diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,6 +1,6 @@
 {
-  "best_metric": 1.445023536682129,
-  "best_model_checkpoint": "/home/co-ou1/rds/hpc-work/models/bart_large_gov/checkpoint-545",
+  "best_metric": 1.4266291856765747,
+  "best_model_checkpoint": "/home/co-ou1/rds/hpc-work/models/bart_large_gov/checkpoint-1227",
   "epoch": 19.935868071461293,
   "eval_steps": 500,
   "global_step": 2720,
@@ -11,8431 +11,8431 @@
     {
       "epoch": 0.01,
       "learning_rate": 4.996323529411765e-05,
-      "loss": 2.9333,
+      "loss": 3.1458,
       "step": 2
     },
     {
       "epoch": 0.03,
       "learning_rate": 4.99264705882353e-05,
-      "loss": 2.467,
+      "loss": 2.5311,
       "step": 4
     },
     {
       "epoch": 0.04,
       "learning_rate": 4.9889705882352945e-05,
-      "loss": 2.1631,
+      "loss": 2.3371,
       "step": 6
     },
     {
       "epoch": 0.06,
       "learning_rate": 4.985294117647059e-05,
-      "loss": 1.9596,
+      "loss": 2.119,
       "step": 8
     },
     {
       "epoch": 0.07,
       "learning_rate": 4.9816176470588236e-05,
-      "loss": 1.9644,
+      "loss": 2.0765,
       "step": 10
     },
     {
       "epoch": 0.09,
       "learning_rate": 4.977941176470589e-05,
-      "loss": 1.9089,
+      "loss": 1.9836,
       "step": 12
     },
     {
       "epoch": 0.1,
       "learning_rate": 4.9742647058823534e-05,
-      "loss": 1.9869,
+      "loss": 2.0458,
       "step": 14
     },
     {
       "epoch": 0.12,
       "learning_rate": 4.970588235294118e-05,
-      "loss": 1.9179,
+      "loss": 1.9844,
       "step": 16
     },
     {
       "epoch": 0.13,
       "learning_rate": 4.9669117647058824e-05,
-      "loss": 1.9178,
+      "loss": 1.9514,
       "step": 18
     },
     {
       "epoch": 0.15,
       "learning_rate": 4.9632352941176476e-05,
-      "loss": 1.8694,
+      "loss": 1.8552,
       "step": 20
     },
     {
       "epoch": 0.16,
       "learning_rate": 4.959558823529412e-05,
-      "loss": 1.8608,
+      "loss": 1.8923,
       "step": 22
     },
     {
       "epoch": 0.18,
       "learning_rate": 4.955882352941177e-05,
-      "loss": 1.8558,
+      "loss": 1.8408,
       "step": 24
     },
     {
       "epoch": 0.19,
       "learning_rate": 4.952205882352941e-05,
-      "loss": 1.7973,
+      "loss": 1.8763,
       "step": 26
     },
     {
       "epoch": 0.21,
       "learning_rate": 4.9485294117647065e-05,
-      "loss": 1.819,
+      "loss": 1.8369,
       "step": 28
     },
     {
       "epoch": 0.22,
       "learning_rate": 4.944852941176471e-05,
-      "loss": 1.8119,
+      "loss": 1.8329,
       "step": 30
     },
     {
       "epoch": 0.23,
       "learning_rate": 4.9411764705882355e-05,
-      "loss": 1.8298,
+      "loss": 1.865,
       "step": 32
     },
     {
       "epoch": 0.25,
       "learning_rate": 4.937500000000001e-05,
-      "loss": 1.722,
+      "loss": 1.8123,
       "step": 34
     },
     {
       "epoch": 0.26,
       "learning_rate": 4.933823529411765e-05,
-      "loss": 1.7837,
+      "loss": 1.8076,
       "step": 36
     },
     {
       "epoch": 0.28,
       "learning_rate": 4.93014705882353e-05,
-      "loss": 1.7793,
+      "loss": 1.8127,
       "step": 38
     },
     {
       "epoch": 0.29,
       "learning_rate": 4.9264705882352944e-05,
-      "loss": 1.7033,
+      "loss": 1.7886,
       "step": 40
     },
     {
       "epoch": 0.31,
       "learning_rate": 4.9227941176470596e-05,
-      "loss": 1.7449,
+      "loss": 1.7742,
       "step": 42
     },
     {
       "epoch": 0.32,
       "learning_rate": 4.9191176470588234e-05,
-      "loss": 1.7904,
+      "loss": 1.8307,
       "step": 44
     },
     {
       "epoch": 0.34,
       "learning_rate": 4.915441176470588e-05,
-      "loss": 1.7559,
+      "loss": 1.8065,
       "step": 46
     },
     {
       "epoch": 0.35,
       "learning_rate": 4.911764705882353e-05,
-      "loss": 1.7203,
+      "loss": 1.8092,
       "step": 48
     },
     {
       "epoch": 0.37,
       "learning_rate": 4.908088235294118e-05,
-      "loss": 1.7266,
+      "loss": 1.7918,
       "step": 50
     },
     {
       "epoch": 0.38,
       "learning_rate": 4.904411764705882e-05,
-      "loss": 1.707,
+      "loss": 1.7952,
       "step": 52
     },
     {
       "epoch": 0.4,
       "learning_rate": 4.900735294117647e-05,
-      "loss": 1.777,
+      "loss": 1.7627,
       "step": 54
     },
     {
       "epoch": 0.41,
       "learning_rate": 4.897058823529412e-05,
-      "loss": 1.7624,
+      "loss": 1.745,
       "step": 56
     },
     {
       "epoch": 0.43,
       "learning_rate": 4.8933823529411766e-05,
-      "loss": 1.7569,
+      "loss": 1.7852,
       "step": 58
     },
     {
       "epoch": 0.44,
       "learning_rate": 4.889705882352941e-05,
-      "loss": 1.7643,
+      "loss": 1.7929,
       "step": 60
     },
     {
       "epoch": 0.45,
       "learning_rate": 4.886029411764706e-05,
-      "loss": 1.7315,
+      "loss": 1.7667,
       "step": 62
     },
     {
       "epoch": 0.47,
       "learning_rate": 4.882352941176471e-05,
-      "loss": 1.774,
+      "loss": 1.7801,
       "step": 64
     },
     {
       "epoch": 0.48,
       "learning_rate": 4.8786764705882354e-05,
-      "loss": 1.7839,
+      "loss": 1.7506,
       "step": 66
     },
     {
       "epoch": 0.5,
       "learning_rate": 4.875e-05,
-      "loss": 1.6725,
+      "loss": 1.7355,
       "step": 68
     },
     {
       "epoch": 0.51,
       "learning_rate": 4.871323529411765e-05,
-      "loss": 1.7551,
+      "loss": 1.7368,
       "step": 70
     },
     {
       "epoch": 0.53,
       "learning_rate": 4.86764705882353e-05,
-      "loss": 1.7377,
+      "loss": 1.7198,
       "step": 72
     },
     {
       "epoch": 0.54,
       "learning_rate": 4.863970588235294e-05,
-      "loss": 1.6739,
+      "loss": 1.732,
       "step": 74
     },
     {
       "epoch": 0.56,
       "learning_rate": 4.860294117647059e-05,
-      "loss": 1.7351,
+      "loss": 1.7899,
       "step": 76
     },
     {
       "epoch": 0.57,
       "learning_rate": 4.856617647058824e-05,
-      "loss": 1.6912,
+      "loss": 1.7788,
       "step": 78
     },
     {
       "epoch": 0.59,
       "learning_rate": 4.8529411764705885e-05,
-      "loss": 1.63,
+      "loss": 1.7476,
       "step": 80
     },
     {
       "epoch": 0.6,
       "learning_rate": 4.849264705882353e-05,
-      "loss": 1.7231,
+      "loss": 1.7499,
       "step": 82
     },
     {
       "epoch": 0.62,
       "learning_rate": 4.8455882352941176e-05,
-      "loss": 1.741,
+      "loss": 1.7246,
       "step": 84
     },
     {
       "epoch": 0.63,
       "learning_rate": 4.841911764705883e-05,
-      "loss": 1.6945,
+      "loss": 1.7551,
       "step": 86
     },
     {
       "epoch": 0.64,
       "learning_rate": 4.838235294117647e-05,
-      "loss": 1.7284,
+      "loss": 1.7388,
       "step": 88
     },
     {
       "epoch": 0.66,
       "learning_rate": 4.834558823529412e-05,
-      "loss": 1.7246,
+      "loss": 1.7604,
       "step": 90
     },
     {
       "epoch": 0.67,
       "learning_rate": 4.830882352941177e-05,
-      "loss": 1.6913,
+      "loss": 1.6976,
       "step": 92
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.8272058823529416e-05,
-      "loss": 1.6901,
+      "loss": 1.7179,
       "step": 94
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.823529411764706e-05,
-      "loss": 1.6237,
+      "loss": 1.6908,
       "step": 96
     },
     {
       "epoch": 0.72,
       "learning_rate": 4.819852941176471e-05,
-      "loss": 1.6793,
+      "loss": 1.7271,
       "step": 98
     },
     {
       "epoch": 0.73,
       "learning_rate": 4.816176470588236e-05,
-      "loss": 1.7057,
+      "loss": 1.7638,
       "step": 100
     },
     {
       "epoch": 0.75,
       "learning_rate": 4.8125000000000004e-05,
-      "loss": 1.7349,
+      "loss": 1.7888,
       "step": 102
     },
     {
       "epoch": 0.76,
       "learning_rate": 4.808823529411765e-05,
-      "loss": 1.7279,
+      "loss": 1.7234,
       "step": 104
     },
     {
       "epoch": 0.78,
       "learning_rate": 4.8051470588235295e-05,
-      "loss": 1.6742,
+      "loss": 1.6952,
       "step": 106
     },
     {
       "epoch": 0.79,
       "learning_rate": 4.801470588235295e-05,
-      "loss": 1.6748,
+      "loss": 1.6991,
       "step": 108
     },
     {
       "epoch": 0.81,
       "learning_rate": 4.797794117647059e-05,
-      "loss": 1.6322,
+      "loss": 1.7197,
       "step": 110
     },
     {
       "epoch": 0.82,
       "learning_rate": 4.794117647058824e-05,
-      "loss": 1.6596,
+      "loss": 1.721,
       "step": 112
     },
     {
       "epoch": 0.84,
       "learning_rate": 4.790441176470588e-05,
-      "loss": 1.6613,
+      "loss": 1.6757,
       "step": 114
     },
     {
       "epoch": 0.85,
       "learning_rate": 4.7867647058823535e-05,
-      "loss": 1.639,
+      "loss": 1.6877,
       "step": 116
     },
     {
       "epoch": 0.86,
       "learning_rate": 4.783088235294118e-05,
-      "loss": 1.6603,
+      "loss": 1.6713,
       "step": 118
     },
     {
       "epoch": 0.88,
       "learning_rate": 4.7794117647058826e-05,
-      "loss": 1.5827,
+      "loss": 1.6566,
       "step": 120
     },
     {
       "epoch": 0.89,
       "learning_rate": 4.775735294117648e-05,
-      "loss": 1.693,
+      "loss": 1.7646,
       "step": 122
     },
     {
       "epoch": 0.91,
       "learning_rate": 4.7720588235294124e-05,
-      "loss": 1.6421,
+      "loss": 1.698,
       "step": 124
     },
     {
       "epoch": 0.92,
       "learning_rate": 4.768382352941177e-05,
-      "loss": 1.6872,
+      "loss": 1.6811,
       "step": 126
     },
     {
       "epoch": 0.94,
       "learning_rate": 4.7647058823529414e-05,
-      "loss": 1.6206,
+      "loss": 1.6475,
       "step": 128
     },
     {
       "epoch": 0.95,
       "learning_rate": 4.761029411764706e-05,
-      "loss": 1.6894,
+      "loss": 1.7184,
       "step": 130
     },
     {
       "epoch": 0.97,
       "learning_rate": 4.7573529411764705e-05,
-      "loss": 1.6132,
+      "loss": 1.6903,
       "step": 132
     },
     {
       "epoch": 0.98,
       "learning_rate": 4.753676470588235e-05,
-      "loss": 1.6449,
+      "loss": 1.696,
       "step": 134
     },
     {
       "epoch": 1.0,
       "learning_rate": 4.75e-05,
-      "loss": 1.6694,
+      "loss": 1.7352,
       "step": 136
     },
     {
       "epoch": 1.0,
-      "eval_gen_len": 139.72530864197532,
-      "eval_loss": 1.533786654472351,
-      "eval_rouge1": 54.2061,
-      "eval_rouge2": 29.3577,
-      "eval_rougeL": 37.2911,
-      "eval_rougeLsum": 49.8337,
-      "eval_runtime": 275.7588,
-      "eval_samples_per_second": 3.525,
-      "eval_steps_per_second": 0.442,
+      "eval_gen_len": 685.9300411522634,
+      "eval_loss": 1.5224188566207886,
+      "eval_rouge1": 72.0472,
+      "eval_rouge2": 41.3267,
+      "eval_rougeL": 36.4817,
+      "eval_rougeLsum": 69.4011,
+      "eval_runtime": 1507.0932,
+      "eval_samples_per_second": 0.645,
+      "eval_steps_per_second": 0.081,
       "step": 136
     },
     {
       "epoch": 1.01,
       "learning_rate": 4.746323529411765e-05,
-      "loss": 1.5791,
+      "loss": 1.6969,
       "step": 138
     },
     {
       "epoch": 1.03,
       "learning_rate": 4.742647058823529e-05,
-      "loss": 1.5997,
+      "loss": 1.6445,
       "step": 140
     },
     {
       "epoch": 1.04,
       "learning_rate": 4.738970588235294e-05,
-      "loss": 1.5539,
+      "loss": 1.6552,
       "step": 142
     },
     {
       "epoch": 1.06,
       "learning_rate": 4.735294117647059e-05,
-      "loss": 1.6384,
+      "loss": 1.6619,
       "step": 144
     },
     {
       "epoch": 1.07,
       "learning_rate": 4.7316176470588236e-05,
-      "loss": 1.5839,
+      "loss": 1.6811,
       "step": 146
     },
     {
       "epoch": 1.08,
       "learning_rate": 4.727941176470588e-05,
-      "loss": 1.5375,
+      "loss": 1.6565,
       "step": 148
     },
     {
       "epoch": 1.1,
       "learning_rate": 4.7242647058823534e-05,
-      "loss": 1.5505,
+      "loss": 1.6777,
       "step": 150
     },
     {
       "epoch": 1.11,
       "learning_rate": 4.720588235294118e-05,
-      "loss": 1.5994,
+      "loss": 1.6993,
       "step": 152
     },
     {
       "epoch": 1.13,
       "learning_rate": 4.7169117647058824e-05,
-      "loss": 1.6364,
+      "loss": 1.6585,
       "step": 154
     },
     {
       "epoch": 1.14,
       "learning_rate": 4.713235294117647e-05,
-      "loss": 1.5387,
+      "loss": 1.6274,
       "step": 156
     },
     {
       "epoch": 1.16,
       "learning_rate": 4.709558823529412e-05,
-      "loss": 1.5614,
+      "loss": 1.6659,
       "step": 158
     },
     {
       "epoch": 1.17,
       "learning_rate": 4.705882352941177e-05,
-      "loss": 1.572,
+      "loss": 1.6835,
       "step": 160
     },
     {
       "epoch": 1.19,
       "learning_rate": 4.702205882352941e-05,
-      "loss": 1.5236,
+      "loss": 1.6414,
       "step": 162
     },
     {
       "epoch": 1.2,
       "learning_rate": 4.698529411764706e-05,
-      "loss": 1.5675,
+      "loss": 1.6614,
       "step": 164
     },
     {
       "epoch": 1.22,
       "learning_rate": 4.694852941176471e-05,
-      "loss": 1.5518,
+      "loss": 1.6693,
       "step": 166
     },
     {
       "epoch": 1.23,
       "learning_rate": 4.6911764705882356e-05,
-      "loss": 1.5171,
+      "loss": 1.6631,
       "step": 168
     },
     {
       "epoch": 1.25,
       "learning_rate": 4.6875e-05,
-      "loss": 1.6097,
+      "loss": 1.6745,
       "step": 170
     },
     {
       "epoch": 1.26,
       "learning_rate": 4.6838235294117646e-05,
-      "loss": 1.486,
+      "loss": 1.642,
       "step": 172
     },
     {
       "epoch": 1.28,
       "learning_rate": 4.68014705882353e-05,
-      "loss": 1.5021,
+      "loss": 1.6089,
       "step": 174
     },
     {
       "epoch": 1.29,
       "learning_rate": 4.6764705882352944e-05,
-      "loss": 1.5326,
+      "loss": 1.6192,
       "step": 176
     },
     {
       "epoch": 1.3,
       "learning_rate": 4.672794117647059e-05,
-      "loss": 1.551,
+      "loss": 1.5909,
       "step": 178
     },
     {
       "epoch": 1.32,
       "learning_rate": 4.669117647058824e-05,
-      "loss": 1.5579,
+      "loss": 1.679,
       "step": 180
     },
     {
       "epoch": 1.33,
       "learning_rate": 4.665441176470589e-05,
-      "loss": 1.5516,
+      "loss": 1.6633,
       "step": 182
     },
     {
       "epoch": 1.35,
       "learning_rate": 4.661764705882353e-05,
-      "loss": 1.5211,
+      "loss": 1.6667,
       "step": 184
     },
     {
       "epoch": 1.36,
       "learning_rate": 4.658088235294118e-05,
-      "loss": 1.4823,
+      "loss": 1.5912,
       "step": 186
     },
     {
       "epoch": 1.38,
       "learning_rate": 4.654411764705883e-05,
-      "loss": 1.5625,
+      "loss": 1.6803,
       "step": 188
     },
     {
       "epoch": 1.39,
       "learning_rate": 4.6507352941176475e-05,
-      "loss": 1.4965,
+      "loss": 1.5855,
       "step": 190
     },
     {
       "epoch": 1.41,
       "learning_rate": 4.647058823529412e-05,
-      "loss": 1.5516,
+      "loss": 1.6265,
       "step": 192
     },
     {
       "epoch": 1.42,
       "learning_rate": 4.6433823529411766e-05,
-      "loss": 1.5452,
+      "loss": 1.6296,
       "step": 194
     },
     {
       "epoch": 1.44,
       "learning_rate": 4.639705882352942e-05,
-      "loss": 1.558,
+      "loss": 1.6503,
       "step": 196
     },
     {
       "epoch": 1.45,
       "learning_rate": 4.636029411764706e-05,
-      "loss": 1.4985,
+      "loss": 1.6359,
       "step": 198
     },
     {
       "epoch": 1.47,
       "learning_rate": 4.632352941176471e-05,
-      "loss": 1.5132,
+      "loss": 1.5901,
       "step": 200
     },
     {
       "epoch": 1.48,
       "learning_rate": 4.6286764705882354e-05,
-      "loss": 1.5821,
+      "loss": 1.6452,
       "step": 202
     },
     {
       "epoch": 1.5,
       "learning_rate": 4.6250000000000006e-05,
-      "loss": 1.5036,
+      "loss": 1.6119,
       "step": 204
     },
     {
       "epoch": 1.51,
       "learning_rate": 4.621323529411765e-05,
-      "loss": 1.55,
+      "loss": 1.6387,
       "step": 206
     },
     {
       "epoch": 1.52,
       "learning_rate": 4.61764705882353e-05,
-      "loss": 1.456,
+      "loss": 1.6565,
       "step": 208
     },
     {
       "epoch": 1.54,
       "learning_rate": 4.613970588235295e-05,
-      "loss": 1.5497,
+      "loss": 1.6407,
       "step": 210
     },
     {
       "epoch": 1.55,
       "learning_rate": 4.6102941176470594e-05,
-      "loss": 1.4507,
+      "loss": 1.5864,
       "step": 212
     },
     {
       "epoch": 1.57,
       "learning_rate": 4.606617647058823e-05,
-      "loss": 1.498,
+      "loss": 1.5997,
       "step": 214
     },
     {
       "epoch": 1.58,
       "learning_rate": 4.6029411764705885e-05,
-      "loss": 1.4755,
+      "loss": 1.6041,
       "step": 216
     },
     {
       "epoch": 1.6,
       "learning_rate": 4.599264705882353e-05,
-      "loss": 1.5222,
+      "loss": 1.6505,
       "step": 218
     },
     {
       "epoch": 1.61,
       "learning_rate": 4.5955882352941176e-05,
-      "loss": 1.5111,
+      "loss": 1.6734,
       "step": 220
     },
     {
       "epoch": 1.63,
       "learning_rate": 4.591911764705882e-05,
-      "loss": 1.4812,
+      "loss": 1.5972,
       "step": 222
     },
     {
       "epoch": 1.64,
       "learning_rate": 4.588235294117647e-05,
-      "loss": 1.5365,
+      "loss": 1.6595,
       "step": 224
     },
     {
       "epoch": 1.66,
       "learning_rate": 4.584558823529412e-05,
-      "loss": 1.5188,
+      "loss": 1.5991,
       "step": 226
     },
     {
       "epoch": 1.67,
       "learning_rate": 4.5808823529411764e-05,
-      "loss": 1.4869,
+      "loss": 1.6267,
       "step": 228
     },
     {
       "epoch": 1.69,
       "learning_rate": 4.577205882352941e-05,
-      "loss": 1.5374,
+      "loss": 1.6275,
       "step": 230
     },
     {
       "epoch": 1.7,
       "learning_rate": 4.573529411764706e-05,
-      "loss": 1.5932,
+      "loss": 1.6901,
       "step": 232
     },
     {
       "epoch": 1.72,
       "learning_rate": 4.569852941176471e-05,
-      "loss": 1.5479,
+      "loss": 1.635,
       "step": 234
     },
     {
       "epoch": 1.73,
       "learning_rate": 4.566176470588235e-05,
-      "loss": 1.5238,
+      "loss": 1.6775,
       "step": 236
     },
     {
       "epoch": 1.74,
       "learning_rate": 4.5625e-05,
-      "loss": 1.5053,
+      "loss": 1.6446,
       "step": 238
     },
     {
       "epoch": 1.76,
       "learning_rate": 4.558823529411765e-05,
-      "loss": 1.5273,
+      "loss": 1.6315,
       "step": 240
     },
     {
       "epoch": 1.77,
       "learning_rate": 4.5551470588235295e-05,
-      "loss": 1.5534,
+      "loss": 1.6885,
       "step": 242
     },
     {
       "epoch": 1.79,
       "learning_rate": 4.551470588235294e-05,
-      "loss": 1.5814,
+      "loss": 1.6688,
       "step": 244
     },
     {
       "epoch": 1.8,
       "learning_rate": 4.547794117647059e-05,
-      "loss": 1.517,
+      "loss": 1.6464,
       "step": 246
     },
     {
       "epoch": 1.82,
       "learning_rate": 4.544117647058824e-05,
-      "loss": 1.521,
+      "loss": 1.6047,
       "step": 248
     },
     {
       "epoch": 1.83,
       "learning_rate": 4.5404411764705883e-05,
-      "loss": 1.4901,
+      "loss": 1.6206,
       "step": 250
     },
     {
       "epoch": 1.85,
       "learning_rate": 4.536764705882353e-05,
-      "loss": 1.5161,
+      "loss": 1.6146,
       "step": 252
     },
     {
       "epoch": 1.86,
       "learning_rate": 4.533088235294118e-05,
-      "loss": 1.5796,
+      "loss": 1.6104,
       "step": 254
     },
     {
       "epoch": 1.88,
       "learning_rate": 4.5294117647058826e-05,
-      "loss": 1.4371,
+      "loss": 1.618,
       "step": 256
     },
     {
       "epoch": 1.89,
       "learning_rate": 4.525735294117647e-05,
-      "loss": 1.5034,
+      "loss": 1.6014,
       "step": 258
     },
     {
       "epoch": 1.91,
       "learning_rate": 4.522058823529412e-05,
-      "loss": 1.5708,
+      "loss": 1.6227,
       "step": 260
     },
     {
       "epoch": 1.92,
       "learning_rate": 4.518382352941177e-05,
-      "loss": 1.4524,
+      "loss": 1.5916,
       "step": 262
     },
     {
       "epoch": 1.93,
       "learning_rate": 4.5147058823529415e-05,
-      "loss": 1.4931,
+      "loss": 1.6464,
       "step": 264
     },
     {
       "epoch": 1.95,
       "learning_rate": 4.511029411764706e-05,
-      "loss": 1.5326,
+      "loss": 1.6178,
       "step": 266
     },
     {
       "epoch": 1.96,
       "learning_rate": 4.507352941176471e-05,
-      "loss": 1.5344,
+      "loss": 1.6777,
       "step": 268
     },
     {
       "epoch": 1.98,
       "learning_rate": 4.503676470588236e-05,
-      "loss": 1.5018,
+      "loss": 1.6599,
       "step": 270
     },
     {
       "epoch": 1.99,
       "learning_rate": 4.5e-05,
-      "loss": 1.5178,
+      "loss": 1.6874,
       "step": 272
     },
     {
       "epoch": 1.99,
-      "eval_gen_len": 128.94753086419752,
-      "eval_loss": 1.4697880744934082,
-      "eval_rouge1": 55.6621,
-      "eval_rouge2": 30.6254,
-      "eval_rougeL": 38.7491,
-      "eval_rougeLsum": 51.2934,
-      "eval_runtime": 225.1875,
-      "eval_samples_per_second": 4.316,
-      "eval_steps_per_second": 0.542,
+      "eval_gen_len": 699.4866255144033,
+      "eval_loss": 1.4778804779052734,
+      "eval_rouge1": 71.7737,
+      "eval_rouge2": 40.8546,
+      "eval_rougeL": 36.8472,
+      "eval_rougeLsum": 69.2034,
+      "eval_runtime": 1548.1543,
+      "eval_samples_per_second": 0.628,
+      "eval_steps_per_second": 0.079,
       "step": 272
     },
     {
       "epoch": 2.01,
       "learning_rate": 4.496323529411765e-05,
-      "loss": 1.4707,
+      "loss": 1.5938,
       "step": 274
     },
     {
       "epoch": 2.02,
       "learning_rate": 4.49264705882353e-05,
-      "loss": 1.4499,
+      "loss": 1.6166,
       "step": 276
     },
     {
       "epoch": 2.04,
       "learning_rate": 4.4889705882352946e-05,
-      "loss": 1.4416,
+      "loss": 1.5896,
       "step": 278
     },
     {
       "epoch": 2.05,
       "learning_rate": 4.485294117647059e-05,
-      "loss": 1.4304,
+      "loss": 1.5819,
       "step": 280
     },
     {
       "epoch": 2.07,
       "learning_rate": 4.4816176470588236e-05,
-      "loss": 1.3961,
+      "loss": 1.5363,
       "step": 282
     },
     {
       "epoch": 2.08,
       "learning_rate": 4.477941176470589e-05,
-      "loss": 1.4177,
+      "loss": 1.6237,
       "step": 284
     },
     {
       "epoch": 2.1,
       "learning_rate": 4.4742647058823534e-05,
-      "loss": 1.426,
+      "loss": 1.5904,
       "step": 286
     },
     {
       "epoch": 2.11,
       "learning_rate": 4.470588235294118e-05,
-      "loss": 1.411,
+      "loss": 1.5691,
       "step": 288
     },
     {
       "epoch": 2.13,
       "learning_rate": 4.4669117647058825e-05,
-      "loss": 1.4288,
+      "loss": 1.5736,
       "step": 290
     },
     {
       "epoch": 2.14,
       "learning_rate": 4.463235294117648e-05,
-      "loss": 1.4189,
+      "loss": 1.5828,
       "step": 292
     },
     {
       "epoch": 2.15,
       "learning_rate": 4.459558823529412e-05,
-      "loss": 1.4396,
+      "loss": 1.626,
       "step": 294
     },
     {
       "epoch": 2.17,
       "learning_rate": 4.455882352941177e-05,
-      "loss": 1.448,
+      "loss": 1.5758,
       "step": 296
     },
     {
       "epoch": 2.18,
       "learning_rate": 4.452205882352941e-05,
-      "loss": 1.4147,
+      "loss": 1.5686,
       "step": 298
     },
     {
       "epoch": 2.2,
       "learning_rate": 4.448529411764706e-05,
-      "loss": 1.4604,
+      "loss": 1.594,
       "step": 300
     },
     {
       "epoch": 2.21,
       "learning_rate": 4.4448529411764704e-05,
-      "loss": 1.4261,
+      "loss": 1.605,
       "step": 302
     },
     {
       "epoch": 2.23,
       "learning_rate": 4.4411764705882356e-05,
-      "loss": 1.39,
+      "loss": 1.5628,
       "step": 304
     },
     {
       "epoch": 2.24,
       "learning_rate": 4.4375e-05,
-      "loss": 1.4104,
+      "loss": 1.5738,
       "step": 306
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.4338235294117647e-05,
-      "loss": 1.4143,
+      "loss": 1.5813,
       "step": 308
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.430147058823529e-05,
-      "loss": 1.3803,
+      "loss": 1.558,
       "step": 310
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.4264705882352944e-05,
-      "loss": 1.4426,
+      "loss": 1.5974,
       "step": 312
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.422794117647059e-05,
-      "loss": 1.4429,
+      "loss": 1.582,
       "step": 314
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.4191176470588235e-05,
-      "loss": 1.4092,
+      "loss": 1.5758,
       "step": 316
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.415441176470588e-05,
-      "loss": 1.4195,
+      "loss": 1.5467,
       "step": 318
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.411764705882353e-05,
-      "loss": 1.4206,
+      "loss": 1.5621,
       "step": 320
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.408088235294118e-05,
-      "loss": 1.4465,
+      "loss": 1.5402,
       "step": 322
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.404411764705882e-05,
-      "loss": 1.4232,
+      "loss": 1.6221,
       "step": 324
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.400735294117647e-05,
-      "loss": 1.3775,
+      "loss": 1.5728,
       "step": 326
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.397058823529412e-05,
-      "loss": 1.4158,
+      "loss": 1.5801,
       "step": 328
     },
     {
       "epoch": 2.42,
       "learning_rate": 4.3933823529411766e-05,
-      "loss": 1.4791,
+      "loss": 1.6026,
       "step": 330
     },
     {
       "epoch": 2.43,
       "learning_rate": 4.389705882352941e-05,
-      "loss": 1.442,
+      "loss": 1.623,
       "step": 332
     },
     {
       "epoch": 2.45,
       "learning_rate": 4.3860294117647063e-05,
-      "loss": 1.4081,
+      "loss": 1.5657,
       "step": 334
     },
     {
       "epoch": 2.46,
       "learning_rate": 4.382352941176471e-05,
-      "loss": 1.3624,
+      "loss": 1.5637,
       "step": 336
     },
     {
       "epoch": 2.48,
       "learning_rate": 4.3786764705882354e-05,
-      "loss": 1.4339,
+      "loss": 1.6006,
       "step": 338
     },
     {
       "epoch": 2.49,
       "learning_rate": 4.375e-05,
-      "loss": 1.4567,
+      "loss": 1.6032,
       "step": 340
     },
     {
       "epoch": 2.51,
       "learning_rate": 4.371323529411765e-05,
-      "loss": 1.4509,
+      "loss": 1.5816,
       "step": 342
     },
     {
       "epoch": 2.52,
       "learning_rate": 4.36764705882353e-05,
-      "loss": 1.4989,
+      "loss": 1.5769,
       "step": 344
     },
     {
       "epoch": 2.54,
       "learning_rate": 4.363970588235294e-05,
-      "loss": 1.4039,
+      "loss": 1.6042,
       "step": 346
     },
     {
       "epoch": 2.55,
       "learning_rate": 4.360294117647059e-05,
-      "loss": 1.4554,
+      "loss": 1.5442,
       "step": 348
     },
     {
       "epoch": 2.57,
       "learning_rate": 4.356617647058824e-05,
-      "loss": 1.3791,
+      "loss": 1.5834,
       "step": 350
     },
     {
       "epoch": 2.58,
       "learning_rate": 4.3529411764705885e-05,
-      "loss": 1.4329,
+      "loss": 1.5664,
       "step": 352
     },
     {
       "epoch": 2.59,
       "learning_rate": 4.349264705882353e-05,
-      "loss": 1.414,
+      "loss": 1.5611,
       "step": 354
     },
     {
       "epoch": 2.61,
       "learning_rate": 4.345588235294118e-05,
-      "loss": 1.4346,
+      "loss": 1.5661,
       "step": 356
     },
     {
       "epoch": 2.62,
       "learning_rate": 4.341911764705883e-05,
-      "loss": 1.3559,
+      "loss": 1.5466,
       "step": 358
     },
     {
       "epoch": 2.64,
       "learning_rate": 4.3382352941176474e-05,
-      "loss": 1.4023,
+      "loss": 1.5953,
       "step": 360
     },
     {
       "epoch": 2.65,
       "learning_rate": 4.334558823529412e-05,
-      "loss": 1.3509,
+      "loss": 1.5733,
       "step": 362
     },
     {
       "epoch": 2.67,
       "learning_rate": 4.330882352941177e-05,
-      "loss": 1.3932,
+      "loss": 1.5854,
       "step": 364
     },
     {
       "epoch": 2.68,
       "learning_rate": 4.3272058823529416e-05,
-      "loss": 1.4956,
+      "loss": 1.615,
       "step": 366
     },
     {
       "epoch": 2.7,
       "learning_rate": 4.323529411764706e-05,
-      "loss": 1.4287,
+      "loss": 1.5459,
       "step": 368
     },
     {
       "epoch": 2.71,
       "learning_rate": 4.319852941176471e-05,
-      "loss": 1.4228,
+      "loss": 1.5725,
       "step": 370
     },
     {
       "epoch": 2.73,
       "learning_rate": 4.316176470588236e-05,
-      "loss": 1.3941,
+      "loss": 1.5714,
       "step": 372
     },
     {
       "epoch": 2.74,
       "learning_rate": 4.3125000000000005e-05,
-      "loss": 1.4925,
+      "loss": 1.6121,
       "step": 374
     },
     {
       "epoch": 2.76,
       "learning_rate": 4.308823529411765e-05,
-      "loss": 1.39,
+      "loss": 1.5585,
       "step": 376
     },
     {
       "epoch": 2.77,
       "learning_rate": 4.3051470588235295e-05,
-      "loss": 1.3544,
+      "loss": 1.5815,
       "step": 378
     },
     {
       "epoch": 2.79,
       "learning_rate": 4.301470588235295e-05,
-      "loss": 1.3676,
+      "loss": 1.5459,
       "step": 380
     },
     {
       "epoch": 2.8,
       "learning_rate": 4.297794117647059e-05,
-      "loss": 1.4096,
+      "loss": 1.5533,
       "step": 382
     },
     {
       "epoch": 2.81,
       "learning_rate": 4.294117647058823e-05,
-      "loss": 1.4409,
+      "loss": 1.6328,
       "step": 384
     },
     {
       "epoch": 2.83,
       "learning_rate": 4.2904411764705884e-05,
-      "loss": 1.3851,
+      "loss": 1.5658,
       "step": 386
     },
     {
       "epoch": 2.84,
       "learning_rate": 4.286764705882353e-05,
-      "loss": 1.4253,
+      "loss": 1.5903,
       "step": 388
     },
     {
       "epoch": 2.86,
       "learning_rate": 4.2830882352941174e-05,
-      "loss": 1.3793,
+      "loss": 1.5538,
       "step": 390
     },
     {
       "epoch": 2.87,
       "learning_rate": 4.2794117647058827e-05,
-      "loss": 1.4005,
+      "loss": 1.5852,
       "step": 392
     },
     {
       "epoch": 2.89,
       "learning_rate": 4.275735294117647e-05,
-      "loss": 1.463,
+      "loss": 1.5633,
       "step": 394
     },
     {
       "epoch": 2.9,
       "learning_rate": 4.272058823529412e-05,
-      "loss": 1.4231,
+      "loss": 1.5322,
       "step": 396
     },
     {
       "epoch": 2.92,
       "learning_rate": 4.268382352941176e-05,
-      "loss": 1.4883,
+      "loss": 1.649,
       "step": 398
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.2647058823529415e-05,
-      "loss": 1.4709,
+      "loss": 1.5893,
       "step": 400
     },
     {
       "epoch": 2.95,
       "learning_rate": 4.261029411764706e-05,
-      "loss": 1.3951,
+      "loss": 1.5398,
       "step": 402
     },
     {
       "epoch": 2.96,
       "learning_rate": 4.2573529411764706e-05,
-      "loss": 1.4843,
+      "loss": 1.5964,
       "step": 404
     },
     {
       "epoch": 2.98,
       "learning_rate": 4.253676470588235e-05,
-      "loss": 1.4263,
+      "loss": 1.5827,
       "step": 406
     },
     {
       "epoch": 2.99,
       "learning_rate": 4.25e-05,
-      "loss": 1.4208,
+      "loss": 1.5695,
       "step": 408
     },
     {
       "epoch": 3.0,
-      "eval_gen_len": 129.56584362139918,
-      "eval_loss": 1.4486558437347412,
-      "eval_rouge1": 55.4905,
-      "eval_rouge2": 30.4201,
-      "eval_rougeL": 38.412,
-      "eval_rougeLsum": 51.1108,
-      "eval_runtime": 255.4888,
-      "eval_samples_per_second": 3.804,
-      "eval_steps_per_second": 0.478,
+      "eval_gen_len": 695.0977366255144,
+      "eval_loss": 1.4583299160003662,
+      "eval_rouge1": 72.2243,
+      "eval_rouge2": 41.372,
+      "eval_rougeL": 37.8382,
+      "eval_rougeLsum": 69.6295,
+      "eval_runtime": 1521.0423,
+      "eval_samples_per_second": 0.639,
+      "eval_steps_per_second": 0.08,
       "step": 409
     },
     {
       "epoch": 3.01,
       "learning_rate": 4.246323529411765e-05,
-      "loss": 1.3905,
+      "loss": 1.6104,
       "step": 410
     },
     {
       "epoch": 3.02,
       "learning_rate": 4.2426470588235294e-05,
-      "loss": 1.3456,
+      "loss": 1.5129,
       "step": 412
     },
     {
       "epoch": 3.03,
       "learning_rate": 4.238970588235294e-05,
-      "loss": 1.3687,
+      "loss": 1.5183,
       "step": 414
     },
     {
       "epoch": 3.05,
       "learning_rate": 4.235294117647059e-05,
-      "loss": 1.326,
+      "loss": 1.4872,
       "step": 416
     },
     {
       "epoch": 3.06,
       "learning_rate": 4.231617647058824e-05,
-      "loss": 1.3037,
+      "loss": 1.54,
       "step": 418
     },
     {
       "epoch": 3.08,
       "learning_rate": 4.227941176470588e-05,
-      "loss": 1.2895,
+      "loss": 1.4846,
       "step": 420
     },
     {
       "epoch": 3.09,
       "learning_rate": 4.2242647058823534e-05,
-      "loss": 1.2996,
+      "loss": 1.5372,
       "step": 422
     },
     {
       "epoch": 3.11,
       "learning_rate": 4.220588235294118e-05,
-      "loss": 1.3182,
+      "loss": 1.5029,
       "step": 424
     },
     {
       "epoch": 3.12,
       "learning_rate": 4.2169117647058825e-05,
-      "loss": 1.3633,
+      "loss": 1.5842,
       "step": 426
     },
     {
       "epoch": 3.14,
       "learning_rate": 4.213235294117647e-05,
-      "loss": 1.3328,
+      "loss": 1.5388,
       "step": 428
     },
     {
       "epoch": 3.15,
       "learning_rate": 4.209558823529412e-05,
-      "loss": 1.4186,
+      "loss": 1.5548,
       "step": 430
     },
     {
       "epoch": 3.17,
       "learning_rate": 4.205882352941177e-05,
-      "loss": 1.3802,
+      "loss": 1.5404,
       "step": 432
     },
     {
       "epoch": 3.18,
       "learning_rate": 4.202205882352941e-05,
-      "loss": 1.3181,
+      "loss": 1.5176,
       "step": 434
     },
     {
       "epoch": 3.2,
       "learning_rate": 4.198529411764706e-05,
-      "loss": 1.3821,
+      "loss": 1.5236,
       "step": 436
     },
     {
       "epoch": 3.21,
       "learning_rate": 4.194852941176471e-05,
-      "loss": 1.3545,
+      "loss": 1.5629,
       "step": 438
     },
     {
       "epoch": 3.22,
       "learning_rate": 4.1911764705882356e-05,
-      "loss": 1.3861,
+      "loss": 1.5498,
       "step": 440
     },
     {
       "epoch": 3.24,
       "learning_rate": 4.1875e-05,
-      "loss": 1.2915,
+      "loss": 1.5055,
       "step": 442
     },
     {
       "epoch": 3.25,
       "learning_rate": 4.1838235294117654e-05,
-      "loss": 1.3581,
+      "loss": 1.5415,
       "step": 444
     },
     {
       "epoch": 3.27,
       "learning_rate": 4.18014705882353e-05,
-      "loss": 1.3574,
+      "loss": 1.5641,
       "step": 446
     },
     {
       "epoch": 3.28,
       "learning_rate": 4.1764705882352944e-05,
-      "loss": 1.3463,
+      "loss": 1.5706,
       "step": 448
     },
     {
       "epoch": 3.3,
       "learning_rate": 4.172794117647059e-05,
-      "loss": 1.3353,
+      "loss": 1.5415,
       "step": 450
     },
     {
       "epoch": 3.31,
       "learning_rate": 4.169117647058824e-05,
-      "loss": 1.3506,
+      "loss": 1.5429,
       "step": 452
     },
     {
       "epoch": 3.33,
       "learning_rate": 4.165441176470589e-05,
-      "loss": 1.3615,
+      "loss": 1.5656,
       "step": 454
     },
     {
       "epoch": 3.34,
       "learning_rate": 4.161764705882353e-05,
-      "loss": 1.3299,
+      "loss": 1.5368,
       "step": 456
     },
     {
       "epoch": 3.36,
       "learning_rate": 4.158088235294118e-05,
-      "loss": 1.3412,
+      "loss": 1.5456,
       "step": 458
     },
     {
       "epoch": 3.37,
       "learning_rate": 4.154411764705883e-05,
-      "loss": 1.3079,
+      "loss": 1.5175,
       "step": 460
     },
     {
       "epoch": 3.39,
       "learning_rate": 4.1507352941176475e-05,
-      "loss": 1.3701,
+      "loss": 1.4974,
       "step": 462
     },
     {
       "epoch": 3.4,
       "learning_rate": 4.147058823529412e-05,
-      "loss": 1.3815,
+      "loss": 1.5959,
       "step": 464
     },
     {
       "epoch": 3.42,
       "learning_rate": 4.1433823529411766e-05,
-      "loss": 1.4218,
+      "loss": 1.5371,
       "step": 466
     },
     {
       "epoch": 3.43,
       "learning_rate": 4.139705882352941e-05,
-      "loss": 1.3304,
+      "loss": 1.5161,
       "step": 468
     },
     {
       "epoch": 3.44,
       "learning_rate": 4.136029411764706e-05,
-      "loss": 1.3453,
+      "loss": 1.5434,
       "step": 470
     },
     {
       "epoch": 3.46,
       "learning_rate": 4.13235294117647e-05,
-      "loss": 1.3061,
+      "loss": 1.5237,
       "step": 472
     },
     {
       "epoch": 3.47,
       "learning_rate": 4.1286764705882354e-05,
-      "loss": 1.3804,
+      "loss": 1.5675,
       "step": 474
     },
     {
       "epoch": 3.49,
       "learning_rate": 4.125e-05,
-      "loss": 1.3491,
+      "loss": 1.5648,
       "step": 476
     },
     {
       "epoch": 3.5,
       "learning_rate": 4.1213235294117645e-05,
-      "loss": 1.357,
+      "loss": 1.5291,
       "step": 478
     },
     {
       "epoch": 3.52,
       "learning_rate": 4.11764705882353e-05,
-      "loss": 1.3392,
+      "loss": 1.5356,
       "step": 480
     },
     {
       "epoch": 3.53,
       "learning_rate": 4.113970588235294e-05,
-      "loss": 1.3016,
+      "loss": 1.5438,
       "step": 482
     },
     {
       "epoch": 3.55,
       "learning_rate": 4.110294117647059e-05,
-      "loss": 1.3255,
+      "loss": 1.5403,
       "step": 484
     },
     {
       "epoch": 3.56,
       "learning_rate": 4.1066176470588233e-05,
-      "loss": 1.2793,
+      "loss": 1.5521,
       "step": 486
     },
     {
       "epoch": 3.58,
       "learning_rate": 4.1029411764705886e-05,
-      "loss": 1.3481,
+      "loss": 1.5517,
       "step": 488
     },
     {
       "epoch": 3.59,
       "learning_rate": 4.099264705882353e-05,
-      "loss": 1.337,
+      "loss": 1.5388,
       "step": 490
     },
     {
       "epoch": 3.61,
       "learning_rate": 4.0955882352941176e-05,
-      "loss": 1.3136,
+      "loss": 1.5106,
       "step": 492
     },
     {
       "epoch": 3.62,
       "learning_rate": 4.091911764705882e-05,
-      "loss": 1.3408,
+      "loss": 1.4889,
       "step": 494
     },
     {
       "epoch": 3.64,
       "learning_rate": 4.0882352941176474e-05,
-      "loss": 1.3054,
+      "loss": 1.5282,
       "step": 496
     },
     {
       "epoch": 3.65,
       "learning_rate": 4.084558823529412e-05,
-      "loss": 1.3519,
+      "loss": 1.5273,
       "step": 498
     },
     {
       "epoch": 3.66,
       "learning_rate": 4.0808823529411765e-05,
-      "loss": 1.3548,
+      "loss": 1.5599,
       "step": 500
     },
     {
       "epoch": 3.68,
       "learning_rate": 4.077205882352941e-05,
-      "loss": 1.3478,
+      "loss": 1.5173,
       "step": 502
     },
     {
       "epoch": 3.69,
       "learning_rate": 4.073529411764706e-05,
-      "loss": 1.325,
+      "loss": 1.5329,
       "step": 504
     },
     {
       "epoch": 3.71,
       "learning_rate": 4.069852941176471e-05,
-      "loss": 1.3395,
+      "loss": 1.5685,
       "step": 506
     },
     {
       "epoch": 3.72,
       "learning_rate": 4.066176470588235e-05,
-      "loss": 1.3672,
+      "loss": 1.584,
       "step": 508
     },
     {
       "epoch": 3.74,
       "learning_rate": 4.0625000000000005e-05,
-      "loss": 1.2985,
+      "loss": 1.5307,
       "step": 510
     },
     {
       "epoch": 3.75,
       "learning_rate": 4.058823529411765e-05,
-      "loss": 1.3651,
+      "loss": 1.5374,
       "step": 512
     },
     {
       "epoch": 3.77,
       "learning_rate": 4.0551470588235296e-05,
-      "loss": 1.3319,
+      "loss": 1.5278,
       "step": 514
     },
     {
       "epoch": 3.78,
       "learning_rate": 4.051470588235294e-05,
-      "loss": 1.3808,
+      "loss": 1.5813,
       "step": 516
     },
     {
       "epoch": 3.8,
       "learning_rate": 4.047794117647059e-05,
-      "loss": 1.3148,
+      "loss": 1.4977,
       "step": 518
     },
     {
       "epoch": 3.81,
       "learning_rate": 4.044117647058824e-05,
-      "loss": 1.3408,
+      "loss": 1.5187,
       "step": 520
     },
     {
       "epoch": 3.83,
       "learning_rate": 4.0404411764705884e-05,
-      "loss": 1.3475,
+      "loss": 1.5442,
       "step": 522
     },
     {
       "epoch": 3.84,
       "learning_rate": 4.036764705882353e-05,
-      "loss": 1.2896,
+      "loss": 1.529,
       "step": 524
     },
     {
       "epoch": 3.86,
       "learning_rate": 4.033088235294118e-05,
-      "loss": 1.319,
+      "loss": 1.5556,
       "step": 526
     },
     {
       "epoch": 3.87,
       "learning_rate": 4.029411764705883e-05,
-      "loss": 1.3878,
+      "loss": 1.5494,
       "step": 528
     },
     {
       "epoch": 3.88,
       "learning_rate": 4.025735294117647e-05,
-      "loss": 1.3426,
+      "loss": 1.5284,
       "step": 530
     },
     {
       "epoch": 3.9,
       "learning_rate": 4.0220588235294124e-05,
-      "loss": 1.3433,
+      "loss": 1.5657,
       "step": 532
     },
     {
       "epoch": 3.91,
       "learning_rate": 4.018382352941177e-05,
-      "loss": 1.3111,
+      "loss": 1.5148,
       "step": 534
     },
     {
       "epoch": 3.93,
       "learning_rate": 4.0147058823529415e-05,
-      "loss": 1.3679,
+      "loss": 1.5335,
       "step": 536
     },
     {
       "epoch": 3.94,
       "learning_rate": 4.011029411764706e-05,
-      "loss": 1.342,
+      "loss": 1.5635,
       "step": 538
     },
     {
       "epoch": 3.96,
       "learning_rate": 4.007352941176471e-05,
-      "loss": 1.3492,
+      "loss": 1.5113,
       "step": 540
     },
     {
       "epoch": 3.97,
       "learning_rate": 4.003676470588236e-05,
-      "loss": 1.3343,
+      "loss": 1.4989,
       "step": 542
     },
     {
       "epoch": 3.99,
       "learning_rate": 4e-05,
-      "loss": 1.3399,
+      "loss": 1.4951,
       "step": 544
     },
     {
       "epoch": 3.99,
-      "eval_gen_len": 128.50617283950618,
-      "eval_loss": 1.445023536682129,
-      "eval_rouge1": 56.2783,
-      "eval_rouge2": 31.1387,
-      "eval_rougeL": 39.2121,
-      "eval_rougeLsum": 51.8068,
-      "eval_runtime": 215.4293,
-      "eval_samples_per_second": 4.512,
-      "eval_steps_per_second": 0.566,
+      "eval_gen_len": 753.59670781893,
+      "eval_loss": 1.4494504928588867,
+      "eval_rouge1": 71.5808,
+      "eval_rouge2": 40.5556,
+      "eval_rougeL": 37.152,
+      "eval_rougeLsum": 69.0536,
+      "eval_runtime": 1658.1467,
+      "eval_samples_per_second": 0.586,
+      "eval_steps_per_second": 0.074,
       "step": 545
     },
     {
       "epoch": 4.0,
       "learning_rate": 3.996323529411765e-05,
-      "loss": 1.2389,
+      "loss": 1.491,
       "step": 546
     },
     {
       "epoch": 4.02,
       "learning_rate": 3.99264705882353e-05,
-      "loss": 1.3121,
+      "loss": 1.5411,
       "step": 548
     },
     {
       "epoch": 4.03,
       "learning_rate": 3.9889705882352946e-05,
-      "loss": 1.1987,
+      "loss": 1.4722,
       "step": 550
     },
     {
       "epoch": 4.05,
       "learning_rate": 3.985294117647059e-05,
-      "loss": 1.2706,
+      "loss": 1.543,
       "step": 552
     },
     {
       "epoch": 4.06,
       "learning_rate": 3.981617647058824e-05,
-      "loss": 1.3032,
+      "loss": 1.5291,
       "step": 554
     },
     {
       "epoch": 4.08,
       "learning_rate": 3.977941176470588e-05,
-      "loss": 1.1786,
+      "loss": 1.5114,
       "step": 556
     },
     {
       "epoch": 4.09,
       "learning_rate": 3.974264705882353e-05,
-      "loss": 1.1998,
+      "loss": 1.455,
       "step": 558
     },
     {
       "epoch": 4.1,
       "learning_rate": 3.970588235294117e-05,
-      "loss": 1.2755,
+      "loss": 1.5099,
       "step": 560
     },
     {
       "epoch": 4.12,
       "learning_rate": 3.9669117647058825e-05,
-      "loss": 1.2423,
+      "loss": 1.435,
       "step": 562
     },
     {
       "epoch": 4.13,
       "learning_rate": 3.963235294117647e-05,
-      "loss": 1.2877,
+      "loss": 1.5318,
       "step": 564
     },
     {
       "epoch": 4.15,
       "learning_rate": 3.9595588235294116e-05,
-      "loss": 1.2659,
+      "loss": 1.459,
       "step": 566
     },
     {
       "epoch": 4.16,
       "learning_rate": 3.955882352941177e-05,
-      "loss": 1.319,
+      "loss": 1.5056,
       "step": 568
     },
     {
       "epoch": 4.18,
       "learning_rate": 3.952205882352941e-05,
-      "loss": 1.2898,
+      "loss": 1.5043,
       "step": 570
     },
     {
       "epoch": 4.19,
       "learning_rate": 3.948529411764706e-05,
-      "loss": 1.298,
+      "loss": 1.4988,
       "step": 572
     },
     {
       "epoch": 4.21,
       "learning_rate": 3.9448529411764704e-05,
-      "loss": 1.2663,
+      "loss": 1.4962,
       "step": 574
     },
     {
       "epoch": 4.22,
       "learning_rate": 3.9411764705882356e-05,
-      "loss": 1.2646,
+      "loss": 1.4977,
       "step": 576
     },
     {
       "epoch": 4.24,
       "learning_rate": 3.9375e-05,
-      "loss": 1.3008,
+      "loss": 1.5485,
       "step": 578
     },
     {
       "epoch": 4.25,
       "learning_rate": 3.933823529411765e-05,
-      "loss": 1.2065,
+      "loss": 1.4396,
       "step": 580
     },
     {
       "epoch": 4.27,
       "learning_rate": 3.930147058823529e-05,
-      "loss": 1.3354,
+      "loss": 1.5226,
       "step": 582
     },
     {
       "epoch": 4.28,
       "learning_rate": 3.9264705882352945e-05,
-      "loss": 1.2798,
+      "loss": 1.4878,
       "step": 584
     },
     {
       "epoch": 4.3,
       "learning_rate": 3.922794117647059e-05,
-      "loss": 1.2848,
+      "loss": 1.5179,
       "step": 586
     },
     {
       "epoch": 4.31,
       "learning_rate": 3.9191176470588235e-05,
-      "loss": 1.3071,
+      "loss": 1.5323,
       "step": 588
     },
     {
       "epoch": 4.32,
       "learning_rate": 3.915441176470588e-05,
-      "loss": 1.2356,
+      "loss": 1.4906,
       "step": 590
     },
     {
       "epoch": 4.34,
       "learning_rate": 3.911764705882353e-05,
-      "loss": 1.2597,
+      "loss": 1.4704,
       "step": 592
     },
     {
       "epoch": 4.35,
       "learning_rate": 3.908088235294118e-05,
-      "loss": 1.2158,
+      "loss": 1.4559,
       "step": 594
     },
     {
       "epoch": 4.37,
       "learning_rate": 3.9044117647058823e-05,
-      "loss": 1.2774,
+      "loss": 1.5595,
       "step": 596
     },
     {
       "epoch": 4.38,
       "learning_rate": 3.9007352941176476e-05,
-      "loss": 1.2969,
+      "loss": 1.5314,
       "step": 598
     },
     {
       "epoch": 4.4,
       "learning_rate": 3.897058823529412e-05,
-      "loss": 1.2796,
+      "loss": 1.4872,
       "step": 600
     },
     {
       "epoch": 4.41,
       "learning_rate": 3.8933823529411766e-05,
-      "loss": 1.222,
+      "loss": 1.5049,
       "step": 602
     },
     {
       "epoch": 4.43,
       "learning_rate": 3.889705882352941e-05,
-      "loss": 1.2782,
+      "loss": 1.4839,
       "step": 604
     },
     {
       "epoch": 4.44,
       "learning_rate": 3.8860294117647064e-05,
-      "loss": 1.2615,
+      "loss": 1.5166,
       "step": 606
     },
     {
       "epoch": 4.46,
       "learning_rate": 3.882352941176471e-05,
-      "loss": 1.2746,
+      "loss": 1.4871,
       "step": 608
     },
     {
       "epoch": 4.47,
       "learning_rate": 3.8786764705882355e-05,
-      "loss": 1.2141,
+      "loss": 1.5005,
       "step": 610
     },
     {
       "epoch": 4.49,
       "learning_rate": 3.875e-05,
-      "loss": 1.333,
+      "loss": 1.4969,
       "step": 612
     },
     {
       "epoch": 4.5,
       "learning_rate": 3.871323529411765e-05,
-      "loss": 1.2931,
+      "loss": 1.5049,
       "step": 614
     },
     {
       "epoch": 4.51,
       "learning_rate": 3.86764705882353e-05,
-      "loss": 1.3037,
+      "loss": 1.5228,
       "step": 616
     },
     {
       "epoch": 4.53,
       "learning_rate": 3.863970588235294e-05,
-      "loss": 1.2758,
+      "loss": 1.4748,
       "step": 618
     },
     {
       "epoch": 4.54,
       "learning_rate": 3.8602941176470595e-05,
-      "loss": 1.2348,
+      "loss": 1.5033,
       "step": 620
     },
     {
       "epoch": 4.56,
       "learning_rate": 3.856617647058824e-05,
-      "loss": 1.3094,
+      "loss": 1.4935,
       "step": 622
     },
     {
       "epoch": 4.57,
       "learning_rate": 3.8529411764705886e-05,
-      "loss": 1.2596,
+      "loss": 1.4777,
       "step": 624
     },
     {
       "epoch": 4.59,
       "learning_rate": 3.849264705882353e-05,
-      "loss": 1.2625,
+      "loss": 1.4853,
       "step": 626
     },
     {
       "epoch": 4.6,
       "learning_rate": 3.845588235294118e-05,
-      "loss": 1.2935,
+      "loss": 1.5046,
       "step": 628
     },
     {
       "epoch": 4.62,
       "learning_rate": 3.841911764705883e-05,
-      "loss": 1.2953,
+      "loss": 1.5109,
       "step": 630
     },
     {
       "epoch": 4.63,
       "learning_rate": 3.8382352941176474e-05,
-      "loss": 1.3194,
+      "loss": 1.5174,
       "step": 632
     },
     {
       "epoch": 4.65,
       "learning_rate": 3.834558823529412e-05,
-      "loss": 1.2109,
+      "loss": 1.465,
       "step": 634
     },
     {
       "epoch": 4.66,
       "learning_rate": 3.830882352941177e-05,
-      "loss": 1.2562,
+      "loss": 1.5373,
       "step": 636
     },
     {
       "epoch": 4.68,
       "learning_rate": 3.827205882352941e-05,
-      "loss": 1.2706,
+      "loss": 1.4863,
       "step": 638
     },
     {
       "epoch": 4.69,
       "learning_rate": 3.8235294117647055e-05,
-      "loss": 1.2471,
+      "loss": 1.4979,
       "step": 640
     },
     {
       "epoch": 4.71,
       "learning_rate": 3.819852941176471e-05,
-      "loss": 1.2963,
+      "loss": 1.5164,
       "step": 642
     },
     {
       "epoch": 4.72,
       "learning_rate": 3.816176470588235e-05,
-      "loss": 1.3085,
+      "loss": 1.5135,
       "step": 644
     },
     {
       "epoch": 4.73,
       "learning_rate": 3.8125e-05,
-      "loss": 1.3405,
+      "loss": 1.5337,
       "step": 646
     },
     {
       "epoch": 4.75,
       "learning_rate": 3.8088235294117644e-05,
-      "loss": 1.2403,
+      "loss": 1.4749,
       "step": 648
     },
     {
       "epoch": 4.76,
       "learning_rate": 3.8051470588235296e-05,
-      "loss": 1.3096,
+      "loss": 1.5061,
       "step": 650
     },
     {
       "epoch": 4.78,
       "learning_rate": 3.801470588235294e-05,
-      "loss": 1.2445,
+      "loss": 1.4992,
       "step": 652
     },
     {
       "epoch": 4.79,
       "learning_rate": 3.7977941176470587e-05,
-      "loss": 1.2315,
+      "loss": 1.4527,
       "step": 654
     },
     {
       "epoch": 4.81,
       "learning_rate": 3.794117647058824e-05,
-      "loss": 1.2387,
+      "loss": 1.4815,
       "step": 656
     },
     {
       "epoch": 4.82,
       "learning_rate": 3.7904411764705884e-05,
-      "loss": 1.2679,
+      "loss": 1.4803,
       "step": 658
     },
     {
       "epoch": 4.84,
       "learning_rate": 3.786764705882353e-05,
-      "loss": 1.2949,
+      "loss": 1.5093,
       "step": 660
     },
     {
       "epoch": 4.85,
       "learning_rate": 3.7830882352941175e-05,
-      "loss": 1.2746,
+      "loss": 1.5159,
       "step": 662
     },
     {
       "epoch": 4.87,
       "learning_rate": 3.779411764705883e-05,
-      "loss": 1.2897,
+      "loss": 1.5017,
       "step": 664
     },
     {
       "epoch": 4.88,
       "learning_rate": 3.775735294117647e-05,
-      "loss": 1.2759,
+      "loss": 1.5075,
       "step": 666
     },
     {
       "epoch": 4.9,
       "learning_rate": 3.772058823529412e-05,
-      "loss": 1.2655,
+      "loss": 1.5284,
       "step": 668
     },
     {
       "epoch": 4.91,
       "learning_rate": 3.768382352941176e-05,
-      "loss": 1.2816,
+      "loss": 1.4895,
       "step": 670
     },
     {
       "epoch": 4.93,
       "learning_rate": 3.7647058823529415e-05,
-      "loss": 1.2932,
+      "loss": 1.4893,
       "step": 672
     },
     {
       "epoch": 4.94,
       "learning_rate": 3.761029411764706e-05,
-      "loss": 1.2306,
+      "loss": 1.4695,
       "step": 674
     },
     {
       "epoch": 4.95,
       "learning_rate": 3.7573529411764706e-05,
-      "loss": 1.2941,
+      "loss": 1.4988,
       "step": 676
     },
     {
       "epoch": 4.97,
       "learning_rate": 3.753676470588235e-05,
-      "loss": 1.2228,
+      "loss": 1.4898,
       "step": 678
     },
     {
       "epoch": 4.98,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 1.2257,
+      "loss": 1.4561,
       "step": 680
     },
     {
       "epoch": 5.0,
       "learning_rate": 3.746323529411765e-05,
-      "loss": 1.2326,
+      "loss": 1.496,
       "step": 682
     },
     {
       "epoch": 5.0,
-      "eval_gen_len": 129.1358024691358,
-      "eval_loss": 1.4477734565734863,
-      "eval_rouge1": 56.0182,
-      "eval_rouge2": 30.7104,
-      "eval_rougeL": 38.8337,
-      "eval_rougeLsum": 51.6162,
-      "eval_runtime": 220.6236,
-      "eval_samples_per_second": 4.406,
-      "eval_steps_per_second": 0.553,
+      "eval_gen_len": 700.216049382716,
+      "eval_loss": 1.438555359840393,
+      "eval_rouge1": 72.1271,
+      "eval_rouge2": 41.1645,
+      "eval_rougeL": 38.4096,
+      "eval_rougeLsum": 69.6176,
+      "eval_runtime": 1598.9004,
+      "eval_samples_per_second": 0.608,
+      "eval_steps_per_second": 0.076,
       "step": 682
     },
     {
       "epoch": 5.01,
       "learning_rate": 3.7426470588235294e-05,
-      "loss": 1.2188,
+      "loss": 1.4707,
       "step": 684
     },
     {
       "epoch": 5.03,
       "learning_rate": 3.7389705882352946e-05,
-      "loss": 1.2168,
+      "loss": 1.5299,
       "step": 686
     },
     {
       "epoch": 5.04,
       "learning_rate": 3.735294117647059e-05,
-      "loss": 1.2027,
+      "loss": 1.46,
       "step": 688
     },
     {
       "epoch": 5.06,
       "learning_rate": 3.731617647058824e-05,
-      "loss": 1.223,
+      "loss": 1.4996,
       "step": 690
     },
     {
       "epoch": 5.07,
       "learning_rate": 3.727941176470588e-05,
-      "loss": 1.1577,
+      "loss": 1.4146,
       "step": 692
     },
     {
       "epoch": 5.09,
       "learning_rate": 3.7242647058823535e-05,
-      "loss": 1.2015,
+      "loss": 1.3875,
       "step": 694
     },
     {
       "epoch": 5.1,
       "learning_rate": 3.720588235294118e-05,
-      "loss": 1.15,
+      "loss": 1.4534,
       "step": 696
     },
     {
       "epoch": 5.12,
       "learning_rate": 3.7169117647058825e-05,
-      "loss": 1.2414,
+      "loss": 1.4974,
       "step": 698
     },
     {
       "epoch": 5.13,
       "learning_rate": 3.713235294117647e-05,
-      "loss": 1.2267,
+      "loss": 1.4793,
       "step": 700
     },
     {
       "epoch": 5.15,
       "learning_rate": 3.709558823529412e-05,
-      "loss": 1.2259,
+      "loss": 1.4864,
       "step": 702
     },
     {
       "epoch": 5.16,
       "learning_rate": 3.705882352941177e-05,
-      "loss": 1.2137,
+      "loss": 1.4895,
       "step": 704
     },
     {
       "epoch": 5.17,
       "learning_rate": 3.7022058823529414e-05,
-      "loss": 1.2525,
+      "loss": 1.4974,
       "step": 706
     },
     {
       "epoch": 5.19,
       "learning_rate": 3.6985294117647066e-05,
-      "loss": 1.2228,
+      "loss": 1.4878,
       "step": 708
     },
     {
       "epoch": 5.2,
       "learning_rate": 3.694852941176471e-05,
-      "loss": 1.2692,
+      "loss": 1.5243,
       "step": 710
     },
     {
       "epoch": 5.22,
       "learning_rate": 3.6911764705882356e-05,
-      "loss": 1.2168,
+      "loss": 1.4038,
       "step": 712
     },
     {
       "epoch": 5.23,
       "learning_rate": 3.6875e-05,
-      "loss": 1.2035,
+      "loss": 1.4994,
       "step": 714
     },
     {
       "epoch": 5.25,
       "learning_rate": 3.6838235294117654e-05,
-      "loss": 1.1863,
+      "loss": 1.4496,
       "step": 716
     },
     {
       "epoch": 5.26,
       "learning_rate": 3.68014705882353e-05,
-      "loss": 1.2229,
+      "loss": 1.4704,
       "step": 718
     },
     {
       "epoch": 5.28,
       "learning_rate": 3.6764705882352945e-05,
-      "loss": 1.2149,
+      "loss": 1.4691,
       "step": 720
     },
     {
       "epoch": 5.29,
       "learning_rate": 3.672794117647059e-05,
-      "loss": 1.218,
+      "loss": 1.4459,
       "step": 722
     },
     {
       "epoch": 5.31,
       "learning_rate": 3.6691176470588235e-05,
-      "loss": 1.2226,
+      "loss": 1.4827,
       "step": 724
     },
     {
       "epoch": 5.32,
       "learning_rate": 3.665441176470588e-05,
-      "loss": 1.207,
+      "loss": 1.4771,
       "step": 726
     },
     {
       "epoch": 5.34,
       "learning_rate": 3.6617647058823526e-05,
-      "loss": 1.2022,
+      "loss": 1.4671,
       "step": 728
     },
     {
       "epoch": 5.35,
       "learning_rate": 3.658088235294118e-05,
-      "loss": 1.1717,
+      "loss": 1.4551,
       "step": 730
     },
     {
       "epoch": 5.37,
       "learning_rate": 3.6544117647058824e-05,
-      "loss": 1.1861,
+      "loss": 1.4747,
       "step": 732
     },
     {
       "epoch": 5.38,
       "learning_rate": 3.650735294117647e-05,
-      "loss": 1.2465,
+      "loss": 1.4824,
       "step": 734
     },
     {
       "epoch": 5.39,
       "learning_rate": 3.6470588235294114e-05,
-      "loss": 1.2025,
+      "loss": 1.4738,
       "step": 736
     },
     {
       "epoch": 5.41,
       "learning_rate": 3.6433823529411767e-05,
-      "loss": 1.2396,
+      "loss": 1.445,
       "step": 738
     },
     {
       "epoch": 5.42,
       "learning_rate": 3.639705882352941e-05,
-      "loss": 1.2238,
+      "loss": 1.4796,
       "step": 740
     },
     {
       "epoch": 5.44,
       "learning_rate": 3.636029411764706e-05,
-      "loss": 1.187,
+      "loss": 1.4632,
       "step": 742
     },
     {
       "epoch": 5.45,
       "learning_rate": 3.632352941176471e-05,
-      "loss": 1.2013,
+      "loss": 1.4111,
       "step": 744
     },
     {
       "epoch": 5.47,
       "learning_rate": 3.6286764705882355e-05,
-      "loss": 1.2093,
+      "loss": 1.497,
       "step": 746
     },
     {
       "epoch": 5.48,
       "learning_rate": 3.625e-05,
-      "loss": 1.2343,
+      "loss": 1.502,
       "step": 748
     },
     {
       "epoch": 5.5,
       "learning_rate": 3.6213235294117646e-05,
-      "loss": 1.2589,
+      "loss": 1.4615,
       "step": 750
     },
     {
       "epoch": 5.51,
       "learning_rate": 3.61764705882353e-05,
-      "loss": 1.2349,
+      "loss": 1.4528,
       "step": 752
     },
     {
       "epoch": 5.53,
       "learning_rate": 3.613970588235294e-05,
-      "loss": 1.1512,
+      "loss": 1.4344,
       "step": 754
     },
     {
       "epoch": 5.54,
       "learning_rate": 3.610294117647059e-05,
-      "loss": 1.2158,
+      "loss": 1.4592,
       "step": 756
     },
     {
       "epoch": 5.56,
       "learning_rate": 3.6066176470588234e-05,
-      "loss": 1.1684,
+      "loss": 1.4368,
       "step": 758
     },
     {
       "epoch": 5.57,
       "learning_rate": 3.6029411764705886e-05,
-      "loss": 1.255,
+      "loss": 1.4946,
       "step": 760
     },
     {
       "epoch": 5.58,
       "learning_rate": 3.599264705882353e-05,
-      "loss": 1.2116,
+      "loss": 1.5049,
       "step": 762
     },
     {
       "epoch": 5.6,
       "learning_rate": 3.595588235294118e-05,
-      "loss": 1.2747,
+      "loss": 1.4952,
       "step": 764
     },
     {
       "epoch": 5.61,
       "learning_rate": 3.591911764705882e-05,
-      "loss": 1.1318,
+      "loss": 1.4093,
       "step": 766
     },
     {
       "epoch": 5.63,
       "learning_rate": 3.5882352941176474e-05,
-      "loss": 1.1944,
+      "loss": 1.4677,
       "step": 768
     },
     {
       "epoch": 5.64,
       "learning_rate": 3.584558823529412e-05,
-      "loss": 1.2051,
+      "loss": 1.4667,
       "step": 770
     },
     {
       "epoch": 5.66,
       "learning_rate": 3.5808823529411765e-05,
-      "loss": 1.2404,
+      "loss": 1.4669,
       "step": 772
     },
     {
       "epoch": 5.67,
       "learning_rate": 3.577205882352942e-05,
-      "loss": 1.1943,
+      "loss": 1.495,
       "step": 774
     },
     {
       "epoch": 5.69,
       "learning_rate": 3.573529411764706e-05,
-      "loss": 1.1969,
+      "loss": 1.4924,
       "step": 776
     },
     {
       "epoch": 5.7,
       "learning_rate": 3.569852941176471e-05,
-      "loss": 1.1994,
+      "loss": 1.4797,
       "step": 778
     },
     {
       "epoch": 5.72,
       "learning_rate": 3.566176470588235e-05,
-      "loss": 1.2162,
+      "loss": 1.4765,
       "step": 780
     },
     {
       "epoch": 5.73,
       "learning_rate": 3.5625000000000005e-05,
-      "loss": 1.2033,
+      "loss": 1.4418,
       "step": 782
     },
     {
       "epoch": 5.75,
       "learning_rate": 3.558823529411765e-05,
-      "loss": 1.1769,
+      "loss": 1.4843,
       "step": 784
     },
     {
       "epoch": 5.76,
       "learning_rate": 3.5551470588235296e-05,
-      "loss": 1.1423,
+      "loss": 1.456,
       "step": 786
     },
     {
       "epoch": 5.78,
       "learning_rate": 3.551470588235294e-05,
-      "loss": 1.2333,
+      "loss": 1.5136,
       "step": 788
     },
     {
       "epoch": 5.79,
       "learning_rate": 3.5477941176470594e-05,
-      "loss": 1.2148,
+      "loss": 1.4719,
       "step": 790
     },
     {
       "epoch": 5.8,
       "learning_rate": 3.544117647058824e-05,
-      "loss": 1.1786,
+      "loss": 1.4212,
       "step": 792
     },
     {
       "epoch": 5.82,
       "learning_rate": 3.5404411764705884e-05,
-      "loss": 1.1703,
+      "loss": 1.4173,
       "step": 794
     },
     {
       "epoch": 5.83,
       "learning_rate": 3.5367647058823536e-05,
-      "loss": 1.1893,
+      "loss": 1.4575,
       "step": 796
     },
     {
       "epoch": 5.85,
       "learning_rate": 3.533088235294118e-05,
-      "loss": 1.1783,
+      "loss": 1.4826,
       "step": 798
     },
     {
       "epoch": 5.86,
       "learning_rate": 3.529411764705883e-05,
-      "loss": 1.1627,
+      "loss": 1.421,
       "step": 800
     },
     {
       "epoch": 5.88,
       "learning_rate": 3.525735294117647e-05,
-      "loss": 1.2095,
+      "loss": 1.4342,
       "step": 802
     },
     {
       "epoch": 5.89,
       "learning_rate": 3.5220588235294125e-05,
-      "loss": 1.1953,
+      "loss": 1.4659,
       "step": 804
     },
     {
       "epoch": 5.91,
       "learning_rate": 3.518382352941177e-05,
-      "loss": 1.1971,
+      "loss": 1.4755,
       "step": 806
     },
     {
       "epoch": 5.92,
       "learning_rate": 3.514705882352941e-05,
-      "loss": 1.2699,
+      "loss": 1.474,
       "step": 808
     },
     {
       "epoch": 5.94,
       "learning_rate": 3.511029411764706e-05,
-      "loss": 1.2029,
+      "loss": 1.4797,
       "step": 810
     },
     {
       "epoch": 5.95,
       "learning_rate": 3.5073529411764706e-05,
-      "loss": 1.2145,
+      "loss": 1.4609,
       "step": 812
     },
     {
       "epoch": 5.97,
       "learning_rate": 3.503676470588235e-05,
-      "loss": 1.2096,
+      "loss": 1.4638,
       "step": 814
     },
     {
       "epoch": 5.98,
       "learning_rate": 3.5e-05,
-      "loss": 1.2102,
+      "loss": 1.4634,
       "step": 816
     },
     {
       "epoch": 6.0,
       "learning_rate": 3.496323529411765e-05,
-      "loss": 1.1784,
+      "loss": 1.4258,
       "step": 818
     },
     {
       "epoch": 6.0,
-      "eval_gen_len": 128.7314814814815,
-      "eval_loss": 1.4533237218856812,
-      "eval_rouge1": 56.4333,
-      "eval_rouge2": 31.4483,
-      "eval_rougeL": 39.5546,
-      "eval_rougeLsum": 52.1347,
-      "eval_runtime": 216.1787,
-      "eval_samples_per_second": 4.496,
-      "eval_steps_per_second": 0.564,
+      "eval_gen_len": 743.7067901234568,
+      "eval_loss": 1.4373914003372192,
+      "eval_rouge1": 71.9975,
+      "eval_rouge2": 41.0013,
+      "eval_rougeL": 37.9947,
+      "eval_rougeLsum": 69.449,
+      "eval_runtime": 1641.7585,
+      "eval_samples_per_second": 0.592,
+      "eval_steps_per_second": 0.074,
       "step": 818
     },
     {
       "epoch": 6.01,
       "learning_rate": 3.4926470588235294e-05,
-      "loss": 1.1578,
+      "loss": 1.4952,
       "step": 820
     },
     {
       "epoch": 6.02,
       "learning_rate": 3.488970588235294e-05,
-      "loss": 1.1257,
+      "loss": 1.4216,
       "step": 822
     },
     {
       "epoch": 6.04,
       "learning_rate": 3.4852941176470585e-05,
-      "loss": 1.1946,
+      "loss": 1.4445,
       "step": 824
     },
     {
       "epoch": 6.05,
       "learning_rate": 3.481617647058824e-05,
-      "loss": 1.1224,
+      "loss": 1.4159,
       "step": 826
     },
     {
       "epoch": 6.07,
       "learning_rate": 3.477941176470588e-05,
-      "loss": 1.1055,
+      "loss": 1.3788,
       "step": 828
     },
     {
       "epoch": 6.08,
       "learning_rate": 3.474264705882353e-05,
-      "loss": 1.1109,
+      "loss": 1.4245,
       "step": 830
     },
     {
       "epoch": 6.1,
       "learning_rate": 3.470588235294118e-05,
-      "loss": 1.1532,
+      "loss": 1.4599,
       "step": 832
     },
     {
       "epoch": 6.11,
       "learning_rate": 3.4669117647058826e-05,
-      "loss": 1.0739,
+      "loss": 1.4125,
       "step": 834
     },
     {
       "epoch": 6.13,
       "learning_rate": 3.463235294117647e-05,
-      "loss": 1.1052,
+      "loss": 1.4442,
       "step": 836
     },
     {
       "epoch": 6.14,
       "learning_rate": 3.4595588235294116e-05,
-      "loss": 1.1529,
+      "loss": 1.41,
       "step": 838
     },
     {
       "epoch": 6.16,
       "learning_rate": 3.455882352941177e-05,
-      "loss": 1.1235,
+      "loss": 1.4069,
       "step": 840
     },
     {
       "epoch": 6.17,
       "learning_rate": 3.4522058823529414e-05,
-      "loss": 1.1584,
+      "loss": 1.4513,
       "step": 842
     },
     {
       "epoch": 6.19,
       "learning_rate": 3.448529411764706e-05,
-      "loss": 1.1058,
+      "loss": 1.4286,
       "step": 844
     },
     {
       "epoch": 6.2,
       "learning_rate": 3.4448529411764705e-05,
-      "loss": 1.2066,
+      "loss": 1.472,
       "step": 846
     },
     {
       "epoch": 6.22,
       "learning_rate": 3.441176470588236e-05,
-      "loss": 1.1805,
+      "loss": 1.4066,
       "step": 848
     },
     {
       "epoch": 6.23,
       "learning_rate": 3.4375e-05,
-      "loss": 1.1599,
+      "loss": 1.4513,
       "step": 850
     },
     {
       "epoch": 6.24,
       "learning_rate": 3.433823529411765e-05,
-      "loss": 1.1682,
+      "loss": 1.4255,
       "step": 852
     },
     {
       "epoch": 6.26,
       "learning_rate": 3.430147058823529e-05,
-      "loss": 1.21,
+      "loss": 1.479,
       "step": 854
     },
     {
       "epoch": 6.27,
       "learning_rate": 3.4264705882352945e-05,
-      "loss": 1.1669,
+      "loss": 1.4979,
       "step": 856
     },
     {
       "epoch": 6.29,
       "learning_rate": 3.422794117647059e-05,
-      "loss": 1.1205,
+      "loss": 1.4266,
       "step": 858
     },
     {
       "epoch": 6.3,
       "learning_rate": 3.4191176470588236e-05,
-      "loss": 1.1411,
+      "loss": 1.4678,
       "step": 860
     },
     {
       "epoch": 6.32,
       "learning_rate": 3.415441176470589e-05,
-      "loss": 1.1476,
+      "loss": 1.3908,
       "step": 862
     },
     {
       "epoch": 6.33,
       "learning_rate": 3.411764705882353e-05,
-      "loss": 1.1084,
+      "loss": 1.445,
       "step": 864
     },
     {
       "epoch": 6.35,
       "learning_rate": 3.408088235294118e-05,
-      "loss": 1.1273,
+      "loss": 1.4338,
       "step": 866
     },
     {
       "epoch": 6.36,
       "learning_rate": 3.4044117647058824e-05,
-      "loss": 1.119,
+      "loss": 1.439,
       "step": 868
     },
     {
       "epoch": 6.38,
       "learning_rate": 3.4007352941176476e-05,
-      "loss": 1.1379,
+      "loss": 1.4452,
       "step": 870
     },
     {
       "epoch": 6.39,
       "learning_rate": 3.397058823529412e-05,
-      "loss": 1.1741,
+      "loss": 1.4424,
       "step": 872
     },
     {
       "epoch": 6.41,
       "learning_rate": 3.393382352941177e-05,
-      "loss": 1.1785,
+      "loss": 1.4427,
       "step": 874
     },
     {
       "epoch": 6.42,
       "learning_rate": 3.389705882352941e-05,
-      "loss": 1.1535,
+      "loss": 1.4691,
       "step": 876
     },
     {
       "epoch": 6.44,
       "learning_rate": 3.3860294117647064e-05,
-      "loss": 1.1566,
+      "loss": 1.4458,
       "step": 878
     },
     {
       "epoch": 6.45,
       "learning_rate": 3.382352941176471e-05,
-      "loss": 1.1341,
+      "loss": 1.3844,
       "step": 880
     },
     {
       "epoch": 6.46,
       "learning_rate": 3.3786764705882355e-05,
-      "loss": 1.1289,
+      "loss": 1.4202,
       "step": 882
     },
     {
       "epoch": 6.48,
       "learning_rate": 3.375000000000001e-05,
-      "loss": 1.1759,
+      "loss": 1.4793,
       "step": 884
     },
     {
       "epoch": 6.49,
       "learning_rate": 3.371323529411765e-05,
-      "loss": 1.1628,
+      "loss": 1.4329,
       "step": 886
     },
     {
       "epoch": 6.51,
       "learning_rate": 3.36764705882353e-05,
-      "loss": 1.1737,
+      "loss": 1.4631,
       "step": 888
     },
     {
       "epoch": 6.52,
       "learning_rate": 3.363970588235294e-05,
-      "loss": 1.1817,
+      "loss": 1.4744,
       "step": 890
     },
     {
       "epoch": 6.54,
       "learning_rate": 3.3602941176470595e-05,
-      "loss": 1.1518,
+      "loss": 1.4477,
       "step": 892
     },
     {
       "epoch": 6.55,
       "learning_rate": 3.3566176470588234e-05,
-      "loss": 1.1865,
+      "loss": 1.4235,
       "step": 894
     },
     {
       "epoch": 6.57,
       "learning_rate": 3.352941176470588e-05,
-      "loss": 1.1342,
+      "loss": 1.4051,
       "step": 896
     },
     {
       "epoch": 6.58,
       "learning_rate": 3.349264705882353e-05,
-      "loss": 1.1437,
+      "loss": 1.4575,
       "step": 898
     },
     {
       "epoch": 6.6,
       "learning_rate": 3.345588235294118e-05,
-      "loss": 1.2004,
+      "loss": 1.4539,
       "step": 900
     },
     {
       "epoch": 6.61,
       "learning_rate": 3.341911764705882e-05,
-      "loss": 1.148,
+      "loss": 1.4338,
       "step": 902
     },
     {
       "epoch": 6.63,
       "learning_rate": 3.338235294117647e-05,
-      "loss": 1.1445,
+      "loss": 1.4074,
       "step": 904
     },
     {
       "epoch": 6.64,
       "learning_rate": 3.334558823529412e-05,
-      "loss": 1.1467,
+      "loss": 1.4157,
       "step": 906
     },
     {
       "epoch": 6.66,
       "learning_rate": 3.3308823529411765e-05,
-      "loss": 1.1816,
+      "loss": 1.4574,
       "step": 908
     },
     {
       "epoch": 6.67,
       "learning_rate": 3.327205882352941e-05,
-      "loss": 1.204,
+      "loss": 1.4211,
       "step": 910
     },
     {
       "epoch": 6.68,
       "learning_rate": 3.3235294117647056e-05,
-      "loss": 1.1821,
+      "loss": 1.4321,
       "step": 912
     },
     {
       "epoch": 6.7,
       "learning_rate": 3.319852941176471e-05,
-      "loss": 1.1906,
+      "loss": 1.439,
       "step": 914
     },
     {
       "epoch": 6.71,
       "learning_rate": 3.3161764705882353e-05,
-      "loss": 1.138,
+      "loss": 1.4324,
       "step": 916
     },
     {
       "epoch": 6.73,
       "learning_rate": 3.3125e-05,
-      "loss": 1.1876,
+      "loss": 1.4745,
       "step": 918
     },
     {
       "epoch": 6.74,
       "learning_rate": 3.308823529411765e-05,
-      "loss": 1.1551,
+      "loss": 1.4624,
       "step": 920
     },
     {
       "epoch": 6.76,
       "learning_rate": 3.3051470588235296e-05,
-      "loss": 1.1642,
+      "loss": 1.4574,
       "step": 922
     },
     {
       "epoch": 6.77,
       "learning_rate": 3.301470588235294e-05,
-      "loss": 1.1545,
+      "loss": 1.4399,
       "step": 924
     },
     {
       "epoch": 6.79,
       "learning_rate": 3.297794117647059e-05,
-      "loss": 1.2166,
+      "loss": 1.4982,
       "step": 926
     },
     {
       "epoch": 6.8,
       "learning_rate": 3.294117647058824e-05,
-      "loss": 1.1004,
+      "loss": 1.4166,
       "step": 928
     },
     {
       "epoch": 6.82,
       "learning_rate": 3.2904411764705885e-05,
-      "loss": 1.1681,
+      "loss": 1.4914,
       "step": 930
     },
     {
       "epoch": 6.83,
       "learning_rate": 3.286764705882353e-05,
-      "loss": 1.1348,
+      "loss": 1.4439,
       "step": 932
     },
     {
       "epoch": 6.85,
       "learning_rate": 3.2830882352941175e-05,
-      "loss": 1.1448,
+      "loss": 1.4482,
       "step": 934
     },
     {
       "epoch": 6.86,
       "learning_rate": 3.279411764705883e-05,
-      "loss": 1.1787,
+      "loss": 1.413,
       "step": 936
     },
     {
       "epoch": 6.87,
       "learning_rate": 3.275735294117647e-05,
-      "loss": 1.1697,
+      "loss": 1.4354,
       "step": 938
     },
     {
       "epoch": 6.89,
       "learning_rate": 3.272058823529412e-05,
-      "loss": 1.1161,
+      "loss": 1.4421,
       "step": 940
     },
     {
       "epoch": 6.9,
       "learning_rate": 3.2683823529411764e-05,
-      "loss": 1.1297,
+      "loss": 1.4091,
       "step": 942
     },
     {
       "epoch": 6.92,
       "learning_rate": 3.2647058823529416e-05,
-      "loss": 1.1287,
+      "loss": 1.4231,
       "step": 944
     },
     {
       "epoch": 6.93,
       "learning_rate": 3.261029411764706e-05,
-      "loss": 1.1702,
+      "loss": 1.4688,
       "step": 946
     },
     {
       "epoch": 6.95,
       "learning_rate": 3.2573529411764706e-05,
-      "loss": 1.1213,
+      "loss": 1.4393,
       "step": 948
     },
     {
       "epoch": 6.96,
       "learning_rate": 3.253676470588236e-05,
-      "loss": 1.1846,
+      "loss": 1.4945,
       "step": 950
     },
     {
       "epoch": 6.98,
       "learning_rate": 3.2500000000000004e-05,
-      "loss": 1.0957,
+      "loss": 1.4128,
       "step": 952
     },
     {
       "epoch": 6.99,
       "learning_rate": 3.246323529411765e-05,
-      "loss": 1.1739,
+      "loss": 1.4301,
       "step": 954
     },
     {
       "epoch": 7.0,
-      "eval_gen_len": 128.82407407407408,
-      "eval_loss": 1.4607478380203247,
-      "eval_rouge1": 56.3636,
-      "eval_rouge2": 31.1125,
-      "eval_rougeL": 39.4055,
-      "eval_rougeLsum": 51.9709,
-      "eval_runtime": 217.708,
-      "eval_samples_per_second": 4.465,
-      "eval_steps_per_second": 0.56,
+      "eval_gen_len": 724.5061728395061,
+      "eval_loss": 1.4296016693115234,
+      "eval_rouge1": 71.8896,
+      "eval_rouge2": 40.8303,
+      "eval_rougeL": 38.346,
+      "eval_rougeLsum": 69.357,
+      "eval_runtime": 1562.869,
+      "eval_samples_per_second": 0.622,
+      "eval_steps_per_second": 0.078,
       "step": 955
     },
     {
       "epoch": 7.01,
       "learning_rate": 3.2426470588235295e-05,
-      "loss": 1.1437,
+      "loss": 1.4917,
       "step": 956
     },
     {
       "epoch": 7.02,
       "learning_rate": 3.238970588235295e-05,
-      "loss": 1.054,
+      "loss": 1.3842,
       "step": 958
     },
     {
       "epoch": 7.04,
       "learning_rate": 3.235294117647059e-05,
-      "loss": 1.0563,
+      "loss": 1.3968,
       "step": 960
     },
     {
       "epoch": 7.05,
       "learning_rate": 3.231617647058824e-05,
-      "loss": 1.0716,
+      "loss": 1.4171,
       "step": 962
     },
     {
       "epoch": 7.07,
       "learning_rate": 3.227941176470588e-05,
-      "loss": 1.1708,
+      "loss": 1.4565,
       "step": 964
     },
     {
       "epoch": 7.08,
       "learning_rate": 3.2242647058823535e-05,
-      "loss": 1.1274,
+      "loss": 1.4157,
       "step": 966
     },
     {
       "epoch": 7.09,
       "learning_rate": 3.220588235294118e-05,
-      "loss": 1.0927,
+      "loss": 1.4288,
       "step": 968
     },
     {
       "epoch": 7.11,
       "learning_rate": 3.2169117647058826e-05,
-      "loss": 1.0811,
+      "loss": 1.4051,
       "step": 970
     },
     {
       "epoch": 7.12,
       "learning_rate": 3.213235294117647e-05,
-      "loss": 1.0686,
+      "loss": 1.4092,
       "step": 972
     },
     {
       "epoch": 7.14,
       "learning_rate": 3.209558823529412e-05,
-      "loss": 1.0668,
+      "loss": 1.4243,
       "step": 974
     },
     {
       "epoch": 7.15,
       "learning_rate": 3.205882352941177e-05,
-      "loss": 1.0802,
+      "loss": 1.4042,
       "step": 976
     },
     {
       "epoch": 7.17,
       "learning_rate": 3.202205882352941e-05,
-      "loss": 1.129,
+      "loss": 1.4389,
       "step": 978
     },
     {
       "epoch": 7.18,
       "learning_rate": 3.198529411764706e-05,
-      "loss": 1.1028,
+      "loss": 1.4031,
       "step": 980
     },
     {
       "epoch": 7.2,
       "learning_rate": 3.1948529411764705e-05,
-      "loss": 1.069,
+      "loss": 1.3706,
       "step": 982
     },
     {
       "epoch": 7.21,
       "learning_rate": 3.191176470588235e-05,
-      "loss": 1.0973,
+      "loss": 1.4254,
       "step": 984
     },
     {
       "epoch": 7.23,
       "learning_rate": 3.1875e-05,
-      "loss": 1.0982,
+      "loss": 1.4442,
       "step": 986
     },
     {
       "epoch": 7.24,
       "learning_rate": 3.183823529411765e-05,
-      "loss": 1.0913,
+      "loss": 1.384,
       "step": 988
     },
     {
       "epoch": 7.26,
       "learning_rate": 3.180147058823529e-05,
-      "loss": 1.072,
+      "loss": 1.4263,
       "step": 990
     },
     {
       "epoch": 7.27,
       "learning_rate": 3.176470588235294e-05,
-      "loss": 1.1118,
+      "loss": 1.4057,
       "step": 992
     },
     {
       "epoch": 7.29,
       "learning_rate": 3.172794117647059e-05,
-      "loss": 1.0822,
+      "loss": 1.4213,
       "step": 994
     },
     {
       "epoch": 7.3,
       "learning_rate": 3.1691176470588236e-05,
-      "loss": 1.1018,
+      "loss": 1.4214,
       "step": 996
     },
     {
       "epoch": 7.31,
       "learning_rate": 3.165441176470588e-05,
-      "loss": 1.0897,
+      "loss": 1.4503,
       "step": 998
     },
     {
       "epoch": 7.33,
       "learning_rate": 3.161764705882353e-05,
-      "loss": 1.0972,
+      "loss": 1.3982,
       "step": 1000
     },
     {
       "epoch": 7.34,
       "learning_rate": 3.158088235294118e-05,
-      "loss": 1.1716,
+      "loss": 1.431,
       "step": 1002
     },
     {
       "epoch": 7.36,
       "learning_rate": 3.1544117647058824e-05,
-      "loss": 1.06,
+      "loss": 1.3804,
       "step": 1004
     },
     {
       "epoch": 7.37,
       "learning_rate": 3.150735294117647e-05,
-      "loss": 1.1046,
+      "loss": 1.4094,
       "step": 1006
     },
     {
       "epoch": 7.39,
       "learning_rate": 3.147058823529412e-05,
-      "loss": 1.0623,
+      "loss": 1.3905,
       "step": 1008
     },
     {
       "epoch": 7.4,
       "learning_rate": 3.143382352941177e-05,
-      "loss": 1.1041,
+      "loss": 1.4601,
       "step": 1010
     },
     {
       "epoch": 7.42,
       "learning_rate": 3.139705882352941e-05,
-      "loss": 1.0833,
+      "loss": 1.3997,
       "step": 1012
     },
     {
       "epoch": 7.43,
       "learning_rate": 3.136029411764706e-05,
-      "loss": 1.1318,
+      "loss": 1.4487,
       "step": 1014
     },
     {
       "epoch": 7.45,
       "learning_rate": 3.132352941176471e-05,
-      "loss": 1.0885,
+      "loss": 1.4067,
       "step": 1016
     },
     {
       "epoch": 7.46,
       "learning_rate": 3.1286764705882355e-05,
-      "loss": 1.0923,
+      "loss": 1.4207,
       "step": 1018
     },
     {
       "epoch": 7.48,
       "learning_rate": 3.125e-05,
-      "loss": 1.0532,
+      "loss": 1.4311,
       "step": 1020
     },
     {
       "epoch": 7.49,
       "learning_rate": 3.1213235294117646e-05,
-      "loss": 1.1709,
+      "loss": 1.4227,
       "step": 1022
     },
     {
       "epoch": 7.51,
       "learning_rate": 3.11764705882353e-05,
-      "loss": 1.0687,
+      "loss": 1.376,
       "step": 1024
     },
     {
       "epoch": 7.52,
       "learning_rate": 3.1139705882352944e-05,
-      "loss": 1.0702,
+      "loss": 1.4088,
       "step": 1026
     },
     {
       "epoch": 7.53,
       "learning_rate": 3.110294117647059e-05,
-      "loss": 1.1208,
+      "loss": 1.424,
       "step": 1028
     },
     {
       "epoch": 7.55,
       "learning_rate": 3.1066176470588234e-05,
-      "loss": 1.1507,
+      "loss": 1.4597,
       "step": 1030
     },
     {
       "epoch": 7.56,
       "learning_rate": 3.1029411764705886e-05,
-      "loss": 1.1507,
+      "loss": 1.438,
       "step": 1032
     },
     {
       "epoch": 7.58,
       "learning_rate": 3.099264705882353e-05,
-      "loss": 1.1436,
+      "loss": 1.4579,
       "step": 1034
     },
     {
       "epoch": 7.59,
       "learning_rate": 3.095588235294118e-05,
-      "loss": 1.1445,
+      "loss": 1.4094,
       "step": 1036
     },
     {
       "epoch": 7.61,
       "learning_rate": 3.091911764705883e-05,
-      "loss": 1.0973,
+      "loss": 1.4407,
       "step": 1038
     },
     {
       "epoch": 7.62,
       "learning_rate": 3.0882352941176475e-05,
-      "loss": 1.0896,
+      "loss": 1.4369,
       "step": 1040
     },
     {
       "epoch": 7.64,
       "learning_rate": 3.084558823529412e-05,
-      "loss": 1.1257,
+      "loss": 1.4274,
       "step": 1042
     },
     {
       "epoch": 7.65,
       "learning_rate": 3.0808823529411765e-05,
-      "loss": 1.0905,
+      "loss": 1.3982,
       "step": 1044
     },
     {
       "epoch": 7.67,
       "learning_rate": 3.077205882352942e-05,
-      "loss": 1.0523,
+      "loss": 1.4094,
       "step": 1046
     },
     {
       "epoch": 7.68,
       "learning_rate": 3.073529411764706e-05,
-      "loss": 1.1089,
+      "loss": 1.4243,
       "step": 1048
     },
     {
       "epoch": 7.7,
       "learning_rate": 3.069852941176471e-05,
-      "loss": 1.1549,
+      "loss": 1.4223,
       "step": 1050
     },
     {
       "epoch": 7.71,
       "learning_rate": 3.0661764705882354e-05,
-      "loss": 1.1035,
+      "loss": 1.3813,
       "step": 1052
     },
     {
       "epoch": 7.73,
       "learning_rate": 3.0625000000000006e-05,
-      "loss": 1.1345,
+      "loss": 1.4707,
       "step": 1054
     },
     {
       "epoch": 7.74,
       "learning_rate": 3.058823529411765e-05,
-      "loss": 1.1102,
+      "loss": 1.4154,
       "step": 1056
     },
     {
       "epoch": 7.75,
       "learning_rate": 3.0551470588235297e-05,
-      "loss": 1.0805,
+      "loss": 1.417,
       "step": 1058
     },
     {
       "epoch": 7.77,
       "learning_rate": 3.0514705882352945e-05,
-      "loss": 1.0786,
+      "loss": 1.4043,
       "step": 1060
     },
     {
       "epoch": 7.78,
       "learning_rate": 3.0477941176470594e-05,
-      "loss": 1.1292,
+      "loss": 1.3876,
       "step": 1062
     },
     {
       "epoch": 7.8,
       "learning_rate": 3.0441176470588233e-05,
-      "loss": 1.0522,
+      "loss": 1.3565,
       "step": 1064
     },
     {
       "epoch": 7.81,
       "learning_rate": 3.040441176470588e-05,
-      "loss": 1.1068,
+      "loss": 1.405,
       "step": 1066
     },
     {
       "epoch": 7.83,
       "learning_rate": 3.036764705882353e-05,
-      "loss": 1.1305,
+      "loss": 1.4422,
       "step": 1068
     },
     {
       "epoch": 7.84,
       "learning_rate": 3.0330882352941176e-05,
-      "loss": 1.1219,
+      "loss": 1.4294,
       "step": 1070
     },
     {
       "epoch": 7.86,
       "learning_rate": 3.0294117647058824e-05,
-      "loss": 1.1672,
+      "loss": 1.4392,
       "step": 1072
     },
     {
       "epoch": 7.87,
       "learning_rate": 3.025735294117647e-05,
-      "loss": 1.062,
+      "loss": 1.3836,
       "step": 1074
     },
     {
       "epoch": 7.89,
       "learning_rate": 3.022058823529412e-05,
-      "loss": 1.0983,
+      "loss": 1.3834,
       "step": 1076
     },
     {
       "epoch": 7.9,
       "learning_rate": 3.0183823529411764e-05,
-      "loss": 1.158,
+      "loss": 1.4444,
       "step": 1078
     },
     {
       "epoch": 7.92,
       "learning_rate": 3.0147058823529413e-05,
-      "loss": 1.0447,
+      "loss": 1.374,
       "step": 1080
     },
     {
       "epoch": 7.93,
       "learning_rate": 3.0110294117647058e-05,
-      "loss": 1.1328,
+      "loss": 1.4511,
       "step": 1082
     },
     {
       "epoch": 7.95,
       "learning_rate": 3.0073529411764707e-05,
-      "loss": 1.1709,
+      "loss": 1.4613,
       "step": 1084
     },
     {
       "epoch": 7.96,
       "learning_rate": 3.0036764705882352e-05,
-      "loss": 1.0755,
+      "loss": 1.362,
       "step": 1086
     },
     {
       "epoch": 7.97,
       "learning_rate": 3e-05,
-      "loss": 1.1316,
+      "loss": 1.4456,
       "step": 1088
     },
     {
       "epoch": 7.99,
       "learning_rate": 2.9963235294117646e-05,
-      "loss": 1.1585,
+      "loss": 1.4015,
       "step": 1090
     },
     {
       "epoch": 8.0,
-      "eval_gen_len": 128.96399176954733,
-      "eval_loss": 1.477362871170044,
-      "eval_rouge1": 55.9356,
-      "eval_rouge2": 30.7012,
-      "eval_rougeL": 38.7824,
-      "eval_rougeLsum": 51.5664,
-      "eval_runtime": 219.0461,
-      "eval_samples_per_second": 4.437,
-      "eval_steps_per_second": 0.557,
+      "eval_gen_len": 731.2685185185185,
+      "eval_loss": 1.4312541484832764,
+      "eval_rouge1": 72.0031,
+      "eval_rouge2": 40.9229,
+      "eval_rougeL": 38.2581,
+      "eval_rougeLsum": 69.4154,
+      "eval_runtime": 1579.5755,
+      "eval_samples_per_second": 0.615,
+      "eval_steps_per_second": 0.077,
       "step": 1091
     },
     {
       "epoch": 8.0,
       "learning_rate": 2.9926470588235295e-05,
-      "loss": 1.064,
+      "loss": 1.3968,
       "step": 1092
     },
     {
       "epoch": 8.02,
       "learning_rate": 2.988970588235294e-05,
-      "loss": 1.0127,
+      "loss": 1.3402,
       "step": 1094
     },
     {
       "epoch": 8.03,
       "learning_rate": 2.985294117647059e-05,
-      "loss": 1.041,
+      "loss": 1.3487,
       "step": 1096
     },
     {
       "epoch": 8.05,
       "learning_rate": 2.9816176470588238e-05,
-      "loss": 1.0275,
+      "loss": 1.3428,
       "step": 1098
     },
     {
       "epoch": 8.06,
       "learning_rate": 2.9779411764705883e-05,
-      "loss": 1.0411,
+      "loss": 1.3522,
       "step": 1100
     },
     {
       "epoch": 8.08,
       "learning_rate": 2.9742647058823532e-05,
-      "loss": 0.9834,
+      "loss": 1.3956,
       "step": 1102
     },
     {
       "epoch": 8.09,
       "learning_rate": 2.9705882352941177e-05,
-      "loss": 1.062,
+      "loss": 1.3808,
       "step": 1104
     },
     {
       "epoch": 8.11,
       "learning_rate": 2.9669117647058826e-05,
-      "loss": 1.0085,
+      "loss": 1.3745,
       "step": 1106
     },
     {
       "epoch": 8.12,
       "learning_rate": 2.963235294117647e-05,
-      "loss": 1.0523,
+      "loss": 1.4278,
       "step": 1108
     },
     {
       "epoch": 8.14,
       "learning_rate": 2.959558823529412e-05,
-      "loss": 1.0417,
+      "loss": 1.3555,
       "step": 1110
     },
     {
       "epoch": 8.15,
       "learning_rate": 2.9558823529411766e-05,
-      "loss": 1.0556,
+      "loss": 1.4131,
       "step": 1112
     },
     {
       "epoch": 8.16,
       "learning_rate": 2.9522058823529414e-05,
-      "loss": 1.0554,
+      "loss": 1.3813,
       "step": 1114
     },
     {
       "epoch": 8.18,
       "learning_rate": 2.948529411764706e-05,
-      "loss": 1.0356,
+      "loss": 1.4082,
       "step": 1116
     },
     {
       "epoch": 8.19,
       "learning_rate": 2.944852941176471e-05,
-      "loss": 1.0252,
+      "loss": 1.3813,
       "step": 1118
     },
     {
       "epoch": 8.21,
       "learning_rate": 2.9411764705882354e-05,
-      "loss": 1.026,
+      "loss": 1.3513,
       "step": 1120
     },
     {
       "epoch": 8.22,
       "learning_rate": 2.9375000000000003e-05,
-      "loss": 1.0843,
+      "loss": 1.4245,
       "step": 1122
     },
     {
       "epoch": 8.24,
       "learning_rate": 2.933823529411765e-05,
-      "loss": 1.0634,
+      "loss": 1.3952,
       "step": 1124
     },
     {
       "epoch": 8.25,
       "learning_rate": 2.9301470588235297e-05,
-      "loss": 1.0619,
+      "loss": 1.4263,
       "step": 1126
     },
     {
       "epoch": 8.27,
       "learning_rate": 2.9264705882352945e-05,
-      "loss": 1.0327,
+      "loss": 1.3879,
       "step": 1128
     },
     {
       "epoch": 8.28,
       "learning_rate": 2.922794117647059e-05,
-      "loss": 1.0512,
+      "loss": 1.4151,
       "step": 1130
     },
     {
       "epoch": 8.3,
       "learning_rate": 2.919117647058824e-05,
-      "loss": 1.097,
+      "loss": 1.413,
       "step": 1132
     },
     {
       "epoch": 8.31,
       "learning_rate": 2.9154411764705885e-05,
-      "loss": 1.0198,
+      "loss": 1.3681,
       "step": 1134
     },
     {
       "epoch": 8.33,
       "learning_rate": 2.9117647058823534e-05,
-      "loss": 1.046,
+      "loss": 1.4315,
       "step": 1136
     },
     {
       "epoch": 8.34,
       "learning_rate": 2.908088235294118e-05,
-      "loss": 1.0727,
+      "loss": 1.4176,
       "step": 1138
     },
     {
       "epoch": 8.36,
       "learning_rate": 2.9044117647058828e-05,
-      "loss": 1.0551,
+      "loss": 1.3543,
       "step": 1140
     },
     {
       "epoch": 8.37,
       "learning_rate": 2.9007352941176473e-05,
-      "loss": 1.0817,
+      "loss": 1.3848,
       "step": 1142
     },
     {
       "epoch": 8.38,
       "learning_rate": 2.8970588235294122e-05,
-      "loss": 1.061,
+      "loss": 1.4014,
       "step": 1144
     },
     {
       "epoch": 8.4,
       "learning_rate": 2.8933823529411767e-05,
-      "loss": 1.0196,
+      "loss": 1.3431,
       "step": 1146
     },
     {
       "epoch": 8.41,
       "learning_rate": 2.889705882352941e-05,
-      "loss": 1.0931,
+      "loss": 1.4267,
       "step": 1148
     },
     {
       "epoch": 8.43,
       "learning_rate": 2.8860294117647058e-05,
-      "loss": 1.0846,
+      "loss": 1.3691,
       "step": 1150
     },
     {
       "epoch": 8.44,
       "learning_rate": 2.8823529411764703e-05,
-      "loss": 1.0732,
+      "loss": 1.431,
       "step": 1152
     },
     {
       "epoch": 8.46,
       "learning_rate": 2.8786764705882352e-05,
-      "loss": 1.0944,
+      "loss": 1.407,
       "step": 1154
     },
     {
       "epoch": 8.47,
       "learning_rate": 2.8749999999999997e-05,
-      "loss": 1.0691,
+      "loss": 1.3792,
       "step": 1156
     },
     {
       "epoch": 8.49,
       "learning_rate": 2.8713235294117646e-05,
-      "loss": 1.0971,
+      "loss": 1.4238,
       "step": 1158
     },
     {
       "epoch": 8.5,
       "learning_rate": 2.8676470588235295e-05,
-      "loss": 1.0711,
+      "loss": 1.3587,
       "step": 1160
     },
     {
       "epoch": 8.52,
       "learning_rate": 2.863970588235294e-05,
-      "loss": 1.1213,
+      "loss": 1.4583,
       "step": 1162
     },
     {
       "epoch": 8.53,
       "learning_rate": 2.860294117647059e-05,
-      "loss": 1.0351,
+      "loss": 1.3723,
       "step": 1164
     },
     {
       "epoch": 8.55,
       "learning_rate": 2.8566176470588234e-05,
-      "loss": 1.0728,
+      "loss": 1.3941,
       "step": 1166
     },
     {
       "epoch": 8.56,
       "learning_rate": 2.8529411764705883e-05,
-      "loss": 1.0392,
+      "loss": 1.3643,
       "step": 1168
     },
     {
       "epoch": 8.58,
       "learning_rate": 2.849264705882353e-05,
-      "loss": 1.0651,
+      "loss": 1.395,
       "step": 1170
     },
     {
       "epoch": 8.59,
       "learning_rate": 2.8455882352941177e-05,
-      "loss": 1.0739,
+      "loss": 1.4268,
       "step": 1172
     },
     {
       "epoch": 8.6,
       "learning_rate": 2.8419117647058823e-05,
-      "loss": 1.0238,
+      "loss": 1.3778,
       "step": 1174
     },
     {
       "epoch": 8.62,
       "learning_rate": 2.838235294117647e-05,
-      "loss": 1.0343,
+      "loss": 1.3982,
       "step": 1176
     },
     {
       "epoch": 8.63,
       "learning_rate": 2.8345588235294117e-05,
-      "loss": 1.085,
+      "loss": 1.4177,
       "step": 1178
     },
     {
       "epoch": 8.65,
       "learning_rate": 2.8308823529411766e-05,
-      "loss": 1.028,
+      "loss": 1.4038,
       "step": 1180
     },
     {
       "epoch": 8.66,
       "learning_rate": 2.827205882352941e-05,
-      "loss": 1.051,
+      "loss": 1.3917,
       "step": 1182
     },
     {
       "epoch": 8.68,
       "learning_rate": 2.823529411764706e-05,
-      "loss": 1.0861,
+      "loss": 1.3826,
       "step": 1184
     },
     {
       "epoch": 8.69,
       "learning_rate": 2.819852941176471e-05,
-      "loss": 1.0662,
+      "loss": 1.3897,
       "step": 1186
     },
     {
       "epoch": 8.71,
       "learning_rate": 2.8161764705882354e-05,
-      "loss": 1.0152,
+      "loss": 1.4077,
       "step": 1188
     },
     {
       "epoch": 8.72,
       "learning_rate": 2.8125000000000003e-05,
-      "loss": 1.0921,
+      "loss": 1.3959,
       "step": 1190
     },
     {
       "epoch": 8.74,
       "learning_rate": 2.8088235294117648e-05,
-      "loss": 1.055,
+      "loss": 1.4188,
       "step": 1192
     },
     {
       "epoch": 8.75,
       "learning_rate": 2.8051470588235297e-05,
-      "loss": 1.091,
+      "loss": 1.42,
       "step": 1194
     },
     {
       "epoch": 8.77,
       "learning_rate": 2.8014705882352942e-05,
-      "loss": 1.0633,
+      "loss": 1.4458,
       "step": 1196
     },
     {
       "epoch": 8.78,
       "learning_rate": 2.797794117647059e-05,
-      "loss": 1.0423,
+      "loss": 1.4319,
       "step": 1198
     },
     {
       "epoch": 8.8,
       "learning_rate": 2.7941176470588236e-05,
-      "loss": 1.0793,
+      "loss": 1.3902,
       "step": 1200
     },
     {
       "epoch": 8.81,
       "learning_rate": 2.7904411764705885e-05,
-      "loss": 1.0861,
+      "loss": 1.4131,
       "step": 1202
     },
     {
       "epoch": 8.82,
       "learning_rate": 2.786764705882353e-05,
-      "loss": 1.0503,
+      "loss": 1.3798,
       "step": 1204
     },
     {
       "epoch": 8.84,
       "learning_rate": 2.783088235294118e-05,
-      "loss": 1.1103,
+      "loss": 1.4616,
       "step": 1206
     },
     {
       "epoch": 8.85,
       "learning_rate": 2.7794117647058824e-05,
-      "loss": 1.0901,
+      "loss": 1.3843,
       "step": 1208
     },
     {
       "epoch": 8.87,
       "learning_rate": 2.7757352941176473e-05,
-      "loss": 1.0669,
+      "loss": 1.4045,
       "step": 1210
     },
     {
       "epoch": 8.88,
       "learning_rate": 2.7720588235294122e-05,
-      "loss": 1.04,
+      "loss": 1.3807,
       "step": 1212
     },
     {
       "epoch": 8.9,
       "learning_rate": 2.7683823529411767e-05,
-      "loss": 1.0737,
+      "loss": 1.4065,
       "step": 1214
     },
     {
       "epoch": 8.91,
       "learning_rate": 2.7647058823529416e-05,
-      "loss": 1.0484,
+      "loss": 1.384,
       "step": 1216
     },
     {
       "epoch": 8.93,
       "learning_rate": 2.761029411764706e-05,
-      "loss": 1.0745,
+      "loss": 1.3956,
       "step": 1218
     },
     {
       "epoch": 8.94,
       "learning_rate": 2.757352941176471e-05,
-      "loss": 1.0958,
+      "loss": 1.4194,
       "step": 1220
     },
     {
       "epoch": 8.96,
       "learning_rate": 2.7536764705882356e-05,
-      "loss": 1.0877,
+      "loss": 1.3842,
       "step": 1222
     },
     {
       "epoch": 8.97,
       "learning_rate": 2.7500000000000004e-05,
-      "loss": 1.0696,
+      "loss": 1.3934,
       "step": 1224
     },
     {
       "epoch": 8.99,
       "learning_rate": 2.746323529411765e-05,
-      "loss": 1.0297,
+      "loss": 1.391,
       "step": 1226
     },
     {
       "epoch": 8.99,
-      "eval_gen_len": 128.65534979423867,
-      "eval_loss": 1.4938671588897705,
-      "eval_rouge1": 56.7487,
-      "eval_rouge2": 31.552,
-      "eval_rougeL": 39.6461,
-      "eval_rougeLsum": 52.411,
-      "eval_runtime": 219.1861,
-      "eval_samples_per_second": 4.435,
-      "eval_steps_per_second": 0.557,
+      "eval_gen_len": 751.0288065843621,
+      "eval_loss": 1.4266291856765747,
+      "eval_rouge1": 71.9948,
+      "eval_rouge2": 41.0084,
+      "eval_rougeL": 38.0938,
+      "eval_rougeLsum": 69.4488,
+      "eval_runtime": 1628.8819,
+      "eval_samples_per_second": 0.597,
+      "eval_steps_per_second": 0.075,
       "step": 1227
     },
     {
       "epoch": 9.0,
       "learning_rate": 2.74264705882353e-05,
-      "loss": 1.0531,
+      "loss": 1.3653,
       "step": 1228
     },
     {
       "epoch": 9.02,
       "learning_rate": 2.7389705882352944e-05,
-      "loss": 1.0446,
+      "loss": 1.3824,
       "step": 1230
     },
     {
       "epoch": 9.03,
       "learning_rate": 2.7352941176470593e-05,
-      "loss": 1.0474,
+      "loss": 1.3889,
       "step": 1232
     },
     {
       "epoch": 9.04,
       "learning_rate": 2.7316176470588235e-05,
-      "loss": 0.9544,
+      "loss": 1.3367,
       "step": 1234
     },
     {
       "epoch": 9.06,
       "learning_rate": 2.727941176470588e-05,
-      "loss": 1.0093,
+      "loss": 1.3948,
       "step": 1236
     },
     {
       "epoch": 9.07,
       "learning_rate": 2.724264705882353e-05,
-      "loss": 0.9961,
+      "loss": 1.3637,
       "step": 1238
     },
     {
       "epoch": 9.09,
       "learning_rate": 2.7205882352941174e-05,
-      "loss": 1.0057,
+      "loss": 1.3765,
       "step": 1240
     },
     {
       "epoch": 9.1,
       "learning_rate": 2.7169117647058823e-05,
-      "loss": 1.0148,
+      "loss": 1.4013,
       "step": 1242
     },
     {
       "epoch": 9.12,
       "learning_rate": 2.7132352941176468e-05,
-      "loss": 0.99,
+      "loss": 1.3769,
       "step": 1244
     },
     {
       "epoch": 9.13,
       "learning_rate": 2.7095588235294117e-05,
-      "loss": 1.0108,
+      "loss": 1.3869,
       "step": 1246
     },
     {
       "epoch": 9.15,
       "learning_rate": 2.7058823529411766e-05,
-      "loss": 1.0202,
+      "loss": 1.411,
       "step": 1248
     },
     {
       "epoch": 9.16,
       "learning_rate": 2.702205882352941e-05,
-      "loss": 1.0083,
+      "loss": 1.3683,
       "step": 1250
     },
     {
       "epoch": 9.18,
       "learning_rate": 2.698529411764706e-05,
-      "loss": 1.0384,
+      "loss": 1.3908,
       "step": 1252
     },
     {
       "epoch": 9.19,
       "learning_rate": 2.6948529411764705e-05,
-      "loss": 1.002,
+      "loss": 1.3476,
       "step": 1254
     },
     {
       "epoch": 9.21,
       "learning_rate": 2.6911764705882354e-05,
-      "loss": 1.0202,
+      "loss": 1.3895,
       "step": 1256
     },
     {
       "epoch": 9.22,
       "learning_rate": 2.6875e-05,
-      "loss": 1.0125,
+      "loss": 1.3803,
       "step": 1258
     },
     {
       "epoch": 9.23,
       "learning_rate": 2.6838235294117648e-05,
-      "loss": 0.9531,
+      "loss": 1.3206,
       "step": 1260
     },
     {
       "epoch": 9.25,
       "learning_rate": 2.6801470588235293e-05,
-      "loss": 1.0564,
+      "loss": 1.4003,
       "step": 1262
     },
     {
       "epoch": 9.26,
       "learning_rate": 2.6764705882352942e-05,
-      "loss": 0.973,
+      "loss": 1.3715,
       "step": 1264
     },
     {
       "epoch": 9.28,
       "learning_rate": 2.6727941176470588e-05,
-      "loss": 0.9966,
+      "loss": 1.3392,
       "step": 1266
     },
     {
       "epoch": 9.29,
       "learning_rate": 2.6691176470588236e-05,
-      "loss": 1.0315,
+      "loss": 1.3589,
       "step": 1268
     },
     {
       "epoch": 9.31,
       "learning_rate": 2.6654411764705882e-05,
-      "loss": 1.0552,
+      "loss": 1.3877,
       "step": 1270
     },
     {
       "epoch": 9.32,
       "learning_rate": 2.661764705882353e-05,
-      "loss": 0.9671,
+      "loss": 1.3362,
       "step": 1272
     },
     {
       "epoch": 9.34,
       "learning_rate": 2.658088235294118e-05,
-      "loss": 0.9798,
+      "loss": 1.4065,
       "step": 1274
     },
     {
       "epoch": 9.35,
       "learning_rate": 2.6544117647058825e-05,
-      "loss": 0.9898,
+      "loss": 1.3609,
       "step": 1276
     },
     {
       "epoch": 9.37,
       "learning_rate": 2.6507352941176473e-05,
-      "loss": 1.0526,
+      "loss": 1.3775,
       "step": 1278
     },
     {
       "epoch": 9.38,
       "learning_rate": 2.647058823529412e-05,
-      "loss": 1.0261,
+      "loss": 1.3911,
       "step": 1280
     },
     {
       "epoch": 9.4,
       "learning_rate": 2.6433823529411767e-05,
-      "loss": 0.9646,
+      "loss": 1.3185,
       "step": 1282
     },
     {
       "epoch": 9.41,
       "learning_rate": 2.6397058823529413e-05,
-      "loss": 0.9961,
+      "loss": 1.3545,
       "step": 1284
     },
     {
       "epoch": 9.43,
       "learning_rate": 2.636029411764706e-05,
-      "loss": 1.0317,
+      "loss": 1.3931,
       "step": 1286
     },
     {
       "epoch": 9.44,
       "learning_rate": 2.6323529411764707e-05,
-      "loss": 1.0495,
+      "loss": 1.3904,
       "step": 1288
     },
     {
       "epoch": 9.45,
       "learning_rate": 2.6286764705882356e-05,
-      "loss": 1.0679,
+      "loss": 1.3639,
       "step": 1290
     },
     {
       "epoch": 9.47,
       "learning_rate": 2.625e-05,
-      "loss": 1.0365,
+      "loss": 1.3596,
       "step": 1292
     },
     {
       "epoch": 9.48,
       "learning_rate": 2.621323529411765e-05,
-      "loss": 0.9912,
+      "loss": 1.3503,
       "step": 1294
     },
     {
       "epoch": 9.5,
       "learning_rate": 2.6176470588235295e-05,
-      "loss": 1.0174,
+      "loss": 1.3422,
       "step": 1296
     },
     {
       "epoch": 9.51,
       "learning_rate": 2.6139705882352944e-05,
-      "loss": 1.0361,
+      "loss": 1.3685,
       "step": 1298
     },
     {
       "epoch": 9.53,
       "learning_rate": 2.6102941176470593e-05,
-      "loss": 0.9918,
+      "loss": 1.3588,
       "step": 1300
     },
     {
       "epoch": 9.54,
       "learning_rate": 2.6066176470588238e-05,
-      "loss": 0.9941,
+      "loss": 1.3447,
       "step": 1302
     },
     {
       "epoch": 9.56,
       "learning_rate": 2.6029411764705887e-05,
-      "loss": 1.0732,
+      "loss": 1.3905,
       "step": 1304
     },
     {
       "epoch": 9.57,
       "learning_rate": 2.5992647058823532e-05,
-      "loss": 1.0018,
+      "loss": 1.355,
       "step": 1306
     },
     {
       "epoch": 9.59,
       "learning_rate": 2.595588235294118e-05,
-      "loss": 0.9767,
+      "loss": 1.4111,
       "step": 1308
     },
     {
       "epoch": 9.6,
       "learning_rate": 2.5919117647058826e-05,
-      "loss": 0.9916,
+      "loss": 1.3899,
       "step": 1310
     },
     {
       "epoch": 9.62,
       "learning_rate": 2.5882352941176475e-05,
-      "loss": 1.0272,
+      "loss": 1.3392,
       "step": 1312
     },
     {
       "epoch": 9.63,
       "learning_rate": 2.584558823529412e-05,
-      "loss": 1.0605,
+      "loss": 1.4007,
       "step": 1314
     },
     {
       "epoch": 9.65,
       "learning_rate": 2.580882352941177e-05,
-      "loss": 0.9939,
+      "loss": 1.3787,
       "step": 1316
     },
     {
       "epoch": 9.66,
       "learning_rate": 2.577205882352941e-05,
-      "loss": 1.0565,
+      "loss": 1.3836,
       "step": 1318
     },
     {
       "epoch": 9.67,
       "learning_rate": 2.5735294117647057e-05,
-      "loss": 1.0013,
+      "loss": 1.3484,
       "step": 1320
     },
     {
       "epoch": 9.69,
       "learning_rate": 2.5698529411764705e-05,
-      "loss": 1.0204,
+      "loss": 1.3679,
       "step": 1322
     },
     {
       "epoch": 9.7,
       "learning_rate": 2.566176470588235e-05,
-      "loss": 1.0265,
+      "loss": 1.3896,
       "step": 1324
     },
     {
       "epoch": 9.72,
       "learning_rate": 2.5625e-05,
-      "loss": 1.0473,
+      "loss": 1.3651,
       "step": 1326
     },
     {
       "epoch": 9.73,
       "learning_rate": 2.5588235294117645e-05,
-      "loss": 1.0479,
+      "loss": 1.3793,
       "step": 1328
     },
     {
       "epoch": 9.75,
       "learning_rate": 2.5551470588235294e-05,
-      "loss": 1.0774,
+      "loss": 1.4282,
       "step": 1330
     },
     {
       "epoch": 9.76,
       "learning_rate": 2.551470588235294e-05,
-      "loss": 0.9653,
+      "loss": 1.3675,
       "step": 1332
     },
     {
       "epoch": 9.78,
       "learning_rate": 2.5477941176470588e-05,
-      "loss": 1.0668,
+      "loss": 1.4051,
       "step": 1334
     },
     {
       "epoch": 9.79,
       "learning_rate": 2.5441176470588236e-05,
-      "loss": 1.0469,
+      "loss": 1.3748,
       "step": 1336
     },
     {
       "epoch": 9.81,
       "learning_rate": 2.5404411764705882e-05,
-      "loss": 1.0151,
+      "loss": 1.3787,
       "step": 1338
     },
     {
       "epoch": 9.82,
       "learning_rate": 2.536764705882353e-05,
-      "loss": 0.9794,
+      "loss": 1.3327,
       "step": 1340
     },
     {
       "epoch": 9.84,
       "learning_rate": 2.5330882352941176e-05,
-      "loss": 1.0478,
+      "loss": 1.4099,
       "step": 1342
     },
     {
       "epoch": 9.85,
       "learning_rate": 2.5294117647058825e-05,
-      "loss": 1.0012,
+      "loss": 1.3376,
       "step": 1344
     },
     {
       "epoch": 9.87,
       "learning_rate": 2.525735294117647e-05,
-      "loss": 0.9858,
+      "loss": 1.3869,
       "step": 1346
     },
     {
       "epoch": 9.88,
       "learning_rate": 2.522058823529412e-05,
-      "loss": 1.0248,
+      "loss": 1.375,
       "step": 1348
     },
     {
       "epoch": 9.89,
       "learning_rate": 2.5183823529411764e-05,
-      "loss": 1.0285,
+      "loss": 1.3683,
       "step": 1350
     },
     {
       "epoch": 9.91,
       "learning_rate": 2.5147058823529413e-05,
-      "loss": 1.0664,
+      "loss": 1.4441,
       "step": 1352
     },
     {
       "epoch": 9.92,
       "learning_rate": 2.511029411764706e-05,
-      "loss": 1.0532,
+      "loss": 1.3933,
       "step": 1354
     },
     {
       "epoch": 9.94,
       "learning_rate": 2.5073529411764707e-05,
-      "loss": 1.0642,
+      "loss": 1.395,
       "step": 1356
     },
     {
       "epoch": 9.95,
       "learning_rate": 2.5036764705882352e-05,
-      "loss": 1.0381,
+      "loss": 1.3576,
       "step": 1358
     },
     {
       "epoch": 9.97,
       "learning_rate": 2.5e-05,
-      "loss": 0.9934,
+      "loss": 1.3596,
       "step": 1360
     },
     {
       "epoch": 9.98,
       "learning_rate": 2.496323529411765e-05,
-      "loss": 1.0075,
+      "loss": 1.3993,
       "step": 1362
     },
     {
       "epoch": 10.0,
       "learning_rate": 2.4926470588235295e-05,
-      "loss": 1.0085,
+      "loss": 1.3642,
       "step": 1364
     },
     {
       "epoch": 10.0,
-      "eval_gen_len": 128.62654320987653,
-      "eval_loss": 1.507472276687622,
-      "eval_rouge1": 56.3918,
-      "eval_rouge2": 31.2201,
-      "eval_rougeL": 39.4213,
-      "eval_rougeLsum": 51.9449,
-      "eval_runtime": 217.0015,
-      "eval_samples_per_second": 4.479,
-      "eval_steps_per_second": 0.562,
+      "eval_gen_len": 756.9567901234568,
+      "eval_loss": 1.4286822080612183,
+      "eval_rouge1": 71.9115,
+      "eval_rouge2": 40.8683,
+      "eval_rougeL": 38.1602,
+      "eval_rougeLsum": 69.3514,
+      "eval_runtime": 1650.4445,
+      "eval_samples_per_second": 0.589,
+      "eval_steps_per_second": 0.074,
       "step": 1364
     },
     {
       "epoch": 10.01,
       "learning_rate": 2.4889705882352944e-05,
-      "loss": 0.9832,
+      "loss": 1.3945,
       "step": 1366
     },
     {
       "epoch": 10.03,
       "learning_rate": 2.485294117647059e-05,
-      "loss": 0.9834,
+      "loss": 1.363,
       "step": 1368
     },
     {
       "epoch": 10.04,
       "learning_rate": 2.4816176470588238e-05,
-      "loss": 0.939,
+      "loss": 1.3679,
       "step": 1370
     },
     {
       "epoch": 10.06,
       "learning_rate": 2.4779411764705884e-05,
-      "loss": 0.9535,
+      "loss": 1.3388,
       "step": 1372
     },
     {
       "epoch": 10.07,
       "learning_rate": 2.4742647058823532e-05,
-      "loss": 0.941,
+      "loss": 1.3329,
       "step": 1374
     },
     {
       "epoch": 10.09,
       "learning_rate": 2.4705882352941178e-05,
-      "loss": 0.955,
+      "loss": 1.3307,
       "step": 1376
     },
     {
       "epoch": 10.1,
       "learning_rate": 2.4669117647058826e-05,
-      "loss": 0.9496,
+      "loss": 1.3527,
       "step": 1378
     },
     {
       "epoch": 10.11,
       "learning_rate": 2.4632352941176472e-05,
-      "loss": 0.9929,
+      "loss": 1.3653,
       "step": 1380
     },
     {
       "epoch": 10.13,
       "learning_rate": 2.4595588235294117e-05,
-      "loss": 1.0207,
+      "loss": 1.382,
       "step": 1382
     },
     {
       "epoch": 10.14,
       "learning_rate": 2.4558823529411766e-05,
-      "loss": 0.9971,
+      "loss": 1.3889,
       "step": 1384
     },
     {
       "epoch": 10.16,
       "learning_rate": 2.452205882352941e-05,
-      "loss": 0.9831,
+      "loss": 1.3916,
       "step": 1386
     },
     {
       "epoch": 10.17,
       "learning_rate": 2.448529411764706e-05,
-      "loss": 0.933,
+      "loss": 1.3198,
       "step": 1388
     },
     {
       "epoch": 10.19,
       "learning_rate": 2.4448529411764705e-05,
-      "loss": 0.9764,
+      "loss": 1.3155,
       "step": 1390
     },
     {
       "epoch": 10.2,
       "learning_rate": 2.4411764705882354e-05,
-      "loss": 0.9694,
+      "loss": 1.3847,
       "step": 1392
     },
     {
       "epoch": 10.22,
       "learning_rate": 2.4375e-05,
-      "loss": 0.9751,
+      "loss": 1.349,
       "step": 1394
     },
     {
       "epoch": 10.23,
       "learning_rate": 2.433823529411765e-05,
-      "loss": 0.9312,
+      "loss": 1.352,
       "step": 1396
     },
     {
       "epoch": 10.25,
       "learning_rate": 2.4301470588235294e-05,
-      "loss": 0.9863,
+      "loss": 1.3717,
       "step": 1398
     },
     {
       "epoch": 10.26,
       "learning_rate": 2.4264705882352942e-05,
-      "loss": 1.0406,
+      "loss": 1.3683,
       "step": 1400
     },
     {
       "epoch": 10.28,
       "learning_rate": 2.4227941176470588e-05,
-      "loss": 0.95,
+      "loss": 1.3213,
       "step": 1402
     },
     {
       "epoch": 10.29,
       "learning_rate": 2.4191176470588237e-05,
-      "loss": 0.9429,
+      "loss": 1.3255,
       "step": 1404
     },
     {
       "epoch": 10.31,
       "learning_rate": 2.4154411764705885e-05,
-      "loss": 0.985,
+      "loss": 1.3526,
       "step": 1406
     },
     {
       "epoch": 10.32,
       "learning_rate": 2.411764705882353e-05,
-      "loss": 0.9571,
+      "loss": 1.3597,
       "step": 1408
     },
     {
       "epoch": 10.33,
       "learning_rate": 2.408088235294118e-05,
-      "loss": 1.0004,
+      "loss": 1.3518,
       "step": 1410
     },
     {
       "epoch": 10.35,
       "learning_rate": 2.4044117647058825e-05,
-      "loss": 0.9771,
+      "loss": 1.3943,
       "step": 1412
     },
     {
       "epoch": 10.36,
       "learning_rate": 2.4007352941176474e-05,
-      "loss": 1.0061,
+      "loss": 1.373,
       "step": 1414
     },
     {
       "epoch": 10.38,
       "learning_rate": 2.397058823529412e-05,
-      "loss": 0.9816,
+      "loss": 1.3419,
       "step": 1416
     },
     {
       "epoch": 10.39,
       "learning_rate": 2.3933823529411768e-05,
-      "loss": 1.0354,
+      "loss": 1.3706,
       "step": 1418
     },
     {
       "epoch": 10.41,
       "learning_rate": 2.3897058823529413e-05,
-      "loss": 1.0185,
+      "loss": 1.3642,
       "step": 1420
     },
     {
       "epoch": 10.42,
       "learning_rate": 2.3860294117647062e-05,
-      "loss": 0.9801,
+      "loss": 1.3543,
       "step": 1422
     },
     {
       "epoch": 10.44,
       "learning_rate": 2.3823529411764707e-05,
-      "loss": 0.9537,
+      "loss": 1.3694,
       "step": 1424
     },
     {
       "epoch": 10.45,
       "learning_rate": 2.3786764705882353e-05,
-      "loss": 0.947,
+      "loss": 1.309,
       "step": 1426
     },
     {
       "epoch": 10.47,
       "learning_rate": 2.375e-05,
-      "loss": 0.9349,
+      "loss": 1.3165,
       "step": 1428
     },
     {
       "epoch": 10.48,
       "learning_rate": 2.3713235294117647e-05,
-      "loss": 0.9977,
+      "loss": 1.3724,
       "step": 1430
     },
     {
       "epoch": 10.5,
       "learning_rate": 2.3676470588235295e-05,
-      "loss": 0.9688,
+      "loss": 1.3499,
       "step": 1432
     },
     {
       "epoch": 10.51,
       "learning_rate": 2.363970588235294e-05,
-      "loss": 0.9573,
+      "loss": 1.3617,
       "step": 1434
     },
     {
       "epoch": 10.52,
       "learning_rate": 2.360294117647059e-05,
-      "loss": 0.9805,
+      "loss": 1.3622,
       "step": 1436
     },
     {
       "epoch": 10.54,
       "learning_rate": 2.3566176470588235e-05,
-      "loss": 0.9624,
+      "loss": 1.3893,
       "step": 1438
     },
     {
       "epoch": 10.55,
       "learning_rate": 2.3529411764705884e-05,
-      "loss": 1.0043,
+      "loss": 1.3935,
       "step": 1440
     },
     {
       "epoch": 10.57,
       "learning_rate": 2.349264705882353e-05,
-      "loss": 1.0019,
+      "loss": 1.3686,
       "step": 1442
     },
     {
       "epoch": 10.58,
       "learning_rate": 2.3455882352941178e-05,
-      "loss": 0.9988,
+      "loss": 1.3825,
       "step": 1444
     },
     {
       "epoch": 10.6,
       "learning_rate": 2.3419117647058823e-05,
-      "loss": 0.9651,
+      "loss": 1.3421,
       "step": 1446
     },
     {
       "epoch": 10.61,
       "learning_rate": 2.3382352941176472e-05,
-      "loss": 0.9789,
+      "loss": 1.322,
       "step": 1448
     },
     {
       "epoch": 10.63,
       "learning_rate": 2.334558823529412e-05,
-      "loss": 1.0235,
+      "loss": 1.4137,
       "step": 1450
     },
     {
       "epoch": 10.64,
       "learning_rate": 2.3308823529411766e-05,
-      "loss": 0.983,
+      "loss": 1.3694,
       "step": 1452
     },
     {
       "epoch": 10.66,
       "learning_rate": 2.3272058823529415e-05,
-      "loss": 0.9842,
+      "loss": 1.3782,
       "step": 1454
     },
     {
       "epoch": 10.67,
       "learning_rate": 2.323529411764706e-05,
-      "loss": 1.0099,
+      "loss": 1.3616,
       "step": 1456
     },
     {
       "epoch": 10.69,
       "learning_rate": 2.319852941176471e-05,
-      "loss": 0.9968,
+      "loss": 1.3411,
       "step": 1458
     },
     {
       "epoch": 10.7,
       "learning_rate": 2.3161764705882354e-05,
-      "loss": 0.9875,
+      "loss": 1.3463,
       "step": 1460
     },
     {
       "epoch": 10.72,
       "learning_rate": 2.3125000000000003e-05,
-      "loss": 0.9934,
+      "loss": 1.3614,
       "step": 1462
     },
     {
       "epoch": 10.73,
       "learning_rate": 2.308823529411765e-05,
-      "loss": 0.9903,
+      "loss": 1.3385,
       "step": 1464
     },
     {
       "epoch": 10.74,
       "learning_rate": 2.3051470588235297e-05,
-      "loss": 0.9859,
+      "loss": 1.3826,
       "step": 1466
     },
     {
       "epoch": 10.76,
       "learning_rate": 2.3014705882352943e-05,
-      "loss": 0.9909,
+      "loss": 1.3593,
       "step": 1468
     },
     {
       "epoch": 10.77,
       "learning_rate": 2.2977941176470588e-05,
-      "loss": 0.9937,
+      "loss": 1.3619,
       "step": 1470
     },
     {
       "epoch": 10.79,
       "learning_rate": 2.2941176470588237e-05,
-      "loss": 0.9585,
+      "loss": 1.34,
       "step": 1472
     },
     {
       "epoch": 10.8,
       "learning_rate": 2.2904411764705882e-05,
-      "loss": 0.9539,
+      "loss": 1.3325,
       "step": 1474
     },
     {
       "epoch": 10.82,
       "learning_rate": 2.286764705882353e-05,
-      "loss": 1.0016,
+      "loss": 1.3662,
       "step": 1476
     },
     {
       "epoch": 10.83,
       "learning_rate": 2.2830882352941176e-05,
-      "loss": 0.9746,
+      "loss": 1.3124,
       "step": 1478
     },
     {
       "epoch": 10.85,
       "learning_rate": 2.2794117647058825e-05,
-      "loss": 0.9698,
+      "loss": 1.3394,
       "step": 1480
     },
     {
       "epoch": 10.86,
       "learning_rate": 2.275735294117647e-05,
-      "loss": 1.0281,
+      "loss": 1.3887,
       "step": 1482
     },
     {
       "epoch": 10.88,
       "learning_rate": 2.272058823529412e-05,
-      "loss": 0.9775,
+      "loss": 1.3633,
       "step": 1484
     },
     {
       "epoch": 10.89,
       "learning_rate": 2.2683823529411764e-05,
-      "loss": 0.9692,
+      "loss": 1.3102,
       "step": 1486
     },
     {
       "epoch": 10.91,
       "learning_rate": 2.2647058823529413e-05,
-      "loss": 0.9502,
+      "loss": 1.3374,
       "step": 1488
     },
     {
       "epoch": 10.92,
       "learning_rate": 2.261029411764706e-05,
-      "loss": 1.0112,
+      "loss": 1.3936,
       "step": 1490
     },
     {
       "epoch": 10.94,
       "learning_rate": 2.2573529411764707e-05,
-      "loss": 0.9891,
+      "loss": 1.3102,
       "step": 1492
     },
     {
       "epoch": 10.95,
       "learning_rate": 2.2536764705882356e-05,
-      "loss": 1.0143,
+      "loss": 1.3755,
       "step": 1494
     },
     {
       "epoch": 10.96,
       "learning_rate": 2.25e-05,
-      "loss": 1.0172,
+      "loss": 1.3736,
       "step": 1496
     },
     {
       "epoch": 10.98,
       "learning_rate": 2.246323529411765e-05,
-      "loss": 0.9478,
+      "loss": 1.3589,
       "step": 1498
     },
     {
       "epoch": 10.99,
       "learning_rate": 2.2426470588235296e-05,
-      "loss": 0.9738,
+      "loss": 1.3516,
       "step": 1500
     },
     {
       "epoch": 10.99,
-      "eval_gen_len": 128.8281893004115,
-      "eval_loss": 1.5237066745758057,
-      "eval_rouge1": 56.3041,
-      "eval_rouge2": 30.9239,
-      "eval_rougeL": 39.2625,
-      "eval_rougeLsum": 51.8217,
-      "eval_runtime": 217.7346,
-      "eval_samples_per_second": 4.464,
-      "eval_steps_per_second": 0.56,
+      "eval_gen_len": 719.2798353909465,
+      "eval_loss": 1.4288647174835205,
+      "eval_rouge1": 72.3822,
+      "eval_rouge2": 41.5074,
+      "eval_rougeL": 38.8088,
+      "eval_rougeLsum": 69.8232,
+      "eval_runtime": 1557.6648,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.078,
       "step": 1500
     },
     {
       "epoch": 11.01,
       "learning_rate": 2.2389705882352944e-05,
-      "loss": 0.9912,
+      "loss": 1.3657,
       "step": 1502
     },
     {
       "epoch": 11.02,
       "learning_rate": 2.235294117647059e-05,
-      "loss": 0.9189,
+      "loss": 1.3212,
       "step": 1504
     },
     {
       "epoch": 11.04,
       "learning_rate": 2.231617647058824e-05,
-      "loss": 0.9605,
+      "loss": 1.3978,
       "step": 1506
     },
     {
       "epoch": 11.05,
       "learning_rate": 2.2279411764705884e-05,
-      "loss": 0.9208,
+      "loss": 1.3518,
       "step": 1508
     },
     {
       "epoch": 11.07,
       "learning_rate": 2.224264705882353e-05,
-      "loss": 0.9377,
+      "loss": 1.346,
       "step": 1510
     },
     {
       "epoch": 11.08,
       "learning_rate": 2.2205882352941178e-05,
-      "loss": 0.912,
+      "loss": 1.3151,
       "step": 1512
     },
     {
       "epoch": 11.1,
       "learning_rate": 2.2169117647058823e-05,
-      "loss": 0.9114,
+      "loss": 1.319,
       "step": 1514
     },
     {
       "epoch": 11.11,
       "learning_rate": 2.2132352941176472e-05,
-      "loss": 0.9191,
+      "loss": 1.3212,
       "step": 1516
     },
     {
       "epoch": 11.13,
       "learning_rate": 2.2095588235294117e-05,
-      "loss": 0.9428,
+      "loss": 1.3204,
       "step": 1518
     },
     {
       "epoch": 11.14,
       "learning_rate": 2.2058823529411766e-05,
-      "loss": 0.9373,
+      "loss": 1.327,
       "step": 1520
     },
     {
       "epoch": 11.16,
       "learning_rate": 2.202205882352941e-05,
-      "loss": 0.9407,
+      "loss": 1.3346,
       "step": 1522
     },
     {
       "epoch": 11.17,
       "learning_rate": 2.198529411764706e-05,
-      "loss": 0.9159,
+      "loss": 1.317,
       "step": 1524
     },
     {
       "epoch": 11.18,
       "learning_rate": 2.1948529411764706e-05,
-      "loss": 0.9831,
+      "loss": 1.3602,
       "step": 1526
     },
     {
       "epoch": 11.2,
       "learning_rate": 2.1911764705882354e-05,
-      "loss": 0.9569,
+      "loss": 1.3381,
       "step": 1528
     },
     {
       "epoch": 11.21,
       "learning_rate": 2.1875e-05,
-      "loss": 0.9624,
+      "loss": 1.3612,
       "step": 1530
     },
     {
       "epoch": 11.23,
       "learning_rate": 2.183823529411765e-05,
-      "loss": 0.9384,
+      "loss": 1.354,
       "step": 1532
     },
     {
       "epoch": 11.24,
       "learning_rate": 2.1801470588235294e-05,
-      "loss": 0.9646,
+      "loss": 1.3243,
       "step": 1534
     },
     {
       "epoch": 11.26,
       "learning_rate": 2.1764705882352943e-05,
-      "loss": 0.9266,
+      "loss": 1.3307,
       "step": 1536
     },
     {
       "epoch": 11.27,
       "learning_rate": 2.172794117647059e-05,
-      "loss": 0.9742,
+      "loss": 1.3548,
       "step": 1538
     },
     {
       "epoch": 11.29,
       "learning_rate": 2.1691176470588237e-05,
-      "loss": 0.9294,
+      "loss": 1.3255,
       "step": 1540
     },
     {
       "epoch": 11.3,
       "learning_rate": 2.1654411764705886e-05,
-      "loss": 0.9703,
+      "loss": 1.315,
       "step": 1542
     },
     {
       "epoch": 11.32,
       "learning_rate": 2.161764705882353e-05,
-      "loss": 0.9507,
+      "loss": 1.3557,
       "step": 1544
     },
     {
       "epoch": 11.33,
       "learning_rate": 2.158088235294118e-05,
-      "loss": 0.9248,
+      "loss": 1.3271,
       "step": 1546
     },
     {
       "epoch": 11.35,
       "learning_rate": 2.1544117647058825e-05,
-      "loss": 0.956,
+      "loss": 1.3445,
       "step": 1548
     },
     {
       "epoch": 11.36,
       "learning_rate": 2.1507352941176474e-05,
-      "loss": 0.945,
+      "loss": 1.3348,
       "step": 1550
     },
     {
       "epoch": 11.38,
       "learning_rate": 2.1470588235294116e-05,
-      "loss": 0.9612,
+      "loss": 1.3343,
       "step": 1552
     },
     {
       "epoch": 11.39,
       "learning_rate": 2.1433823529411765e-05,
-      "loss": 0.9711,
+      "loss": 1.349,
       "step": 1554
     },
     {
       "epoch": 11.4,
       "learning_rate": 2.1397058823529413e-05,
-      "loss": 0.9508,
+      "loss": 1.3468,
       "step": 1556
     },
     {
       "epoch": 11.42,
       "learning_rate": 2.136029411764706e-05,
-      "loss": 1.001,
+      "loss": 1.3431,
       "step": 1558
     },
     {
       "epoch": 11.43,
       "learning_rate": 2.1323529411764707e-05,
-      "loss": 0.9397,
+      "loss": 1.3461,
       "step": 1560
     },
     {
       "epoch": 11.45,
       "learning_rate": 2.1286764705882353e-05,
-      "loss": 0.9023,
+      "loss": 1.3521,
       "step": 1562
     },
     {
       "epoch": 11.46,
       "learning_rate": 2.125e-05,
-      "loss": 0.9255,
+      "loss": 1.3195,
       "step": 1564
     },
     {
       "epoch": 11.48,
       "learning_rate": 2.1213235294117647e-05,
-      "loss": 0.9306,
+      "loss": 1.3161,
       "step": 1566
     },
     {
       "epoch": 11.49,
       "learning_rate": 2.1176470588235296e-05,
-      "loss": 0.9276,
+      "loss": 1.3336,
       "step": 1568
     },
     {
       "epoch": 11.51,
       "learning_rate": 2.113970588235294e-05,
-      "loss": 0.927,
+      "loss": 1.3297,
       "step": 1570
     },
     {
       "epoch": 11.52,
       "learning_rate": 2.110294117647059e-05,
-      "loss": 0.9508,
+      "loss": 1.3476,
       "step": 1572
     },
     {
       "epoch": 11.54,
       "learning_rate": 2.1066176470588235e-05,
-      "loss": 0.9379,
+      "loss": 1.3284,
       "step": 1574
     },
     {
       "epoch": 11.55,
       "learning_rate": 2.1029411764705884e-05,
-      "loss": 0.9608,
+      "loss": 1.3674,
       "step": 1576
     },
     {
       "epoch": 11.57,
       "learning_rate": 2.099264705882353e-05,
-      "loss": 0.9044,
+      "loss": 1.3074,
       "step": 1578
     },
     {
       "epoch": 11.58,
       "learning_rate": 2.0955882352941178e-05,
-      "loss": 0.9432,
+      "loss": 1.3501,
       "step": 1580
     },
     {
       "epoch": 11.6,
       "learning_rate": 2.0919117647058827e-05,
-      "loss": 0.9489,
+      "loss": 1.3241,
       "step": 1582
     },
     {
       "epoch": 11.61,
       "learning_rate": 2.0882352941176472e-05,
-      "loss": 0.9585,
+      "loss": 1.3565,
       "step": 1584
     },
     {
       "epoch": 11.62,
       "learning_rate": 2.084558823529412e-05,
-      "loss": 0.9307,
+      "loss": 1.3133,
       "step": 1586
     },
     {
       "epoch": 11.64,
       "learning_rate": 2.0808823529411766e-05,
-      "loss": 0.9143,
+      "loss": 1.3516,
       "step": 1588
     },
     {
       "epoch": 11.65,
       "learning_rate": 2.0772058823529415e-05,
-      "loss": 0.9661,
+      "loss": 1.3538,
       "step": 1590
     },
     {
       "epoch": 11.67,
       "learning_rate": 2.073529411764706e-05,
-      "loss": 0.9378,
+      "loss": 1.3515,
       "step": 1592
     },
     {
       "epoch": 11.68,
       "learning_rate": 2.0698529411764706e-05,
-      "loss": 0.9438,
+      "loss": 1.3462,
       "step": 1594
     },
     {
       "epoch": 11.7,
       "learning_rate": 2.066176470588235e-05,
-      "loss": 0.9832,
+      "loss": 1.3422,
       "step": 1596
     },
     {
       "epoch": 11.71,
       "learning_rate": 2.0625e-05,
-      "loss": 0.9139,
+      "loss": 1.3305,
       "step": 1598
     },
     {
       "epoch": 11.73,
       "learning_rate": 2.058823529411765e-05,
-      "loss": 0.973,
+      "loss": 1.3876,
       "step": 1600
     },
     {
       "epoch": 11.74,
       "learning_rate": 2.0551470588235294e-05,
-      "loss": 0.9838,
+      "loss": 1.3291,
       "step": 1602
     },
     {
       "epoch": 11.76,
       "learning_rate": 2.0514705882352943e-05,
-      "loss": 1.0008,
+      "loss": 1.3781,
       "step": 1604
     },
     {
       "epoch": 11.77,
       "learning_rate": 2.0477941176470588e-05,
-      "loss": 0.941,
+      "loss": 1.3234,
       "step": 1606
     },
     {
       "epoch": 11.79,
       "learning_rate": 2.0441176470588237e-05,
-      "loss": 0.9632,
+      "loss": 1.3116,
       "step": 1608
     },
     {
       "epoch": 11.8,
       "learning_rate": 2.0404411764705882e-05,
-      "loss": 0.9263,
+      "loss": 1.3286,
       "step": 1610
     },
     {
       "epoch": 11.81,
       "learning_rate": 2.036764705882353e-05,
-      "loss": 0.9247,
+      "loss": 1.3293,
       "step": 1612
     },
     {
       "epoch": 11.83,
       "learning_rate": 2.0330882352941176e-05,
-      "loss": 0.9677,
+      "loss": 1.3454,
       "step": 1614
     },
     {
       "epoch": 11.84,
       "learning_rate": 2.0294117647058825e-05,
-      "loss": 0.9709,
+      "loss": 1.3543,
       "step": 1616
     },
     {
       "epoch": 11.86,
       "learning_rate": 2.025735294117647e-05,
-      "loss": 0.9259,
+      "loss": 1.3454,
       "step": 1618
     },
     {
       "epoch": 11.87,
       "learning_rate": 2.022058823529412e-05,
-      "loss": 0.9373,
+      "loss": 1.3094,
       "step": 1620
     },
     {
       "epoch": 11.89,
       "learning_rate": 2.0183823529411765e-05,
-      "loss": 0.9283,
+      "loss": 1.3677,
       "step": 1622
     },
     {
       "epoch": 11.9,
       "learning_rate": 2.0147058823529413e-05,
-      "loss": 0.9833,
+      "loss": 1.3973,
       "step": 1624
     },
     {
       "epoch": 11.92,
       "learning_rate": 2.0110294117647062e-05,
-      "loss": 0.9238,
+      "loss": 1.3076,
       "step": 1626
     },
     {
       "epoch": 11.93,
       "learning_rate": 2.0073529411764708e-05,
-      "loss": 0.9651,
+      "loss": 1.3677,
       "step": 1628
     },
     {
       "epoch": 11.95,
       "learning_rate": 2.0036764705882356e-05,
-      "loss": 0.9671,
+      "loss": 1.3758,
       "step": 1630
     },
     {
       "epoch": 11.96,
       "learning_rate": 2e-05,
-      "loss": 1.0051,
+      "loss": 1.3601,
       "step": 1632
     },
     {
       "epoch": 11.98,
       "learning_rate": 1.996323529411765e-05,
-      "loss": 0.9691,
+      "loss": 1.3611,
       "step": 1634
     },
     {
       "epoch": 11.99,
       "learning_rate": 1.9926470588235296e-05,
-      "loss": 0.9583,
+      "loss": 1.3243,
       "step": 1636
     },
     {
       "epoch": 12.0,
-      "eval_gen_len": 128.91358024691357,
-      "eval_loss": 1.544446587562561,
-      "eval_rouge1": 55.6539,
-      "eval_rouge2": 30.2395,
-      "eval_rougeL": 38.5901,
-      "eval_rougeLsum": 51.1518,
-      "eval_runtime": 218.8835,
-      "eval_samples_per_second": 4.441,
-      "eval_steps_per_second": 0.557,
+      "eval_gen_len": 749.9475308641976,
+      "eval_loss": 1.4301081895828247,
+      "eval_rouge1": 71.83,
+      "eval_rouge2": 40.764,
+      "eval_rougeL": 38.1124,
+      "eval_rougeLsum": 69.2767,
+      "eval_runtime": 1620.9293,
+      "eval_samples_per_second": 0.6,
+      "eval_steps_per_second": 0.075,
       "step": 1637
     },
     {
       "epoch": 12.01,
       "learning_rate": 1.988970588235294e-05,
-      "loss": 0.978,
+      "loss": 1.357,
       "step": 1638
     },
     {
       "epoch": 12.02,
       "learning_rate": 1.9852941176470586e-05,
-      "loss": 0.9212,
+      "loss": 1.3135,
       "step": 1640
     },
     {
       "epoch": 12.03,
       "learning_rate": 1.9816176470588235e-05,
-      "loss": 0.9258,
+      "loss": 1.3555,
       "step": 1642
     },
     {
       "epoch": 12.05,
       "learning_rate": 1.9779411764705884e-05,
-      "loss": 0.8803,
+      "loss": 1.2697,
       "step": 1644
     },
     {
       "epoch": 12.06,
       "learning_rate": 1.974264705882353e-05,
-      "loss": 0.9238,
+      "loss": 1.3532,
       "step": 1646
     },
     {
       "epoch": 12.08,
       "learning_rate": 1.9705882352941178e-05,
-      "loss": 0.9472,
+      "loss": 1.3258,
       "step": 1648
     },
     {
       "epoch": 12.09,
       "learning_rate": 1.9669117647058824e-05,
-      "loss": 0.9415,
+      "loss": 1.325,
       "step": 1650
     },
     {
       "epoch": 12.11,
       "learning_rate": 1.9632352941176472e-05,
-      "loss": 0.9185,
+      "loss": 1.2975,
       "step": 1652
     },
     {
       "epoch": 12.12,
       "learning_rate": 1.9595588235294118e-05,
-      "loss": 0.9,
+      "loss": 1.2828,
       "step": 1654
     },
     {
       "epoch": 12.14,
       "learning_rate": 1.9558823529411766e-05,
-      "loss": 0.8914,
+      "loss": 1.312,
       "step": 1656
     },
     {
       "epoch": 12.15,
       "learning_rate": 1.9522058823529412e-05,
-      "loss": 0.8579,
+      "loss": 1.3121,
       "step": 1658
     },
     {
       "epoch": 12.17,
       "learning_rate": 1.948529411764706e-05,
-      "loss": 0.8696,
+      "loss": 1.3396,
       "step": 1660
     },
     {
       "epoch": 12.18,
       "learning_rate": 1.9448529411764706e-05,
-      "loss": 0.8479,
+      "loss": 1.2769,
       "step": 1662
     },
     {
       "epoch": 12.2,
       "learning_rate": 1.9411764705882355e-05,
-      "loss": 0.9053,
+      "loss": 1.3263,
       "step": 1664
     },
     {
       "epoch": 12.21,
       "learning_rate": 1.9375e-05,
-      "loss": 0.9217,
+      "loss": 1.326,
       "step": 1666
     },
     {
       "epoch": 12.23,
       "learning_rate": 1.933823529411765e-05,
-      "loss": 0.8974,
+      "loss": 1.3461,
       "step": 1668
     },
     {
       "epoch": 12.24,
       "learning_rate": 1.9301470588235298e-05,
-      "loss": 0.922,
+      "loss": 1.3487,
       "step": 1670
     },
     {
       "epoch": 12.25,
       "learning_rate": 1.9264705882352943e-05,
-      "loss": 0.9335,
+      "loss": 1.3022,
       "step": 1672
     },
     {
       "epoch": 12.27,
       "learning_rate": 1.922794117647059e-05,
-      "loss": 0.9173,
+      "loss": 1.3561,
       "step": 1674
     },
     {
       "epoch": 12.28,
       "learning_rate": 1.9191176470588237e-05,
-      "loss": 0.951,
+      "loss": 1.3453,
       "step": 1676
     },
     {
       "epoch": 12.3,
       "learning_rate": 1.9154411764705886e-05,
-      "loss": 0.9103,
+      "loss": 1.3329,
       "step": 1678
     },
     {
       "epoch": 12.31,
       "learning_rate": 1.9117647058823528e-05,
-      "loss": 0.882,
+      "loss": 1.2951,
       "step": 1680
     },
     {
       "epoch": 12.33,
       "learning_rate": 1.9080882352941176e-05,
-      "loss": 0.9081,
+      "loss": 1.3186,
       "step": 1682
     },
     {
       "epoch": 12.34,
       "learning_rate": 1.9044117647058822e-05,
-      "loss": 0.9444,
+      "loss": 1.3403,
       "step": 1684
     },
     {
       "epoch": 12.36,
       "learning_rate": 1.900735294117647e-05,
-      "loss": 0.8887,
+      "loss": 1.2859,
       "step": 1686
     },
     {
       "epoch": 12.37,
       "learning_rate": 1.897058823529412e-05,
-      "loss": 0.912,
+      "loss": 1.3411,
       "step": 1688
     },
     {
       "epoch": 12.39,
       "learning_rate": 1.8933823529411765e-05,
-      "loss": 0.9049,
+      "loss": 1.3516,
       "step": 1690
     },
     {
       "epoch": 12.4,
       "learning_rate": 1.8897058823529413e-05,
-      "loss": 0.9131,
+      "loss": 1.3255,
       "step": 1692
     },
     {
       "epoch": 12.42,
       "learning_rate": 1.886029411764706e-05,
-      "loss": 0.8734,
+      "loss": 1.3153,
       "step": 1694
     },
     {
       "epoch": 12.43,
       "learning_rate": 1.8823529411764708e-05,
-      "loss": 0.8905,
+      "loss": 1.2845,
       "step": 1696
     },
     {
       "epoch": 12.45,
       "learning_rate": 1.8786764705882353e-05,
-      "loss": 0.9443,
+      "loss": 1.3377,
       "step": 1698
     },
     {
       "epoch": 12.46,
       "learning_rate": 1.8750000000000002e-05,
-      "loss": 0.9412,
+      "loss": 1.373,
       "step": 1700
     },
     {
       "epoch": 12.47,
       "learning_rate": 1.8713235294117647e-05,
-      "loss": 0.8967,
+      "loss": 1.2894,
       "step": 1702
     },
     {
       "epoch": 12.49,
       "learning_rate": 1.8676470588235296e-05,
-      "loss": 0.9488,
+      "loss": 1.3133,
       "step": 1704
     },
     {
       "epoch": 12.5,
       "learning_rate": 1.863970588235294e-05,
-      "loss": 0.8957,
+      "loss": 1.3286,
       "step": 1706
     },
     {
       "epoch": 12.52,
       "learning_rate": 1.860294117647059e-05,
-      "loss": 0.9519,
+      "loss": 1.3867,
       "step": 1708
     },
     {
       "epoch": 12.53,
       "learning_rate": 1.8566176470588235e-05,
-      "loss": 0.8753,
+      "loss": 1.2944,
       "step": 1710
     },
     {
       "epoch": 12.55,
       "learning_rate": 1.8529411764705884e-05,
-      "loss": 0.9227,
+      "loss": 1.3457,
       "step": 1712
     },
     {
       "epoch": 12.56,
       "learning_rate": 1.8492647058823533e-05,
-      "loss": 0.9583,
+      "loss": 1.3665,
       "step": 1714
     },
     {
       "epoch": 12.58,
       "learning_rate": 1.8455882352941178e-05,
-      "loss": 0.9345,
+      "loss": 1.3161,
       "step": 1716
     },
     {
       "epoch": 12.59,
       "learning_rate": 1.8419117647058827e-05,
-      "loss": 0.9291,
+      "loss": 1.3251,
       "step": 1718
     },
     {
       "epoch": 12.61,
       "learning_rate": 1.8382352941176472e-05,
-      "loss": 0.9343,
+      "loss": 1.36,
       "step": 1720
     },
     {
       "epoch": 12.62,
       "learning_rate": 1.8345588235294118e-05,
-      "loss": 0.9206,
+      "loss": 1.3419,
       "step": 1722
     },
     {
       "epoch": 12.64,
       "learning_rate": 1.8308823529411763e-05,
-      "loss": 0.9619,
+      "loss": 1.3806,
       "step": 1724
     },
     {
       "epoch": 12.65,
       "learning_rate": 1.8272058823529412e-05,
-      "loss": 0.9643,
+      "loss": 1.3691,
       "step": 1726
     },
     {
       "epoch": 12.67,
       "learning_rate": 1.8235294117647057e-05,
-      "loss": 0.8768,
+      "loss": 1.3219,
       "step": 1728
     },
     {
       "epoch": 12.68,
       "learning_rate": 1.8198529411764706e-05,
-      "loss": 0.8937,
+      "loss": 1.3191,
       "step": 1730
     },
     {
       "epoch": 12.69,
       "learning_rate": 1.8161764705882355e-05,
-      "loss": 0.9184,
+      "loss": 1.2994,
       "step": 1732
     },
     {
       "epoch": 12.71,
       "learning_rate": 1.8125e-05,
-      "loss": 0.9213,
+      "loss": 1.3314,
       "step": 1734
     },
     {
       "epoch": 12.72,
       "learning_rate": 1.808823529411765e-05,
-      "loss": 0.9434,
+      "loss": 1.2967,
       "step": 1736
     },
     {
       "epoch": 12.74,
       "learning_rate": 1.8051470588235294e-05,
-      "loss": 0.952,
+      "loss": 1.3496,
       "step": 1738
     },
     {
       "epoch": 12.75,
       "learning_rate": 1.8014705882352943e-05,
-      "loss": 0.9266,
+      "loss": 1.3272,
       "step": 1740
     },
     {
       "epoch": 12.77,
       "learning_rate": 1.797794117647059e-05,
-      "loss": 0.8499,
+      "loss": 1.2768,
       "step": 1742
     },
     {
       "epoch": 12.78,
       "learning_rate": 1.7941176470588237e-05,
-      "loss": 0.9336,
+      "loss": 1.3192,
       "step": 1744
     },
     {
       "epoch": 12.8,
       "learning_rate": 1.7904411764705882e-05,
-      "loss": 0.9199,
+      "loss": 1.344,
       "step": 1746
     },
     {
       "epoch": 12.81,
       "learning_rate": 1.786764705882353e-05,
-      "loss": 0.9746,
+      "loss": 1.3883,
       "step": 1748
     },
     {
       "epoch": 12.83,
       "learning_rate": 1.7830882352941177e-05,
-      "loss": 0.9535,
+      "loss": 1.3605,
       "step": 1750
     },
     {
       "epoch": 12.84,
       "learning_rate": 1.7794117647058825e-05,
-      "loss": 0.9126,
+      "loss": 1.3313,
       "step": 1752
     },
     {
       "epoch": 12.86,
       "learning_rate": 1.775735294117647e-05,
-      "loss": 0.9181,
+      "loss": 1.2923,
       "step": 1754
     },
     {
       "epoch": 12.87,
       "learning_rate": 1.772058823529412e-05,
-      "loss": 0.8901,
+      "loss": 1.2899,
       "step": 1756
     },
     {
       "epoch": 12.89,
       "learning_rate": 1.7683823529411768e-05,
-      "loss": 0.887,
+      "loss": 1.3147,
       "step": 1758
     },
     {
       "epoch": 12.9,
       "learning_rate": 1.7647058823529414e-05,
-      "loss": 0.9095,
+      "loss": 1.3371,
       "step": 1760
     },
     {
       "epoch": 12.91,
       "learning_rate": 1.7610294117647062e-05,
-      "loss": 0.8842,
+      "loss": 1.3207,
       "step": 1762
     },
     {
       "epoch": 12.93,
       "learning_rate": 1.7573529411764704e-05,
-      "loss": 0.9332,
+      "loss": 1.3458,
       "step": 1764
     },
     {
       "epoch": 12.94,
       "learning_rate": 1.7536764705882353e-05,
-      "loss": 0.93,
+      "loss": 1.3439,
       "step": 1766
     },
     {
       "epoch": 12.96,
       "learning_rate": 1.75e-05,
-      "loss": 0.9431,
+      "loss": 1.3251,
       "step": 1768
     },
     {
       "epoch": 12.97,
       "learning_rate": 1.7463235294117647e-05,
-      "loss": 0.8755,
+      "loss": 1.3033,
       "step": 1770
     },
     {
       "epoch": 12.99,
       "learning_rate": 1.7426470588235293e-05,
-      "loss": 0.9601,
+      "loss": 1.3582,
       "step": 1772
     },
     {
       "epoch": 12.99,
-      "eval_gen_len": 128.9783950617284,
-      "eval_loss": 1.5515539646148682,
-      "eval_rouge1": 55.9154,
-      "eval_rouge2": 30.5471,
-      "eval_rougeL": 38.8607,
-      "eval_rougeLsum": 51.2856,
-      "eval_runtime": 219.5635,
-      "eval_samples_per_second": 4.427,
-      "eval_steps_per_second": 0.556,
+      "eval_gen_len": 736.6697530864197,
+      "eval_loss": 1.4282923936843872,
+      "eval_rouge1": 71.9495,
+      "eval_rouge2": 40.9556,
+      "eval_rougeL": 38.4201,
+      "eval_rougeLsum": 69.4394,
+      "eval_runtime": 1583.479,
+      "eval_samples_per_second": 0.614,
+      "eval_steps_per_second": 0.077,
       "step": 1773
     },
     {
       "epoch": 13.0,
       "learning_rate": 1.738970588235294e-05,
-      "loss": 0.9193,
+      "loss": 1.2927,
       "step": 1774
     },
     {
       "epoch": 13.02,
       "learning_rate": 1.735294117647059e-05,
-      "loss": 0.8611,
+      "loss": 1.3201,
       "step": 1776
     },
     {
       "epoch": 13.03,
       "learning_rate": 1.7316176470588235e-05,
-      "loss": 0.9041,
+      "loss": 1.3495,
       "step": 1778
     },
     {
       "epoch": 13.05,
       "learning_rate": 1.7279411764705884e-05,
-      "loss": 0.8886,
+      "loss": 1.3304,
       "step": 1780
     },
     {
       "epoch": 13.06,
       "learning_rate": 1.724264705882353e-05,
-      "loss": 0.8797,
+      "loss": 1.313,
       "step": 1782
     },
     {
       "epoch": 13.08,
       "learning_rate": 1.720588235294118e-05,
-      "loss": 0.9299,
+      "loss": 1.3057,
       "step": 1784
     },
     {
       "epoch": 13.09,
       "learning_rate": 1.7169117647058824e-05,
-      "loss": 0.8481,
+      "loss": 1.2878,
       "step": 1786
     },
     {
       "epoch": 13.1,
       "learning_rate": 1.7132352941176472e-05,
-      "loss": 0.9105,
+      "loss": 1.3116,
       "step": 1788
     },
     {
       "epoch": 13.12,
       "learning_rate": 1.7095588235294118e-05,
-      "loss": 0.8449,
+      "loss": 1.3044,
       "step": 1790
     },
     {
       "epoch": 13.13,
       "learning_rate": 1.7058823529411767e-05,
-      "loss": 0.9001,
+      "loss": 1.3104,
       "step": 1792
     },
     {
       "epoch": 13.15,
       "learning_rate": 1.7022058823529412e-05,
-      "loss": 0.8804,
+      "loss": 1.3194,
       "step": 1794
     },
     {
       "epoch": 13.16,
       "learning_rate": 1.698529411764706e-05,
-      "loss": 0.8577,
+      "loss": 1.2856,
       "step": 1796
     },
     {
       "epoch": 13.18,
       "learning_rate": 1.6948529411764706e-05,
-      "loss": 0.8812,
+      "loss": 1.3316,
       "step": 1798
     },
     {
       "epoch": 13.19,
       "learning_rate": 1.6911764705882355e-05,
-      "loss": 0.9037,
+      "loss": 1.2928,
       "step": 1800
     },
     {
       "epoch": 13.21,
       "learning_rate": 1.6875000000000004e-05,
-      "loss": 0.8882,
+      "loss": 1.3015,
       "step": 1802
     },
     {
       "epoch": 13.22,
       "learning_rate": 1.683823529411765e-05,
-      "loss": 0.883,
+      "loss": 1.2799,
       "step": 1804
     },
     {
       "epoch": 13.24,
       "learning_rate": 1.6801470588235298e-05,
-      "loss": 0.8434,
+      "loss": 1.2931,
       "step": 1806
     },
     {
       "epoch": 13.25,
       "learning_rate": 1.676470588235294e-05,
-      "loss": 0.9052,
+      "loss": 1.3208,
       "step": 1808
     },
     {
       "epoch": 13.27,
       "learning_rate": 1.672794117647059e-05,
-      "loss": 0.875,
+      "loss": 1.2931,
       "step": 1810
     },
     {
       "epoch": 13.28,
       "learning_rate": 1.6691176470588234e-05,
-      "loss": 0.8739,
+      "loss": 1.3227,
       "step": 1812
     },
     {
       "epoch": 13.3,
       "learning_rate": 1.6654411764705883e-05,
-      "loss": 0.8779,
+      "loss": 1.2942,
       "step": 1814
     },
     {
       "epoch": 13.31,
       "learning_rate": 1.6617647058823528e-05,
-      "loss": 0.8743,
+      "loss": 1.2788,
       "step": 1816
     },
     {
       "epoch": 13.32,
       "learning_rate": 1.6580882352941177e-05,
-      "loss": 0.8741,
+      "loss": 1.3047,
       "step": 1818
     },
     {
       "epoch": 13.34,
       "learning_rate": 1.6544117647058825e-05,
-      "loss": 0.8905,
+      "loss": 1.3175,
       "step": 1820
     },
     {
       "epoch": 13.35,
       "learning_rate": 1.650735294117647e-05,
-      "loss": 0.8617,
+      "loss": 1.2905,
       "step": 1822
     },
     {
       "epoch": 13.37,
       "learning_rate": 1.647058823529412e-05,
-      "loss": 0.8905,
+      "loss": 1.3043,
       "step": 1824
     },
     {
       "epoch": 13.38,
       "learning_rate": 1.6433823529411765e-05,
-      "loss": 0.8572,
+      "loss": 1.2759,
       "step": 1826
     },
     {
       "epoch": 13.4,
       "learning_rate": 1.6397058823529414e-05,
-      "loss": 0.8647,
+      "loss": 1.3425,
       "step": 1828
     },
     {
       "epoch": 13.41,
       "learning_rate": 1.636029411764706e-05,
-      "loss": 0.9023,
+      "loss": 1.3027,
       "step": 1830
     },
     {
       "epoch": 13.43,
       "learning_rate": 1.6323529411764708e-05,
-      "loss": 0.905,
+      "loss": 1.3053,
       "step": 1832
     },
     {
       "epoch": 13.44,
       "learning_rate": 1.6286764705882353e-05,
-      "loss": 0.9219,
+      "loss": 1.3119,
       "step": 1834
     },
     {
       "epoch": 13.46,
       "learning_rate": 1.6250000000000002e-05,
-      "loss": 0.9263,
+      "loss": 1.3279,
       "step": 1836
     },
     {
       "epoch": 13.47,
       "learning_rate": 1.6213235294117647e-05,
-      "loss": 0.8915,
+      "loss": 1.3614,
       "step": 1838
     },
     {
       "epoch": 13.49,
       "learning_rate": 1.6176470588235296e-05,
-      "loss": 0.9464,
+      "loss": 1.3282,
       "step": 1840
     },
     {
       "epoch": 13.5,
       "learning_rate": 1.613970588235294e-05,
-      "loss": 0.8873,
+      "loss": 1.3251,
       "step": 1842
     },
     {
       "epoch": 13.52,
       "learning_rate": 1.610294117647059e-05,
-      "loss": 0.8944,
+      "loss": 1.326,
       "step": 1844
     },
     {
       "epoch": 13.53,
       "learning_rate": 1.6066176470588236e-05,
-      "loss": 0.8724,
+      "loss": 1.3205,
       "step": 1846
     },
     {
       "epoch": 13.54,
       "learning_rate": 1.6029411764705884e-05,
-      "loss": 0.9189,
+      "loss": 1.3226,
       "step": 1848
     },
     {
       "epoch": 13.56,
       "learning_rate": 1.599264705882353e-05,
-      "loss": 0.8386,
+      "loss": 1.3032,
       "step": 1850
     },
     {
       "epoch": 13.57,
       "learning_rate": 1.5955882352941175e-05,
-      "loss": 0.868,
+      "loss": 1.3281,
       "step": 1852
     },
     {
       "epoch": 13.59,
       "learning_rate": 1.5919117647058824e-05,
-      "loss": 0.8934,
+      "loss": 1.3202,
       "step": 1854
     },
     {
       "epoch": 13.6,
       "learning_rate": 1.588235294117647e-05,
-      "loss": 0.8914,
+      "loss": 1.3221,
       "step": 1856
     },
     {
       "epoch": 13.62,
       "learning_rate": 1.5845588235294118e-05,
-      "loss": 0.948,
+      "loss": 1.3169,
       "step": 1858
     },
     {
       "epoch": 13.63,
       "learning_rate": 1.5808823529411763e-05,
-      "loss": 0.9076,
+      "loss": 1.3271,
       "step": 1860
     },
     {
       "epoch": 13.65,
       "learning_rate": 1.5772058823529412e-05,
-      "loss": 0.8992,
+      "loss": 1.316,
       "step": 1862
     },
     {
       "epoch": 13.66,
       "learning_rate": 1.573529411764706e-05,
-      "loss": 0.8784,
+      "loss": 1.3288,
       "step": 1864
     },
     {
       "epoch": 13.68,
       "learning_rate": 1.5698529411764706e-05,
-      "loss": 0.9055,
+      "loss": 1.3265,
       "step": 1866
     },
     {
       "epoch": 13.69,
       "learning_rate": 1.5661764705882355e-05,
-      "loss": 0.927,
+      "loss": 1.288,
       "step": 1868
     },
     {
       "epoch": 13.71,
       "learning_rate": 1.5625e-05,
-      "loss": 0.8678,
+      "loss": 1.3173,
       "step": 1870
     },
     {
       "epoch": 13.72,
       "learning_rate": 1.558823529411765e-05,
-      "loss": 0.8993,
+      "loss": 1.2725,
       "step": 1872
     },
     {
       "epoch": 13.74,
       "learning_rate": 1.5551470588235294e-05,
-      "loss": 0.8649,
+      "loss": 1.3313,
       "step": 1874
     },
     {
       "epoch": 13.75,
       "learning_rate": 1.5514705882352943e-05,
-      "loss": 0.9001,
+      "loss": 1.3461,
       "step": 1876
     },
     {
       "epoch": 13.76,
       "learning_rate": 1.547794117647059e-05,
-      "loss": 0.8923,
+      "loss": 1.3167,
       "step": 1878
     },
     {
       "epoch": 13.78,
       "learning_rate": 1.5441176470588237e-05,
-      "loss": 0.9079,
+      "loss": 1.31,
       "step": 1880
     },
     {
       "epoch": 13.79,
       "learning_rate": 1.5404411764705883e-05,
-      "loss": 0.8799,
+      "loss": 1.3019,
       "step": 1882
     },
     {
       "epoch": 13.81,
       "learning_rate": 1.536764705882353e-05,
-      "loss": 0.8619,
+      "loss": 1.2956,
       "step": 1884
     },
     {
       "epoch": 13.82,
       "learning_rate": 1.5330882352941177e-05,
-      "loss": 0.9564,
+      "loss": 1.3674,
       "step": 1886
     },
     {
       "epoch": 13.84,
       "learning_rate": 1.5294117647058826e-05,
-      "loss": 0.8944,
+      "loss": 1.3059,
       "step": 1888
     },
     {
       "epoch": 13.85,
       "learning_rate": 1.5257352941176473e-05,
-      "loss": 0.8868,
+      "loss": 1.3666,
       "step": 1890
     },
     {
       "epoch": 13.87,
       "learning_rate": 1.5220588235294116e-05,
-      "loss": 0.8411,
+      "loss": 1.3184,
       "step": 1892
     },
     {
       "epoch": 13.88,
       "learning_rate": 1.5183823529411765e-05,
-      "loss": 0.9226,
+      "loss": 1.3217,
       "step": 1894
     },
     {
       "epoch": 13.9,
       "learning_rate": 1.5147058823529412e-05,
-      "loss": 0.9408,
+      "loss": 1.3442,
       "step": 1896
     },
     {
       "epoch": 13.91,
       "learning_rate": 1.511029411764706e-05,
-      "loss": 0.8672,
+      "loss": 1.3049,
       "step": 1898
     },
     {
       "epoch": 13.93,
       "learning_rate": 1.5073529411764706e-05,
-      "loss": 0.9427,
+      "loss": 1.3225,
       "step": 1900
     },
     {
       "epoch": 13.94,
       "learning_rate": 1.5036764705882353e-05,
-      "loss": 0.9099,
+      "loss": 1.2776,
       "step": 1902
     },
     {
       "epoch": 13.96,
       "learning_rate": 1.5e-05,
-      "loss": 0.9027,
+      "loss": 1.3137,
       "step": 1904
     },
     {
       "epoch": 13.97,
       "learning_rate": 1.4963235294117647e-05,
-      "loss": 0.8715,
+      "loss": 1.3005,
       "step": 1906
     },
     {
       "epoch": 13.98,
       "learning_rate": 1.4926470588235295e-05,
-      "loss": 0.9116,
+      "loss": 1.3646,
       "step": 1908
     },
     {
       "epoch": 14.0,
       "learning_rate": 1.4889705882352942e-05,
-      "loss": 0.8882,
+      "loss": 1.3149,
       "step": 1910
     },
     {
       "epoch": 14.0,
-      "eval_gen_len": 128.52057613168725,
-      "eval_loss": 1.5736336708068848,
-      "eval_rouge1": 56.3282,
-      "eval_rouge2": 30.9807,
-      "eval_rougeL": 39.2351,
-      "eval_rougeLsum": 51.8022,
-      "eval_runtime": 218.5676,
-      "eval_samples_per_second": 4.447,
-      "eval_steps_per_second": 0.558,
+      "eval_gen_len": 753.3230452674898,
+      "eval_loss": 1.4298332929611206,
+      "eval_rouge1": 71.9599,
+      "eval_rouge2": 40.8875,
+      "eval_rougeL": 38.2722,
+      "eval_rougeLsum": 69.4209,
+      "eval_runtime": 1612.8705,
+      "eval_samples_per_second": 0.603,
+      "eval_steps_per_second": 0.076,
       "step": 1910
     },
     {
       "epoch": 14.01,
       "learning_rate": 1.4852941176470589e-05,
-      "loss": 0.8246,
+      "loss": 1.2861,
       "step": 1912
     },
     {
       "epoch": 14.03,
       "learning_rate": 1.4816176470588236e-05,
-      "loss": 0.9006,
+      "loss": 1.3035,
       "step": 1914
     },
     {
       "epoch": 14.04,
       "learning_rate": 1.4779411764705883e-05,
-      "loss": 0.8617,
+      "loss": 1.3005,
       "step": 1916
     },
     {
       "epoch": 14.06,
       "learning_rate": 1.474264705882353e-05,
-      "loss": 0.9181,
+      "loss": 1.3289,
       "step": 1918
     },
     {
       "epoch": 14.07,
       "learning_rate": 1.4705882352941177e-05,
-      "loss": 0.848,
+      "loss": 1.2952,
       "step": 1920
     },
     {
       "epoch": 14.09,
       "learning_rate": 1.4669117647058826e-05,
-      "loss": 0.8767,
+      "loss": 1.3101,
       "step": 1922
     },
     {
       "epoch": 14.1,
       "learning_rate": 1.4632352941176473e-05,
-      "loss": 0.8644,
+      "loss": 1.29,
       "step": 1924
     },
     {
       "epoch": 14.12,
       "learning_rate": 1.459558823529412e-05,
-      "loss": 0.8462,
+      "loss": 1.3145,
       "step": 1926
     },
     {
       "epoch": 14.13,
       "learning_rate": 1.4558823529411767e-05,
-      "loss": 0.8573,
+      "loss": 1.294,
       "step": 1928
     },
     {
       "epoch": 14.15,
       "learning_rate": 1.4522058823529414e-05,
-      "loss": 0.8592,
+      "loss": 1.3079,
       "step": 1930
     },
     {
       "epoch": 14.16,
       "learning_rate": 1.4485294117647061e-05,
-      "loss": 0.8611,
+      "loss": 1.2614,
       "step": 1932
     },
     {
       "epoch": 14.17,
       "learning_rate": 1.4448529411764705e-05,
-      "loss": 0.8848,
+      "loss": 1.3538,
       "step": 1934
     },
     {
       "epoch": 14.19,
       "learning_rate": 1.4411764705882352e-05,
-      "loss": 0.8964,
+      "loss": 1.3405,
       "step": 1936
     },
     {
       "epoch": 14.2,
       "learning_rate": 1.4374999999999999e-05,
-      "loss": 0.8316,
+      "loss": 1.309,
       "step": 1938
     },
     {
       "epoch": 14.22,
       "learning_rate": 1.4338235294117647e-05,
-      "loss": 0.8804,
+      "loss": 1.3272,
       "step": 1940
     },
     {
       "epoch": 14.23,
       "learning_rate": 1.4301470588235295e-05,
-      "loss": 0.8706,
+      "loss": 1.3025,
       "step": 1942
     },
     {
       "epoch": 14.25,
       "learning_rate": 1.4264705882352942e-05,
-      "loss": 0.8235,
+      "loss": 1.2779,
       "step": 1944
     },
     {
       "epoch": 14.26,
       "learning_rate": 1.4227941176470589e-05,
-      "loss": 0.8962,
+      "loss": 1.3401,
       "step": 1946
     },
     {
       "epoch": 14.28,
       "learning_rate": 1.4191176470588236e-05,
-      "loss": 0.8734,
+      "loss": 1.2977,
       "step": 1948
     },
     {
       "epoch": 14.29,
       "learning_rate": 1.4154411764705883e-05,
-      "loss": 0.8944,
+      "loss": 1.3437,
       "step": 1950
     },
     {
       "epoch": 14.31,
       "learning_rate": 1.411764705882353e-05,
-      "loss": 0.8675,
+      "loss": 1.282,
       "step": 1952
     },
     {
       "epoch": 14.32,
       "learning_rate": 1.4080882352941177e-05,
-      "loss": 0.8344,
+      "loss": 1.2567,
       "step": 1954
     },
     {
       "epoch": 14.34,
       "learning_rate": 1.4044117647058824e-05,
-      "loss": 0.876,
+      "loss": 1.3013,
       "step": 1956
     },
     {
       "epoch": 14.35,
       "learning_rate": 1.4007352941176471e-05,
-      "loss": 0.8512,
+      "loss": 1.2499,
       "step": 1958
     },
     {
       "epoch": 14.37,
       "learning_rate": 1.3970588235294118e-05,
-      "loss": 0.8957,
+      "loss": 1.2942,
       "step": 1960
     },
     {
       "epoch": 14.38,
       "learning_rate": 1.3933823529411765e-05,
-      "loss": 0.8604,
+      "loss": 1.2772,
       "step": 1962
     },
     {
       "epoch": 14.39,
       "learning_rate": 1.3897058823529412e-05,
-      "loss": 0.8587,
+      "loss": 1.3157,
       "step": 1964
     },
     {
       "epoch": 14.41,
       "learning_rate": 1.3860294117647061e-05,
-      "loss": 0.8611,
+      "loss": 1.3004,
       "step": 1966
     },
     {
       "epoch": 14.42,
       "learning_rate": 1.3823529411764708e-05,
-      "loss": 0.8726,
+      "loss": 1.2854,
       "step": 1968
     },
     {
       "epoch": 14.44,
       "learning_rate": 1.3786764705882355e-05,
-      "loss": 0.872,
+      "loss": 1.3026,
       "step": 1970
     },
     {
       "epoch": 14.45,
       "learning_rate": 1.3750000000000002e-05,
-      "loss": 0.8704,
+      "loss": 1.3356,
       "step": 1972
     },
     {
       "epoch": 14.47,
       "learning_rate": 1.371323529411765e-05,
-      "loss": 0.8747,
+      "loss": 1.3398,
       "step": 1974
     },
     {
       "epoch": 14.48,
       "learning_rate": 1.3676470588235296e-05,
-      "loss": 0.8648,
+      "loss": 1.2938,
       "step": 1976
     },
     {
       "epoch": 14.5,
       "learning_rate": 1.363970588235294e-05,
-      "loss": 0.8985,
+      "loss": 1.3315,
       "step": 1978
     },
     {
       "epoch": 14.51,
       "learning_rate": 1.3602941176470587e-05,
-      "loss": 0.864,
+      "loss": 1.311,
       "step": 1980
     },
     {
       "epoch": 14.53,
       "learning_rate": 1.3566176470588234e-05,
-      "loss": 0.8825,
+      "loss": 1.3192,
       "step": 1982
     },
     {
       "epoch": 14.54,
       "learning_rate": 1.3529411764705883e-05,
-      "loss": 0.8681,
+      "loss": 1.3083,
       "step": 1984
     },
     {
       "epoch": 14.56,
       "learning_rate": 1.349264705882353e-05,
-      "loss": 0.8663,
+      "loss": 1.2973,
       "step": 1986
     },
     {
       "epoch": 14.57,
       "learning_rate": 1.3455882352941177e-05,
-      "loss": 0.8551,
+      "loss": 1.3182,
       "step": 1988
     },
     {
       "epoch": 14.59,
       "learning_rate": 1.3419117647058824e-05,
-      "loss": 0.8468,
+      "loss": 1.2913,
       "step": 1990
     },
     {
       "epoch": 14.6,
       "learning_rate": 1.3382352941176471e-05,
-      "loss": 0.8105,
+      "loss": 1.2591,
       "step": 1992
     },
     {
       "epoch": 14.61,
       "learning_rate": 1.3345588235294118e-05,
-      "loss": 0.8792,
+      "loss": 1.3413,
       "step": 1994
     },
     {
       "epoch": 14.63,
       "learning_rate": 1.3308823529411765e-05,
-      "loss": 0.859,
+      "loss": 1.2683,
       "step": 1996
     },
     {
       "epoch": 14.64,
       "learning_rate": 1.3272058823529412e-05,
-      "loss": 0.8592,
+      "loss": 1.2815,
       "step": 1998
     },
     {
       "epoch": 14.66,
       "learning_rate": 1.323529411764706e-05,
-      "loss": 0.8974,
+      "loss": 1.2921,
       "step": 2000
     },
     {
       "epoch": 14.67,
       "learning_rate": 1.3198529411764706e-05,
-      "loss": 0.8388,
+      "loss": 1.2715,
       "step": 2002
     },
     {
       "epoch": 14.69,
       "learning_rate": 1.3161764705882353e-05,
-      "loss": 0.8662,
+      "loss": 1.3028,
       "step": 2004
     },
     {
       "epoch": 14.7,
       "learning_rate": 1.3125e-05,
-      "loss": 0.8543,
+      "loss": 1.2744,
       "step": 2006
     },
     {
       "epoch": 14.72,
       "learning_rate": 1.3088235294117648e-05,
-      "loss": 0.8434,
+      "loss": 1.2491,
       "step": 2008
     },
     {
       "epoch": 14.73,
       "learning_rate": 1.3051470588235296e-05,
-      "loss": 0.8778,
+      "loss": 1.305,
       "step": 2010
     },
     {
       "epoch": 14.75,
       "learning_rate": 1.3014705882352943e-05,
-      "loss": 0.865,
+      "loss": 1.3338,
       "step": 2012
     },
     {
       "epoch": 14.76,
       "learning_rate": 1.297794117647059e-05,
-      "loss": 0.8724,
+      "loss": 1.2969,
       "step": 2014
     },
     {
       "epoch": 14.78,
       "learning_rate": 1.2941176470588238e-05,
-      "loss": 0.8574,
+      "loss": 1.3163,
       "step": 2016
     },
     {
       "epoch": 14.79,
       "learning_rate": 1.2904411764705885e-05,
-      "loss": 0.8757,
+      "loss": 1.3305,
       "step": 2018
     },
     {
       "epoch": 14.81,
       "learning_rate": 1.2867647058823528e-05,
-      "loss": 0.9153,
+      "loss": 1.3379,
       "step": 2020
     },
     {
       "epoch": 14.82,
       "learning_rate": 1.2830882352941175e-05,
-      "loss": 0.8359,
+      "loss": 1.3051,
       "step": 2022
     },
     {
       "epoch": 14.83,
       "learning_rate": 1.2794117647058822e-05,
-      "loss": 0.8779,
+      "loss": 1.3007,
       "step": 2024
     },
     {
       "epoch": 14.85,
       "learning_rate": 1.275735294117647e-05,
-      "loss": 0.8764,
+      "loss": 1.2793,
       "step": 2026
     },
     {
       "epoch": 14.86,
       "learning_rate": 1.2720588235294118e-05,
-      "loss": 0.8796,
+      "loss": 1.2757,
       "step": 2028
     },
     {
       "epoch": 14.88,
       "learning_rate": 1.2683823529411765e-05,
-      "loss": 0.8569,
+      "loss": 1.3085,
       "step": 2030
     },
     {
       "epoch": 14.89,
       "learning_rate": 1.2647058823529412e-05,
-      "loss": 0.8489,
+      "loss": 1.2862,
       "step": 2032
     },
     {
       "epoch": 14.91,
       "learning_rate": 1.261029411764706e-05,
-      "loss": 0.9111,
+      "loss": 1.3401,
       "step": 2034
     },
     {
       "epoch": 14.92,
       "learning_rate": 1.2573529411764706e-05,
-      "loss": 0.83,
+      "loss": 1.3122,
       "step": 2036
     },
     {
       "epoch": 14.94,
       "learning_rate": 1.2536764705882354e-05,
-      "loss": 0.9074,
+      "loss": 1.3264,
       "step": 2038
     },
     {
       "epoch": 14.95,
       "learning_rate": 1.25e-05,
-      "loss": 0.8712,
+      "loss": 1.3266,
       "step": 2040
     },
     {
       "epoch": 14.97,
       "learning_rate": 1.2463235294117648e-05,
-      "loss": 0.8541,
+      "loss": 1.2872,
       "step": 2042
     },
     {
       "epoch": 14.98,
       "learning_rate": 1.2426470588235295e-05,
-      "loss": 0.8579,
+      "loss": 1.3349,
       "step": 2044
     },
     {
       "epoch": 15.0,
       "learning_rate": 1.2389705882352942e-05,
-      "loss": 0.851,
+      "loss": 1.288,
       "step": 2046
     },
     {
       "epoch": 15.0,
-      "eval_gen_len": 128.76234567901236,
-      "eval_loss": 1.5890727043151855,
-      "eval_rouge1": 56.0531,
-      "eval_rouge2": 30.6748,
-      "eval_rougeL": 38.8847,
-      "eval_rougeLsum": 51.5739,
-      "eval_runtime": 218.1756,
-      "eval_samples_per_second": 4.455,
-      "eval_steps_per_second": 0.559,
+      "eval_gen_len": 744.8858024691358,
+      "eval_loss": 1.4325816631317139,
+      "eval_rouge1": 72.1615,
+      "eval_rouge2": 41.1549,
+      "eval_rougeL": 38.611,
+      "eval_rougeLsum": 69.5977,
+      "eval_runtime": 1630.2451,
+      "eval_samples_per_second": 0.596,
+      "eval_steps_per_second": 0.075,
       "step": 2046
     },
     {
       "epoch": 15.01,
       "learning_rate": 1.2352941176470589e-05,
-      "loss": 0.8108,
+      "loss": 1.3295,
       "step": 2048
     },
     {
       "epoch": 15.03,
       "learning_rate": 1.2316176470588236e-05,
-      "loss": 0.9012,
+      "loss": 1.2984,
       "step": 2050
     },
     {
       "epoch": 15.04,
       "learning_rate": 1.2279411764705883e-05,
-      "loss": 0.796,
+      "loss": 1.263,
       "step": 2052
     },
     {
       "epoch": 15.05,
       "learning_rate": 1.224264705882353e-05,
-      "loss": 0.8676,
+      "loss": 1.3002,
       "step": 2054
     },
     {
       "epoch": 15.07,
       "learning_rate": 1.2205882352941177e-05,
-      "loss": 0.8537,
+      "loss": 1.2865,
       "step": 2056
     },
     {
       "epoch": 15.08,
       "learning_rate": 1.2169117647058824e-05,
-      "loss": 0.8722,
+      "loss": 1.3056,
       "step": 2058
     },
     {
       "epoch": 15.1,
       "learning_rate": 1.2132352941176471e-05,
-      "loss": 0.872,
+      "loss": 1.3438,
       "step": 2060
     },
     {
       "epoch": 15.11,
       "learning_rate": 1.2095588235294118e-05,
-      "loss": 0.83,
+      "loss": 1.3021,
       "step": 2062
     },
     {
       "epoch": 15.13,
       "learning_rate": 1.2058823529411765e-05,
-      "loss": 0.8522,
+      "loss": 1.3076,
       "step": 2064
     },
     {
       "epoch": 15.14,
       "learning_rate": 1.2022058823529412e-05,
-      "loss": 0.8791,
+      "loss": 1.3062,
       "step": 2066
     },
     {
       "epoch": 15.16,
       "learning_rate": 1.198529411764706e-05,
-      "loss": 0.8428,
+      "loss": 1.2882,
       "step": 2068
     },
     {
       "epoch": 15.17,
       "learning_rate": 1.1948529411764707e-05,
-      "loss": 0.8357,
+      "loss": 1.2838,
       "step": 2070
     },
     {
       "epoch": 15.19,
       "learning_rate": 1.1911764705882354e-05,
-      "loss": 0.832,
+      "loss": 1.2722,
       "step": 2072
     },
     {
       "epoch": 15.2,
       "learning_rate": 1.1875e-05,
-      "loss": 0.8356,
+      "loss": 1.2971,
       "step": 2074
     },
     {
       "epoch": 15.22,
       "learning_rate": 1.1838235294117648e-05,
-      "loss": 0.7895,
+      "loss": 1.2587,
       "step": 2076
     },
     {
       "epoch": 15.23,
       "learning_rate": 1.1801470588235295e-05,
-      "loss": 0.8756,
+      "loss": 1.2874,
       "step": 2078
     },
     {
       "epoch": 15.25,
       "learning_rate": 1.1764705882352942e-05,
-      "loss": 0.8746,
+      "loss": 1.3102,
       "step": 2080
     },
     {
       "epoch": 15.26,
       "learning_rate": 1.1727941176470589e-05,
-      "loss": 0.8305,
+      "loss": 1.2912,
       "step": 2082
     },
     {
       "epoch": 15.27,
       "learning_rate": 1.1691176470588236e-05,
-      "loss": 0.8031,
+      "loss": 1.2675,
       "step": 2084
     },
     {
       "epoch": 15.29,
       "learning_rate": 1.1654411764705883e-05,
-      "loss": 0.8222,
+      "loss": 1.2793,
       "step": 2086
     },
     {
       "epoch": 15.3,
       "learning_rate": 1.161764705882353e-05,
-      "loss": 0.8424,
+      "loss": 1.2745,
       "step": 2088
     },
     {
       "epoch": 15.32,
       "learning_rate": 1.1580882352941177e-05,
-      "loss": 0.8262,
+      "loss": 1.3025,
       "step": 2090
     },
     {
       "epoch": 15.33,
       "learning_rate": 1.1544117647058824e-05,
-      "loss": 0.8239,
+      "loss": 1.3012,
       "step": 2092
     },
     {
       "epoch": 15.35,
       "learning_rate": 1.1507352941176471e-05,
-      "loss": 0.8205,
+      "loss": 1.2981,
       "step": 2094
     },
     {
       "epoch": 15.36,
       "learning_rate": 1.1470588235294118e-05,
-      "loss": 0.875,
+      "loss": 1.3368,
       "step": 2096
     },
     {
       "epoch": 15.38,
       "learning_rate": 1.1433823529411765e-05,
-      "loss": 0.9005,
+      "loss": 1.3038,
       "step": 2098
     },
     {
       "epoch": 15.39,
       "learning_rate": 1.1397058823529412e-05,
-      "loss": 0.8394,
+      "loss": 1.3169,
       "step": 2100
     },
     {
       "epoch": 15.41,
       "learning_rate": 1.136029411764706e-05,
-      "loss": 0.8192,
+      "loss": 1.2868,
       "step": 2102
     },
     {
       "epoch": 15.42,
       "learning_rate": 1.1323529411764707e-05,
-      "loss": 0.7979,
+      "loss": 1.3159,
       "step": 2104
     },
     {
       "epoch": 15.44,
       "learning_rate": 1.1286764705882354e-05,
-      "loss": 0.859,
+      "loss": 1.2993,
       "step": 2106
     },
     {
       "epoch": 15.45,
       "learning_rate": 1.125e-05,
-      "loss": 0.8507,
+      "loss": 1.2819,
       "step": 2108
     },
     {
       "epoch": 15.46,
       "learning_rate": 1.1213235294117648e-05,
-      "loss": 0.8197,
+      "loss": 1.2649,
       "step": 2110
     },
     {
       "epoch": 15.48,
       "learning_rate": 1.1176470588235295e-05,
-      "loss": 0.8413,
+      "loss": 1.2955,
       "step": 2112
     },
     {
       "epoch": 15.49,
       "learning_rate": 1.1139705882352942e-05,
-      "loss": 0.882,
+      "loss": 1.2832,
       "step": 2114
     },
     {
       "epoch": 15.51,
       "learning_rate": 1.1102941176470589e-05,
-      "loss": 0.8315,
+      "loss": 1.2808,
       "step": 2116
     },
     {
       "epoch": 15.52,
       "learning_rate": 1.1066176470588236e-05,
-      "loss": 0.8411,
+      "loss": 1.2859,
       "step": 2118
     },
     {
       "epoch": 15.54,
       "learning_rate": 1.1029411764705883e-05,
-      "loss": 0.8753,
+      "loss": 1.3093,
       "step": 2120
     },
     {
       "epoch": 15.55,
       "learning_rate": 1.099264705882353e-05,
-      "loss": 0.8606,
+      "loss": 1.3145,
       "step": 2122
     },
     {
       "epoch": 15.57,
       "learning_rate": 1.0955882352941177e-05,
-      "loss": 0.85,
+      "loss": 1.3115,
       "step": 2124
     },
     {
       "epoch": 15.58,
       "learning_rate": 1.0919117647058824e-05,
-      "loss": 0.8528,
+      "loss": 1.2994,
       "step": 2126
     },
     {
       "epoch": 15.6,
       "learning_rate": 1.0882352941176471e-05,
-      "loss": 0.867,
+      "loss": 1.3344,
       "step": 2128
     },
     {
       "epoch": 15.61,
       "learning_rate": 1.0845588235294118e-05,
-      "loss": 0.8622,
+      "loss": 1.2821,
       "step": 2130
     },
     {
       "epoch": 15.63,
       "learning_rate": 1.0808823529411765e-05,
-      "loss": 0.8359,
+      "loss": 1.3163,
       "step": 2132
     },
     {
       "epoch": 15.64,
       "learning_rate": 1.0772058823529413e-05,
-      "loss": 0.8579,
+      "loss": 1.3224,
       "step": 2134
     },
     {
       "epoch": 15.66,
       "learning_rate": 1.0735294117647058e-05,
-      "loss": 0.8362,
+      "loss": 1.3034,
       "step": 2136
     },
     {
       "epoch": 15.67,
       "learning_rate": 1.0698529411764707e-05,
-      "loss": 0.832,
+      "loss": 1.2512,
       "step": 2138
     },
     {
       "epoch": 15.68,
       "learning_rate": 1.0661764705882354e-05,
-      "loss": 0.8286,
+      "loss": 1.2868,
       "step": 2140
     },
     {
       "epoch": 15.7,
       "learning_rate": 1.0625e-05,
-      "loss": 0.8245,
+      "loss": 1.2871,
       "step": 2142
     },
     {
       "epoch": 15.71,
       "learning_rate": 1.0588235294117648e-05,
-      "loss": 0.894,
+      "loss": 1.2935,
       "step": 2144
     },
     {
       "epoch": 15.73,
       "learning_rate": 1.0551470588235295e-05,
-      "loss": 0.8512,
+      "loss": 1.3185,
       "step": 2146
     },
     {
       "epoch": 15.74,
       "learning_rate": 1.0514705882352942e-05,
-      "loss": 0.9047,
+      "loss": 1.3362,
       "step": 2148
     },
     {
       "epoch": 15.76,
       "learning_rate": 1.0477941176470589e-05,
-      "loss": 0.8459,
+      "loss": 1.305,
       "step": 2150
     },
     {
       "epoch": 15.77,
       "learning_rate": 1.0441176470588236e-05,
-      "loss": 0.8563,
+      "loss": 1.284,
       "step": 2152
     },
     {
       "epoch": 15.79,
       "learning_rate": 1.0404411764705883e-05,
-      "loss": 0.8453,
+      "loss": 1.2949,
       "step": 2154
     },
     {
       "epoch": 15.8,
       "learning_rate": 1.036764705882353e-05,
-      "loss": 0.8583,
+      "loss": 1.3266,
       "step": 2156
     },
     {
       "epoch": 15.82,
       "learning_rate": 1.0330882352941176e-05,
-      "loss": 0.8689,
+      "loss": 1.3154,
       "step": 2158
     },
     {
       "epoch": 15.83,
       "learning_rate": 1.0294117647058824e-05,
-      "loss": 0.8375,
+      "loss": 1.2759,
       "step": 2160
     },
     {
       "epoch": 15.85,
       "learning_rate": 1.0257352941176471e-05,
-      "loss": 0.8454,
+      "loss": 1.297,
       "step": 2162
     },
     {
       "epoch": 15.86,
       "learning_rate": 1.0220588235294118e-05,
-      "loss": 0.8112,
+      "loss": 1.276,
       "step": 2164
     },
     {
       "epoch": 15.88,
       "learning_rate": 1.0183823529411766e-05,
-      "loss": 0.8637,
+      "loss": 1.287,
       "step": 2166
     },
     {
       "epoch": 15.89,
       "learning_rate": 1.0147058823529413e-05,
-      "loss": 0.8704,
+      "loss": 1.2727,
       "step": 2168
     },
     {
       "epoch": 15.9,
       "learning_rate": 1.011029411764706e-05,
-      "loss": 0.8203,
+      "loss": 1.2757,
       "step": 2170
     },
     {
       "epoch": 15.92,
       "learning_rate": 1.0073529411764707e-05,
-      "loss": 0.8463,
+      "loss": 1.2815,
       "step": 2172
     },
     {
       "epoch": 15.93,
       "learning_rate": 1.0036764705882354e-05,
-      "loss": 0.8285,
+      "loss": 1.2777,
       "step": 2174
     },
     {
       "epoch": 15.95,
       "learning_rate": 1e-05,
-      "loss": 0.8072,
+      "loss": 1.2677,
       "step": 2176
     },
     {
       "epoch": 15.96,
       "learning_rate": 9.963235294117648e-06,
-      "loss": 0.8673,
+      "loss": 1.2776,
       "step": 2178
     },
     {
       "epoch": 15.98,
       "learning_rate": 9.926470588235293e-06,
-      "loss": 0.8373,
+      "loss": 1.2892,
       "step": 2180
     },
     {
       "epoch": 15.99,
       "learning_rate": 9.889705882352942e-06,
-      "loss": 0.8825,
+      "loss": 1.2937,
       "step": 2182
     },
     {
       "epoch": 16.0,
-      "eval_gen_len": 128.7798353909465,
-      "eval_loss": 1.5978367328643799,
-      "eval_rouge1": 56.0084,
-      "eval_rouge2": 30.7943,
-      "eval_rougeL": 38.9692,
-      "eval_rougeLsum": 51.5587,
-      "eval_runtime": 220.9798,
-      "eval_samples_per_second": 4.399,
-      "eval_steps_per_second": 0.552,
+      "eval_gen_len": 755.533950617284,
+      "eval_loss": 1.431522250175476,
+      "eval_rouge1": 71.9783,
+      "eval_rouge2": 40.9073,
+      "eval_rougeL": 38.4263,
+      "eval_rougeLsum": 69.4109,
+      "eval_runtime": 1634.7834,
+      "eval_samples_per_second": 0.595,
+      "eval_steps_per_second": 0.075,
       "step": 2183
     },
     {
       "epoch": 16.01,
       "learning_rate": 9.852941176470589e-06,
-      "loss": 0.8308,
+      "loss": 1.2779,
       "step": 2184
     },
     {
       "epoch": 16.02,
       "learning_rate": 9.816176470588236e-06,
-      "loss": 0.8076,
+      "loss": 1.2125,
       "step": 2186
     },
     {
       "epoch": 16.04,
       "learning_rate": 9.779411764705883e-06,
-      "loss": 0.827,
+      "loss": 1.2782,
       "step": 2188
     },
     {
       "epoch": 16.05,
       "learning_rate": 9.74264705882353e-06,
-      "loss": 0.8192,
+      "loss": 1.2828,
       "step": 2190
     },
     {
       "epoch": 16.07,
       "learning_rate": 9.705882352941177e-06,
-      "loss": 0.8323,
+      "loss": 1.2797,
       "step": 2192
     },
     {
       "epoch": 16.08,
       "learning_rate": 9.669117647058824e-06,
-      "loss": 0.8062,
+      "loss": 1.2381,
       "step": 2194
     },
     {
       "epoch": 16.1,
       "learning_rate": 9.632352941176471e-06,
-      "loss": 0.8208,
+      "loss": 1.2742,
       "step": 2196
     },
     {
       "epoch": 16.11,
       "learning_rate": 9.595588235294119e-06,
-      "loss": 0.8459,
+      "loss": 1.2858,
       "step": 2198
     },
     {
       "epoch": 16.12,
       "learning_rate": 9.558823529411764e-06,
-      "loss": 0.8087,
+      "loss": 1.2895,
       "step": 2200
     },
     {
       "epoch": 16.14,
       "learning_rate": 9.522058823529411e-06,
-      "loss": 0.84,
+      "loss": 1.3103,
       "step": 2202
     },
     {
       "epoch": 16.15,
       "learning_rate": 9.48529411764706e-06,
-      "loss": 0.791,
+      "loss": 1.2326,
       "step": 2204
     },
     {
       "epoch": 16.17,
       "learning_rate": 9.448529411764707e-06,
-      "loss": 0.8096,
+      "loss": 1.2753,
       "step": 2206
     },
     {
       "epoch": 16.18,
       "learning_rate": 9.411764705882354e-06,
-      "loss": 0.837,
+      "loss": 1.2954,
       "step": 2208
     },
     {
       "epoch": 16.2,
       "learning_rate": 9.375000000000001e-06,
-      "loss": 0.85,
+      "loss": 1.2823,
       "step": 2210
     },
     {
       "epoch": 16.21,
       "learning_rate": 9.338235294117648e-06,
-      "loss": 0.8501,
+      "loss": 1.3075,
       "step": 2212
     },
     {
       "epoch": 16.23,
       "learning_rate": 9.301470588235295e-06,
-      "loss": 0.8556,
+      "loss": 1.3029,
       "step": 2214
     },
     {
       "epoch": 16.24,
       "learning_rate": 9.264705882352942e-06,
-      "loss": 0.8493,
+      "loss": 1.2876,
       "step": 2216
     },
     {
       "epoch": 16.26,
       "learning_rate": 9.227941176470589e-06,
-      "loss": 0.8204,
+      "loss": 1.2668,
       "step": 2218
     },
     {
       "epoch": 16.27,
       "learning_rate": 9.191176470588236e-06,
-      "loss": 0.8436,
+      "loss": 1.3387,
       "step": 2220
     },
     {
       "epoch": 16.29,
       "learning_rate": 9.154411764705882e-06,
-      "loss": 0.8219,
+      "loss": 1.3048,
       "step": 2222
     },
     {
       "epoch": 16.3,
       "learning_rate": 9.117647058823529e-06,
-      "loss": 0.8244,
+      "loss": 1.305,
       "step": 2224
     },
     {
       "epoch": 16.32,
       "learning_rate": 9.080882352941177e-06,
-      "loss": 0.8164,
+      "loss": 1.2966,
       "step": 2226
     },
     {
       "epoch": 16.33,
       "learning_rate": 9.044117647058824e-06,
-      "loss": 0.8271,
+      "loss": 1.2869,
       "step": 2228
     },
     {
       "epoch": 16.34,
       "learning_rate": 9.007352941176471e-06,
-      "loss": 0.819,
+      "loss": 1.2822,
       "step": 2230
     },
     {
       "epoch": 16.36,
       "learning_rate": 8.970588235294119e-06,
-      "loss": 0.8254,
+      "loss": 1.2886,
       "step": 2232
     },
     {
       "epoch": 16.37,
       "learning_rate": 8.933823529411766e-06,
-      "loss": 0.8555,
+      "loss": 1.3117,
       "step": 2234
     },
     {
       "epoch": 16.39,
       "learning_rate": 8.897058823529413e-06,
-      "loss": 0.8445,
+      "loss": 1.2609,
       "step": 2236
     },
     {
       "epoch": 16.4,
       "learning_rate": 8.86029411764706e-06,
-      "loss": 0.8336,
+      "loss": 1.2904,
       "step": 2238
     },
     {
       "epoch": 16.42,
       "learning_rate": 8.823529411764707e-06,
-      "loss": 0.8301,
+      "loss": 1.3034,
       "step": 2240
     },
     {
       "epoch": 16.43,
       "learning_rate": 8.786764705882352e-06,
-      "loss": 0.9075,
+      "loss": 1.3141,
       "step": 2242
     },
     {
       "epoch": 16.45,
       "learning_rate": 8.75e-06,
-      "loss": 0.8457,
+      "loss": 1.3099,
       "step": 2244
     },
     {
       "epoch": 16.46,
       "learning_rate": 8.713235294117646e-06,
-      "loss": 0.818,
+      "loss": 1.2894,
       "step": 2246
     },
     {
       "epoch": 16.48,
       "learning_rate": 8.676470588235295e-06,
-      "loss": 0.8632,
+      "loss": 1.3491,
       "step": 2248
     },
     {
       "epoch": 16.49,
       "learning_rate": 8.639705882352942e-06,
-      "loss": 0.8443,
+      "loss": 1.288,
       "step": 2250
     },
     {
       "epoch": 16.51,
       "learning_rate": 8.60294117647059e-06,
-      "loss": 0.8487,
+      "loss": 1.2939,
       "step": 2252
     },
     {
       "epoch": 16.52,
       "learning_rate": 8.566176470588236e-06,
-      "loss": 0.8031,
+      "loss": 1.238,
       "step": 2254
     },
     {
       "epoch": 16.54,
       "learning_rate": 8.529411764705883e-06,
-      "loss": 0.7671,
+      "loss": 1.2663,
       "step": 2256
     },
     {
       "epoch": 16.55,
       "learning_rate": 8.49264705882353e-06,
-      "loss": 0.8483,
+      "loss": 1.2732,
       "step": 2258
     },
     {
       "epoch": 16.56,
       "learning_rate": 8.455882352941177e-06,
-      "loss": 0.8293,
+      "loss": 1.294,
       "step": 2260
     },
     {
       "epoch": 16.58,
       "learning_rate": 8.419117647058824e-06,
-      "loss": 0.8185,
+      "loss": 1.3043,
       "step": 2262
     },
     {
       "epoch": 16.59,
       "learning_rate": 8.38235294117647e-06,
-      "loss": 0.7964,
+      "loss": 1.2902,
       "step": 2264
     },
     {
       "epoch": 16.61,
       "learning_rate": 8.345588235294117e-06,
-      "loss": 0.8519,
+      "loss": 1.3337,
       "step": 2266
     },
     {
       "epoch": 16.62,
       "learning_rate": 8.308823529411764e-06,
-      "loss": 0.7689,
+      "loss": 1.2245,
       "step": 2268
     },
     {
       "epoch": 16.64,
       "learning_rate": 8.272058823529413e-06,
-      "loss": 0.8181,
+      "loss": 1.3017,
       "step": 2270
     },
     {
       "epoch": 16.65,
       "learning_rate": 8.23529411764706e-06,
-      "loss": 0.8239,
+      "loss": 1.3169,
       "step": 2272
     },
     {
       "epoch": 16.67,
       "learning_rate": 8.198529411764707e-06,
-      "loss": 0.8342,
+      "loss": 1.2641,
       "step": 2274
     },
     {
       "epoch": 16.68,
       "learning_rate": 8.161764705882354e-06,
-      "loss": 0.8618,
+      "loss": 1.2978,
       "step": 2276
     },
     {
       "epoch": 16.7,
       "learning_rate": 8.125000000000001e-06,
-      "loss": 0.7793,
+      "loss": 1.276,
       "step": 2278
     },
     {
       "epoch": 16.71,
       "learning_rate": 8.088235294117648e-06,
-      "loss": 0.8342,
+      "loss": 1.3104,
       "step": 2280
     },
     {
       "epoch": 16.73,
       "learning_rate": 8.051470588235295e-06,
-      "loss": 0.7744,
+      "loss": 1.2719,
       "step": 2282
     },
     {
       "epoch": 16.74,
       "learning_rate": 8.014705882352942e-06,
-      "loss": 0.8342,
+      "loss": 1.3139,
       "step": 2284
     },
     {
       "epoch": 16.75,
       "learning_rate": 7.977941176470588e-06,
-      "loss": 0.8492,
+      "loss": 1.2831,
       "step": 2286
     },
     {
       "epoch": 16.77,
       "learning_rate": 7.941176470588235e-06,
-      "loss": 0.8114,
+      "loss": 1.2787,
       "step": 2288
     },
     {
       "epoch": 16.78,
       "learning_rate": 7.904411764705882e-06,
-      "loss": 0.7961,
+      "loss": 1.2411,
       "step": 2290
     },
     {
       "epoch": 16.8,
       "learning_rate": 7.86764705882353e-06,
-      "loss": 0.8321,
+      "loss": 1.2503,
       "step": 2292
     },
     {
       "epoch": 16.81,
       "learning_rate": 7.830882352941177e-06,
-      "loss": 0.8185,
+      "loss": 1.2947,
       "step": 2294
     },
     {
       "epoch": 16.83,
       "learning_rate": 7.794117647058825e-06,
-      "loss": 0.8419,
+      "loss": 1.2967,
       "step": 2296
     },
     {
       "epoch": 16.84,
       "learning_rate": 7.757352941176472e-06,
-      "loss": 0.8241,
+      "loss": 1.2755,
       "step": 2298
     },
     {
       "epoch": 16.86,
       "learning_rate": 7.720588235294119e-06,
-      "loss": 0.8294,
+      "loss": 1.2765,
       "step": 2300
     },
     {
       "epoch": 16.87,
       "learning_rate": 7.683823529411766e-06,
-      "loss": 0.8553,
+      "loss": 1.2852,
       "step": 2302
     },
     {
       "epoch": 16.89,
       "learning_rate": 7.647058823529413e-06,
-      "loss": 0.7851,
+      "loss": 1.2997,
       "step": 2304
     },
     {
       "epoch": 16.9,
       "learning_rate": 7.610294117647058e-06,
-      "loss": 0.8508,
+      "loss": 1.2993,
       "step": 2306
     },
     {
       "epoch": 16.92,
       "learning_rate": 7.573529411764706e-06,
-      "loss": 0.8455,
+      "loss": 1.2849,
       "step": 2308
     },
     {
       "epoch": 16.93,
       "learning_rate": 7.536764705882353e-06,
-      "loss": 0.8724,
+      "loss": 1.2966,
       "step": 2310
     },
     {
       "epoch": 16.95,
       "learning_rate": 7.5e-06,
-      "loss": 0.8606,
+      "loss": 1.3007,
       "step": 2312
     },
     {
       "epoch": 16.96,
       "learning_rate": 7.463235294117647e-06,
-      "loss": 0.8677,
+      "loss": 1.3096,
       "step": 2314
     },
     {
       "epoch": 16.97,
       "learning_rate": 7.426470588235294e-06,
-      "loss": 0.8488,
+      "loss": 1.2897,
       "step": 2316
     },
     {
       "epoch": 16.99,
       "learning_rate": 7.389705882352941e-06,
-      "loss": 0.8169,
+      "loss": 1.258,
       "step": 2318
     },
     {
       "epoch": 17.0,
-      "eval_gen_len": 128.86316872427983,
-      "eval_loss": 1.6075981855392456,
-      "eval_rouge1": 55.8274,
-      "eval_rouge2": 30.41,
-      "eval_rougeL": 38.6258,
-      "eval_rougeLsum": 51.3009,
-      "eval_runtime": 219.3849,
-      "eval_samples_per_second": 4.431,
-      "eval_steps_per_second": 0.556,
+      "eval_gen_len": 734.6399176954733,
+      "eval_loss": 1.4328014850616455,
+      "eval_rouge1": 72.0298,
+      "eval_rouge2": 40.931,
+      "eval_rougeL": 38.4845,
+      "eval_rougeLsum": 69.4823,
+      "eval_runtime": 1579.2093,
+      "eval_samples_per_second": 0.615,
+      "eval_steps_per_second": 0.077,
       "step": 2319
     },
     {
       "epoch": 17.0,
       "learning_rate": 7.3529411764705884e-06,
-      "loss": 0.7751,
+      "loss": 1.2402,
       "step": 2320
     },
     {
       "epoch": 17.02,
       "learning_rate": 7.316176470588236e-06,
-      "loss": 0.8066,
+      "loss": 1.3125,
       "step": 2322
     },
     {
       "epoch": 17.03,
       "learning_rate": 7.279411764705883e-06,
-      "loss": 0.8462,
+      "loss": 1.2644,
       "step": 2324
     },
     {
       "epoch": 17.05,
       "learning_rate": 7.2426470588235305e-06,
-      "loss": 0.8035,
+      "loss": 1.2538,
       "step": 2326
     },
     {
       "epoch": 17.06,
       "learning_rate": 7.205882352941176e-06,
-      "loss": 0.7691,
+      "loss": 1.3019,
       "step": 2328
     },
     {
       "epoch": 17.08,
       "learning_rate": 7.169117647058824e-06,
-      "loss": 0.7958,
+      "loss": 1.283,
       "step": 2330
     },
     {
       "epoch": 17.09,
       "learning_rate": 7.132352941176471e-06,
-      "loss": 0.7804,
+      "loss": 1.2648,
       "step": 2332
     },
     {
       "epoch": 17.11,
       "learning_rate": 7.095588235294118e-06,
-      "loss": 0.8048,
+      "loss": 1.2731,
       "step": 2334
     },
     {
       "epoch": 17.12,
       "learning_rate": 7.058823529411765e-06,
-      "loss": 0.7775,
+      "loss": 1.2447,
       "step": 2336
     },
     {
       "epoch": 17.14,
       "learning_rate": 7.022058823529412e-06,
-      "loss": 0.8318,
+      "loss": 1.3294,
       "step": 2338
     },
     {
       "epoch": 17.15,
       "learning_rate": 6.985294117647059e-06,
-      "loss": 0.8566,
+      "loss": 1.326,
       "step": 2340
     },
     {
       "epoch": 17.17,
       "learning_rate": 6.948529411764706e-06,
-      "loss": 0.797,
+      "loss": 1.285,
       "step": 2342
     },
     {
       "epoch": 17.18,
       "learning_rate": 6.911764705882354e-06,
-      "loss": 0.8252,
+      "loss": 1.3157,
       "step": 2344
     },
     {
       "epoch": 17.19,
       "learning_rate": 6.875000000000001e-06,
-      "loss": 0.7782,
+      "loss": 1.2731,
       "step": 2346
     },
     {
       "epoch": 17.21,
       "learning_rate": 6.838235294117648e-06,
-      "loss": 0.8611,
+      "loss": 1.2805,
       "step": 2348
     },
     {
       "epoch": 17.22,
       "learning_rate": 6.8014705882352935e-06,
-      "loss": 0.8206,
+      "loss": 1.3002,
       "step": 2350
     },
     {
       "epoch": 17.24,
       "learning_rate": 6.7647058823529414e-06,
-      "loss": 0.829,
+      "loss": 1.2717,
       "step": 2352
     },
     {
       "epoch": 17.25,
       "learning_rate": 6.7279411764705885e-06,
-      "loss": 0.7828,
+      "loss": 1.2338,
       "step": 2354
     },
     {
       "epoch": 17.27,
       "learning_rate": 6.6911764705882356e-06,
-      "loss": 0.7837,
+      "loss": 1.2363,
       "step": 2356
     },
     {
       "epoch": 17.28,
       "learning_rate": 6.654411764705883e-06,
-      "loss": 0.8112,
+      "loss": 1.2774,
       "step": 2358
     },
     {
       "epoch": 17.3,
       "learning_rate": 6.61764705882353e-06,
-      "loss": 0.8087,
+      "loss": 1.2517,
       "step": 2360
     },
     {
       "epoch": 17.31,
       "learning_rate": 6.580882352941177e-06,
-      "loss": 0.7931,
+      "loss": 1.2792,
       "step": 2362
     },
     {
       "epoch": 17.33,
       "learning_rate": 6.544117647058824e-06,
-      "loss": 0.8214,
+      "loss": 1.2314,
       "step": 2364
     },
     {
       "epoch": 17.34,
       "learning_rate": 6.507352941176472e-06,
-      "loss": 0.7835,
+      "loss": 1.2584,
       "step": 2366
     },
     {
       "epoch": 17.36,
       "learning_rate": 6.470588235294119e-06,
-      "loss": 0.8415,
+      "loss": 1.3062,
       "step": 2368
     },
     {
       "epoch": 17.37,
       "learning_rate": 6.433823529411764e-06,
-      "loss": 0.8249,
+      "loss": 1.2593,
       "step": 2370
     },
     {
       "epoch": 17.39,
       "learning_rate": 6.397058823529411e-06,
-      "loss": 0.7932,
+      "loss": 1.2988,
       "step": 2372
     },
     {
       "epoch": 17.4,
       "learning_rate": 6.360294117647059e-06,
-      "loss": 0.844,
+      "loss": 1.2704,
       "step": 2374
     },
     {
       "epoch": 17.41,
       "learning_rate": 6.323529411764706e-06,
-      "loss": 0.8582,
+      "loss": 1.2924,
       "step": 2376
     },
     {
       "epoch": 17.43,
       "learning_rate": 6.286764705882353e-06,
-      "loss": 0.8629,
+      "loss": 1.2964,
       "step": 2378
     },
     {
       "epoch": 17.44,
       "learning_rate": 6.25e-06,
-      "loss": 0.828,
+      "loss": 1.2825,
       "step": 2380
     },
     {
       "epoch": 17.46,
       "learning_rate": 6.213235294117647e-06,
-      "loss": 0.8536,
+      "loss": 1.2956,
       "step": 2382
     },
     {
       "epoch": 17.47,
       "learning_rate": 6.1764705882352944e-06,
-      "loss": 0.79,
+      "loss": 1.2423,
       "step": 2384
     },
     {
       "epoch": 17.49,
       "learning_rate": 6.1397058823529415e-06,
-      "loss": 0.7974,
+      "loss": 1.2824,
       "step": 2386
     },
     {
       "epoch": 17.5,
       "learning_rate": 6.1029411764705885e-06,
-      "loss": 0.8383,
+      "loss": 1.2939,
       "step": 2388
     },
     {
       "epoch": 17.52,
       "learning_rate": 6.066176470588236e-06,
-      "loss": 0.8324,
+      "loss": 1.2747,
       "step": 2390
     },
     {
       "epoch": 17.53,
       "learning_rate": 6.029411764705883e-06,
-      "loss": 0.786,
+      "loss": 1.2501,
       "step": 2392
     },
     {
       "epoch": 17.55,
       "learning_rate": 5.99264705882353e-06,
-      "loss": 0.8319,
+      "loss": 1.2715,
       "step": 2394
     },
     {
       "epoch": 17.56,
       "learning_rate": 5.955882352941177e-06,
-      "loss": 0.8172,
+      "loss": 1.2974,
       "step": 2396
     },
     {
       "epoch": 17.58,
       "learning_rate": 5.919117647058824e-06,
-      "loss": 0.8089,
+      "loss": 1.2486,
       "step": 2398
     },
     {
       "epoch": 17.59,
       "learning_rate": 5.882352941176471e-06,
-      "loss": 0.7822,
+      "loss": 1.2593,
       "step": 2400
     },
     {
       "epoch": 17.61,
       "learning_rate": 5.845588235294118e-06,
-      "loss": 0.8161,
+      "loss": 1.2523,
       "step": 2402
     },
     {
       "epoch": 17.62,
       "learning_rate": 5.808823529411765e-06,
-      "loss": 0.8333,
+      "loss": 1.2977,
       "step": 2404
     },
     {
       "epoch": 17.63,
       "learning_rate": 5.772058823529412e-06,
-      "loss": 0.7785,
+      "loss": 1.2788,
       "step": 2406
     },
     {
       "epoch": 17.65,
       "learning_rate": 5.735294117647059e-06,
-      "loss": 0.7676,
+      "loss": 1.2569,
       "step": 2408
     },
     {
       "epoch": 17.66,
       "learning_rate": 5.698529411764706e-06,
-      "loss": 0.8038,
+      "loss": 1.2709,
       "step": 2410
     },
     {
       "epoch": 17.68,
       "learning_rate": 5.661764705882353e-06,
-      "loss": 0.8318,
+      "loss": 1.2801,
       "step": 2412
     },
     {
       "epoch": 17.69,
       "learning_rate": 5.625e-06,
-      "loss": 0.8082,
+      "loss": 1.2747,
       "step": 2414
     },
     {
       "epoch": 17.71,
       "learning_rate": 5.588235294117647e-06,
-      "loss": 0.8248,
+      "loss": 1.2934,
       "step": 2416
     },
     {
       "epoch": 17.72,
       "learning_rate": 5.5514705882352945e-06,
-      "loss": 0.7761,
+      "loss": 1.2478,
       "step": 2418
     },
     {
       "epoch": 17.74,
       "learning_rate": 5.5147058823529415e-06,
-      "loss": 0.8286,
+      "loss": 1.3025,
       "step": 2420
     },
     {
       "epoch": 17.75,
       "learning_rate": 5.477941176470589e-06,
-      "loss": 0.8185,
+      "loss": 1.303,
       "step": 2422
     },
     {
       "epoch": 17.77,
       "learning_rate": 5.441176470588236e-06,
-      "loss": 0.8263,
+      "loss": 1.2874,
       "step": 2424
     },
     {
       "epoch": 17.78,
       "learning_rate": 5.404411764705883e-06,
-      "loss": 0.8006,
+      "loss": 1.2947,
       "step": 2426
     },
     {
       "epoch": 17.8,
       "learning_rate": 5.367647058823529e-06,
-      "loss": 0.8668,
+      "loss": 1.3189,
       "step": 2428
     },
     {
       "epoch": 17.81,
       "learning_rate": 5.330882352941177e-06,
-      "loss": 0.7969,
+      "loss": 1.2738,
       "step": 2430
     },
     {
       "epoch": 17.83,
       "learning_rate": 5.294117647058824e-06,
-      "loss": 0.8191,
+      "loss": 1.3261,
       "step": 2432
     },
     {
       "epoch": 17.84,
       "learning_rate": 5.257352941176471e-06,
-      "loss": 0.8315,
+      "loss": 1.313,
       "step": 2434
     },
     {
       "epoch": 17.85,
       "learning_rate": 5.220588235294118e-06,
-      "loss": 0.7925,
+      "loss": 1.2975,
       "step": 2436
     },
     {
       "epoch": 17.87,
       "learning_rate": 5.183823529411765e-06,
-      "loss": 0.8377,
+      "loss": 1.2727,
       "step": 2438
     },
     {
       "epoch": 17.88,
       "learning_rate": 5.147058823529412e-06,
-      "loss": 0.8196,
+      "loss": 1.2827,
       "step": 2440
     },
     {
       "epoch": 17.9,
       "learning_rate": 5.110294117647059e-06,
-      "loss": 0.7612,
+      "loss": 1.2173,
       "step": 2442
     },
     {
       "epoch": 17.91,
       "learning_rate": 5.073529411764706e-06,
-      "loss": 0.8375,
+      "loss": 1.2882,
       "step": 2444
     },
     {
       "epoch": 17.93,
       "learning_rate": 5.036764705882353e-06,
-      "loss": 0.8251,
+      "loss": 1.2659,
       "step": 2446
     },
     {
       "epoch": 17.94,
       "learning_rate": 5e-06,
-      "loss": 0.8507,
+      "loss": 1.2641,
       "step": 2448
     },
     {
       "epoch": 17.96,
       "learning_rate": 4.963235294117647e-06,
-      "loss": 0.8701,
+      "loss": 1.2967,
       "step": 2450
     },
     {
       "epoch": 17.97,
       "learning_rate": 4.9264705882352945e-06,
-      "loss": 0.8294,
+      "loss": 1.302,
       "step": 2452
     },
     {
       "epoch": 17.99,
       "learning_rate": 4.889705882352942e-06,
-      "loss": 0.8194,
+      "loss": 1.2617,
       "step": 2454
     },
     {
       "epoch": 17.99,
-      "eval_gen_len": 128.64609053497944,
-      "eval_loss": 1.6177321672439575,
-      "eval_rouge1": 56.3214,
-      "eval_rouge2": 30.9896,
-      "eval_rougeL": 39.4754,
-      "eval_rougeLsum": 51.9525,
-      "eval_runtime": 216.2607,
-      "eval_samples_per_second": 4.495,
-      "eval_steps_per_second": 0.564,
+      "eval_gen_len": 744.7067901234568,
+      "eval_loss": 1.4336270093917847,
+      "eval_rouge1": 71.9488,
+      "eval_rouge2": 40.8816,
+      "eval_rougeL": 38.4521,
+      "eval_rougeLsum": 69.4151,
+      "eval_runtime": 1622.2621,
+      "eval_samples_per_second": 0.599,
+      "eval_steps_per_second": 0.075,
       "step": 2455
     },
     {
       "epoch": 18.0,
       "learning_rate": 4.852941176470589e-06,
-      "loss": 0.8531,
+      "loss": 1.3316,
       "step": 2456
     },
     {
       "epoch": 18.02,
       "learning_rate": 4.816176470588236e-06,
-      "loss": 0.7938,
+      "loss": 1.271,
       "step": 2458
     },
     {
       "epoch": 18.03,
       "learning_rate": 4.779411764705882e-06,
-      "loss": 0.8304,
+      "loss": 1.2762,
       "step": 2460
     },
     {
       "epoch": 18.04,
       "learning_rate": 4.74264705882353e-06,
-      "loss": 0.7959,
+      "loss": 1.2747,
       "step": 2462
     },
     {
       "epoch": 18.06,
       "learning_rate": 4.705882352941177e-06,
-      "loss": 0.8089,
+      "loss": 1.2935,
       "step": 2464
     },
     {
       "epoch": 18.07,
       "learning_rate": 4.669117647058824e-06,
-      "loss": 0.8112,
+      "loss": 1.2907,
       "step": 2466
     },
     {
       "epoch": 18.09,
       "learning_rate": 4.632352941176471e-06,
-      "loss": 0.8237,
+      "loss": 1.2581,
       "step": 2468
     },
     {
       "epoch": 18.1,
       "learning_rate": 4.595588235294118e-06,
-      "loss": 0.8255,
+      "loss": 1.3023,
       "step": 2470
     },
     {
       "epoch": 18.12,
       "learning_rate": 4.558823529411764e-06,
-      "loss": 0.8026,
+      "loss": 1.2795,
       "step": 2472
     },
     {
       "epoch": 18.13,
       "learning_rate": 4.522058823529412e-06,
-      "loss": 0.8219,
+      "loss": 1.2715,
       "step": 2474
     },
     {
       "epoch": 18.15,
       "learning_rate": 4.485294117647059e-06,
-      "loss": 0.7687,
+      "loss": 1.2621,
       "step": 2476
     },
     {
       "epoch": 18.16,
       "learning_rate": 4.448529411764706e-06,
-      "loss": 0.7815,
+      "loss": 1.2882,
       "step": 2478
     },
     {
       "epoch": 18.18,
       "learning_rate": 4.411764705882353e-06,
-      "loss": 0.7863,
+      "loss": 1.2765,
       "step": 2480
     },
     {
       "epoch": 18.19,
       "learning_rate": 4.375e-06,
-      "loss": 0.805,
+      "loss": 1.2678,
       "step": 2482
     },
     {
       "epoch": 18.21,
       "learning_rate": 4.3382352941176475e-06,
-      "loss": 0.7961,
+      "loss": 1.2843,
       "step": 2484
     },
     {
       "epoch": 18.22,
       "learning_rate": 4.301470588235295e-06,
-      "loss": 0.8162,
+      "loss": 1.3001,
       "step": 2486
     },
     {
       "epoch": 18.24,
       "learning_rate": 4.264705882352942e-06,
-      "loss": 0.8307,
+      "loss": 1.2784,
       "step": 2488
     },
     {
       "epoch": 18.25,
       "learning_rate": 4.227941176470589e-06,
-      "loss": 0.8335,
+      "loss": 1.3116,
       "step": 2490
     },
     {
       "epoch": 18.26,
       "learning_rate": 4.191176470588235e-06,
-      "loss": 0.7927,
+      "loss": 1.2721,
       "step": 2492
     },
     {
       "epoch": 18.28,
       "learning_rate": 4.154411764705882e-06,
-      "loss": 0.7851,
+      "loss": 1.2575,
       "step": 2494
     },
     {
       "epoch": 18.29,
       "learning_rate": 4.11764705882353e-06,
-      "loss": 0.7844,
+      "loss": 1.2532,
       "step": 2496
     },
     {
       "epoch": 18.31,
       "learning_rate": 4.080882352941177e-06,
-      "loss": 0.8222,
+      "loss": 1.3147,
       "step": 2498
     },
     {
       "epoch": 18.32,
       "learning_rate": 4.044117647058824e-06,
-      "loss": 0.8013,
+      "loss": 1.2965,
       "step": 2500
     },
     {
       "epoch": 18.34,
       "learning_rate": 4.007352941176471e-06,
-      "loss": 0.8087,
+      "loss": 1.2836,
       "step": 2502
     },
     {
       "epoch": 18.35,
       "learning_rate": 3.970588235294117e-06,
-      "loss": 0.8239,
+      "loss": 1.276,
       "step": 2504
     },
     {
       "epoch": 18.37,
       "learning_rate": 3.933823529411765e-06,
-      "loss": 0.8153,
+      "loss": 1.2781,
       "step": 2506
     },
     {
       "epoch": 18.38,
       "learning_rate": 3.897058823529412e-06,
-      "loss": 0.8534,
+      "loss": 1.2817,
       "step": 2508
     },
     {
       "epoch": 18.4,
       "learning_rate": 3.860294117647059e-06,
-      "loss": 0.7795,
+      "loss": 1.256,
       "step": 2510
     },
     {
       "epoch": 18.41,
       "learning_rate": 3.823529411764706e-06,
-      "loss": 0.7732,
+      "loss": 1.2359,
       "step": 2512
     },
     {
       "epoch": 18.43,
       "learning_rate": 3.786764705882353e-06,
-      "loss": 0.8287,
+      "loss": 1.2867,
       "step": 2514
     },
     {
       "epoch": 18.44,
       "learning_rate": 3.75e-06,
-      "loss": 0.7886,
+      "loss": 1.2644,
       "step": 2516
     },
     {
       "epoch": 18.46,
       "learning_rate": 3.713235294117647e-06,
-      "loss": 0.7978,
+      "loss": 1.2818,
       "step": 2518
     },
     {
       "epoch": 18.47,
       "learning_rate": 3.6764705882352942e-06,
-      "loss": 0.8232,
+      "loss": 1.2902,
       "step": 2520
     },
     {
       "epoch": 18.48,
       "learning_rate": 3.6397058823529417e-06,
-      "loss": 0.7782,
+      "loss": 1.2513,
       "step": 2522
     },
     {
       "epoch": 18.5,
       "learning_rate": 3.602941176470588e-06,
-      "loss": 0.8201,
+      "loss": 1.2876,
       "step": 2524
     },
     {
       "epoch": 18.51,
       "learning_rate": 3.5661764705882354e-06,
-      "loss": 0.7897,
+      "loss": 1.2579,
       "step": 2526
     },
     {
       "epoch": 18.53,
       "learning_rate": 3.5294117647058825e-06,
-      "loss": 0.7692,
+      "loss": 1.2587,
       "step": 2528
     },
     {
       "epoch": 18.54,
       "learning_rate": 3.4926470588235295e-06,
-      "loss": 0.784,
+      "loss": 1.2613,
       "step": 2530
     },
     {
       "epoch": 18.56,
       "learning_rate": 3.455882352941177e-06,
-      "loss": 0.8474,
+      "loss": 1.3092,
       "step": 2532
     },
     {
       "epoch": 18.57,
       "learning_rate": 3.419117647058824e-06,
-      "loss": 0.7644,
+      "loss": 1.2454,
       "step": 2534
     },
     {
       "epoch": 18.59,
       "learning_rate": 3.3823529411764707e-06,
-      "loss": 0.7707,
+      "loss": 1.254,
       "step": 2536
     },
     {
       "epoch": 18.6,
       "learning_rate": 3.3455882352941178e-06,
-      "loss": 0.7818,
+      "loss": 1.2796,
       "step": 2538
     },
     {
       "epoch": 18.62,
       "learning_rate": 3.308823529411765e-06,
-      "loss": 0.797,
+      "loss": 1.2328,
       "step": 2540
     },
     {
       "epoch": 18.63,
       "learning_rate": 3.272058823529412e-06,
-      "loss": 0.8142,
+      "loss": 1.2987,
       "step": 2542
     },
     {
       "epoch": 18.65,
       "learning_rate": 3.2352941176470594e-06,
-      "loss": 0.7797,
+      "loss": 1.2678,
       "step": 2544
     },
     {
       "epoch": 18.66,
       "learning_rate": 3.1985294117647056e-06,
-      "loss": 0.8282,
+      "loss": 1.2742,
       "step": 2546
     },
     {
       "epoch": 18.68,
       "learning_rate": 3.161764705882353e-06,
-      "loss": 0.7816,
+      "loss": 1.285,
       "step": 2548
     },
     {
       "epoch": 18.69,
       "learning_rate": 3.125e-06,
-      "loss": 0.7697,
+      "loss": 1.2575,
       "step": 2550
     },
     {
       "epoch": 18.7,
       "learning_rate": 3.0882352941176472e-06,
-      "loss": 0.8033,
+      "loss": 1.2807,
       "step": 2552
     },
     {
       "epoch": 18.72,
       "learning_rate": 3.0514705882352943e-06,
-      "loss": 0.8221,
+      "loss": 1.2584,
       "step": 2554
     },
     {
       "epoch": 18.73,
       "learning_rate": 3.0147058823529413e-06,
-      "loss": 0.7657,
+      "loss": 1.2162,
       "step": 2556
     },
     {
       "epoch": 18.75,
       "learning_rate": 2.9779411764705884e-06,
-      "loss": 0.7999,
+      "loss": 1.2371,
       "step": 2558
     },
     {
       "epoch": 18.76,
       "learning_rate": 2.9411764705882355e-06,
-      "loss": 0.7996,
+      "loss": 1.2401,
       "step": 2560
     },
     {
       "epoch": 18.78,
       "learning_rate": 2.9044117647058825e-06,
-      "loss": 0.7903,
+      "loss": 1.2498,
       "step": 2562
     },
     {
       "epoch": 18.79,
       "learning_rate": 2.8676470588235296e-06,
-      "loss": 0.7895,
+      "loss": 1.2554,
       "step": 2564
     },
     {
       "epoch": 18.81,
       "learning_rate": 2.8308823529411766e-06,
-      "loss": 0.8348,
+      "loss": 1.2683,
       "step": 2566
     },
     {
       "epoch": 18.82,
       "learning_rate": 2.7941176470588237e-06,
-      "loss": 0.7819,
+      "loss": 1.2478,
       "step": 2568
     },
     {
       "epoch": 18.84,
       "learning_rate": 2.7573529411764708e-06,
-      "loss": 0.8005,
+      "loss": 1.2815,
       "step": 2570
     },
     {
       "epoch": 18.85,
       "learning_rate": 2.720588235294118e-06,
-      "loss": 0.7874,
+      "loss": 1.2642,
       "step": 2572
     },
     {
       "epoch": 18.87,
       "learning_rate": 2.6838235294117645e-06,
-      "loss": 0.7964,
+      "loss": 1.264,
       "step": 2574
     },
     {
       "epoch": 18.88,
       "learning_rate": 2.647058823529412e-06,
-      "loss": 0.8508,
+      "loss": 1.2911,
       "step": 2576
     },
     {
       "epoch": 18.9,
       "learning_rate": 2.610294117647059e-06,
-      "loss": 0.8511,
+      "loss": 1.3437,
       "step": 2578
     },
     {
       "epoch": 18.91,
       "learning_rate": 2.573529411764706e-06,
-      "loss": 0.8308,
+      "loss": 1.2873,
       "step": 2580
     },
     {
       "epoch": 18.92,
       "learning_rate": 2.536764705882353e-06,
-      "loss": 0.8333,
+      "loss": 1.3098,
       "step": 2582
     },
     {
       "epoch": 18.94,
       "learning_rate": 2.5e-06,
-      "loss": 0.8121,
+      "loss": 1.2848,
       "step": 2584
     },
     {
       "epoch": 18.95,
       "learning_rate": 2.4632352941176473e-06,
-      "loss": 0.7601,
+      "loss": 1.26,
       "step": 2586
     },
     {
       "epoch": 18.97,
       "learning_rate": 2.4264705882352943e-06,
-      "loss": 0.9098,
+      "loss": 1.2744,
       "step": 2588
     },
     {
       "epoch": 18.98,
       "learning_rate": 2.389705882352941e-06,
-      "loss": 0.8387,
+      "loss": 1.2806,
       "step": 2590
     },
     {
       "epoch": 19.0,
       "learning_rate": 2.3529411764705885e-06,
-      "loss": 0.8441,
+      "loss": 1.2864,
       "step": 2592
     },
     {
       "epoch": 19.0,
-      "eval_gen_len": 128.82407407407408,
-      "eval_loss": 1.625985026359558,
-      "eval_rouge1": 55.9842,
-      "eval_rouge2": 30.6332,
-      "eval_rougeL": 38.999,
-      "eval_rougeLsum": 51.5685,
-      "eval_runtime": 216.0054,
-      "eval_samples_per_second": 4.5,
-      "eval_steps_per_second": 0.565,
+      "eval_gen_len": 744.2448559670781,
+      "eval_loss": 1.4346004724502563,
+      "eval_rouge1": 72.1334,
+      "eval_rouge2": 40.9965,
+      "eval_rougeL": 38.5682,
+      "eval_rougeLsum": 69.5666,
+      "eval_runtime": 1612.4207,
+      "eval_samples_per_second": 0.603,
+      "eval_steps_per_second": 0.076,
       "step": 2592
     },
     {
       "epoch": 19.01,
       "learning_rate": 2.3161764705882355e-06,
-      "loss": 0.7704,
+      "loss": 1.2617,
       "step": 2594
     },
     {
       "epoch": 19.03,
       "learning_rate": 2.279411764705882e-06,
-      "loss": 0.79,
+      "loss": 1.2852,
       "step": 2596
     },
     {
       "epoch": 19.04,
       "learning_rate": 2.2426470588235296e-06,
-      "loss": 0.8014,
+      "loss": 1.2548,
       "step": 2598
     },
     {
       "epoch": 19.06,
       "learning_rate": 2.2058823529411767e-06,
-      "loss": 0.8042,
+      "loss": 1.2467,
       "step": 2600
     },
     {
       "epoch": 19.07,
       "learning_rate": 2.1691176470588238e-06,
-      "loss": 0.7609,
+      "loss": 1.2457,
       "step": 2602
     },
     {
       "epoch": 19.09,
       "learning_rate": 2.132352941176471e-06,
-      "loss": 0.8418,
+      "loss": 1.2929,
       "step": 2604
     },
     {
       "epoch": 19.1,
       "learning_rate": 2.0955882352941175e-06,
-      "loss": 0.7901,
+      "loss": 1.2768,
       "step": 2606
     },
     {
       "epoch": 19.11,
       "learning_rate": 2.058823529411765e-06,
-      "loss": 0.7841,
+      "loss": 1.2755,
       "step": 2608
     },
     {
       "epoch": 19.13,
       "learning_rate": 2.022058823529412e-06,
-      "loss": 0.778,
+      "loss": 1.2621,
       "step": 2610
     },
     {
       "epoch": 19.14,
       "learning_rate": 1.9852941176470586e-06,
-      "loss": 0.8207,
+      "loss": 1.2802,
       "step": 2612
     },
     {
       "epoch": 19.16,
       "learning_rate": 1.948529411764706e-06,
-      "loss": 0.7564,
+      "loss": 1.2528,
       "step": 2614
     },
     {
       "epoch": 19.17,
       "learning_rate": 1.911764705882353e-06,
-      "loss": 0.7915,
+      "loss": 1.2608,
       "step": 2616
     },
     {
       "epoch": 19.19,
       "learning_rate": 1.875e-06,
-      "loss": 0.7916,
+      "loss": 1.286,
       "step": 2618
     },
     {
       "epoch": 19.2,
       "learning_rate": 1.8382352941176471e-06,
-      "loss": 0.8082,
+      "loss": 1.2803,
       "step": 2620
     },
     {
       "epoch": 19.22,
       "learning_rate": 1.801470588235294e-06,
-      "loss": 0.8426,
+      "loss": 1.2816,
       "step": 2622
     },
     {
       "epoch": 19.23,
       "learning_rate": 1.7647058823529412e-06,
-      "loss": 0.8074,
+      "loss": 1.2336,
       "step": 2624
     },
     {
       "epoch": 19.25,
       "learning_rate": 1.7279411764705885e-06,
-      "loss": 0.8121,
+      "loss": 1.278,
       "step": 2626
     },
     {
       "epoch": 19.26,
       "learning_rate": 1.6911764705882354e-06,
-      "loss": 0.8088,
+      "loss": 1.3079,
       "step": 2628
     },
     {
       "epoch": 19.28,
       "learning_rate": 1.6544117647058824e-06,
-      "loss": 0.816,
+      "loss": 1.2659,
       "step": 2630
     },
     {
       "epoch": 19.29,
       "learning_rate": 1.6176470588235297e-06,
-      "loss": 0.7726,
+      "loss": 1.2686,
       "step": 2632
     },
     {
       "epoch": 19.31,
       "learning_rate": 1.5808823529411765e-06,
-      "loss": 0.7635,
+      "loss": 1.2069,
       "step": 2634
     },
     {
       "epoch": 19.32,
       "learning_rate": 1.5441176470588236e-06,
-      "loss": 0.7938,
+      "loss": 1.2774,
       "step": 2636
     },
     {
       "epoch": 19.33,
       "learning_rate": 1.5073529411764707e-06,
-      "loss": 0.7696,
+      "loss": 1.248,
       "step": 2638
     },
     {
       "epoch": 19.35,
       "learning_rate": 1.4705882352941177e-06,
-      "loss": 0.8152,
+      "loss": 1.2786,
       "step": 2640
     },
     {
       "epoch": 19.36,
       "learning_rate": 1.4338235294117648e-06,
-      "loss": 0.8182,
+      "loss": 1.251,
       "step": 2642
     },
     {
       "epoch": 19.38,
       "learning_rate": 1.3970588235294119e-06,
-      "loss": 0.7999,
+      "loss": 1.3055,
       "step": 2644
     },
     {
       "epoch": 19.39,
       "learning_rate": 1.360294117647059e-06,
-      "loss": 0.8111,
+      "loss": 1.2598,
       "step": 2646
     },
     {
       "epoch": 19.41,
       "learning_rate": 1.323529411764706e-06,
-      "loss": 0.7841,
+      "loss": 1.2479,
       "step": 2648
     },
     {
       "epoch": 19.42,
       "learning_rate": 1.286764705882353e-06,
-      "loss": 0.7895,
+      "loss": 1.2415,
       "step": 2650
     },
     {
       "epoch": 19.44,
       "learning_rate": 1.25e-06,
-      "loss": 0.7895,
+      "loss": 1.2595,
       "step": 2652
     },
     {
       "epoch": 19.45,
       "learning_rate": 1.2132352941176472e-06,
-      "loss": 0.8028,
+      "loss": 1.2443,
       "step": 2654
     },
     {
       "epoch": 19.47,
       "learning_rate": 1.1764705882352942e-06,
-      "loss": 0.7947,
+      "loss": 1.318,
       "step": 2656
     },
     {
       "epoch": 19.48,
       "learning_rate": 1.139705882352941e-06,
-      "loss": 0.7839,
+      "loss": 1.2591,
       "step": 2658
     },
     {
       "epoch": 19.5,
       "learning_rate": 1.1029411764705884e-06,
-      "loss": 0.8313,
+      "loss": 1.3034,
       "step": 2660
     },
     {
       "epoch": 19.51,
       "learning_rate": 1.0661764705882354e-06,
-      "loss": 0.803,
+      "loss": 1.2829,
       "step": 2662
     },
     {
       "epoch": 19.53,
       "learning_rate": 1.0294117647058825e-06,
-      "loss": 0.7748,
+      "loss": 1.2802,
       "step": 2664
     },
     {
       "epoch": 19.54,
       "learning_rate": 9.926470588235293e-07,
-      "loss": 0.7863,
+      "loss": 1.2556,
       "step": 2666
     },
     {
       "epoch": 19.55,
       "learning_rate": 9.558823529411766e-07,
-      "loss": 0.838,
+      "loss": 1.3106,
       "step": 2668
     },
     {
       "epoch": 19.57,
       "learning_rate": 9.191176470588236e-07,
-      "loss": 0.7913,
+      "loss": 1.2525,
       "step": 2670
     },
     {
       "epoch": 19.58,
       "learning_rate": 8.823529411764706e-07,
-      "loss": 0.8057,
+      "loss": 1.2987,
       "step": 2672
     },
     {
       "epoch": 19.6,
       "learning_rate": 8.455882352941177e-07,
-      "loss": 0.8014,
+      "loss": 1.2641,
       "step": 2674
     },
     {
       "epoch": 19.61,
       "learning_rate": 8.088235294117648e-07,
-      "loss": 0.8081,
+      "loss": 1.2668,
       "step": 2676
     },
     {
       "epoch": 19.63,
       "learning_rate": 7.720588235294118e-07,
-      "loss": 0.7946,
+      "loss": 1.3005,
       "step": 2678
     },
     {
       "epoch": 19.64,
       "learning_rate": 7.352941176470589e-07,
-      "loss": 0.7908,
+      "loss": 1.282,
       "step": 2680
     },
     {
       "epoch": 19.66,
       "learning_rate": 6.985294117647059e-07,
-      "loss": 0.7736,
+      "loss": 1.2671,
       "step": 2682
     },
     {
       "epoch": 19.67,
       "learning_rate": 6.61764705882353e-07,
-      "loss": 0.7925,
+      "loss": 1.2342,
       "step": 2684
     },
     {
       "epoch": 19.69,
       "learning_rate": 6.25e-07,
-      "loss": 0.7959,
+      "loss": 1.2571,
       "step": 2686
     },
     {
       "epoch": 19.7,
       "learning_rate": 5.882352941176471e-07,
-      "loss": 0.7814,
+      "loss": 1.3025,
       "step": 2688
     },
     {
       "epoch": 19.72,
       "learning_rate": 5.514705882352942e-07,
-      "loss": 0.7872,
+      "loss": 1.2558,
       "step": 2690
     },
     {
       "epoch": 19.73,
       "learning_rate": 5.147058823529412e-07,
-      "loss": 0.804,
+      "loss": 1.2856,
       "step": 2692
     },
     {
       "epoch": 19.75,
       "learning_rate": 4.779411764705883e-07,
-      "loss": 0.7901,
+      "loss": 1.2436,
       "step": 2694
     },
     {
       "epoch": 19.76,
       "learning_rate": 4.411764705882353e-07,
-      "loss": 0.7893,
+      "loss": 1.2663,
       "step": 2696
     },
     {
       "epoch": 19.77,
       "learning_rate": 4.044117647058824e-07,
-      "loss": 0.7764,
+      "loss": 1.251,
       "step": 2698
     },
     {
       "epoch": 19.79,
       "learning_rate": 3.6764705882352943e-07,
-      "loss": 0.8344,
+      "loss": 1.2936,
       "step": 2700
     },
     {
       "epoch": 19.8,
       "learning_rate": 3.308823529411765e-07,
-      "loss": 0.8093,
+      "loss": 1.2658,
       "step": 2702
     },
     {
       "epoch": 19.82,
       "learning_rate": 2.9411764705882356e-07,
-      "loss": 0.7541,
+      "loss": 1.2671,
       "step": 2704
     },
     {
       "epoch": 19.83,
       "learning_rate": 2.573529411764706e-07,
-      "loss": 0.8165,
+      "loss": 1.272,
       "step": 2706
     },
     {
       "epoch": 19.85,
       "learning_rate": 2.2058823529411765e-07,
-      "loss": 0.8012,
+      "loss": 1.2926,
       "step": 2708
     },
     {
       "epoch": 19.86,
       "learning_rate": 1.8382352941176472e-07,
-      "loss": 0.7605,
+      "loss": 1.2426,
       "step": 2710
     },
     {
       "epoch": 19.88,
       "learning_rate": 1.4705882352941178e-07,
-      "loss": 0.8056,
+      "loss": 1.301,
       "step": 2712
     },
     {
       "epoch": 19.89,
       "learning_rate": 1.1029411764705883e-07,
-      "loss": 0.8107,
+      "loss": 1.324,
       "step": 2714
     },
     {
       "epoch": 19.91,
       "learning_rate": 7.352941176470589e-08,
-      "loss": 0.7975,
+      "loss": 1.3007,
       "step": 2716
     },
     {
       "epoch": 19.92,
       "learning_rate": 3.6764705882352945e-08,
-      "loss": 0.8409,
+      "loss": 1.29,
       "step": 2718
     },
     {
       "epoch": 19.94,
       "learning_rate": 0.0,
-      "loss": 0.792,
+      "loss": 1.2936,
       "step": 2720
     },
     {
       "epoch": 19.94,
-      "eval_gen_len": 128.74074074074073,
-      "eval_loss": 1.6327582597732544,
-      "eval_rouge1": 55.8983,
-      "eval_rouge2": 30.5018,
-      "eval_rougeL": 38.7764,
-      "eval_rougeLsum": 51.3611,
-      "eval_runtime": 216.9358,
-      "eval_samples_per_second": 4.481,
-      "eval_steps_per_second": 0.562,
+      "eval_gen_len": 744.4588477366256,
+      "eval_loss": 1.4350636005401611,
+      "eval_rouge1": 72.0397,
+      "eval_rouge2": 40.9431,
+      "eval_rougeL": 38.4161,
+      "eval_rougeLsum": 69.5028,
+      "eval_runtime": 1636.79,
+      "eval_samples_per_second": 0.594,
+      "eval_steps_per_second": 0.075,
       "step": 2720
     },
     {
       "epoch": 19.94,
       "step": 2720,
       "total_flos": 7.21120699038892e+17,
-      "train_loss": 1.0796797680504182,
-      "train_runtime": 51892.12,
-      "train_samples_per_second": 6.728,
-      "train_steps_per_second": 0.052
+      "train_loss": 1.4114765528370352,
+      "train_runtime": 122523.8715,
+      "train_samples_per_second": 2.85,
+      "train_steps_per_second": 0.022
     }
   ],
   "logging_steps": 2,