
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 0.0005,
+      "loss": 1.6385,
+      "step": 300200
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6464,
+      "step": 300300
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 0.0005,
+      "loss": 1.6211,
+      "step": 300400
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 0.0005,
+      "loss": 1.6403,
+      "step": 300500
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 0.0005,
+      "loss": 1.6051,
+      "step": 300600
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 0.0005,
+      "loss": 1.6281,
+      "step": 300700
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 0.0005,
+      "loss": 1.628,
+      "step": 300800
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6414,
+      "step": 300900
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6388,
+      "step": 301000
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 0.0005,
+      "loss": 1.6349,
+      "step": 301100
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 0.0005,
+      "loss": 1.6436,
+      "step": 301200
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 0.0005,
+      "loss": 1.6358,
+      "step": 301300
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 0.0005,
+      "loss": 1.6356,
+      "step": 301400
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 0.0005,
+      "loss": 1.6494,
+      "step": 301500
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 0.0005,
+      "loss": 1.6575,
+      "step": 301600
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 0.0005,
+      "loss": 1.6476,
+      "step": 301700
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 0.0005,
+      "loss": 1.6487,
+      "step": 301800
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 0.0005,
+      "loss": 1.6258,
+      "step": 301900
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 0.0005,
+      "loss": 1.6489,
+      "step": 302000
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6275,
+      "step": 302100
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 0.0005,
+      "loss": 1.6157,
+      "step": 302200
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6603,
+      "step": 302300
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6366,
+      "step": 302400
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 0.0005,
+      "loss": 1.648,
+      "step": 302500
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6612,
+      "step": 302600
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 0.0005,
+      "loss": 1.6281,
+      "step": 302700
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 0.0005,
+      "loss": 1.6699,
+      "step": 302800
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 0.0005,
+      "loss": 1.6539,
+      "step": 302900
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 0.0005,
+      "loss": 1.6383,
+      "step": 303000
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 0.0005,
+      "loss": 1.6336,
+      "step": 303100
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6608,
+      "step": 303200
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6244,
+      "step": 303300
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 0.0005,
+      "loss": 1.6521,
+      "step": 303400
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 0.0005,
+      "loss": 1.6382,
+      "step": 303500
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 0.0005,
+      "loss": 1.6328,
+      "step": 303600
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 0.0005,
+      "loss": 1.6603,
+      "step": 303700
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 0.0005,
+      "loss": 1.6212,
+      "step": 303800
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 0.0005,
+      "loss": 1.6531,
+      "step": 303900
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6388,
+      "step": 304000
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 0.0005,
+      "loss": 1.65,
+      "step": 304100
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 0.0005,
+      "loss": 1.6484,
+      "step": 304200
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 0.0005,
+      "loss": 1.6867,
+      "step": 304300
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6901,
+      "step": 304400
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 0.0005,
+      "loss": 1.6416,
+      "step": 304500
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6623,
+      "step": 304600
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 0.0005,
+      "loss": 1.651,
+      "step": 304700
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 0.0005,
+      "loss": 1.6387,
+      "step": 304800
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 0.0005,
+      "loss": 1.6614,
+      "step": 304900
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 0.0005,
+      "loss": 1.6405,
+      "step": 305000
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6339,
+      "step": 305100
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6873,
+      "step": 305200
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 0.0005,
+      "loss": 1.6774,
+      "step": 305300
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 0.0005,
+      "loss": 1.6678,
+      "step": 305400
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6646,
+      "step": 305500
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6671,
+      "step": 305600
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 0.0005,
+      "loss": 1.6392,
+      "step": 305700
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6524,
+      "step": 305800
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 0.0005,
+      "loss": 1.6699,
+      "step": 305900
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6702,
+      "step": 306000
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6122,
+      "step": 306100
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5641,
+      "step": 306200
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5877,
+      "step": 306300
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 0.0005,
+      "loss": 1.5558,
+      "step": 306400
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 0.0005,
+      "loss": 1.5677,
+      "step": 306500
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 0.0005,
+      "loss": 1.5845,
+      "step": 306600
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 0.0005,
+      "loss": 1.6126,
+      "step": 306700
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 0.0005,
+      "loss": 1.5886,
+      "step": 306800
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5871,
+      "step": 306900
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5685,
+      "step": 307000
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 0.0005,
+      "loss": 1.5822,
+      "step": 307100
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5561,
+      "step": 307200
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 0.0005,
+      "loss": 1.5736,
+      "step": 307300
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 0.0005,
+      "loss": 1.6063,
+      "step": 307400
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5786,
+      "step": 307500
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 0.0005,
+      "loss": 1.5952,
+      "step": 307600
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5937,
+      "step": 307700
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5623,
+      "step": 307800
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 0.0005,
+      "loss": 1.5925,
+      "step": 307900
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 0.0005,
+      "loss": 1.5756,
+      "step": 308000
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5649,
+      "step": 308100
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 0.0005,
+      "loss": 1.5827,
+      "step": 308200
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 0.0005,
+      "loss": 1.5905,
+      "step": 308300
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 0.0005,
+      "loss": 1.6098,
+      "step": 308400
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 0.0005,
+      "loss": 1.5813,
+      "step": 308500
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5959,
+      "step": 308600
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 0.0005,
+      "loss": 1.602,
+      "step": 308700
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 0.0005,
+      "loss": 1.5968,
+      "step": 308800
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 0.0005,
+      "loss": 1.5857,
+      "step": 308900
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5816,
+      "step": 309000
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 0.0005,
+      "loss": 1.6208,
+      "step": 309100
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 0.0005,
+      "loss": 1.6071,
+      "step": 309200
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 0.0005,
+      "loss": 1.6001,
+      "step": 309300
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 0.0005,
+      "loss": 1.5837,
+      "step": 309400
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5884,
+      "step": 309500
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 0.0005,
+      "loss": 1.6021,
+      "step": 309600
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 0.0005,
+      "loss": 1.6263,
+      "step": 309700
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5973,
+      "step": 309800
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 0.0005,
+      "loss": 1.61,
+      "step": 309900
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 0.0005,
+      "loss": 1.6117,
+      "step": 310000
+    },
+    {
+      "epoch": 24.31,
+      "eval_gen_len": 18.821841617374414,
+      "eval_loss": 2.028640031814575,
+      "eval_rouge1": 35.3532,
+      "eval_rouge2": 14.2979,
+      "eval_rougeL": 29.0784,
+      "eval_rougeLsum": 29.0753,
+      "eval_runtime": 359.726,
+      "eval_samples_per_second": 31.488,
+      "eval_steps_per_second": 1.968,
+      "step": 310000
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 0.0005,
+      "loss": 1.6213,
+      "step": 310100
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6144,
+      "step": 310200
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6167,
+      "step": 310300
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 0.0005,
+      "loss": 1.5916,
+      "step": 310400
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 0.0005,
+      "loss": 1.6089,
+      "step": 310500
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 0.0005,
+      "loss": 1.6133,
+      "step": 310600
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5997,
+      "step": 310700
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 0.0005,
+      "loss": 1.5821,
+      "step": 310800
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5782,
+      "step": 310900
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 0.0005,
+      "loss": 1.6115,
+      "step": 311000
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 0.0005,
+      "loss": 1.586,
+      "step": 311100
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 0.0005,
+      "loss": 1.608,
+      "step": 311200
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5944,
+      "step": 311300
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 0.0005,
+      "loss": 1.6191,
+      "step": 311400
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5982,
+      "step": 311500
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6038,
+      "step": 311600
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6403,
+      "step": 311700
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 0.0005,
+      "loss": 1.6233,
+      "step": 311800
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 0.0005,
+      "loss": 1.6071,
+      "step": 311900
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 0.0005,
+      "loss": 1.6255,
+      "step": 312000
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 0.0005,
+      "loss": 1.6234,
+      "step": 312100
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 0.0005,
+      "loss": 1.6308,
+      "step": 312200
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 0.0005,
+      "loss": 1.6172,
+      "step": 312300
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 0.0005,
+      "loss": 1.6077,
+      "step": 312400
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6378,
+      "step": 312500
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6038,
+      "step": 312600
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 0.0005,
+      "loss": 1.6411,
+      "step": 312700
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 0.0005,
+      "loss": 1.6063,
+      "step": 312800
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 0.0005,
+      "loss": 1.638,
+      "step": 312900
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6137,
+      "step": 313000
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6173,
+      "step": 313100
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5994,
+      "step": 313200
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 0.0005,
+      "loss": 1.6321,
+      "step": 313300
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 0.0005,
+      "loss": 1.6248,
+      "step": 313400
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 0.0005,
+      "loss": 1.622,
+      "step": 313500
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 0.0005,
+      "loss": 1.623,
+      "step": 313600
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6185,
+      "step": 313700
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 0.0005,
+      "loss": 1.6278,
+      "step": 313800
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 0.0005,
+      "loss": 1.6302,
+      "step": 313900
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5845,
+      "step": 314000
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 0.0005,
+      "loss": 1.6256,
+      "step": 314100
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 0.0005,
+      "loss": 1.6112,
+      "step": 314200
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 0.0005,
+      "loss": 1.6205,
+      "step": 314300
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 0.0005,
+      "loss": 1.6244,
+      "step": 314400
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 0.0005,
+      "loss": 1.6834,
+      "step": 314500
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 0.0005,
+      "loss": 1.6155,
+      "step": 314600
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 0.0005,
+      "loss": 1.6176,
+      "step": 314700
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6207,
+      "step": 314800
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6335,
+      "step": 314900
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 0.0005,
+      "loss": 1.6422,
+      "step": 315000
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 0.0005,
+      "loss": 1.612,
+      "step": 315100
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 0.0005,
+      "loss": 1.6276,
+      "step": 315200
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6261,
+      "step": 315300
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 0.0005,
+      "loss": 1.6383,
+      "step": 315400
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 0.0005,
+      "loss": 1.6474,
+      "step": 315500
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 0.0005,
+      "loss": 1.6421,
+      "step": 315600
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 0.0005,
+      "loss": 1.6137,
+      "step": 315700
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 0.0005,
+      "loss": 1.6436,
+      "step": 315800
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 0.0005,
+      "loss": 1.6318,
+      "step": 315900
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6004,
+      "step": 316000
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 0.0005,
+      "loss": 1.6413,
+      "step": 316100
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5908,
+      "step": 316200
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 0.0005,
+      "loss": 1.6252,
+      "step": 316300
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 0.0005,
+      "loss": 1.6307,
+      "step": 316400
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 0.0005,
+      "loss": 1.6039,
+      "step": 316500
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 0.0005,
+      "loss": 1.635,
+      "step": 316600
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6595,
+      "step": 316700
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6373,
+      "step": 316800
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 0.0005,
+      "loss": 1.6513,
+      "step": 316900
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 0.0005,
+      "loss": 1.6329,
+      "step": 317000
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6245,
+      "step": 317100
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6224,
+      "step": 317200
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 0.0005,
+      "loss": 1.6333,
+      "step": 317300
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6248,
+      "step": 317400
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 0.0005,
+      "loss": 1.6371,
+      "step": 317500
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 0.0005,
+      "loss": 1.6373,
+      "step": 317600
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 0.0005,
+      "loss": 1.6378,
+      "step": 317700
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 0.0005,
+      "loss": 1.6357,
+      "step": 317800
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6325,
+      "step": 317900
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 0.0005,
+      "loss": 1.6113,
+      "step": 318000
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 0.0005,
+      "loss": 1.6174,
+      "step": 318100
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 0.0005,
+      "loss": 1.6451,
+      "step": 318200
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6558,
+      "step": 318300
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 0.0005,
+      "loss": 1.662,
+      "step": 318400
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6724,
+      "step": 318500
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6367,
+      "step": 318600
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 0.0005,
+      "loss": 1.6426,
+      "step": 318700
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6336,
+      "step": 318800
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5282,
+      "step": 318900
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5774,
+      "step": 319000
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5832,
+      "step": 319100
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 0.0005,
+      "loss": 1.5514,
+      "step": 319200
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 0.0005,
+      "loss": 1.5626,
+      "step": 319300
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5512,
+      "step": 319400
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5384,
+      "step": 319500
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 0.0005,
+      "loss": 1.5862,
+      "step": 319600
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5757,
+      "step": 319700
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 0.0005,
+      "loss": 1.5636,
+      "step": 319800
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5657,
+      "step": 319900
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5626,
+      "step": 320000
+    },
+    {
+      "epoch": 25.09,
+      "eval_gen_len": 18.80259556811159,
+      "eval_loss": 2.050901174545288,
+      "eval_rouge1": 35.454,
+      "eval_rouge2": 14.3262,
+      "eval_rougeL": 29.1638,
+      "eval_rougeLsum": 29.1575,
+      "eval_runtime": 378.8445,
+      "eval_samples_per_second": 29.899,
+      "eval_steps_per_second": 1.869,
+      "step": 320000
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 0.0005,
+      "loss": 1.5716,
+      "step": 320100
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5771,
+      "step": 320200
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 0.0005,
+      "loss": 1.5626,
+      "step": 320300
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5642,
+      "step": 320400
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5571,
+      "step": 320500
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5582,
+      "step": 320600
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 0.0005,
+      "loss": 1.5973,
+      "step": 320700
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5949,
+      "step": 320800
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 0.0005,
+      "loss": 1.6062,
+      "step": 320900
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 0.0005,
+      "loss": 1.5625,
+      "step": 321000
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 0.0005,
+      "loss": 1.6049,
+      "step": 321100
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 0.0005,
+      "loss": 1.5804,
+      "step": 321200
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5686,
+      "step": 321300
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5573,
+      "step": 321400
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 0.0005,
+      "loss": 1.5394,
+      "step": 321500
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 0.0005,
+      "loss": 1.5871,
+      "step": 321600
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5794,
+      "step": 321700
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 0.0005,
+      "loss": 1.5796,
+      "step": 321800
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 0.0005,
+      "loss": 1.581,
+      "step": 321900
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 0.0005,
+      "loss": 1.5899,
+      "step": 322000
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 0.0005,
+      "loss": 1.5909,
+      "step": 322100
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5914,
+      "step": 322200
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 0.0005,
+      "loss": 1.564,
+      "step": 322300
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 0.0005,
+      "loss": 1.595,
+      "step": 322400
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5983,
+      "step": 322500
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 0.0005,
+      "loss": 1.5696,
+      "step": 322600
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 0.0005,
+      "loss": 1.5909,
+      "step": 322700
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 0.0005,
+      "loss": 1.5952,
+      "step": 322800
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5782,
+      "step": 322900
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 0.0005,
+      "loss": 1.6011,
+      "step": 323000
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 0.0005,
+      "loss": 1.5609,
+      "step": 323100
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5914,
+      "step": 323200
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5832,
+      "step": 323300
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5871,
+      "step": 323400
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 0.0005,
+      "loss": 1.5713,
+      "step": 323500
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 0.0005,
+      "loss": 1.6133,
+      "step": 323600
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 0.0005,
+      "loss": 1.6008,
+      "step": 323700
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 0.0005,
+      "loss": 1.6153,
+      "step": 323800
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 0.0005,
+      "loss": 1.6015,
+      "step": 323900
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5809,
+      "step": 324000
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5851,
+      "step": 324100
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5882,
+      "step": 324200
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5865,
+      "step": 324300
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 0.0005,
+      "loss": 1.6289,
+      "step": 324400
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 0.0005,
+      "loss": 1.5973,
+      "step": 324500
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 0.0005,
+      "loss": 1.6141,
+      "step": 324600
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 0.0005,
+      "loss": 1.5939,
+      "step": 324700
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5722,
+      "step": 324800
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 0.0005,
+      "loss": 1.587,
+      "step": 324900
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5885,
+      "step": 325000
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 0.0005,
+      "loss": 1.6158,
+      "step": 325100
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 0.0005,
+      "loss": 1.5998,
+      "step": 325200
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 0.0005,
+      "loss": 1.6076,
+      "step": 325300
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 0.0005,
+      "loss": 1.6122,
+      "step": 325400
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 0.0005,
+      "loss": 1.598,
+      "step": 325500
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 0.0005,
+      "loss": 1.6054,
+      "step": 325600
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 0.0005,
+      "loss": 1.6074,
+      "step": 325700
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 0.0005,
+      "loss": 1.5875,
+      "step": 325800
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5863,
+      "step": 325900
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 0.0005,
+      "loss": 1.6041,
+      "step": 326000
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 0.0005,
+      "loss": 1.6147,
+      "step": 326100
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 0.0005,
+      "loss": 1.6279,
+      "step": 326200
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 0.0005,
+      "loss": 1.6338,
+      "step": 326300
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6134,
+      "step": 326400
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5858,
+      "step": 326500
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 0.0005,
+      "loss": 1.6179,
+      "step": 326600
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 0.0005,
+      "loss": 1.6083,
+      "step": 326700
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 0.0005,
+      "loss": 1.6161,
+      "step": 326800
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 0.0005,
+      "loss": 1.6253,
+      "step": 326900
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 0.0005,
+      "loss": 1.6429,
+      "step": 327000
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 0.0005,
+      "loss": 1.6,
+      "step": 327100
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 0.0005,
+      "loss": 1.5922,
+      "step": 327200
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 0.0005,
+      "loss": 1.6395,
+      "step": 327300
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 0.0005,
+      "loss": 1.61,
+      "step": 327400
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 0.0005,
+      "loss": 1.6241,
+      "step": 327500
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 0.0005,
+      "loss": 1.6458,
+      "step": 327600
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 0.0005,
+      "loss": 1.597,
+      "step": 327700
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6225,
+      "step": 327800
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6317,
+      "step": 327900
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 0.0005,
+      "loss": 1.6394,
+      "step": 328000
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5723,
+      "step": 328100
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 0.0005,
+      "loss": 1.6114,
+      "step": 328200
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5943,
+      "step": 328300
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 0.0005,
+      "loss": 1.6357,
+      "step": 328400
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5943,
+      "step": 328500
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 0.0005,
+      "loss": 1.6326,
+      "step": 328600
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 0.0005,
+      "loss": 1.62,
+      "step": 328700
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6162,
+      "step": 328800
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 0.0005,
+      "loss": 1.6115,
+      "step": 328900
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5891,
+      "step": 329000
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 0.0005,
+      "loss": 1.6339,
+      "step": 329100
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 0.0005,
+      "loss": 1.6204,
+      "step": 329200
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 0.0005,
+      "loss": 1.6211,
+      "step": 329300
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 0.0005,
+      "loss": 1.6182,
+      "step": 329400
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5841,
+      "step": 329500
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 0.0005,
+      "loss": 1.6014,
+      "step": 329600
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 0.0005,
+      "loss": 1.6241,
+      "step": 329700
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 0.0005,
+      "loss": 1.6218,
+      "step": 329800
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6071,
+      "step": 329900
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 0.0005,
+      "loss": 1.6151,
+      "step": 330000
+    },
+    {
+      "epoch": 25.88,
+      "eval_gen_len": 18.749359936435066,
+      "eval_loss": 2.0329372882843018,
+      "eval_rouge1": 35.4835,
+      "eval_rouge2": 14.3358,
+      "eval_rougeL": 29.1764,
+      "eval_rougeLsum": 29.1679,
+      "eval_runtime": 354.0201,
+      "eval_samples_per_second": 31.995,
+      "eval_steps_per_second": 2.0,
+      "step": 330000
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6346,
+      "step": 330100
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5966,
+      "step": 330200
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 0.0005,
+      "loss": 1.6239,
+      "step": 330300
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 0.0005,
+      "loss": 1.6197,
+      "step": 330400
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 0.0005,
+      "loss": 1.6218,
+      "step": 330500
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6206,
+      "step": 330600
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 0.0005,
+      "loss": 1.627,
+      "step": 330700
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 0.0005,
+      "loss": 1.6337,
+      "step": 330800
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 0.0005,
+      "loss": 1.6318,
+      "step": 330900
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6188,
+      "step": 331000
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6249,
+      "step": 331100
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 0.0005,
+      "loss": 1.6343,
+      "step": 331200
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 0.0005,
+      "loss": 1.629,
+      "step": 331300
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 0.0005,
+      "loss": 1.611,
+      "step": 331400
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6585,
+      "step": 331500
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5997,
+      "step": 331600
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5408,
+      "step": 331700
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5374,
+      "step": 331800
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 0.0005,
+      "loss": 1.5412,
+      "step": 331900
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 0.0005,
+      "loss": 1.5719,
+      "step": 332000
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 0.0005,
+      "loss": 1.5398,
+      "step": 332100
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5591,
+      "step": 332200
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 0.0005,
+      "loss": 1.5313,
+      "step": 332300
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5362,
+      "step": 332400
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5486,
+      "step": 332500
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 0.0005,
+      "loss": 1.5609,
+      "step": 332600
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5272,
+      "step": 332700
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 0.0005,
+      "loss": 1.5446,
+      "step": 332800
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5752,
+      "step": 332900
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5672,
+      "step": 333000
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 0.0005,
+      "loss": 1.5693,
+      "step": 333100
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5406,
+      "step": 333200
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5762,
+      "step": 333300
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5586,
+      "step": 333400
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 0.0005,
+      "loss": 1.5743,
+      "step": 333500
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5519,
+      "step": 333600
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 0.0005,
+      "loss": 1.5945,
+      "step": 333700
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 0.0005,
+      "loss": 1.5771,
+      "step": 333800
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 0.0005,
+      "loss": 1.5625,
+      "step": 333900
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 0.0005,
+      "loss": 1.5387,
+      "step": 334000
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5574,
+      "step": 334100
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 0.0005,
+      "loss": 1.562,
+      "step": 334200
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 0.0005,
+      "loss": 1.5775,
+      "step": 334300
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 0.0005,
+      "loss": 1.5707,
+      "step": 334400
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5796,
+      "step": 334500
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 0.0005,
+      "loss": 1.5527,
+      "step": 334600
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 0.0005,
+      "loss": 1.5768,
+      "step": 334700
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 0.0005,
+      "loss": 1.5504,
+      "step": 334800
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 0.0005,
+      "loss": 1.5557,
+      "step": 334900
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5739,
+      "step": 335000
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 0.0005,
+      "loss": 1.5477,
+      "step": 335100
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 0.0005,
+      "loss": 1.566,
+      "step": 335200
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5769,
+      "step": 335300
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 0.0005,
+      "loss": 1.5689,
+      "step": 335400
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 0.0005,
+      "loss": 1.5647,
+      "step": 335500
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5846,
+      "step": 335600
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 0.0005,
+      "loss": 1.5499,
+      "step": 335700
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 0.0005,
+      "loss": 1.5697,
+      "step": 335800
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 0.0005,
+      "loss": 1.6024,
+      "step": 335900
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5963,
+      "step": 336000
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5966,
+      "step": 336100
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5425,
+      "step": 336200
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 0.0005,
+      "loss": 1.6052,
+      "step": 336300
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5511,
+      "step": 336400
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 0.0005,
+      "loss": 1.57,
+      "step": 336500
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5823,
+      "step": 336600
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5938,
+      "step": 336700
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5801,
+      "step": 336800
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5825,
+      "step": 336900
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5734,
+      "step": 337000
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 0.0005,
+      "loss": 1.5825,
+      "step": 337100
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 0.0005,
+      "loss": 1.5665,
+      "step": 337200
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 0.0005,
+      "loss": 1.6113,
+      "step": 337300
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 0.0005,
+      "loss": 1.6016,
+      "step": 337400
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5879,
+      "step": 337500
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5739,
+      "step": 337600
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 0.0005,
+      "loss": 1.5764,
+      "step": 337700
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 0.0005,
+      "loss": 1.6103,
+      "step": 337800
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 0.0005,
+      "loss": 1.5855,
+      "step": 337900
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5896,
+      "step": 338000
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5746,
+      "step": 338100
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 0.0005,
+      "loss": 1.579,
+      "step": 338200
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 0.0005,
+      "loss": 1.614,
+      "step": 338300
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 0.0005,
+      "loss": 1.6163,
+      "step": 338400
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5803,
+      "step": 338500
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 0.0005,
+      "loss": 1.6032,
+      "step": 338600
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5849,
+      "step": 338700
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 0.0005,
+      "loss": 1.6291,
+      "step": 338800
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5912,
+      "step": 338900
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 0.0005,
+      "loss": 1.6117,
+      "step": 339000
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 0.0005,
+      "loss": 1.5926,
+      "step": 339100
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5644,
+      "step": 339200
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 0.0005,
+      "loss": 1.6154,
+      "step": 339300
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5896,
+      "step": 339400
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5688,
+      "step": 339500
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 0.0005,
+      "loss": 1.6337,
+      "step": 339600
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 0.0005,
+      "loss": 1.5815,
+      "step": 339700
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 0.0005,
+      "loss": 1.6126,
+      "step": 339800
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5892,
+      "step": 339900
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 0.0005,
+      "loss": 1.6247,
+      "step": 340000
+    },
+    {
+      "epoch": 26.66,
+      "eval_gen_len": 18.760395515140814,
+      "eval_loss": 2.032465696334839,
+      "eval_rouge1": 35.301,
+      "eval_rouge2": 14.2909,
+      "eval_rougeL": 29.0546,
+      "eval_rougeLsum": 29.0366,
+      "eval_runtime": 354.3826,
+      "eval_samples_per_second": 31.963,
+      "eval_steps_per_second": 1.998,
+      "step": 340000
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 0.0005,
+      "loss": 1.6087,
+      "step": 340100
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 0.0005,
+      "loss": 1.5795,
+      "step": 340200
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5813,
+      "step": 340300
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5897,
+      "step": 340400
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 0.0005,
+      "loss": 1.6026,
+      "step": 340500
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 0.0005,
+      "loss": 1.6048,
+      "step": 340600
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 0.0005,
+      "loss": 1.6041,
+      "step": 340700
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5818,
+      "step": 340800
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5973,
+      "step": 340900
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 0.0005,
+      "loss": 1.6304,
+      "step": 341000
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5943,
+      "step": 341100
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 0.0005,
+      "loss": 1.604,
+      "step": 341200
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5891,
+      "step": 341300
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 0.0005,
+      "loss": 1.6235,
+      "step": 341400
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6246,
+      "step": 341500
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5878,
+      "step": 341600
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 0.0005,
+      "loss": 1.6338,
+      "step": 341700
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5986,
+      "step": 341800
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5755,
+      "step": 341900
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5932,
+      "step": 342000
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 0.0005,
+      "loss": 1.6145,
+      "step": 342100
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5958,
+      "step": 342200
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5869,
+      "step": 342300
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 0.0005,
+      "loss": 1.6158,
+      "step": 342400
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 0.0005,
+      "loss": 1.6216,
+      "step": 342500
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6138,
+      "step": 342600
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5909,
+      "step": 342700
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 0.0005,
+      "loss": 1.6228,
+      "step": 342800
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 0.0005,
+      "loss": 1.6369,
+      "step": 342900
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 0.0005,
+      "loss": 1.5997,
+      "step": 343000
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 0.0005,
+      "loss": 1.6028,
+      "step": 343100
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5992,
+      "step": 343200
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 0.0005,
+      "loss": 1.6149,
+      "step": 343300
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6065,
+      "step": 343400
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5902,
+      "step": 343500
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 0.0005,
+      "loss": 1.633,
+      "step": 343600
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 0.0005,
+      "loss": 1.5864,
+      "step": 343700
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6256,
+      "step": 343800
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 0.0005,
+      "loss": 1.6103,
+      "step": 343900
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6103,
+      "step": 344000
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6147,
+      "step": 344100
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 0.0005,
+      "loss": 1.6163,
+      "step": 344200
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 0.0005,
+      "loss": 1.6354,
+      "step": 344300
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5376,
+      "step": 344400
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5317,
+      "step": 344500
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5318,
+      "step": 344600
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 0.0005,
+      "loss": 1.5563,
+      "step": 344700
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 0.0005,
+      "loss": 1.5286,
+      "step": 344800
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 0.0005,
+      "loss": 1.519,
+      "step": 344900
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5067,
+      "step": 345000
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 0.0005,
+      "loss": 1.5485,
+      "step": 345100
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5473,
+      "step": 345200
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 0.0005,
+      "loss": 1.5467,
+      "step": 345300
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5235,
+      "step": 345400
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5264,
+      "step": 345500
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 0.0005,
+      "loss": 1.5376,
+      "step": 345600
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5383,
+      "step": 345700
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 0.0005,
+      "loss": 1.5499,
+      "step": 345800
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5365,
+      "step": 345900
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5262,
+      "step": 346000
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5533,
+      "step": 346100
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 0.0005,
+      "loss": 1.5306,
+      "step": 346200
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5767,
+      "step": 346300
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5315,
+      "step": 346400
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 0.0005,
+      "loss": 1.5352,
+      "step": 346500
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 0.0005,
+      "loss": 1.534,
+      "step": 346600
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 0.0005,
+      "loss": 1.5646,
+      "step": 346700
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5394,
+      "step": 346800
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5655,
+      "step": 346900
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 0.0005,
+      "loss": 1.5497,
+      "step": 347000
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 0.0005,
+      "loss": 1.56,
+      "step": 347100
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5485,
+      "step": 347200
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5658,
+      "step": 347300
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 0.0005,
+      "loss": 1.5506,
+      "step": 347400
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 0.0005,
+      "loss": 1.5561,
+      "step": 347500
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 0.0005,
+      "loss": 1.5674,
+      "step": 347600
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5509,
+      "step": 347700
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5593,
+      "step": 347800
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 0.0005,
+      "loss": 1.557,
+      "step": 347900
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5817,
+      "step": 348000
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 0.0005,
+      "loss": 1.5781,
+      "step": 348100
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 0.0005,
+      "loss": 1.574,
+      "step": 348200
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 0.0005,
+      "loss": 1.5613,
+      "step": 348300
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5685,
+      "step": 348400
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 0.0005,
+      "loss": 1.5287,
+      "step": 348500
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 0.0005,
+      "loss": 1.5573,
+      "step": 348600
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 0.0005,
+      "loss": 1.565,
+      "step": 348700
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5758,
+      "step": 348800
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5418,
+      "step": 348900
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 0.0005,
+      "loss": 1.557,
+      "step": 349000
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5715,
+      "step": 349100
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5413,
+      "step": 349200
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5829,
+      "step": 349300
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 0.0005,
+      "loss": 1.6001,
+      "step": 349400
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5802,
+      "step": 349500
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5757,
+      "step": 349600
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 0.0005,
+      "loss": 1.566,
+      "step": 349700
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5486,
+      "step": 349800
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 0.0005,
+      "loss": 1.5868,
+      "step": 349900
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 0.0005,
+      "loss": 1.5729,
+      "step": 350000
+    },
+    {
+      "epoch": 27.45,
+      "eval_gen_len": 18.799858744592566,
+      "eval_loss": 2.0437819957733154,
+      "eval_rouge1": 35.5653,
+      "eval_rouge2": 14.397,
+      "eval_rougeL": 29.2537,
+      "eval_rougeLsum": 29.2463,
+      "eval_runtime": 364.5616,
+      "eval_samples_per_second": 31.07,
+      "eval_steps_per_second": 1.942,
+      "step": 350000
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 0.0005,
+      "loss": 1.567,
+      "step": 350100
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 0.0005,
+      "loss": 1.5654,
+      "step": 350200
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5725,
+      "step": 350300
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 0.0005,
+      "loss": 1.5994,
+      "step": 350400
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5819,
+      "step": 350500
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5937,
+      "step": 350600
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 0.0005,
+      "loss": 1.5933,
+      "step": 350700
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5684,
+      "step": 350800
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5684,
+      "step": 350900
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 0.0005,
+      "loss": 1.5823,
+      "step": 351000
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 0.0005,
+      "loss": 1.5831,
+      "step": 351100
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5852,
+      "step": 351200
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 0.0005,
+      "loss": 1.5816,
+      "step": 351300
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5839,
+      "step": 351400
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5611,
+      "step": 351500
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 0.0005,
+      "loss": 1.5772,
+      "step": 351600
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5868,
+      "step": 351700
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 0.0005,
+      "loss": 1.5839,
+      "step": 351800
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5971,
+      "step": 351900
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 0.0005,
+      "loss": 1.6076,
+      "step": 352000
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5971,
+      "step": 352100
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5659,
+      "step": 352200
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 0.0005,
+      "loss": 1.589,
+      "step": 352300
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 0.0005,
+      "loss": 1.5862,
+      "step": 352400
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 0.0005,
+      "loss": 1.5739,
+      "step": 352500
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 0.0005,
+      "loss": 1.6004,
+      "step": 352600
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 0.0005,
+      "loss": 1.558,
+      "step": 352700
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5733,
+      "step": 352800
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 0.0005,
+      "loss": 1.573,
+      "step": 352900
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 0.0005,
+      "loss": 1.5761,
+      "step": 353000
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5865,
+      "step": 353100
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5739,
+      "step": 353200
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5902,
+      "step": 353300
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5813,
+      "step": 353400
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 0.0005,
+      "loss": 1.5722,
+      "step": 353500
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5662,
+      "step": 353600
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5982,
+      "step": 353700
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5977,
+      "step": 353800
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5929,
+      "step": 353900
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5885,
+      "step": 354000
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 0.0005,
+      "loss": 1.6001,
+      "step": 354100
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6063,
+      "step": 354200
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5976,
+      "step": 354300
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5982,
+      "step": 354400
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 0.0005,
+      "loss": 1.603,
+      "step": 354500
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5693,
+      "step": 354600
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5969,
+      "step": 354700
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5818,
+      "step": 354800
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5917,
+      "step": 354900
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 0.0005,
+      "loss": 1.6187,
+      "step": 355000
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5905,
+      "step": 355100
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5871,
+      "step": 355200
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 0.0005,
+      "loss": 1.5929,
+      "step": 355300
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6266,
+      "step": 355400
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5843,
+      "step": 355500
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5929,
+      "step": 355600
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5763,
+      "step": 355700
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 0.0005,
+      "loss": 1.5725,
+      "step": 355800
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5876,
+      "step": 355900
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 0.0005,
+      "loss": 1.6009,
+      "step": 356000
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5801,
+      "step": 356100
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 0.0005,
+      "loss": 1.6225,
+      "step": 356200
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 0.0005,
+      "loss": 1.6086,
+      "step": 356300
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 0.0005,
+      "loss": 1.5756,
+      "step": 356400
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6061,
+      "step": 356500
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 0.0005,
+      "loss": 1.6086,
+      "step": 356600
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 0.0005,
+      "loss": 1.573,
+      "step": 356700
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6276,
+      "step": 356800
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5962,
+      "step": 356900
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5872,
+      "step": 357000
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5682,
+      "step": 357100
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5024,
+      "step": 357200
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5107,
+      "step": 357300
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 0.0005,
+      "loss": 1.5163,
+      "step": 357400
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 0.0005,
+      "loss": 1.526,
+      "step": 357500
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 0.0005,
+      "loss": 1.5352,
+      "step": 357600
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5269,
+      "step": 357700
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 0.0005,
+      "loss": 1.5396,
+      "step": 357800
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 0.0005,
+      "loss": 1.523,
+      "step": 357900
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5314,
+      "step": 358000
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 0.0005,
+      "loss": 1.525,
+      "step": 358100
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5458,
+      "step": 358200
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 0.0005,
+      "loss": 1.5383,
+      "step": 358300
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5293,
+      "step": 358400
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5114,
+      "step": 358500
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 0.0005,
+      "loss": 1.5141,
+      "step": 358600
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5213,
+      "step": 358700
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5199,
+      "step": 358800
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 0.0005,
+      "loss": 1.542,
+      "step": 358900
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 0.0005,
+      "loss": 1.5349,
+      "step": 359000
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 0.0005,
+      "loss": 1.494,
+      "step": 359100
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 0.0005,
+      "loss": 1.5497,
+      "step": 359200
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 0.0005,
+      "loss": 1.5573,
+      "step": 359300
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 0.0005,
+      "loss": 1.533,
+      "step": 359400
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 0.0005,
+      "loss": 1.5426,
+      "step": 359500
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5489,
+      "step": 359600
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 0.0005,
+      "loss": 1.5595,
+      "step": 359700
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 0.0005,
+      "loss": 1.5472,
+      "step": 359800
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 0.0005,
+      "loss": 1.5507,
+      "step": 359900
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 0.0005,
+      "loss": 1.558,
+      "step": 360000
+    },
+    {
+      "epoch": 28.23,
+      "eval_gen_len": 18.821665048115122,
+      "eval_loss": 2.0342040061950684,
+      "eval_rouge1": 35.4848,
+      "eval_rouge2": 14.3471,
+      "eval_rougeL": 29.1774,
+      "eval_rougeLsum": 29.1699,
+      "eval_runtime": 355.9312,
+      "eval_samples_per_second": 31.824,
+      "eval_steps_per_second": 1.989,
+      "step": 360000
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 0.0005,
+      "loss": 1.5385,
+      "step": 360100
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 0.0005,
+      "loss": 1.5381,
+      "step": 360200
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 0.0005,
+      "loss": 1.5552,
+      "step": 360300
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 0.0005,
+      "loss": 1.5452,
+      "step": 360400
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5726,
+      "step": 360500
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 0.0005,
+      "loss": 1.5543,
+      "step": 360600
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5459,
+      "step": 360700
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5494,
+      "step": 360800
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 0.0005,
+      "loss": 1.5852,
+      "step": 360900
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 0.0005,
+      "loss": 1.5807,
+      "step": 361000
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5449,
+      "step": 361100
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5605,
+      "step": 361200
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 0.0005,
+      "loss": 1.5405,
+      "step": 361300
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 0.0005,
+      "loss": 1.5573,
+      "step": 361400
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5437,
+      "step": 361500
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5475,
+      "step": 361600
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5543,
+      "step": 361700
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 0.0005,
+      "loss": 1.5673,
+      "step": 361800
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5429,
+      "step": 361900
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5304,
+      "step": 362000
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5438,
+      "step": 362100
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5253,
+      "step": 362200
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5658,
+      "step": 362300
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5737,
+      "step": 362400
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5934,
+      "step": 362500
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5434,
+      "step": 362600
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 0.0005,
+      "loss": 1.5473,
+      "step": 362700
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 0.0005,
+      "loss": 1.5754,
+      "step": 362800
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 0.0005,
+      "loss": 1.5531,
+      "step": 362900
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5496,
+      "step": 363000
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5591,
+      "step": 363100
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 0.0005,
+      "loss": 1.5777,
+      "step": 363200
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5633,
+      "step": 363300
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 0.0005,
+      "loss": 1.569,
+      "step": 363400
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5578,
+      "step": 363500
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5764,
+      "step": 363600
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5284,
+      "step": 363700
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 0.0005,
+      "loss": 1.5656,
+      "step": 363800
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5843,
+      "step": 363900
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5774,
+      "step": 364000
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 0.0005,
+      "loss": 1.5399,
+      "step": 364100
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5777,
+      "step": 364200
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 0.0005,
+      "loss": 1.5734,
+      "step": 364300
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5538,
+      "step": 364400
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 0.0005,
+      "loss": 1.551,
+      "step": 364500
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 0.0005,
+      "loss": 1.5904,
+      "step": 364600
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5802,
+      "step": 364700
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5516,
+      "step": 364800
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5416,
+      "step": 364900
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5757,
+      "step": 365000
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 0.0005,
+      "loss": 1.5526,
+      "step": 365100
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 0.0005,
+      "loss": 1.5526,
+      "step": 365200
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5748,
+      "step": 365300
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5667,
+      "step": 365400
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 0.0005,
+      "loss": 1.5389,
+      "step": 365500
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5805,
+      "step": 365600
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 0.0005,
+      "loss": 1.5537,
+      "step": 365700
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5464,
+      "step": 365800
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5525,
+      "step": 365900
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5933,
+      "step": 366000
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5686,
+      "step": 366100
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 0.0005,
+      "loss": 1.5916,
+      "step": 366200
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 0.0005,
+      "loss": 1.5913,
+      "step": 366300
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5873,
+      "step": 366400
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5629,
+      "step": 366500
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5586,
+      "step": 366600
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5848,
+      "step": 366700
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 0.0005,
+      "loss": 1.6012,
+      "step": 366800
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 0.0005,
+      "loss": 1.5689,
+      "step": 366900
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 0.0005,
+      "loss": 1.6003,
+      "step": 367000
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 0.0005,
+      "loss": 1.6016,
+      "step": 367100
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 0.0005,
+      "loss": 1.6006,
+      "step": 367200
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5989,
+      "step": 367300
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5512,
+      "step": 367400
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5561,
+      "step": 367500
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5544,
+      "step": 367600
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5827,
+      "step": 367700
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5807,
+      "step": 367800
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5896,
+      "step": 367900
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 0.0005,
+      "loss": 1.5632,
+      "step": 368000
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 0.0005,
+      "loss": 1.6038,
+      "step": 368100
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 0.0005,
+      "loss": 1.572,
+      "step": 368200
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5859,
+      "step": 368300
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5879,
+      "step": 368400
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 0.0005,
+      "loss": 1.5807,
+      "step": 368500
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5997,
+      "step": 368600
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5818,
+      "step": 368700
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5872,
+      "step": 368800
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5478,
+      "step": 368900
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5551,
+      "step": 369000
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 0.0005,
+      "loss": 1.565,
+      "step": 369100
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 0.0005,
+      "loss": 1.5758,
+      "step": 369200
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5759,
+      "step": 369300
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 0.0005,
+      "loss": 1.5761,
+      "step": 369400
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 0.0005,
+      "loss": 1.6011,
+      "step": 369500
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5991,
+      "step": 369600
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5906,
+      "step": 369700
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5794,
+      "step": 369800
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5154,
+      "step": 369900
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5134,
+      "step": 370000
+    },
+    {
+      "epoch": 29.02,
+      "eval_gen_len": 18.791383420146552,
+      "eval_loss": 2.0581727027893066,
+      "eval_rouge1": 35.3426,
+      "eval_rouge2": 14.425,
+      "eval_rougeL": 29.1042,
+      "eval_rougeLsum": 29.0882,
+      "eval_runtime": 354.5816,
+      "eval_samples_per_second": 31.945,
+      "eval_steps_per_second": 1.997,
+      "step": 370000
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5084,
+      "step": 370100
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 0.0005,
+      "loss": 1.5087,
+      "step": 370200
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 0.0005,
+      "loss": 1.5158,
+      "step": 370300
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5025,
+      "step": 370400
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5105,
+      "step": 370500
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4915,
+      "step": 370600
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5036,
+      "step": 370700
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 0.0005,
+      "loss": 1.5132,
+      "step": 370800
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5009,
+      "step": 370900
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5353,
+      "step": 371000
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4914,
+      "step": 371100
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5104,
+      "step": 371200
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 0.0005,
+      "loss": 1.5413,
+      "step": 371300
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 0.0005,
+      "loss": 1.5274,
+      "step": 371400
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5159,
+      "step": 371500
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 0.0005,
+      "loss": 1.54,
+      "step": 371600
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 0.0005,
+      "loss": 1.5592,
+      "step": 371700
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5061,
+      "step": 371800
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5144,
+      "step": 371900
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 0.0005,
+      "loss": 1.5369,
+      "step": 372000
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 0.0005,
+      "loss": 1.5371,
+      "step": 372100
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 0.0005,
+      "loss": 1.5418,
+      "step": 372200
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5116,
+      "step": 372300
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 0.0005,
+      "loss": 1.507,
+      "step": 372400
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 0.0005,
+      "loss": 1.5431,
+      "step": 372500
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 0.0005,
+      "loss": 1.5175,
+      "step": 372600
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5071,
+      "step": 372700
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5442,
+      "step": 372800
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 0.0005,
+      "loss": 1.5561,
+      "step": 372900
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4922,
+      "step": 373000
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 0.0005,
+      "loss": 1.5415,
+      "step": 373100
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 0.0005,
+      "loss": 1.543,
+      "step": 373200
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5519,
+      "step": 373300
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 0.0005,
+      "loss": 1.5147,
+      "step": 373400
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5495,
+      "step": 373500
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 0.0005,
+      "loss": 1.5376,
+      "step": 373600
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 0.0005,
+      "loss": 1.5431,
+      "step": 373700
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 0.0005,
+      "loss": 1.5454,
+      "step": 373800
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5311,
+      "step": 373900
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 0.0005,
+      "loss": 1.531,
+      "step": 374000
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 0.0005,
+      "loss": 1.548,
+      "step": 374100
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 0.0005,
+      "loss": 1.5405,
+      "step": 374200
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5334,
+      "step": 374300
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5425,
+      "step": 374400
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 0.0005,
+      "loss": 1.5558,
+      "step": 374500
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5261,
+      "step": 374600
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5543,
+      "step": 374700
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5462,
+      "step": 374800
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5245,
+      "step": 374900
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5498,
+      "step": 375000
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5805,
+      "step": 375100
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5346,
+      "step": 375200
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5383,
+      "step": 375300
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 0.0005,
+      "loss": 1.5239,
+      "step": 375400
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 0.0005,
+      "loss": 1.588,
+      "step": 375500
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 0.0005,
+      "loss": 1.5122,
+      "step": 375600
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 0.0005,
+      "loss": 1.5565,
+      "step": 375700
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5167,
+      "step": 375800
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 0.0005,
+      "loss": 1.5715,
+      "step": 375900
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5449,
+      "step": 376000
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5567,
+      "step": 376100
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 0.0005,
+      "loss": 1.5546,
+      "step": 376200
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5428,
+      "step": 376300
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5627,
+      "step": 376400
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5605,
+      "step": 376500
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 0.0005,
+      "loss": 1.5493,
+      "step": 376600
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5653,
+      "step": 376700
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 0.0005,
+      "loss": 1.5443,
+      "step": 376800
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5499,
+      "step": 376900
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5444,
+      "step": 377000
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 0.0005,
+      "loss": 1.5792,
+      "step": 377100
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5403,
+      "step": 377200
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 0.0005,
+      "loss": 1.5827,
+      "step": 377300
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5682,
+      "step": 377400
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5276,
+      "step": 377500
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5516,
+      "step": 377600
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5465,
+      "step": 377700
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 0.0005,
+      "loss": 1.577,
+      "step": 377800
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 0.0005,
+      "loss": 1.575,
+      "step": 377900
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 0.0005,
+      "loss": 1.5309,
+      "step": 378000
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5714,
+      "step": 378100
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 0.0005,
+      "loss": 1.5524,
+      "step": 378200
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5659,
+      "step": 378300
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5679,
+      "step": 378400
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 0.0005,
+      "loss": 1.5378,
+      "step": 378500
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5523,
+      "step": 378600
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5675,
+      "step": 378700
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 0.0005,
+      "loss": 1.578,
+      "step": 378800
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5386,
+      "step": 378900
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 0.0005,
+      "loss": 1.5588,
+      "step": 379000
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5356,
+      "step": 379100
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5889,
+      "step": 379200
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5759,
+      "step": 379300
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 0.0005,
+      "loss": 1.556,
+      "step": 379400
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5655,
+      "step": 379500
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 0.0005,
+      "loss": 1.589,
+      "step": 379600
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5837,
+      "step": 379700
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5566,
+      "step": 379800
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5692,
+      "step": 379900
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5468,
+      "step": 380000
+    },
+    {
+      "epoch": 29.8,
+      "eval_gen_len": 18.78440893440452,
+      "eval_loss": 2.0447006225585938,
+      "eval_rouge1": 35.5171,
+      "eval_rouge2": 14.5247,
+      "eval_rougeL": 29.2402,
+      "eval_rougeLsum": 29.2352,
+      "eval_runtime": 354.064,
+      "eval_samples_per_second": 31.991,
+      "eval_steps_per_second": 2.0,
+      "step": 380000
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5943,
+      "step": 380100
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5811,
+      "step": 380200
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5551,
+      "step": 380300
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5615,
+      "step": 380400
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5603,
+      "step": 380500
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5292,
+      "step": 380600
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5513,
+      "step": 380700
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 0.0005,
+      "loss": 1.568,
+      "step": 380800
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5795,
+      "step": 380900
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5701,
+      "step": 381000
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5592,
+      "step": 381100
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5821,
+      "step": 381200
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 0.0005,
+      "loss": 1.5689,
+      "step": 381300
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5638,
+      "step": 381400
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5541,
+      "step": 381500
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5629,
+      "step": 381600
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5603,
+      "step": 381700
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5926,
+      "step": 381800
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 0.0005,
+      "loss": 1.5746,
+      "step": 381900
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5786,
+      "step": 382000
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5756,
+      "step": 382100
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 0.0005,
+      "loss": 1.5677,
+      "step": 382200
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5615,
+      "step": 382300
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5771,
+      "step": 382400
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5749,
+      "step": 382500
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5499,
+      "step": 382600
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4999,
+      "step": 382700
+    },
+    {
+      "epoch": 30.02,
+      "learning_rate": 0.0005,
+      "loss": 1.5096,
+      "step": 382800
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4812,
+      "step": 382900
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4866,
+      "step": 383000
+    },
+    {
+      "epoch": 30.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4996,
+      "step": 383100
+    },
+    {
+      "epoch": 30.05,
+      "learning_rate": 0.0005,
+      "loss": 1.5003,
+      "step": 383200
+    },
+    {
+      "epoch": 30.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4927,
+      "step": 383300
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 0.0005,
+      "loss": 1.4954,
+      "step": 383400
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 0.0005,
+      "loss": 1.5292,
+      "step": 383500
+    },
+    {
+      "epoch": 30.08,
+      "learning_rate": 0.0005,
+      "loss": 1.5011,
+      "step": 383600
+    },
+    {
+      "epoch": 30.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5287,
+      "step": 383700
+    },
+    {
+      "epoch": 30.1,
+      "learning_rate": 0.0005,
+      "loss": 1.485,
+      "step": 383800
+    },
+    {
+      "epoch": 30.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5102,
+      "step": 383900
+    },
+    {
+      "epoch": 30.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5175,
+      "step": 384000
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 0.0005,
+      "loss": 1.5078,
+      "step": 384100
+    },
+    {
+      "epoch": 30.13,
+      "learning_rate": 0.0005,
+      "loss": 1.5168,
+      "step": 384200
+    },
+    {
+      "epoch": 30.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5254,
+      "step": 384300
+    },
+    {
+      "epoch": 30.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5083,
+      "step": 384400
+    },
+    {
+      "epoch": 30.15,
+      "learning_rate": 0.0005,
+      "loss": 1.5071,
+      "step": 384500
+    },
+    {
+      "epoch": 30.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5103,
+      "step": 384600
+    },
+    {
+      "epoch": 30.17,
+      "learning_rate": 0.0005,
+      "loss": 1.4913,
+      "step": 384700
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4917,
+      "step": 384800
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 0.0005,
+      "loss": 1.5229,
+      "step": 384900
+    },
+    {
+      "epoch": 30.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4823,
+      "step": 385000
+    },
+    {
+      "epoch": 30.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5141,
+      "step": 385100
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 0.0005,
+      "loss": 1.5119,
+      "step": 385200
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 0.0005,
+      "loss": 1.5448,
+      "step": 385300
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 0.0005,
+      "loss": 1.5142,
+      "step": 385400
+    },
+    {
+      "epoch": 30.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5171,
+      "step": 385500
+    },
+    {
+      "epoch": 30.24,
+      "learning_rate": 0.0005,
+      "loss": 1.5354,
+      "step": 385600
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4944,
+      "step": 385700
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 0.0005,
+      "loss": 1.523,
+      "step": 385800
+    },
+    {
+      "epoch": 30.26,
+      "learning_rate": 0.0005,
+      "loss": 1.5474,
+      "step": 385900
+    },
+    {
+      "epoch": 30.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5116,
+      "step": 386000
+    },
+    {
+      "epoch": 30.28,
+      "learning_rate": 0.0005,
+      "loss": 1.5224,
+      "step": 386100
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5225,
+      "step": 386200
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5191,
+      "step": 386300
+    },
+    {
+      "epoch": 30.3,
+      "learning_rate": 0.0005,
+      "loss": 1.5531,
+      "step": 386400
+    },
+    {
+      "epoch": 30.31,
+      "learning_rate": 0.0005,
+      "loss": 1.5301,
+      "step": 386500
+    },
+    {
+      "epoch": 30.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5313,
+      "step": 386600
+    },
+    {
+      "epoch": 30.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4778,
+      "step": 386700
+    },
+    {
+      "epoch": 30.33,
+      "learning_rate": 0.0005,
+      "loss": 1.5302,
+      "step": 386800
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 0.0005,
+      "loss": 1.5164,
+      "step": 386900
+    },
+    {
+      "epoch": 30.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5088,
+      "step": 387000
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5215,
+      "step": 387100
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5594,
+      "step": 387200
+    },
+    {
+      "epoch": 30.37,
+      "learning_rate": 0.0005,
+      "loss": 1.534,
+      "step": 387300
+    },
+    {
+      "epoch": 30.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5421,
+      "step": 387400
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5384,
+      "step": 387500
+    },
+    {
+      "epoch": 30.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5201,
+      "step": 387600
+    },
+    {
+      "epoch": 30.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5385,
+      "step": 387700
+    },
+    {
+      "epoch": 30.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5387,
+      "step": 387800
+    },
+    {
+      "epoch": 30.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5556,
+      "step": 387900
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5433,
+      "step": 388000
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5271,
+      "step": 388100
+    },
+    {
+      "epoch": 30.44,
+      "learning_rate": 0.0005,
+      "loss": 1.5368,
+      "step": 388200
+    },
+    {
+      "epoch": 30.45,
+      "learning_rate": 0.0005,
+      "loss": 1.544,
+      "step": 388300
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 0.0005,
+      "loss": 1.5288,
+      "step": 388400
+    },
+    {
+      "epoch": 30.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5404,
+      "step": 388500
+    },
+    {
+      "epoch": 30.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5382,
+      "step": 388600
+    },
+    {
+      "epoch": 30.48,
+      "learning_rate": 0.0005,
+      "loss": 1.5495,
+      "step": 388700
+    },
+    {
+      "epoch": 30.49,
+      "learning_rate": 0.0005,
+      "loss": 1.58,
+      "step": 388800
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 0.0005,
+      "loss": 1.5482,
+      "step": 388900
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 0.0005,
+      "loss": 1.4916,
+      "step": 389000
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5232,
+      "step": 389100
+    },
+    {
+      "epoch": 30.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5288,
+      "step": 389200
+    },
+    {
+      "epoch": 30.53,
+      "learning_rate": 0.0005,
+      "loss": 1.5001,
+      "step": 389300
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5628,
+      "step": 389400
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5341,
+      "step": 389500
+    },
+    {
+      "epoch": 30.55,
+      "learning_rate": 0.0005,
+      "loss": 1.522,
+      "step": 389600
+    },
+    {
+      "epoch": 30.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5442,
+      "step": 389700
+    },
+    {
+      "epoch": 30.57,
+      "learning_rate": 0.0005,
+      "loss": 1.549,
+      "step": 389800
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5188,
+      "step": 389900
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5073,
+      "step": 390000
+    },
+    {
+      "epoch": 30.58,
+      "eval_gen_len": 18.813278008298756,
+      "eval_loss": 2.047095775604248,
+      "eval_rouge1": 35.2735,
+      "eval_rouge2": 14.329,
+      "eval_rougeL": 29.0615,
+      "eval_rougeLsum": 29.0566,
+      "eval_runtime": 354.7807,
+      "eval_samples_per_second": 31.927,
+      "eval_steps_per_second": 1.996,
+      "step": 390000
+    },
+    {
+      "epoch": 30.59,
+      "learning_rate": 0.0005,
+      "loss": 1.5498,
+      "step": 390100
+    },
+    {
+      "epoch": 30.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5313,
+      "step": 390200
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5331,
+      "step": 390300
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5503,
+      "step": 390400
+    },
+    {
+      "epoch": 30.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5392,
+      "step": 390500
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 0.0005,
+      "loss": 1.5448,
+      "step": 390600
+    },
+    {
+      "epoch": 30.64,
+      "learning_rate": 0.0005,
+      "loss": 1.5421,
+      "step": 390700
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5446,
+      "step": 390800
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5466,
+      "step": 390900
+    },
+    {
+      "epoch": 30.66,
+      "learning_rate": 0.0005,
+      "loss": 1.557,
+      "step": 391000
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5456,
+      "step": 391100
+    },
+    {
+      "epoch": 30.68,
+      "learning_rate": 0.0005,
+      "loss": 1.5441,
+      "step": 391200
+    },
+    {
+      "epoch": 30.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5624,
+      "step": 391300
+    },
+    {
+      "epoch": 30.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5449,
+      "step": 391400
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5352,
+      "step": 391500
+    },
+    {
+      "epoch": 30.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5159,
+      "step": 391600
+    },
+    {
+      "epoch": 30.72,
+      "learning_rate": 0.0005,
+      "loss": 1.5566,
+      "step": 391700
+    },
+    {
+      "epoch": 30.72,
+      "learning_rate": 0.0005,
+      "loss": 1.528,
+      "step": 391800
+    },
+    {
+      "epoch": 30.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5379,
+      "step": 391900
+    },
+    {
+      "epoch": 30.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5754,
+      "step": 392000
+    },
+    {
+      "epoch": 30.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5689,
+      "step": 392100
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5678,
+      "step": 392200
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5665,
+      "step": 392300
+    },
+    {
+      "epoch": 30.77,
+      "learning_rate": 0.0005,
+      "loss": 1.5618,
+      "step": 392400
+    },
+    {
+      "epoch": 30.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5661,
+      "step": 392500
+    },
+    {
+      "epoch": 30.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5586,
+      "step": 392600
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 0.0005,
+      "loss": 1.556,
+      "step": 392700
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5416,
+      "step": 392800
+    },
+    {
+      "epoch": 30.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5362,
+      "step": 392900
+    },
+    {
+      "epoch": 30.82,
+      "learning_rate": 0.0005,
+      "loss": 1.554,
+      "step": 393000
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 0.0005,
+      "loss": 1.563,
+      "step": 393100
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5251,
+      "step": 393200
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5873,
+      "step": 393300
+    },
+    {
+      "epoch": 30.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5558,
+      "step": 393400
+    },
+    {
+      "epoch": 30.86,
+      "learning_rate": 0.0005,
+      "loss": 1.5633,
+      "step": 393500
+    },
+    {
+      "epoch": 30.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5633,
+      "step": 393600
+    },
+    {
+      "epoch": 30.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5309,
+      "step": 393700
+    },
+    {
+      "epoch": 30.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5891,
+      "step": 393800
+    },
+    {
+      "epoch": 30.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5933,
+      "step": 393900
+    },
+    {
+      "epoch": 30.9,
+      "learning_rate": 0.0005,
+      "loss": 1.5798,
+      "step": 394000
+    },
+    {
+      "epoch": 30.9,
+      "learning_rate": 0.0005,
+      "loss": 1.5414,
+      "step": 394100
+    },
+    {
+      "epoch": 30.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5658,
+      "step": 394200
+    },
+    {
+      "epoch": 30.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5304,
+      "step": 394300
+    },
+    {
+      "epoch": 30.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5448,
+      "step": 394400
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5352,
+      "step": 394500
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5415,
+      "step": 394600
+    },
+    {
+      "epoch": 30.95,
+      "learning_rate": 0.0005,
+      "loss": 1.5869,
+      "step": 394700
+    },
+    {
+      "epoch": 30.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5442,
+      "step": 394800
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 0.0005,
+      "loss": 1.5851,
+      "step": 394900
+    },
+    {
+      "epoch": 30.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5618,
+      "step": 395000
+    },
+    {
+      "epoch": 30.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5343,
+      "step": 395100
+    },
+    {
+      "epoch": 30.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5599,
+      "step": 395200
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5824,
+      "step": 395300
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4802,
+      "step": 395400
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 0.0005,
+      "loss": 1.5035,
+      "step": 395500
+    },
+    {
+      "epoch": 31.02,
+      "learning_rate": 0.0005,
+      "loss": 1.4814,
+      "step": 395600
+    },
+    {
+      "epoch": 31.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4673,
+      "step": 395700
+    },
+    {
+      "epoch": 31.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4805,
+      "step": 395800
+    },
+    {
+      "epoch": 31.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4997,
+      "step": 395900
+    },
+    {
+      "epoch": 31.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4932,
+      "step": 396000
+    },
+    {
+      "epoch": 31.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4683,
+      "step": 396100
+    },
+    {
+      "epoch": 31.07,
+      "learning_rate": 0.0005,
+      "loss": 1.4829,
+      "step": 396200
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 0.0005,
+      "loss": 1.4926,
+      "step": 396300
+    },
+    {
+      "epoch": 31.09,
+      "learning_rate": 0.0005,
+      "loss": 1.4833,
+      "step": 396400
+    },
+    {
+      "epoch": 31.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5129,
+      "step": 396500
+    },
+    {
+      "epoch": 31.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4949,
+      "step": 396600
+    },
+    {
+      "epoch": 31.11,
+      "learning_rate": 0.0005,
+      "loss": 1.5111,
+      "step": 396700
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4714,
+      "step": 396800
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4798,
+      "step": 396900
+    },
+    {
+      "epoch": 31.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4855,
+      "step": 397000
+    },
+    {
+      "epoch": 31.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4933,
+      "step": 397100
+    },
+    {
+      "epoch": 31.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4793,
+      "step": 397200
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4858,
+      "step": 397300
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 0.0005,
+      "loss": 1.5112,
+      "step": 397400
+    },
+    {
+      "epoch": 31.17,
+      "learning_rate": 0.0005,
+      "loss": 1.5084,
+      "step": 397500
+    },
+    {
+      "epoch": 31.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4982,
+      "step": 397600
+    },
+    {
+      "epoch": 31.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4986,
+      "step": 397700
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5113,
+      "step": 397800
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 0.0005,
+      "loss": 1.5151,
+      "step": 397900
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4965,
+      "step": 398000
+    },
+    {
+      "epoch": 31.22,
+      "learning_rate": 0.0005,
+      "loss": 1.4944,
+      "step": 398100
+    },
+    {
+      "epoch": 31.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4989,
+      "step": 398200
+    },
+    {
+      "epoch": 31.23,
+      "learning_rate": 0.0005,
+      "loss": 1.5089,
+      "step": 398300
+    },
+    {
+      "epoch": 31.24,
+      "learning_rate": 0.0005,
+      "loss": 1.51,
+      "step": 398400
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 0.0005,
+      "loss": 1.5227,
+      "step": 398500
+    },
+    {
+      "epoch": 31.26,
+      "learning_rate": 0.0005,
+      "loss": 1.5136,
+      "step": 398600
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4949,
+      "step": 398700
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 0.0005,
+      "loss": 1.485,
+      "step": 398800
+    },
+    {
+      "epoch": 31.28,
+      "learning_rate": 0.0005,
+      "loss": 1.5228,
+      "step": 398900
+    },
+    {
+      "epoch": 31.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5132,
+      "step": 399000
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 0.0005,
+      "loss": 1.5193,
+      "step": 399100
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 0.0005,
+      "loss": 1.5199,
+      "step": 399200
+    },
+    {
+      "epoch": 31.31,
+      "learning_rate": 0.0005,
+      "loss": 1.5519,
+      "step": 399300
+    },
+    {
+      "epoch": 31.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5422,
+      "step": 399400
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 0.0005,
+      "loss": 1.5042,
+      "step": 399500
+    },
+    {
+      "epoch": 31.34,
+      "learning_rate": 0.0005,
+      "loss": 1.5193,
+      "step": 399600
+    },
+    {
+      "epoch": 31.34,
+      "learning_rate": 0.0005,
+      "loss": 1.5402,
+      "step": 399700
+    },
+    {
+      "epoch": 31.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5366,
+      "step": 399800
+    },
+    {
+      "epoch": 31.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5035,
+      "step": 399900
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 0.0005,
+      "loss": 1.5176,
+      "step": 400000
+    },
+    {
+      "epoch": 31.37,
+      "eval_gen_len": 18.77999470292222,
+      "eval_loss": 2.055631637573242,
+      "eval_rouge1": 35.2942,
+      "eval_rouge2": 14.3432,
+      "eval_rougeL": 29.0501,
+      "eval_rougeLsum": 29.0411,
+      "eval_runtime": 356.7733,
+      "eval_samples_per_second": 31.748,
+      "eval_steps_per_second": 1.984,
+      "step": 400000
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5454,
+      "step": 400100
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5285,
+      "step": 400200
+    },
+    {
+      "epoch": 31.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5118,
+      "step": 400300
+    },
+    {
+      "epoch": 31.4,
+      "learning_rate": 0.0005,
+      "loss": 1.5388,
+      "step": 400400
+    },
+    {
+      "epoch": 31.41,
+      "learning_rate": 0.0005,
+      "loss": 1.516,
+      "step": 400500
+    },
+    {
+      "epoch": 31.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4913,
+      "step": 400600
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5452,
+      "step": 400700
+    },
+    {
+      "epoch": 31.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5255,
+      "step": 400800
+    },
+    {
+      "epoch": 31.44,
+      "learning_rate": 0.0005,
+      "loss": 1.535,
+      "step": 400900
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 0.0005,
+      "loss": 1.5296,
+      "step": 401000
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 0.0005,
+      "loss": 1.5089,
+      "step": 401100
+    },
+    {
+      "epoch": 31.46,
+      "learning_rate": 0.0005,
+      "loss": 1.5081,
+      "step": 401200
+    },
+    {
+      "epoch": 31.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5248,
+      "step": 401300
+    },
+    {
+      "epoch": 31.48,
+      "learning_rate": 0.0005,
+      "loss": 1.5038,
+      "step": 401400
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5106,
+      "step": 401500
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5322,
+      "step": 401600
+    },
+    {
+      "epoch": 31.5,
+      "learning_rate": 0.0005,
+      "loss": 1.523,
+      "step": 401700
+    },
+    {
+      "epoch": 31.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5027,
+      "step": 401800
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5182,
+      "step": 401900
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5425,
+      "step": 402000
+    },
+    {
+      "epoch": 31.53,
+      "learning_rate": 0.0005,
+      "loss": 1.5213,
+      "step": 402100
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 0.0005,
+      "loss": 1.494,
+      "step": 402200
+    },
+    {
+      "epoch": 31.55,
+      "learning_rate": 0.0005,
+      "loss": 1.5569,
+      "step": 402300
+    },
+    {
+      "epoch": 31.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5362,
+      "step": 402400
+    },
+    {
+      "epoch": 31.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5118,
+      "step": 402500
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 0.0005,
+      "loss": 1.5371,
+      "step": 402600
+    },
+    {
+      "epoch": 31.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5366,
+      "step": 402700
+    },
+    {
+      "epoch": 31.59,
+      "learning_rate": 0.0005,
+      "loss": 1.5466,
+      "step": 402800
+    },
+    {
+      "epoch": 31.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5523,
+      "step": 402900
+    },
+    {
+      "epoch": 31.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5207,
+      "step": 403000
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5294,
+      "step": 403100
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5199,
+      "step": 403200
+    },
+    {
+      "epoch": 31.63,
+      "learning_rate": 0.0005,
+      "loss": 1.5206,
+      "step": 403300
+    },
+    {
+      "epoch": 31.63,
+      "learning_rate": 0.0005,
+      "loss": 1.5636,
+      "step": 403400
+    },
+    {
+      "epoch": 31.64,
+      "learning_rate": 0.0005,
+      "loss": 1.5254,
+      "step": 403500
+    },
+    {
+      "epoch": 31.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5464,
+      "step": 403600
+    },
+    {
+      "epoch": 31.66,
+      "learning_rate": 0.0005,
+      "loss": 1.5271,
+      "step": 403700
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5663,
+      "step": 403800
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5366,
+      "step": 403900
+    },
+    {
+      "epoch": 31.68,
+      "learning_rate": 0.0005,
+      "loss": 1.522,
+      "step": 404000
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5086,
+      "step": 404100
+    },
+    {
+      "epoch": 31.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5337,
+      "step": 404200
+    },
+    {
+      "epoch": 31.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5372,
+      "step": 404300
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5346,
+      "step": 404400
+    },
+    {
+      "epoch": 31.72,
+      "learning_rate": 0.0005,
+      "loss": 1.5595,
+      "step": 404500
+    },
+    {
+      "epoch": 31.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5283,
+      "step": 404600
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5531,
+      "step": 404700
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5486,
+      "step": 404800
+    },
+    {
+      "epoch": 31.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5326,
+      "step": 404900
+    },
+    {
+      "epoch": 31.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5473,
+      "step": 405000
+    },
+    {
+      "epoch": 31.77,
+      "learning_rate": 0.0005,
+      "loss": 1.5356,
+      "step": 405100
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5393,
+      "step": 405200
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5449,
+      "step": 405300
+    },
+    {
+      "epoch": 31.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5594,
+      "step": 405400
+    },
+    {
+      "epoch": 31.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5404,
+      "step": 405500
+    },
+    {
+      "epoch": 31.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5441,
+      "step": 405600
+    },
+    {
+      "epoch": 31.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5549,
+      "step": 405700
+    },
+    {
+      "epoch": 31.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5179,
+      "step": 405800
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5099,
+      "step": 405900
+    },
+    {
+      "epoch": 31.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5443,
+      "step": 406000
+    },
+    {
+      "epoch": 31.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5671,
+      "step": 406100
+    },
+    {
+      "epoch": 31.85,
+      "learning_rate": 0.0005,
+      "loss": 1.537,
+      "step": 406200
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 0.0005,
+      "loss": 1.5413,
+      "step": 406300
+    },
+    {
+      "epoch": 31.87,
+      "learning_rate": 0.0005,
+      "loss": 1.548,
+      "step": 406400
+    },
+    {
+      "epoch": 31.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5445,
+      "step": 406500
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5405,
+      "step": 406600
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5748,
+      "step": 406700
+    },
+    {
+      "epoch": 31.9,
+      "learning_rate": 0.0005,
+      "loss": 1.5358,
+      "step": 406800
+    },
+    {
+      "epoch": 31.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5438,
+      "step": 406900
+    },
+    {
+      "epoch": 31.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5229,
+      "step": 407000
+    },
+    {
+      "epoch": 31.92,
+      "learning_rate": 0.0005,
+      "loss": 1.546,
+      "step": 407100
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5315,
+      "step": 407200
+    },
+    {
+      "epoch": 31.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5509,
+      "step": 407300
+    },
+    {
+      "epoch": 31.95,
+      "learning_rate": 0.0005,
+      "loss": 1.5527,
+      "step": 407400
+    },
+    {
+      "epoch": 31.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5307,
+      "step": 407500
+    },
+    {
+      "epoch": 31.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5234,
+      "step": 407600
+    },
+    {
+      "epoch": 31.97,
+      "learning_rate": 0.0005,
+      "loss": 1.5626,
+      "step": 407700
+    },
+    {
+      "epoch": 31.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5508,
+      "step": 407800
+    },
+    {
+      "epoch": 31.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5658,
+      "step": 407900
+    },
+    {
+      "epoch": 31.99,
+      "learning_rate": 0.0005,
+      "loss": 1.541,
+      "step": 408000
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5188,
+      "step": 408100
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4728,
+      "step": 408200
+    },
+    {
+      "epoch": 32.02,
+      "learning_rate": 0.0005,
+      "loss": 1.4386,
+      "step": 408300
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4807,
+      "step": 408400
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4852,
+      "step": 408500
+    },
+    {
+      "epoch": 32.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4643,
+      "step": 408600
+    },
+    {
+      "epoch": 32.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4934,
+      "step": 408700
+    },
+    {
+      "epoch": 32.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4816,
+      "step": 408800
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 0.0005,
+      "loss": 1.4521,
+      "step": 408900
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 0.0005,
+      "loss": 1.4754,
+      "step": 409000
+    },
+    {
+      "epoch": 32.08,
+      "learning_rate": 0.0005,
+      "loss": 1.4847,
+      "step": 409100
+    },
+    {
+      "epoch": 32.09,
+      "learning_rate": 0.0005,
+      "loss": 1.5096,
+      "step": 409200
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4838,
+      "step": 409300
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4722,
+      "step": 409400
+    },
+    {
+      "epoch": 32.11,
+      "learning_rate": 0.0005,
+      "loss": 1.4993,
+      "step": 409500
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4843,
+      "step": 409600
+    },
+    {
+      "epoch": 32.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4763,
+      "step": 409700
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4765,
+      "step": 409800
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 0.0005,
+      "loss": 1.5052,
+      "step": 409900
+    },
+    {
+      "epoch": 32.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4973,
+      "step": 410000
+    },
+    {
+      "epoch": 32.15,
+      "eval_gen_len": 18.798799329036814,
+      "eval_loss": 2.0617787837982178,
+      "eval_rouge1": 35.6624,
+      "eval_rouge2": 14.4996,
+      "eval_rougeL": 29.2625,
+      "eval_rougeLsum": 29.2571,
+      "eval_runtime": 364.7552,
+      "eval_samples_per_second": 31.054,
+      "eval_steps_per_second": 1.941,
+      "step": 410000
+    },
+    {
+      "epoch": 32.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4893,
+      "step": 410100
+    },
+    {
+      "epoch": 32.17,
+      "learning_rate": 0.0005,
+      "loss": 1.51,
+      "step": 410200
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 0.0005,
+      "loss": 1.5004,
+      "step": 410300
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 0.0005,
+      "loss": 1.5074,
+      "step": 410400
+    },
+    {
+      "epoch": 32.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4952,
+      "step": 410500
+    },
+    {
+      "epoch": 32.2,
+      "learning_rate": 0.0005,
+      "loss": 1.4865,
+      "step": 410600
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 0.0005,
+      "loss": 1.5009,
+      "step": 410700
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4766,
+      "step": 410800
+    },
+    {
+      "epoch": 32.22,
+      "learning_rate": 0.0005,
+      "loss": 1.4949,
+      "step": 410900
+    },
+    {
+      "epoch": 32.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4874,
+      "step": 411000
+    },
+    {
+      "epoch": 32.24,
+      "learning_rate": 0.0005,
+      "loss": 1.4945,
+      "step": 411100
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 0.0005,
+      "loss": 1.5005,
+      "step": 411200
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4743,
+      "step": 411300
+    },
+    {
+      "epoch": 32.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4751,
+      "step": 411400
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5003,
+      "step": 411500
+    },
+    {
+      "epoch": 32.28,
+      "learning_rate": 0.0005,
+      "loss": 1.473,
+      "step": 411600
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 0.0005,
+      "loss": 1.477,
+      "step": 411700
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 0.0005,
+      "loss": 1.518,
+      "step": 411800
+    },
+    {
+      "epoch": 32.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4561,
+      "step": 411900
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4887,
+      "step": 412000
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5099,
+      "step": 412100
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 0.0005,
+      "loss": 1.5178,
+      "step": 412200
+    },
+    {
+      "epoch": 32.33,
+      "learning_rate": 0.0005,
+      "loss": 1.5064,
+      "step": 412300
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4879,
+      "step": 412400
+    },
+    {
+      "epoch": 32.35,
+      "learning_rate": 0.0005,
+      "loss": 1.5057,
+      "step": 412500
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5124,
+      "step": 412600
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 0.0005,
+      "loss": 1.5316,
+      "step": 412700
+    },
+    {
+      "epoch": 32.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4788,
+      "step": 412800
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 0.0005,
+      "loss": 1.5265,
+      "step": 412900
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5078,
+      "step": 413000
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5052,
+      "step": 413100
+    },
+    {
+      "epoch": 32.4,
+      "learning_rate": 0.0005,
+      "loss": 1.4874,
+      "step": 413200
+    },
+    {
+      "epoch": 32.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5137,
+      "step": 413300
+    },
+    {
+      "epoch": 32.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5306,
+      "step": 413400
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4981,
+      "step": 413500
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4998,
+      "step": 413600
+    },
+    {
+      "epoch": 32.44,
+      "learning_rate": 0.0005,
+      "loss": 1.496,
+      "step": 413700
+    },
+    {
+      "epoch": 32.45,
+      "learning_rate": 0.0005,
+      "loss": 1.529,
+      "step": 413800
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 0.0005,
+      "loss": 1.5146,
+      "step": 413900
+    },
+    {
+      "epoch": 32.47,
+      "learning_rate": 0.0005,
+      "loss": 1.4946,
+      "step": 414000
+    },
+    {
+      "epoch": 32.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5132,
+      "step": 414100
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 0.0005,
+      "loss": 1.5192,
+      "step": 414200
+    },
+    {
+      "epoch": 32.49,
+      "learning_rate": 0.0005,
+      "loss": 1.498,
+      "step": 414300
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 0.0005,
+      "loss": 1.5389,
+      "step": 414400
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 0.0005,
+      "loss": 1.5097,
+      "step": 414500
+    },
+    {
+      "epoch": 32.51,
+      "learning_rate": 0.0005,
+      "loss": 1.4973,
+      "step": 414600
+    },
+    {
+      "epoch": 32.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5201,
+      "step": 414700
+    },
+    {
+      "epoch": 32.53,
+      "learning_rate": 0.0005,
+      "loss": 1.4835,
+      "step": 414800
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5017,
+      "step": 414900
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 0.0005,
+      "loss": 1.5011,
+      "step": 415000
+    },
+    {
+      "epoch": 32.55,
+      "learning_rate": 0.0005,
+      "loss": 1.5075,
+      "step": 415100
+    },
+    {
+      "epoch": 32.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5067,
+      "step": 415200
+    },
+    {
+      "epoch": 32.57,
+      "learning_rate": 0.0005,
+      "loss": 1.5373,
+      "step": 415300
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5174,
+      "step": 415400
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5264,
+      "step": 415500
+    },
+    {
+      "epoch": 32.59,
+      "learning_rate": 0.0005,
+      "loss": 1.5168,
+      "step": 415600
+    },
+    {
+      "epoch": 32.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5411,
+      "step": 415700
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5051,
+      "step": 415800
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5014,
+      "step": 415900
+    },
+    {
+      "epoch": 32.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5198,
+      "step": 416000
+    },
+    {
+      "epoch": 32.63,
+      "learning_rate": 0.0005,
+      "loss": 1.5211,
+      "step": 416100
+    },
+    {
+      "epoch": 32.64,
+      "learning_rate": 0.0005,
+      "loss": 1.5409,
+      "step": 416200
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5302,
+      "step": 416300
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5224,
+      "step": 416400
+    },
+    {
+      "epoch": 32.66,
+      "learning_rate": 0.0005,
+      "loss": 1.519,
+      "step": 416500
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5162,
+      "step": 416600
+    },
+    {
+      "epoch": 32.68,
+      "learning_rate": 0.0005,
+      "loss": 1.5157,
+      "step": 416700
+    },
+    {
+      "epoch": 32.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5368,
+      "step": 416800
+    },
+    {
+      "epoch": 32.69,
+      "learning_rate": 0.0005,
+      "loss": 1.513,
+      "step": 416900
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5508,
+      "step": 417000
+    },
+    {
+      "epoch": 32.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5104,
+      "step": 417100
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 0.0005,
+      "loss": 1.5185,
+      "step": 417200
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 0.0005,
+      "loss": 1.492,
+      "step": 417300
+    },
+    {
+      "epoch": 32.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5189,
+      "step": 417400
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5375,
+      "step": 417500
+    },
+    {
+      "epoch": 32.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5309,
+      "step": 417600
+    },
+    {
+      "epoch": 32.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5239,
+      "step": 417700
+    },
+    {
+      "epoch": 32.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5401,
+      "step": 417800
+    },
+    {
+      "epoch": 32.77,
+      "learning_rate": 0.0005,
+      "loss": 1.5296,
+      "step": 417900
+    },
+    {
+      "epoch": 32.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5092,
+      "step": 418000
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5272,
+      "step": 418100
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5192,
+      "step": 418200
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5315,
+      "step": 418300
+    },
+    {
+      "epoch": 32.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5335,
+      "step": 418400
+    },
+    {
+      "epoch": 32.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5279,
+      "step": 418500
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5168,
+      "step": 418600
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5587,
+      "step": 418700
+    },
+    {
+      "epoch": 32.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5549,
+      "step": 418800
+    },
+    {
+      "epoch": 32.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5463,
+      "step": 418900
+    },
+    {
+      "epoch": 32.86,
+      "learning_rate": 0.0005,
+      "loss": 1.5279,
+      "step": 419000
+    },
+    {
+      "epoch": 32.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5431,
+      "step": 419100
+    },
+    {
+      "epoch": 32.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5552,
+      "step": 419200
+    },
+    {
+      "epoch": 32.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5442,
+      "step": 419300
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5202,
+      "step": 419400
+    },
+    {
+      "epoch": 32.9,
+      "learning_rate": 0.0005,
+      "loss": 1.543,
+      "step": 419500
+    },
+    {
+      "epoch": 32.9,
+      "learning_rate": 0.0005,
+      "loss": 1.5277,
+      "step": 419600
+    },
+    {
+      "epoch": 32.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5405,
+      "step": 419700
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 0.0005,
+      "loss": 1.557,
+      "step": 419800
+    },
+    {
+      "epoch": 32.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5291,
+      "step": 419900
+    },
+    {
+      "epoch": 32.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5584,
+      "step": 420000
+    },
+    {
+      "epoch": 32.94,
+      "eval_gen_len": 18.746093405138165,
+      "eval_loss": 2.043473720550537,
+      "eval_rouge1": 35.4532,
+      "eval_rouge2": 14.4774,
+      "eval_rougeL": 29.1654,
+      "eval_rougeLsum": 29.1572,
+      "eval_runtime": 355.0094,
+      "eval_samples_per_second": 31.906,
+      "eval_steps_per_second": 1.994,
+      "step": 420000
+    },
+    {
+      "epoch": 32.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5445,
+      "step": 420100
+    },
+    {
+      "epoch": 32.95,
+      "learning_rate": 0.0005,
+      "loss": 1.5559,
+      "step": 420200
+    },
+    {
+      "epoch": 32.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5439,
+      "step": 420300
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 0.0005,
+      "loss": 1.5413,
+      "step": 420400
+    },
+    {
+      "epoch": 32.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5161,
+      "step": 420500
+    },
+    {
+      "epoch": 32.98,
+      "learning_rate": 0.0005,
+      "loss": 1.526,
+      "step": 420600
+    },
+    {
+      "epoch": 32.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5214,
+      "step": 420700
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5649,
+      "step": 420800
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 0.0005,
+      "loss": 1.479,
+      "step": 420900
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4364,
+      "step": 421000
+    },
+    {
+      "epoch": 33.02,
+      "learning_rate": 0.0005,
+      "loss": 1.4831,
+      "step": 421100
+    },
+    {
+      "epoch": 33.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4527,
+      "step": 421200
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4754,
+      "step": 421300
+    },
+    {
+      "epoch": 33.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4573,
+      "step": 421400
+    },
+    {
+      "epoch": 33.05,
+      "learning_rate": 0.0005,
+      "loss": 1.457,
+      "step": 421500
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4525,
+      "step": 421600
+    },
+    {
+      "epoch": 33.07,
+      "learning_rate": 0.0005,
+      "loss": 1.4549,
+      "step": 421700
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 0.0005,
+      "loss": 1.4764,
+      "step": 421800
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 0.0005,
+      "loss": 1.48,
+      "step": 421900
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 0.0005,
+      "loss": 1.4471,
+      "step": 422000
+    },
+    {
+      "epoch": 33.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4762,
+      "step": 422100
+    },
+    {
+      "epoch": 33.11,
+      "learning_rate": 0.0005,
+      "loss": 1.4794,
+      "step": 422200
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4909,
+      "step": 422300
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4878,
+      "step": 422400
+    },
+    {
+      "epoch": 33.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4934,
+      "step": 422500
+    },
+    {
+      "epoch": 33.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4961,
+      "step": 422600
+    },
+    {
+      "epoch": 33.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4815,
+      "step": 422700
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4502,
+      "step": 422800
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4715,
+      "step": 422900
+    },
+    {
+      "epoch": 33.17,
+      "learning_rate": 0.0005,
+      "loss": 1.4665,
+      "step": 423000
+    },
+    {
+      "epoch": 33.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4945,
+      "step": 423100
+    },
+    {
+      "epoch": 33.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4853,
+      "step": 423200
+    },
+    {
+      "epoch": 33.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4898,
+      "step": 423300
+    },
+    {
+      "epoch": 33.2,
+      "learning_rate": 0.0005,
+      "loss": 1.4736,
+      "step": 423400
+    },
+    {
+      "epoch": 33.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4931,
+      "step": 423500
+    },
+    {
+      "epoch": 33.22,
+      "learning_rate": 0.0005,
+      "loss": 1.508,
+      "step": 423600
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 0.0005,
+      "loss": 1.459,
+      "step": 423700
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4718,
+      "step": 423800
+    },
+    {
+      "epoch": 33.24,
+      "learning_rate": 0.0005,
+      "loss": 1.4846,
+      "step": 423900
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4906,
+      "step": 424000
+    },
+    {
+      "epoch": 33.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4812,
+      "step": 424100
+    },
+    {
+      "epoch": 33.27,
+      "learning_rate": 0.0005,
+      "loss": 1.5124,
+      "step": 424200
+    },
+    {
+      "epoch": 33.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4844,
+      "step": 424300
+    },
+    {
+      "epoch": 33.28,
+      "learning_rate": 0.0005,
+      "loss": 1.4951,
+      "step": 424400
+    },
+    {
+      "epoch": 33.29,
+      "learning_rate": 0.0005,
+      "loss": 1.5,
+      "step": 424500
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4982,
+      "step": 424600
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4761,
+      "step": 424700
+    },
+    {
+      "epoch": 33.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4928,
+      "step": 424800
+    },
+    {
+      "epoch": 33.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4939,
+      "step": 424900
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 0.0005,
+      "loss": 1.47,
+      "step": 425000
+    },
+    {
+      "epoch": 33.34,
+      "learning_rate": 0.0005,
+      "loss": 1.506,
+      "step": 425100
+    },
+    {
+      "epoch": 33.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4834,
+      "step": 425200
+    },
+    {
+      "epoch": 33.35,
+      "learning_rate": 0.0005,
+      "loss": 1.4674,
+      "step": 425300
+    },
+    {
+      "epoch": 33.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4815,
+      "step": 425400
+    },
+    {
+      "epoch": 33.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4808,
+      "step": 425500
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 0.0005,
+      "loss": 1.528,
+      "step": 425600
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 0.0005,
+      "loss": 1.4721,
+      "step": 425700
+    },
+    {
+      "epoch": 33.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5043,
+      "step": 425800
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 0.0005,
+      "loss": 1.4849,
+      "step": 425900
+    },
+    {
+      "epoch": 33.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5054,
+      "step": 426000
+    },
+    {
+      "epoch": 33.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4862,
+      "step": 426100
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 0.0005,
+      "loss": 1.4764,
+      "step": 426200
+    },
+    {
+      "epoch": 33.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4996,
+      "step": 426300
+    },
+    {
+      "epoch": 33.44,
+      "learning_rate": 0.0005,
+      "loss": 1.5226,
+      "step": 426400
+    },
+    {
+      "epoch": 33.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4969,
+      "step": 426500
+    },
+    {
+      "epoch": 33.45,
+      "learning_rate": 0.0005,
+      "loss": 1.5382,
+      "step": 426600
+    },
+    {
+      "epoch": 33.46,
+      "learning_rate": 0.0005,
+      "loss": 1.5159,
+      "step": 426700
+    },
+    {
+      "epoch": 33.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5188,
+      "step": 426800
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 0.0005,
+      "loss": 1.5403,
+      "step": 426900
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 0.0005,
+      "loss": 1.5021,
+      "step": 427000
+    },
+    {
+      "epoch": 33.49,
+      "learning_rate": 0.0005,
+      "loss": 1.5213,
+      "step": 427100
+    },
+    {
+      "epoch": 33.5,
+      "learning_rate": 0.0005,
+      "loss": 1.5093,
+      "step": 427200
+    },
+    {
+      "epoch": 33.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5155,
+      "step": 427300
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5006,
+      "step": 427400
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5207,
+      "step": 427500
+    },
+    {
+      "epoch": 33.53,
+      "learning_rate": 0.0005,
+      "loss": 1.526,
+      "step": 427600
+    },
+    {
+      "epoch": 33.54,
+      "learning_rate": 0.0005,
+      "loss": 1.505,
+      "step": 427700
+    },
+    {
+      "epoch": 33.55,
+      "learning_rate": 0.0005,
+      "loss": 1.5,
+      "step": 427800
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 0.0005,
+      "loss": 1.4947,
+      "step": 427900
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 0.0005,
+      "loss": 1.4942,
+      "step": 428000
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 0.0005,
+      "loss": 1.5054,
+      "step": 428100
+    },
+    {
+      "epoch": 33.58,
+      "learning_rate": 0.0005,
+      "loss": 1.5072,
+      "step": 428200
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4926,
+      "step": 428300
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4933,
+      "step": 428400
+    },
+    {
+      "epoch": 33.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5019,
+      "step": 428500
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 0.0005,
+      "loss": 1.503,
+      "step": 428600
+    },
+    {
+      "epoch": 33.62,
+      "learning_rate": 0.0005,
+      "loss": 1.4991,
+      "step": 428700
+    },
+    {
+      "epoch": 33.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4853,
+      "step": 428800
+    },
+    {
+      "epoch": 33.63,
+      "learning_rate": 0.0005,
+      "loss": 1.5268,
+      "step": 428900
+    },
+    {
+      "epoch": 33.64,
+      "learning_rate": 0.0005,
+      "loss": 1.48,
+      "step": 429000
+    },
+    {
+      "epoch": 33.65,
+      "learning_rate": 0.0005,
+      "loss": 1.517,
+      "step": 429100
+    },
+    {
+      "epoch": 33.66,
+      "learning_rate": 0.0005,
+      "loss": 1.5225,
+      "step": 429200
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5214,
+      "step": 429300
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 0.0005,
+      "loss": 1.506,
+      "step": 429400
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 0.0005,
+      "loss": 1.5391,
+      "step": 429500
+    },
+    {
+      "epoch": 33.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5118,
+      "step": 429600
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 0.0005,
+      "loss": 1.4951,
+      "step": 429700
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5314,
+      "step": 429800
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5346,
+      "step": 429900
+    },
+    {
+      "epoch": 33.72,
+      "learning_rate": 0.0005,
+      "loss": 1.5106,
+      "step": 430000
+    },
+    {
+      "epoch": 33.72,
+      "eval_gen_len": 18.773549924958065,
+      "eval_loss": 2.0536205768585205,
+      "eval_rouge1": 35.5461,
+      "eval_rouge2": 14.4953,
+      "eval_rougeL": 29.2671,
+      "eval_rougeLsum": 29.2571,
+      "eval_runtime": 355.0911,
+      "eval_samples_per_second": 31.899,
+      "eval_steps_per_second": 1.994,
+      "step": 430000
+    },
+    {
+      "epoch": 33.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5295,
+      "step": 430100
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5017,
+      "step": 430200
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5423,
+      "step": 430300
+    },
+    {
+      "epoch": 33.75,
+      "learning_rate": 0.0005,
+      "loss": 1.4981,
+      "step": 430400
+    },
+    {
+      "epoch": 33.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5199,
+      "step": 430500
+    },
+    {
+      "epoch": 33.77,
+      "learning_rate": 0.0005,
+      "loss": 1.5,
+      "step": 430600
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5106,
+      "step": 430700
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5251,
+      "step": 430800
+    },
+    {
+      "epoch": 33.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5132,
+      "step": 430900
+    },
+    {
+      "epoch": 33.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5032,
+      "step": 431000
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5136,
+      "step": 431100
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5212,
+      "step": 431200
+    },
+    {
+      "epoch": 33.82,
+      "learning_rate": 0.0005,
+      "loss": 1.5235,
+      "step": 431300
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 0.0005,
+      "loss": 1.4808,
+      "step": 431400
+    },
+    {
+      "epoch": 33.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5457,
+      "step": 431500
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5044,
+      "step": 431600
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5327,
+      "step": 431700
+    },
+    {
+      "epoch": 33.86,
+      "learning_rate": 0.0005,
+      "loss": 1.5262,
+      "step": 431800
+    },
+    {
+      "epoch": 33.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5198,
+      "step": 431900
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5182,
+      "step": 432000
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5306,
+      "step": 432100
+    },
+    {
+      "epoch": 33.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5222,
+      "step": 432200
+    },
+    {
+      "epoch": 33.9,
+      "learning_rate": 0.0005,
+      "loss": 1.488,
+      "step": 432300
+    },
+    {
+      "epoch": 33.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5269,
+      "step": 432400
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5264,
+      "step": 432500
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5207,
+      "step": 432600
+    },
+    {
+      "epoch": 33.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5209,
+      "step": 432700
+    },
+    {
+      "epoch": 33.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5291,
+      "step": 432800
+    },
+    {
+      "epoch": 33.95,
+      "learning_rate": 0.0005,
+      "loss": 1.5382,
+      "step": 432900
+    },
+    {
+      "epoch": 33.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5209,
+      "step": 433000
+    },
+    {
+      "epoch": 33.96,
+      "learning_rate": 0.0005,
+      "loss": 1.5115,
+      "step": 433100
+    },
+    {
+      "epoch": 33.97,
+      "learning_rate": 0.0005,
+      "loss": 1.5056,
+      "step": 433200
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5244,
+      "step": 433300
+    },
+    {
+      "epoch": 33.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5259,
+      "step": 433400
+    },
+    {
+      "epoch": 33.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5489,
+      "step": 433500
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 0.0005,
+      "loss": 1.4757,
+      "step": 433600
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4282,
+      "step": 433700
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 0.0005,
+      "loss": 1.4352,
+      "step": 433800
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4427,
+      "step": 433900
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4662,
+      "step": 434000
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4606,
+      "step": 434100
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4774,
+      "step": 434200
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4842,
+      "step": 434300
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 0.0005,
+      "loss": 1.452,
+      "step": 434400
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 0.0005,
+      "loss": 1.4619,
+      "step": 434500
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 0.0005,
+      "loss": 1.4472,
+      "step": 434600
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 0.0005,
+      "loss": 1.4587,
+      "step": 434700
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4629,
+      "step": 434800
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4664,
+      "step": 434900
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 0.0005,
+      "loss": 1.4545,
+      "step": 435000
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 0.0005,
+      "loss": 1.466,
+      "step": 435100
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4531,
+      "step": 435200
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4366,
+      "step": 435300
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4653,
+      "step": 435400
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4787,
+      "step": 435500
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4584,
+      "step": 435600
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 0.0005,
+      "loss": 1.4645,
+      "step": 435700
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4765,
+      "step": 435800
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4918,
+      "step": 435900
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4779,
+      "step": 436000
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 0.0005,
+      "loss": 1.4586,
+      "step": 436100
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4688,
+      "step": 436200
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4623,
+      "step": 436300
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 0.0005,
+      "loss": 1.4582,
+      "step": 436400
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4566,
+      "step": 436500
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 0.0005,
+      "loss": 1.4839,
+      "step": 436600
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4723,
+      "step": 436700
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4567,
+      "step": 436800
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4795,
+      "step": 436900
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4614,
+      "step": 437000
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 0.0005,
+      "loss": 1.4889,
+      "step": 437100
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 0.0005,
+      "loss": 1.472,
+      "step": 437200
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 0.0005,
+      "loss": 1.4718,
+      "step": 437300
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4689,
+      "step": 437400
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4568,
+      "step": 437500
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4908,
+      "step": 437600
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4646,
+      "step": 437700
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 0.0005,
+      "loss": 1.4859,
+      "step": 437800
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4901,
+      "step": 437900
+    },
+    {
+      "epoch": 34.35,
+      "learning_rate": 0.0005,
+      "loss": 1.4836,
+      "step": 438000
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4968,
+      "step": 438100
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4828,
+      "step": 438200
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4885,
+      "step": 438300
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 0.0005,
+      "loss": 1.4761,
+      "step": 438400
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 0.0005,
+      "loss": 1.5173,
+      "step": 438500
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 0.0005,
+      "loss": 1.4909,
+      "step": 438600
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 0.0005,
+      "loss": 1.4682,
+      "step": 438700
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 0.0005,
+      "loss": 1.5019,
+      "step": 438800
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 0.0005,
+      "loss": 1.5032,
+      "step": 438900
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 0.0005,
+      "loss": 1.5024,
+      "step": 439000
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4704,
+      "step": 439100
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 0.0005,
+      "loss": 1.4926,
+      "step": 439200
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4694,
+      "step": 439300
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 0.0005,
+      "loss": 1.5041,
+      "step": 439400
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 0.0005,
+      "loss": 1.4666,
+      "step": 439500
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 0.0005,
+      "loss": 1.5155,
+      "step": 439600
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4776,
+      "step": 439700
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 0.0005,
+      "loss": 1.4833,
+      "step": 439800
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 0.0005,
+      "loss": 1.4878,
+      "step": 439900
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 0.0005,
+      "loss": 1.5018,
+      "step": 440000
+    },
+    {
+      "epoch": 34.5,
+      "eval_gen_len": 18.794385097554517,
+      "eval_loss": 2.0479230880737305,
+      "eval_rouge1": 35.8012,
+      "eval_rouge2": 14.6585,
+      "eval_rougeL": 29.4215,
+      "eval_rougeLsum": 29.4088,
+      "eval_runtime": 359.8484,
+      "eval_samples_per_second": 31.477,
+      "eval_steps_per_second": 1.967,
+      "step": 440000
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 0.0005,
+      "loss": 1.476,
+      "step": 440100
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 0.0005,
+      "loss": 1.5182,
+      "step": 440200
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 0.0005,
+      "loss": 1.4924,
+      "step": 440300
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 0.0005,
+      "loss": 1.482,
+      "step": 440400
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 0.0005,
+      "loss": 1.4905,
+      "step": 440500
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 0.0005,
+      "loss": 1.4989,
+      "step": 440600
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 0.0005,
+      "loss": 1.5047,
+      "step": 440700
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 0.0005,
+      "loss": 1.5066,
+      "step": 440800
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 0.0005,
+      "loss": 1.4814,
+      "step": 440900
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 0.0005,
+      "loss": 1.4962,
+      "step": 441000
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4896,
+      "step": 441100
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 0.0005,
+      "loss": 1.5113,
+      "step": 441200
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5018,
+      "step": 441300
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 0.0005,
+      "loss": 1.5106,
+      "step": 441400
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 0.0005,
+      "loss": 1.5057,
+      "step": 441500
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4933,
+      "step": 441600
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 0.0005,
+      "loss": 1.4933,
+      "step": 441700
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5129,
+      "step": 441800
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 0.0005,
+      "loss": 1.5032,
+      "step": 441900
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 0.0005,
+      "loss": 1.5064,
+      "step": 442000
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 0.0005,
+      "loss": 1.5261,
+      "step": 442100
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 0.0005,
+      "loss": 1.5168,
+      "step": 442200
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 0.0005,
+      "loss": 1.4994,
+      "step": 442300
+    },
+    {
+      "epoch": 34.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5022,
+      "step": 442400
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5251,
+      "step": 442500
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5243,
+      "step": 442600
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 0.0005,
+      "loss": 1.5109,
+      "step": 442700
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 0.0005,
+      "loss": 1.4982,
+      "step": 442800
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 0.0005,
+      "loss": 1.5158,
+      "step": 442900
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 0.0005,
+      "loss": 1.509,
+      "step": 443000
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5081,
+      "step": 443100
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 0.0005,
+      "loss": 1.5148,
+      "step": 443200
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 0.0005,
+      "loss": 1.4927,
+      "step": 443300
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 0.0005,
+      "loss": 1.5106,
+      "step": 443400
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 0.0005,
+      "loss": 1.4789,
+      "step": 443500
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5104,
+      "step": 443600
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 0.0005,
+      "loss": 1.5178,
+      "step": 443700
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5068,
+      "step": 443800
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 0.0005,
+      "loss": 1.4998,
+      "step": 443900
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 0.0005,
+      "loss": 1.4906,
+      "step": 444000
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 0.0005,
+      "loss": 1.5059,
+      "step": 444100
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 0.0005,
+      "loss": 1.4998,
+      "step": 444200
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5098,
+      "step": 444300
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5164,
+      "step": 444400
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 0.0005,
+      "loss": 1.5034,
+      "step": 444500
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5082,
+      "step": 444600
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5303,
+      "step": 444700
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5157,
+      "step": 444800
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5325,
+      "step": 444900
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 0.0005,
+      "loss": 1.5343,
+      "step": 445000
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 0.0005,
+      "loss": 1.516,
+      "step": 445100
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5298,
+      "step": 445200
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5331,
+      "step": 445300
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5028,
+      "step": 445400
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5004,
+      "step": 445500
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5088,
+      "step": 445600
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 0.0005,
+      "loss": 1.4988,
+      "step": 445700
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 0.0005,
+      "loss": 1.49,
+      "step": 445800
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 0.0005,
+      "loss": 1.5205,
+      "step": 445900
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 0.0005,
+      "loss": 1.4997,
+      "step": 446000
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5047,
+      "step": 446100
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5249,
+      "step": 446200
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5226,
+      "step": 446300
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4523,
+      "step": 446400
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4336,
+      "step": 446500
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 0.0005,
+      "loss": 1.4534,
+      "step": 446600
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4535,
+      "step": 446700
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4416,
+      "step": 446800
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4218,
+      "step": 446900
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4596,
+      "step": 447000
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4474,
+      "step": 447100
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 0.0005,
+      "loss": 1.4481,
+      "step": 447200
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 0.0005,
+      "loss": 1.4636,
+      "step": 447300
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 0.0005,
+      "loss": 1.4575,
+      "step": 447400
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 0.0005,
+      "loss": 1.484,
+      "step": 447500
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 0.0005,
+      "loss": 1.453,
+      "step": 447600
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 0.0005,
+      "loss": 1.4408,
+      "step": 447700
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4777,
+      "step": 447800
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4703,
+      "step": 447900
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4324,
+      "step": 448000
+    },
+    {
+      "epoch": 35.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4664,
+      "step": 448100
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4364,
+      "step": 448200
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4724,
+      "step": 448300
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4512,
+      "step": 448400
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 0.0005,
+      "loss": 1.4756,
+      "step": 448500
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4784,
+      "step": 448600
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4541,
+      "step": 448700
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4614,
+      "step": 448800
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 0.0005,
+      "loss": 1.4543,
+      "step": 448900
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4481,
+      "step": 449000
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 0.0005,
+      "loss": 1.4598,
+      "step": 449100
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4563,
+      "step": 449200
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4464,
+      "step": 449300
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 0.0005,
+      "loss": 1.4748,
+      "step": 449400
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4759,
+      "step": 449500
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4241,
+      "step": 449600
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4666,
+      "step": 449700
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4687,
+      "step": 449800
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 0.0005,
+      "loss": 1.4675,
+      "step": 449900
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 0.0005,
+      "loss": 1.455,
+      "step": 450000
+    },
+    {
+      "epoch": 35.29,
+      "eval_gen_len": 18.781495541626203,
+      "eval_loss": 2.0648272037506104,
+      "eval_rouge1": 35.5722,
+      "eval_rouge2": 14.5725,
+      "eval_rougeL": 29.2816,
+      "eval_rougeLsum": 29.2646,
+      "eval_runtime": 355.571,
+      "eval_samples_per_second": 31.856,
+      "eval_steps_per_second": 1.991,
+      "step": 450000
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4855,
+      "step": 450100
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4581,
+      "step": 450200
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4779,
+      "step": 450300
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4595,
+      "step": 450400
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 0.0005,
+      "loss": 1.4915,
+      "step": 450500
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4983,
+      "step": 450600
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4711,
+      "step": 450700
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 0.0005,
+      "loss": 1.4428,
+      "step": 450800
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4672,
+      "step": 450900
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4455,
+      "step": 451000
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 0.0005,
+      "loss": 1.467,
+      "step": 451100
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 0.0005,
+      "loss": 1.459,
+      "step": 451200
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 0.0005,
+      "loss": 1.4694,
+      "step": 451300
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 0.0005,
+      "loss": 1.4676,
+      "step": 451400
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4714,
+      "step": 451500
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 0.0005,
+      "loss": 1.495,
+      "step": 451600
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 0.0005,
+      "loss": 1.4767,
+      "step": 451700
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4694,
+      "step": 451800
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 0.0005,
+      "loss": 1.4619,
+      "step": 451900
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4719,
+      "step": 452000
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4722,
+      "step": 452100
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 0.0005,
+      "loss": 1.4741,
+      "step": 452200
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 0.0005,
+      "loss": 1.4872,
+      "step": 452300
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4847,
+      "step": 452400
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4803,
+      "step": 452500
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 0.0005,
+      "loss": 1.4905,
+      "step": 452600
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 0.0005,
+      "loss": 1.4646,
+      "step": 452700
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 0.0005,
+      "loss": 1.5042,
+      "step": 452800
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 0.0005,
+      "loss": 1.4808,
+      "step": 452900
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 0.0005,
+      "loss": 1.4739,
+      "step": 453000
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 0.0005,
+      "loss": 1.4859,
+      "step": 453100
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 0.0005,
+      "loss": 1.492,
+      "step": 453200
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 0.0005,
+      "loss": 1.484,
+      "step": 453300
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 0.0005,
+      "loss": 1.4576,
+      "step": 453400
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 0.0005,
+      "loss": 1.4851,
+      "step": 453500
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 0.0005,
+      "loss": 1.4735,
+      "step": 453600
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 0.0005,
+      "loss": 1.4744,
+      "step": 453700
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4876,
+      "step": 453800
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4768,
+      "step": 453900
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 0.0005,
+      "loss": 1.4672,
+      "step": 454000
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 0.0005,
+      "loss": 1.4988,
+      "step": 454100
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 0.0005,
+      "loss": 1.4867,
+      "step": 454200
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4815,
+      "step": 454300
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4974,
+      "step": 454400
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 0.0005,
+      "loss": 1.4821,
+      "step": 454500
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 0.0005,
+      "loss": 1.4884,
+      "step": 454600
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 0.0005,
+      "loss": 1.4819,
+      "step": 454700
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 0.0005,
+      "loss": 1.492,
+      "step": 454800
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 0.0005,
+      "loss": 1.4862,
+      "step": 454900
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 0.0005,
+      "loss": 1.4751,
+      "step": 455000
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 0.0005,
+      "loss": 1.4783,
+      "step": 455100
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 0.0005,
+      "loss": 1.4948,
+      "step": 455200
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 0.0005,
+      "loss": 1.5175,
+      "step": 455300
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 0.0005,
+      "loss": 1.5152,
+      "step": 455400
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 0.0005,
+      "loss": 1.4671,
+      "step": 455500
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 0.0005,
+      "loss": 1.4956,
+      "step": 455600
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 0.0005,
+      "loss": 1.5073,
+      "step": 455700
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 0.0005,
+      "loss": 1.4786,
+      "step": 455800
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5124,
+      "step": 455900
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 0.0005,
+      "loss": 1.4967,
+      "step": 456000
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 0.0005,
+      "loss": 1.4846,
+      "step": 456100
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 0.0005,
+      "loss": 1.4982,
+      "step": 456200
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 0.0005,
+      "loss": 1.5005,
+      "step": 456300
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 0.0005,
+      "loss": 1.504,
+      "step": 456400
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 0.0005,
+      "loss": 1.4895,
+      "step": 456500
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 0.0005,
+      "loss": 1.4705,
+      "step": 456600
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 0.0005,
+      "loss": 1.4976,
+      "step": 456700
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 0.0005,
+      "loss": 1.4959,
+      "step": 456800
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 0.0005,
+      "loss": 1.4716,
+      "step": 456900
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 0.0005,
+      "loss": 1.4709,
+      "step": 457000
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5104,
+      "step": 457100
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 0.0005,
+      "loss": 1.5258,
+      "step": 457200
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 0.0005,
+      "loss": 1.4968,
+      "step": 457300
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 0.0005,
+      "loss": 1.5232,
+      "step": 457400
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5145,
+      "step": 457500
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 0.0005,
+      "loss": 1.5133,
+      "step": 457600
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 0.0005,
+      "loss": 1.4737,
+      "step": 457700
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 0.0005,
+      "loss": 1.513,
+      "step": 457800
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 0.0005,
+      "loss": 1.5141,
+      "step": 457900
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5234,
+      "step": 458000
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5103,
+      "step": 458100
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 0.0005,
+      "loss": 1.5067,
+      "step": 458200
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 0.0005,
+      "loss": 1.4814,
+      "step": 458300
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 0.0005,
+      "loss": 1.527,
+      "step": 458400
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 0.0005,
+      "loss": 1.4959,
+      "step": 458500
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 0.0005,
+      "loss": 1.48,
+      "step": 458600
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 0.0005,
+      "loss": 1.499,
+      "step": 458700
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5181,
+      "step": 458800
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5087,
+      "step": 458900
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5377,
+      "step": 459000
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 0.0005,
+      "loss": 1.5162,
+      "step": 459100
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4301,
+      "step": 459200
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 0.0005,
+      "loss": 1.466,
+      "step": 459300
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 0.0005,
+      "loss": 1.435,
+      "step": 459400
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4635,
+      "step": 459500
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4322,
+      "step": 459600
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4237,
+      "step": 459700
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4357,
+      "step": 459800
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4613,
+      "step": 459900
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 0.0005,
+      "loss": 1.4209,
+      "step": 460000
+    },
+    {
+      "epoch": 36.07,
+      "eval_gen_len": 18.764368323474883,
+      "eval_loss": 2.0837438106536865,
+      "eval_rouge1": 35.6324,
+      "eval_rouge2": 14.6282,
+      "eval_rougeL": 29.3399,
+      "eval_rougeLsum": 29.3299,
+      "eval_runtime": 358.8308,
+      "eval_samples_per_second": 31.566,
+      "eval_steps_per_second": 1.973,
+      "step": 460000
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 0.0005,
+      "loss": 1.4281,
+      "step": 460100
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 0.0005,
+      "loss": 1.4115,
+      "step": 460200
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4535,
+      "step": 460300
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4266,
+      "step": 460400
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 0.0005,
+      "loss": 1.446,
+      "step": 460500
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4321,
+      "step": 460600
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4412,
+      "step": 460700
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4506,
+      "step": 460800
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4445,
+      "step": 460900
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 0.0005,
+      "loss": 1.454,
+      "step": 461000
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4436,
+      "step": 461100
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 0.0005,
+      "loss": 1.445,
+      "step": 461200
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 0.0005,
+      "loss": 1.4234,
+      "step": 461300
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 0.0005,
+      "loss": 1.425,
+      "step": 461400
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4417,
+      "step": 461500
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 0.0005,
+      "loss": 1.4736,
+      "step": 461600
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4525,
+      "step": 461700
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4606,
+      "step": 461800
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 0.0005,
+      "loss": 1.4473,
+      "step": 461900
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4424,
+      "step": 462000
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 0.0005,
+      "loss": 1.4631,
+      "step": 462100
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4441,
+      "step": 462200
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4454,
+      "step": 462300
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4588,
+      "step": 462400
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4737,
+      "step": 462500
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 0.0005,
+      "loss": 1.4574,
+      "step": 462600
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 0.0005,
+      "loss": 1.4646,
+      "step": 462700
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 0.0005,
+      "loss": 1.4646,
+      "step": 462800
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4558,
+      "step": 462900
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4674,
+      "step": 463000
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4519,
+      "step": 463100
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4642,
+      "step": 463200
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 0.0005,
+      "loss": 1.4722,
+      "step": 463300
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4422,
+      "step": 463400
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 0.0005,
+      "loss": 1.4698,
+      "step": 463500
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4571,
+      "step": 463600
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4748,
+      "step": 463700
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4678,
+      "step": 463800
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 0.0005,
+      "loss": 1.4802,
+      "step": 463900
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 0.0005,
+      "loss": 1.4493,
+      "step": 464000
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 0.0005,
+      "loss": 1.4593,
+      "step": 464100
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 0.0005,
+      "loss": 1.4984,
+      "step": 464200
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 0.0005,
+      "loss": 1.456,
+      "step": 464300
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 0.0005,
+      "loss": 1.4591,
+      "step": 464400
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4638,
+      "step": 464500
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4558,
+      "step": 464600
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 0.0005,
+      "loss": 1.465,
+      "step": 464700
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4669,
+      "step": 464800
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 0.0005,
+      "loss": 1.4679,
+      "step": 464900
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 0.0005,
+      "loss": 1.473,
+      "step": 465000
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 0.0005,
+      "loss": 1.4825,
+      "step": 465100
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4514,
+      "step": 465200
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 0.0005,
+      "loss": 1.4839,
+      "step": 465300
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 0.0005,
+      "loss": 1.4616,
+      "step": 465400
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 0.0005,
+      "loss": 1.4662,
+      "step": 465500
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 0.0005,
+      "loss": 1.4781,
+      "step": 465600
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 0.0005,
+      "loss": 1.4787,
+      "step": 465700
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 0.0005,
+      "loss": 1.4712,
+      "step": 465800
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 0.0005,
+      "loss": 1.4667,
+      "step": 465900
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 0.0005,
+      "loss": 1.4688,
+      "step": 466000
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 0.0005,
+      "loss": 1.4499,
+      "step": 466100
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 0.0005,
+      "loss": 1.4675,
+      "step": 466200
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 0.0005,
+      "loss": 1.4699,
+      "step": 466300
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 0.0005,
+      "loss": 1.4869,
+      "step": 466400
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 0.0005,
+      "loss": 1.4657,
+      "step": 466500
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 0.0005,
+      "loss": 1.46,
+      "step": 466600
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 0.0005,
+      "loss": 1.4743,
+      "step": 466700
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 0.0005,
+      "loss": 1.4902,
+      "step": 466800
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 0.0005,
+      "loss": 1.463,
+      "step": 466900
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 0.0005,
+      "loss": 1.4619,
+      "step": 467000
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4819,
+      "step": 467100
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 0.0005,
+      "loss": 1.4381,
+      "step": 467200
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 0.0005,
+      "loss": 1.4876,
+      "step": 467300
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 0.0005,
+      "loss": 1.4934,
+      "step": 467400
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 0.0005,
+      "loss": 1.4673,
+      "step": 467500
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 0.0005,
+      "loss": 1.4811,
+      "step": 467600
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 0.0005,
+      "loss": 1.4584,
+      "step": 467700
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 0.0005,
+      "loss": 1.4824,
+      "step": 467800
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 0.0005,
+      "loss": 1.5057,
+      "step": 467900
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 0.0005,
+      "loss": 1.4781,
+      "step": 468000
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 0.0005,
+      "loss": 1.4791,
+      "step": 468100
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 0.0005,
+      "loss": 1.507,
+      "step": 468200
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 0.0005,
+      "loss": 1.4879,
+      "step": 468300
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 0.0005,
+      "loss": 1.4758,
+      "step": 468400
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 0.0005,
+      "loss": 1.4805,
+      "step": 468500
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 0.0005,
+      "loss": 1.5019,
+      "step": 468600
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 0.0005,
+      "loss": 1.4877,
+      "step": 468700
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 0.0005,
+      "loss": 1.4739,
+      "step": 468800
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 0.0005,
+      "loss": 1.456,
+      "step": 468900
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 0.0005,
+      "loss": 1.4748,
+      "step": 469000
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 0.0005,
+      "loss": 1.4974,
+      "step": 469100
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 0.0005,
+      "loss": 1.4883,
+      "step": 469200
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 0.0005,
+      "loss": 1.4729,
+      "step": 469300
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 0.0005,
+      "loss": 1.485,
+      "step": 469400
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 0.0005,
+      "loss": 1.4733,
+      "step": 469500
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 0.0005,
+      "loss": 1.4927,
+      "step": 469600
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 0.0005,
+      "loss": 1.4826,
+      "step": 469700
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 0.0005,
+      "loss": 1.5009,
+      "step": 469800
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 0.0005,
+      "loss": 1.4938,
+      "step": 469900
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 0.0005,
+      "loss": 1.4894,
+      "step": 470000
+    },
+    {
+      "epoch": 36.86,
+      "eval_gen_len": 18.78617462699744,
+      "eval_loss": 2.0626096725463867,
+      "eval_rouge1": 35.7293,
+      "eval_rouge2": 14.5534,
+      "eval_rougeL": 29.39,
+      "eval_rougeLsum": 29.3792,
+      "eval_runtime": 354.5618,
+      "eval_samples_per_second": 31.946,
+      "eval_steps_per_second": 1.997,
+      "step": 470000
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 0.0005,
+      "loss": 1.506,
+      "step": 470100
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 0.0005,
+      "loss": 1.4744,
+      "step": 470200
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 0.0005,
+      "loss": 1.4895,
+      "step": 470300
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 0.0005,
+      "loss": 1.5299,
+      "step": 470400
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 0.0005,
+      "loss": 1.4758,
+      "step": 470500
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 0.0005,
+      "loss": 1.4826,
+      "step": 470600
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 0.0005,
+      "loss": 1.4854,
+      "step": 470700
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 0.0005,
+      "loss": 1.4919,
+      "step": 470800
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 0.0005,
+      "loss": 1.4969,
+      "step": 470900
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5361,
+      "step": 471000
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 0.0005,
+      "loss": 1.4955,
+      "step": 471100
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 0.0005,
+      "loss": 1.4918,
+      "step": 471200
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 0.0005,
+      "loss": 1.4706,
+      "step": 471300
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 0.0005,
+      "loss": 1.508,
+      "step": 471400
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 0.0005,
+      "loss": 1.4943,
+      "step": 471500
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 0.0005,
+      "loss": 1.5189,
+      "step": 471600
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5013,
+      "step": 471700
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 0.0005,
+      "loss": 1.4928,
+      "step": 471800
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4571,
+      "step": 471900
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 0.0005,
+      "loss": 1.422,
+      "step": 472000
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 0.0005,
+      "loss": 1.3969,
+      "step": 472100
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4296,
+      "step": 472200
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4178,
+      "step": 472300
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4167,
+      "step": 472400
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 0.0005,
+      "loss": 1.4105,
+      "step": 472500
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4229,
+      "step": 472600
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 0.0005,
+      "loss": 1.427,
+      "step": 472700
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 0.0005,
+      "loss": 1.4127,
+      "step": 472800
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 0.0005,
+      "loss": 1.3988,
+      "step": 472900
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 0.0005,
+      "loss": 1.4349,
+      "step": 473000
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 0.0005,
+      "loss": 1.383,
+      "step": 473100
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 0.0005,
+      "loss": 1.4186,
+      "step": 473200
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4331,
+      "step": 473300
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4529,
+      "step": 473400
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4127,
+      "step": 473500
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4333,
+      "step": 473600
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4442,
+      "step": 473700
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4151,
+      "step": 473800
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4192,
+      "step": 473900
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 0.0005,
+      "loss": 1.4437,
+      "step": 474000
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4698,
+      "step": 474100
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4343,
+      "step": 474200
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 0.0005,
+      "loss": 1.456,
+      "step": 474300
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 0.0005,
+      "loss": 1.4433,
+      "step": 474400
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4381,
+      "step": 474500
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 0.0005,
+      "loss": 1.4324,
+      "step": 474600
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4325,
+      "step": 474700
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4719,
+      "step": 474800
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 0.0005,
+      "loss": 1.4383,
+      "step": 474900
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4518,
+      "step": 475000
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 0.0005,
+      "loss": 1.457,
+      "step": 475100
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4485,
+      "step": 475200
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4477,
+      "step": 475300
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 0.0005,
+      "loss": 1.4299,
+      "step": 475400
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 0.0005,
+      "loss": 1.4411,
+      "step": 475500
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4588,
+      "step": 475600
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 0.0005,
+      "loss": 1.453,
+      "step": 475700
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4495,
+      "step": 475800
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4252,
+      "step": 475900
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 0.0005,
+      "loss": 1.4523,
+      "step": 476000
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 0.0005,
+      "loss": 1.434,
+      "step": 476100
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4625,
+      "step": 476200
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 0.0005,
+      "loss": 1.4547,
+      "step": 476300
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4441,
+      "step": 476400
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4308,
+      "step": 476500
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4641,
+      "step": 476600
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 0.0005,
+      "loss": 1.4441,
+      "step": 476700
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 0.0005,
+      "loss": 1.4377,
+      "step": 476800
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 0.0005,
+      "loss": 1.4943,
+      "step": 476900
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4712,
+      "step": 477000
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4618,
+      "step": 477100
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 0.0005,
+      "loss": 1.4653,
+      "step": 477200
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4633,
+      "step": 477300
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 0.0005,
+      "loss": 1.4464,
+      "step": 477400
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4462,
+      "step": 477500
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4831,
+      "step": 477600
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 0.0005,
+      "loss": 1.4498,
+      "step": 477700
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 0.0005,
+      "loss": 1.4278,
+      "step": 477800
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4742,
+      "step": 477900
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4649,
+      "step": 478000
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 0.0005,
+      "loss": 1.4472,
+      "step": 478100
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 0.0005,
+      "loss": 1.4375,
+      "step": 478200
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 0.0005,
+      "loss": 1.4675,
+      "step": 478300
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 0.0005,
+      "loss": 1.4489,
+      "step": 478400
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 0.0005,
+      "loss": 1.476,
+      "step": 478500
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 0.0005,
+      "loss": 1.495,
+      "step": 478600
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 0.0005,
+      "loss": 1.4534,
+      "step": 478700
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 0.0005,
+      "loss": 1.4523,
+      "step": 478800
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 0.0005,
+      "loss": 1.4639,
+      "step": 478900
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 0.0005,
+      "loss": 1.464,
+      "step": 479000
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 0.0005,
+      "loss": 1.4707,
+      "step": 479100
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 0.0005,
+      "loss": 1.4779,
+      "step": 479200
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4698,
+      "step": 479300
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 0.0005,
+      "loss": 1.459,
+      "step": 479400
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 0.0005,
+      "loss": 1.474,
+      "step": 479500
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 0.0005,
+      "loss": 1.4644,
+      "step": 479600
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 0.0005,
+      "loss": 1.4517,
+      "step": 479700
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4844,
+      "step": 479800
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4627,
+      "step": 479900
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 0.0005,
+      "loss": 1.4801,
+      "step": 480000
+    },
+    {
+      "epoch": 37.64,
+      "eval_gen_len": 18.763485477178424,
+      "eval_loss": 2.0635464191436768,
+      "eval_rouge1": 35.7102,
+      "eval_rouge2": 14.5234,
+      "eval_rougeL": 29.4028,
+      "eval_rougeLsum": 29.3844,
+      "eval_runtime": 356.8998,
+      "eval_samples_per_second": 31.737,
+      "eval_steps_per_second": 1.984,
+      "step": 480000
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 0.0005,
+      "loss": 1.4629,
+      "step": 480100
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 0.0005,
+      "loss": 1.4369,
+      "step": 480200
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 0.0005,
+      "loss": 1.5003,
+      "step": 480300
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 0.0005,
+      "loss": 1.4944,
+      "step": 480400
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 0.0005,
+      "loss": 1.4683,
+      "step": 480500
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 0.0005,
+      "loss": 1.4545,
+      "step": 480600
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 0.0005,
+      "loss": 1.4789,
+      "step": 480700
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 0.0005,
+      "loss": 1.4754,
+      "step": 480800
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 0.0005,
+      "loss": 1.4473,
+      "step": 480900
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 0.0005,
+      "loss": 1.4698,
+      "step": 481000
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 0.0005,
+      "loss": 1.4793,
+      "step": 481100
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 0.0005,
+      "loss": 1.4657,
+      "step": 481200
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 0.0005,
+      "loss": 1.4849,
+      "step": 481300
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 0.0005,
+      "loss": 1.4639,
+      "step": 481400
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 0.0005,
+      "loss": 1.4791,
+      "step": 481500
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 0.0005,
+      "loss": 1.4802,
+      "step": 481600
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 0.0005,
+      "loss": 1.4676,
+      "step": 481700
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 0.0005,
+      "loss": 1.4727,
+      "step": 481800
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 0.0005,
+      "loss": 1.4757,
+      "step": 481900
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 0.0005,
+      "loss": 1.5054,
+      "step": 482000
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 0.0005,
+      "loss": 1.4704,
+      "step": 482100
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 0.0005,
+      "loss": 1.4794,
+      "step": 482200
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 0.0005,
+      "loss": 1.4873,
+      "step": 482300
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 0.0005,
+      "loss": 1.4662,
+      "step": 482400
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 0.0005,
+      "loss": 1.4757,
+      "step": 482500
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 0.0005,
+      "loss": 1.4867,
+      "step": 482600
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 0.0005,
+      "loss": 1.4748,
+      "step": 482700
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 0.0005,
+      "loss": 1.5006,
+      "step": 482800
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 0.0005,
+      "loss": 1.4998,
+      "step": 482900
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 0.0005,
+      "loss": 1.4715,
+      "step": 483000
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 0.0005,
+      "loss": 1.4772,
+      "step": 483100
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 0.0005,
+      "loss": 1.477,
+      "step": 483200
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 0.0005,
+      "loss": 1.4816,
+      "step": 483300
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 0.0005,
+      "loss": 1.4715,
+      "step": 483400
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 0.0005,
+      "loss": 1.4821,
+      "step": 483500
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 0.0005,
+      "loss": 1.5009,
+      "step": 483600
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 0.0005,
+      "loss": 1.4916,
+      "step": 483700
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 0.0005,
+      "loss": 1.5157,
+      "step": 483800
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 0.0005,
+      "loss": 1.4821,
+      "step": 483900
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 0.0005,
+      "loss": 1.4669,
+      "step": 484000
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 0.0005,
+      "loss": 1.4763,
+      "step": 484100
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 0.0005,
+      "loss": 1.4832,
+      "step": 484200
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 0.0005,
+      "loss": 1.4875,
+      "step": 484300
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5,
+      "step": 484400
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 0.0005,
+      "loss": 1.4921,
+      "step": 484500
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 0.0005,
+      "loss": 1.4853,
+      "step": 484600
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 0.0005,
+      "loss": 1.4102,
+      "step": 484700
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 0.0005,
+      "loss": 1.4333,
+      "step": 484800
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4171,
+      "step": 484900
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4047,
+      "step": 485000
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4108,
+      "step": 485100
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 0.0005,
+      "loss": 1.3879,
+      "step": 485200
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4144,
+      "step": 485300
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 0.0005,
+      "loss": 1.4304,
+      "step": 485400
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 0.0005,
+      "loss": 1.3913,
+      "step": 485500
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 0.0005,
+      "loss": 1.4303,
+      "step": 485600
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 0.0005,
+      "loss": 1.3798,
+      "step": 485700
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4202,
+      "step": 485800
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 0.0005,
+      "loss": 1.4066,
+      "step": 485900
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 0.0005,
+      "loss": 1.4153,
+      "step": 486000
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4256,
+      "step": 486100
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4468,
+      "step": 486200
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 0.0005,
+      "loss": 1.3938,
+      "step": 486300
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4367,
+      "step": 486400
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4248,
+      "step": 486500
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4351,
+      "step": 486600
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 0.0005,
+      "loss": 1.4488,
+      "step": 486700
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 0.0005,
+      "loss": 1.4195,
+      "step": 486800
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4355,
+      "step": 486900
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4374,
+      "step": 487000
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 0.0005,
+      "loss": 1.4443,
+      "step": 487100
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4462,
+      "step": 487200
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 0.0005,
+      "loss": 1.423,
+      "step": 487300
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 0.0005,
+      "loss": 1.4214,
+      "step": 487400
+    },
+    {
+      "epoch": 38.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4318,
+      "step": 487500
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 0.0005,
+      "loss": 1.4522,
+      "step": 487600
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 0.0005,
+      "loss": 1.4219,
+      "step": 487700
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4298,
+      "step": 487800
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4482,
+      "step": 487900
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4304,
+      "step": 488000
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 0.0005,
+      "loss": 1.4545,
+      "step": 488100
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 0.0005,
+      "loss": 1.4368,
+      "step": 488200
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 0.0005,
+      "loss": 1.4392,
+      "step": 488300
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 0.0005,
+      "loss": 1.435,
+      "step": 488400
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4098,
+      "step": 488500
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4293,
+      "step": 488600
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4606,
+      "step": 488700
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 0.0005,
+      "loss": 1.4354,
+      "step": 488800
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4521,
+      "step": 488900
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 0.0005,
+      "loss": 1.4369,
+      "step": 489000
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 0.0005,
+      "loss": 1.4305,
+      "step": 489100
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4358,
+      "step": 489200
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4482,
+      "step": 489300
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 0.0005,
+      "loss": 1.4456,
+      "step": 489400
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 0.0005,
+      "loss": 1.4476,
+      "step": 489500
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 0.0005,
+      "loss": 1.4623,
+      "step": 489600
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 0.0005,
+      "loss": 1.4577,
+      "step": 489700
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4286,
+      "step": 489800
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 0.0005,
+      "loss": 1.4479,
+      "step": 489900
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4553,
+      "step": 490000
+    },
+    {
+      "epoch": 38.43,
+      "eval_gen_len": 18.818486801447868,
+      "eval_loss": 2.0702340602874756,
+      "eval_rouge1": 35.6617,
+      "eval_rouge2": 14.5328,
+      "eval_rougeL": 29.2952,
+      "eval_rougeLsum": 29.2949,
+      "eval_runtime": 360.1467,
+      "eval_samples_per_second": 31.451,
+      "eval_steps_per_second": 1.966,
+      "step": 490000
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4381,
+      "step": 490100
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 0.0005,
+      "loss": 1.4515,
+      "step": 490200
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4214,
+      "step": 490300
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 0.0005,
+      "loss": 1.4283,
+      "step": 490400
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 0.0005,
+      "loss": 1.435,
+      "step": 490500
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 0.0005,
+      "loss": 1.4338,
+      "step": 490600
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4418,
+      "step": 490700
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 0.0005,
+      "loss": 1.4445,
+      "step": 490800
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 0.0005,
+      "loss": 1.4408,
+      "step": 490900
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 0.0005,
+      "loss": 1.4541,
+      "step": 491000
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 0.0005,
+      "loss": 1.4602,
+      "step": 491100
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 0.0005,
+      "loss": 1.4278,
+      "step": 491200
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 0.0005,
+      "loss": 1.4641,
+      "step": 491300
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 0.0005,
+      "loss": 1.4253,
+      "step": 491400
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 0.0005,
+      "loss": 1.4791,
+      "step": 491500
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 0.0005,
+      "loss": 1.4497,
+      "step": 491600
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 0.0005,
+      "loss": 1.4517,
+      "step": 491700
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 0.0005,
+      "loss": 1.4639,
+      "step": 491800
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 0.0005,
+      "loss": 1.4773,
+      "step": 491900
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 0.0005,
+      "loss": 1.4668,
+      "step": 492000
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4547,
+      "step": 492100
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 0.0005,
+      "loss": 1.4635,
+      "step": 492200
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 0.0005,
+      "loss": 1.4822,
+      "step": 492300
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 0.0005,
+      "loss": 1.4691,
+      "step": 492400
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 0.0005,
+      "loss": 1.4819,
+      "step": 492500
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4483,
+      "step": 492600
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 0.0005,
+      "loss": 1.4592,
+      "step": 492700
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 0.0005,
+      "loss": 1.4713,
+      "step": 492800
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 0.0005,
+      "loss": 1.4678,
+      "step": 492900
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 0.0005,
+      "loss": 1.461,
+      "step": 493000
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 0.0005,
+      "loss": 1.4566,
+      "step": 493100
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 0.0005,
+      "loss": 1.4501,
+      "step": 493200
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 0.0005,
+      "loss": 1.4737,
+      "step": 493300
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 0.0005,
+      "loss": 1.4579,
+      "step": 493400
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 0.0005,
+      "loss": 1.4737,
+      "step": 493500
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 0.0005,
+      "loss": 1.4571,
+      "step": 493600
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 0.0005,
+      "loss": 1.4711,
+      "step": 493700
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 0.0005,
+      "loss": 1.4522,
+      "step": 493800
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 0.0005,
+      "loss": 1.4809,
+      "step": 493900
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 0.0005,
+      "loss": 1.4816,
+      "step": 494000
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 0.0005,
+      "loss": 1.4372,
+      "step": 494100
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 0.0005,
+      "loss": 1.4837,
+      "step": 494200
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 0.0005,
+      "loss": 1.4849,
+      "step": 494300
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 0.0005,
+      "loss": 1.4664,
+      "step": 494400
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 0.0005,
+      "loss": 1.4442,
+      "step": 494500
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 0.0005,
+      "loss": 1.4493,
+      "step": 494600
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 0.0005,
+      "loss": 1.45,
+      "step": 494700
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 0.0005,
+      "loss": 1.4717,
+      "step": 494800
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 0.0005,
+      "loss": 1.4816,
+      "step": 494900
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 0.0005,
+      "loss": 1.4556,
+      "step": 495000
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 0.0005,
+      "loss": 1.4619,
+      "step": 495100
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 0.0005,
+      "loss": 1.4585,
+      "step": 495200
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 0.0005,
+      "loss": 1.4512,
+      "step": 495300
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 0.0005,
+      "loss": 1.4606,
+      "step": 495400
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 0.0005,
+      "loss": 1.4713,
+      "step": 495500
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 0.0005,
+      "loss": 1.4917,
+      "step": 495600
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 0.0005,
+      "loss": 1.4603,
+      "step": 495700
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 0.0005,
+      "loss": 1.4885,
+      "step": 495800
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 0.0005,
+      "loss": 1.4555,
+      "step": 495900
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 0.0005,
+      "loss": 1.4674,
+      "step": 496000
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 0.0005,
+      "loss": 1.4614,
+      "step": 496100
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 0.0005,
+      "loss": 1.4815,
+      "step": 496200
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 0.0005,
+      "loss": 1.4637,
+      "step": 496300
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 0.0005,
+      "loss": 1.475,
+      "step": 496400
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 0.0005,
+      "loss": 1.4696,
+      "step": 496500
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 0.0005,
+      "loss": 1.4896,
+      "step": 496600
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 0.0005,
+      "loss": 1.4555,
+      "step": 496700
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 0.0005,
+      "loss": 1.4873,
+      "step": 496800
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 0.0005,
+      "loss": 1.4663,
+      "step": 496900
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 0.0005,
+      "loss": 1.4983,
+      "step": 497000
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 0.0005,
+      "loss": 1.4715,
+      "step": 497100
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 0.0005,
+      "loss": 1.5052,
+      "step": 497200
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 0.0005,
+      "loss": 1.4862,
+      "step": 497300
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 0.0005,
+      "loss": 1.412,
+      "step": 497400
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 0.0005,
+      "loss": 1.3872,
+      "step": 497500
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 0.0005,
+      "loss": 1.3971,
+      "step": 497600
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 0.0005,
+      "loss": 1.4171,
+      "step": 497700
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 0.0005,
+      "loss": 1.4026,
+      "step": 497800
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 0.0005,
+      "loss": 1.39,
+      "step": 497900
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 0.0005,
+      "loss": 1.3829,
+      "step": 498000
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 0.0005,
+      "loss": 1.3924,
+      "step": 498100
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 0.0005,
+      "loss": 1.396,
+      "step": 498200
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 0.0005,
+      "loss": 1.386,
+      "step": 498300
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 0.0005,
+      "loss": 1.3992,
+      "step": 498400
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 0.0005,
+      "loss": 1.4199,
+      "step": 498500
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 0.0005,
+      "loss": 1.3924,
+      "step": 498600
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 0.0005,
+      "loss": 1.402,
+      "step": 498700
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4241,
+      "step": 498800
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 0.0005,
+      "loss": 1.4045,
+      "step": 498900
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 0.0005,
+      "loss": 1.4163,
+      "step": 499000
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 0.0005,
+      "loss": 1.4071,
+      "step": 499100
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4081,
+      "step": 499200
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 0.0005,
+      "loss": 1.4432,
+      "step": 499300
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 0.0005,
+      "loss": 1.4187,
+      "step": 499400
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 0.0005,
+      "loss": 1.4159,
+      "step": 499500
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 0.0005,
+      "loss": 1.4294,
+      "step": 499600
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4187,
+      "step": 499700
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 0.0005,
+      "loss": 1.4207,
+      "step": 499800
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 0.0005,
+      "loss": 1.429,
+      "step": 499900
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 0.0005,
+      "loss": 1.4295,
+      "step": 500000
+    },
+    {
+      "epoch": 39.21,
+      "eval_gen_len": 18.78025955681116,
+      "eval_loss": 2.1033854484558105,
+      "eval_rouge1": 35.8815,
+      "eval_rouge2": 14.7785,
+      "eval_rougeL": 29.503,
+      "eval_rougeLsum": 29.4979,
+      "eval_runtime": 355.4667,
+      "eval_samples_per_second": 31.865,
+      "eval_steps_per_second": 1.992,
+      "step": 500000
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 0.0005,
+      "loss": 1.4241,
+      "step": 500100
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4048,
+      "step": 500200
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 0.0005,
+      "loss": 1.4098,
+      "step": 500300
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 0.0005,
+      "loss": 1.3955,
+      "step": 500400
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 0.0005,
+      "loss": 1.4262,
+      "step": 500500
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4024,
+      "step": 500600
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 0.0005,
+      "loss": 1.4432,
+      "step": 500700
+    },
+    {
+      "epoch": 39.27,
+      "learning_rate": 0.0005,
+      "loss": 1.4488,
+      "step": 500800
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 0.0005,
+      "loss": 1.4148,
+      "step": 500900
+    },
+    {
+      "epoch": 39.29,
+      "learning_rate": 0.0005,
+      "loss": 1.4618,
+      "step": 501000
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4333,
+      "step": 501100
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 0.0005,
+      "loss": 1.4268,
+      "step": 501200
+    },
+    {
+      "epoch": 39.31,
+      "learning_rate": 0.0005,
+      "loss": 1.4125,
+      "step": 501300
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 0.0005,
+      "loss": 1.4449,
+      "step": 501400
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 0.0005,
+      "loss": 1.4244,
+      "step": 501500
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4344,
+      "step": 501600
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 0.0005,
+      "loss": 1.4306,
+      "step": 501700
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 0.0005,
+      "loss": 1.447,
+      "step": 501800
+    },
+    {
+      "epoch": 39.36,
+      "learning_rate": 0.0005,
+      "loss": 1.4367,
+      "step": 501900
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4568,
+      "step": 502000
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 0.0005,
+      "loss": 1.4365,
+      "step": 502100
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 0.0005,
+      "loss": 1.449,
+      "step": 502200
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 0.0005,
+      "loss": 1.4194,
+      "step": 502300
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 0.0005,
+      "loss": 1.4658,
+      "step": 502400
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4472,
+      "step": 502500
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 0.0005,
+      "loss": 1.4311,
+      "step": 502600
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 0.0005,
+      "loss": 1.448,
+      "step": 502700
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 0.0005,
+      "loss": 1.4346,
+      "step": 502800
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 0.0005,
+      "loss": 1.4348,
+      "step": 502900
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 0.0005,
+      "loss": 1.4081,
+      "step": 503000
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 0.0005,
+      "loss": 1.4255,
+      "step": 503100
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 0.0005,
+      "loss": 1.4441,
+      "step": 503200
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 0.0005,
+      "loss": 1.4332,
+      "step": 503300
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 0.0005,
+      "loss": 1.4609,
+      "step": 503400
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 0.0005,
+      "loss": 1.452,
+      "step": 503500
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 0.0005,
+      "loss": 1.4605,
+      "step": 503600
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 0.0005,
+      "loss": 1.4327,
+      "step": 503700
+    },
+    {
+      "epoch": 39.51,
+      "learning_rate": 0.0005,
+      "loss": 1.4194,
+      "step": 503800
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 0.0005,
+      "loss": 1.4626,
+      "step": 503900
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 0.0005,
+      "loss": 1.4482,
+      "step": 504000
+    },
+    {
+      "epoch": 39.53,
+      "learning_rate": 0.0005,
+      "loss": 1.4396,
+      "step": 504100
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 0.0005,
+      "loss": 1.4632,
+      "step": 504200
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 0.0005,
+      "loss": 1.4469,
+      "step": 504300
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 0.0005,
+      "loss": 1.4564,
+      "step": 504400
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 0.0005,
+      "loss": 1.4522,
+      "step": 504500
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 0.0005,
+      "loss": 1.4582,
+      "step": 504600
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 0.0005,
+      "loss": 1.4384,
+      "step": 504700
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4638,
+      "step": 504800
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 0.0005,
+      "loss": 1.4518,
+      "step": 504900
+    },
+    {
+      "epoch": 39.6,
+      "learning_rate": 0.0005,
+      "loss": 1.4322,
+      "step": 505000
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 0.0005,
+      "loss": 1.4654,
+      "step": 505100
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 0.0005,
+      "loss": 1.4446,
+      "step": 505200
+    },
+    {
+      "epoch": 39.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4646,
+      "step": 505300
+    },
+    {
+      "epoch": 39.63,
+      "learning_rate": 0.0005,
+      "loss": 1.4356,
+      "step": 505400
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 0.0005,
+      "loss": 1.4614,
+      "step": 505500
+    },
+    {
+      "epoch": 39.65,
+      "learning_rate": 0.0005,
+      "loss": 1.4351,
+      "step": 505600
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 0.0005,
+      "loss": 1.4379,
+      "step": 505700
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 0.0005,
+      "loss": 1.4683,
+      "step": 505800
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 0.0005,
+      "loss": 1.4412,
+      "step": 505900
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 0.0005,
+      "loss": 1.4549,
+      "step": 506000
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 0.0005,
+      "loss": 1.4474,
+      "step": 506100
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 0.0005,
+      "loss": 1.4656,
+      "step": 506200
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 0.0005,
+      "loss": 1.4564,
+      "step": 506300
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 0.0005,
+      "loss": 1.4471,
+      "step": 506400
+    },
+    {
+      "epoch": 39.72,
+      "learning_rate": 0.0005,
+      "loss": 1.4516,
+      "step": 506500
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 0.0005,
+      "loss": 1.4619,
+      "step": 506600
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 0.0005,
+      "loss": 1.4515,
+      "step": 506700
+    },
+    {
+      "epoch": 39.74,
+      "learning_rate": 0.0005,
+      "loss": 1.4363,
+      "step": 506800
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 0.0005,
+      "loss": 1.4689,
+      "step": 506900
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 0.0005,
+      "loss": 1.4752,
+      "step": 507000
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 0.0005,
+      "loss": 1.4471,
+      "step": 507100
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 0.0005,
+      "loss": 1.4483,
+      "step": 507200
+    },
+    {
+      "epoch": 39.78,
+      "learning_rate": 0.0005,
+      "loss": 1.43,
+      "step": 507300
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 0.0005,
+      "loss": 1.4408,
+      "step": 507400
+    },
+    {
+      "epoch": 39.8,
+      "learning_rate": 0.0005,
+      "loss": 1.4258,
+      "step": 507500
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 0.0005,
+      "loss": 1.4326,
+      "step": 507600
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 0.0005,
+      "loss": 1.5031,
+      "step": 507700
+    },
+    {
+      "epoch": 39.82,
+      "learning_rate": 0.0005,
+      "loss": 1.461,
+      "step": 507800
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 0.0005,
+      "loss": 1.4695,
+      "step": 507900
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 0.0005,
+      "loss": 1.491,
+      "step": 508000
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 0.0005,
+      "loss": 1.4485,
+      "step": 508100
+    },
+    {
+      "epoch": 39.85,
+      "learning_rate": 0.0005,
+      "loss": 1.4509,
+      "step": 508200
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 0.0005,
+      "loss": 1.4386,
+      "step": 508300
+    },
+    {
+      "epoch": 39.87,
+      "learning_rate": 0.0005,
+      "loss": 1.4672,
+      "step": 508400
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 0.0005,
+      "loss": 1.4693,
+      "step": 508500
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 0.0005,
+      "loss": 1.4867,
+      "step": 508600
+    },
+    {
+      "epoch": 39.89,
+      "learning_rate": 0.0005,
+      "loss": 1.4633,
+      "step": 508700
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 0.0005,
+      "loss": 1.4794,
+      "step": 508800
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 0.0005,
+      "loss": 1.4676,
+      "step": 508900
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 0.0005,
+      "loss": 1.4694,
+      "step": 509000
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 0.0005,
+      "loss": 1.4547,
+      "step": 509100
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 0.0005,
+      "loss": 1.4533,
+      "step": 509200
+    },
+    {
+      "epoch": 39.94,
+      "learning_rate": 0.0005,
+      "loss": 1.4713,
+      "step": 509300
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 0.0005,
+      "loss": 1.455,
+      "step": 509400
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 0.0005,
+      "loss": 1.4833,
+      "step": 509500
+    },
+    {
+      "epoch": 39.96,
+      "learning_rate": 0.0005,
+      "loss": 1.4749,
+      "step": 509600
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 0.0005,
+      "loss": 1.4435,
+      "step": 509700
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 0.0005,
+      "loss": 1.4589,
+      "step": 509800
+    },
+    {
+      "epoch": 39.99,
+      "learning_rate": 0.0005,
+      "loss": 1.469,
+      "step": 509900
+    },
+    {
+      "epoch": 39.99,
+      "learning_rate": 0.0005,
+      "loss": 1.4541,
+      "step": 510000
+    },
+    {
+      "epoch": 39.99,
+      "eval_gen_len": 18.759071245696123,
+      "eval_loss": 2.0739943981170654,
+      "eval_rouge1": 35.8683,
+      "eval_rouge2": 14.7103,
+      "eval_rougeL": 29.5012,
+      "eval_rougeLsum": 29.4958,
+      "eval_runtime": 358.0238,
+      "eval_samples_per_second": 31.638,
+      "eval_steps_per_second": 1.978,
+      "step": 510000
+    }
+  ],
+  "max_steps": 637600,
+  "num_train_epochs": 50,
+  "total_flos": 2.1607039565631652e+18,
+  "trial_name": null,
+  "trial_params": null
+}