diff --git "a/checkpoint-7500/trainer_state.json" "b/checkpoint-7500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-7500/trainer_state.json"
@@ -0,0 +1,5183 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 389.61038961038963,
+  "eval_steps": 500,
+  "global_step": 7500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.99,
+      "eval_gen_len": 14.3217,
+      "eval_loss": 21.81446075439453,
+      "eval_rouge1": 0.0662,
+      "eval_rouge2": 0.0082,
+      "eval_rougeL": 0.0523,
+      "eval_rougeLsum": 0.0526,
+      "eval_runtime": 12.9343,
+      "eval_samples_per_second": 8.891,
+      "eval_steps_per_second": 1.546,
+      "step": 19
+    },
+    {
+      "epoch": 1.97,
+      "eval_gen_len": 14.4,
+      "eval_loss": 21.760791778564453,
+      "eval_rouge1": 0.0661,
+      "eval_rouge2": 0.0082,
+      "eval_rougeL": 0.052,
+      "eval_rougeLsum": 0.0523,
+      "eval_runtime": 11.8223,
+      "eval_samples_per_second": 9.727,
+      "eval_steps_per_second": 1.692,
+      "step": 38
+    },
+    {
+      "epoch": 2.96,
+      "eval_gen_len": 14.4,
+      "eval_loss": 21.68472671508789,
+      "eval_rouge1": 0.0651,
+      "eval_rouge2": 0.0077,
+      "eval_rougeL": 0.051,
+      "eval_rougeLsum": 0.0513,
+      "eval_runtime": 12.1423,
+      "eval_samples_per_second": 9.471,
+      "eval_steps_per_second": 1.647,
+      "step": 57
+    },
+    {
+      "epoch": 4.0,
+      "eval_gen_len": 14.3304,
+      "eval_loss": 21.582918167114258,
+      "eval_rouge1": 0.0666,
+      "eval_rouge2": 0.0081,
+      "eval_rougeL": 0.0525,
+      "eval_rougeLsum": 0.0525,
+      "eval_runtime": 12.0406,
+      "eval_samples_per_second": 9.551,
+      "eval_steps_per_second": 1.661,
+      "step": 77
+    },
+    {
+      "epoch": 4.99,
+      "eval_gen_len": 14.2609,
+      "eval_loss": 21.46294403076172,
+      "eval_rouge1": 0.0668,
+      "eval_rouge2": 0.008,
+      "eval_rougeL": 0.0527,
+      "eval_rougeLsum": 0.0526,
+      "eval_runtime": 12.3725,
+      "eval_samples_per_second": 9.295,
+      "eval_steps_per_second": 1.616,
+      "step": 96
+    },
+    {
+      "epoch": 5.97,
+      "eval_gen_len": 14.2261,
+      "eval_loss": 21.31937599182129,
+      "eval_rouge1": 0.0671,
+      "eval_rouge2": 0.0077,
+      "eval_rougeL": 0.0525,
+      "eval_rougeLsum": 0.0525,
+      "eval_runtime": 12.5289,
+      "eval_samples_per_second": 9.179,
+      "eval_steps_per_second": 1.596,
+      "step": 115
+    },
+    {
+      "epoch": 6.96,
+      "eval_gen_len": 14.3043,
+      "eval_loss": 21.15254783630371,
+      "eval_rouge1": 0.0677,
+      "eval_rouge2": 0.0077,
+      "eval_rougeL": 0.0525,
+      "eval_rougeLsum": 0.0525,
+      "eval_runtime": 12.0936,
+      "eval_samples_per_second": 9.509,
+      "eval_steps_per_second": 1.654,
+      "step": 134
+    },
+    {
+      "epoch": 8.0,
+      "eval_gen_len": 14.2609,
+      "eval_loss": 20.956642150878906,
+      "eval_rouge1": 0.0679,
+      "eval_rouge2": 0.0076,
+      "eval_rougeL": 0.0522,
+      "eval_rougeLsum": 0.0521,
+      "eval_runtime": 12.6351,
+      "eval_samples_per_second": 9.102,
+      "eval_steps_per_second": 1.583,
+      "step": 154
+    },
+    {
+      "epoch": 8.99,
+      "eval_gen_len": 14.1217,
+      "eval_loss": 20.7694091796875,
+      "eval_rouge1": 0.0664,
+      "eval_rouge2": 0.0074,
+      "eval_rougeL": 0.0507,
+      "eval_rougeLsum": 0.0507,
+      "eval_runtime": 12.3661,
+      "eval_samples_per_second": 9.3,
+      "eval_steps_per_second": 1.617,
+      "step": 173
+    },
+    {
+      "epoch": 9.97,
+      "eval_gen_len": 14.0957,
+      "eval_loss": 20.60512351989746,
+      "eval_rouge1": 0.0685,
+      "eval_rouge2": 0.0078,
+      "eval_rougeL": 0.0519,
+      "eval_rougeLsum": 0.0518,
+      "eval_runtime": 12.6094,
+      "eval_samples_per_second": 9.12,
+      "eval_steps_per_second": 1.586,
+      "step": 192
+    },
+    {
+      "epoch": 10.96,
+      "eval_gen_len": 13.9913,
+      "eval_loss": 20.46657371520996,
+      "eval_rouge1": 0.0672,
+      "eval_rouge2": 0.0072,
+      "eval_rougeL": 0.0511,
+      "eval_rougeLsum": 0.0511,
+      "eval_runtime": 12.4343,
+      "eval_samples_per_second": 9.249,
+      "eval_steps_per_second": 1.608,
+      "step": 211
+    },
+    {
+      "epoch": 12.0,
+      "eval_gen_len": 14.2957,
+      "eval_loss": 20.335567474365234,
+      "eval_rouge1": 0.0658,
+      "eval_rouge2": 0.0079,
+      "eval_rougeL": 0.05,
+      "eval_rougeLsum": 0.05,
+      "eval_runtime": 12.4623,
+      "eval_samples_per_second": 9.228,
+      "eval_steps_per_second": 1.605,
+      "step": 231
+    },
+    {
+      "epoch": 12.99,
+      "eval_gen_len": 14.7826,
+      "eval_loss": 20.210546493530273,
+      "eval_rouge1": 0.0729,
+      "eval_rouge2": 0.0108,
+      "eval_rougeL": 0.0562,
+      "eval_rougeLsum": 0.056,
+      "eval_runtime": 12.6201,
+      "eval_samples_per_second": 9.112,
+      "eval_steps_per_second": 1.585,
+      "step": 250
+    },
+    {
+      "epoch": 13.97,
+      "eval_gen_len": 14.8174,
+      "eval_loss": 20.080350875854492,
+      "eval_rouge1": 0.0731,
+      "eval_rouge2": 0.0105,
+      "eval_rougeL": 0.057,
+      "eval_rougeLsum": 0.0566,
+      "eval_runtime": 12.5892,
+      "eval_samples_per_second": 9.135,
+      "eval_steps_per_second": 1.589,
+      "step": 269
+    },
+    {
+      "epoch": 14.96,
+      "eval_gen_len": 15.3826,
+      "eval_loss": 19.947362899780273,
+      "eval_rouge1": 0.0805,
+      "eval_rouge2": 0.0127,
+      "eval_rougeL": 0.0624,
+      "eval_rougeLsum": 0.0619,
+      "eval_runtime": 12.2446,
+      "eval_samples_per_second": 9.392,
+      "eval_steps_per_second": 1.633,
+      "step": 288
+    },
+    {
+      "epoch": 16.0,
+      "eval_gen_len": 15.8261,
+      "eval_loss": 19.80646324157715,
+      "eval_rouge1": 0.0818,
+      "eval_rouge2": 0.0129,
+      "eval_rougeL": 0.0636,
+      "eval_rougeLsum": 0.0633,
+      "eval_runtime": 12.037,
+      "eval_samples_per_second": 9.554,
+      "eval_steps_per_second": 1.662,
+      "step": 308
+    },
+    {
+      "epoch": 16.99,
+      "eval_gen_len": 16.1391,
+      "eval_loss": 19.66917610168457,
+      "eval_rouge1": 0.0837,
+      "eval_rouge2": 0.0139,
+      "eval_rougeL": 0.0648,
+      "eval_rougeLsum": 0.0647,
+      "eval_runtime": 12.6734,
+      "eval_samples_per_second": 9.074,
+      "eval_steps_per_second": 1.578,
+      "step": 327
+    },
+    {
+      "epoch": 17.97,
+      "eval_gen_len": 16.5391,
+      "eval_loss": 19.5202579498291,
+      "eval_rouge1": 0.0915,
+      "eval_rouge2": 0.0168,
+      "eval_rougeL": 0.0713,
+      "eval_rougeLsum": 0.0709,
+      "eval_runtime": 12.3571,
+      "eval_samples_per_second": 9.306,
+      "eval_steps_per_second": 1.619,
+      "step": 346
+    },
+    {
+      "epoch": 18.96,
+      "eval_gen_len": 16.687,
+      "eval_loss": 19.36480712890625,
+      "eval_rouge1": 0.0934,
+      "eval_rouge2": 0.0178,
+      "eval_rougeL": 0.072,
+      "eval_rougeLsum": 0.0719,
+      "eval_runtime": 12.3428,
+      "eval_samples_per_second": 9.317,
+      "eval_steps_per_second": 1.62,
+      "step": 365
+    },
+    {
+      "epoch": 20.0,
+      "eval_gen_len": 17.0522,
+      "eval_loss": 19.185970306396484,
+      "eval_rouge1": 0.0951,
+      "eval_rouge2": 0.0172,
+      "eval_rougeL": 0.0735,
+      "eval_rougeLsum": 0.0735,
+      "eval_runtime": 12.4924,
+      "eval_samples_per_second": 9.206,
+      "eval_steps_per_second": 1.601,
+      "step": 385
+    },
+    {
+      "epoch": 20.99,
+      "eval_gen_len": 17.1913,
+      "eval_loss": 18.998315811157227,
+      "eval_rouge1": 0.0938,
+      "eval_rouge2": 0.0192,
+      "eval_rougeL": 0.0754,
+      "eval_rougeLsum": 0.0755,
+      "eval_runtime": 12.2427,
+      "eval_samples_per_second": 9.393,
+      "eval_steps_per_second": 1.634,
+      "step": 404
+    },
+    {
+      "epoch": 21.97,
+      "eval_gen_len": 17.3304,
+      "eval_loss": 18.782604217529297,
+      "eval_rouge1": 0.0975,
+      "eval_rouge2": 0.0223,
+      "eval_rougeL": 0.0786,
+      "eval_rougeLsum": 0.0788,
+      "eval_runtime": 12.4341,
+      "eval_samples_per_second": 9.249,
+      "eval_steps_per_second": 1.608,
+      "step": 423
+    },
+    {
+      "epoch": 22.96,
+      "eval_gen_len": 17.3304,
+      "eval_loss": 18.529996871948242,
+      "eval_rouge1": 0.0986,
+      "eval_rouge2": 0.0229,
+      "eval_rougeL": 0.0787,
+      "eval_rougeLsum": 0.079,
+      "eval_runtime": 12.138,
+      "eval_samples_per_second": 9.474,
+      "eval_steps_per_second": 1.648,
+      "step": 442
+    },
+    {
+      "epoch": 24.0,
+      "eval_gen_len": 17.4696,
+      "eval_loss": 18.21288299560547,
+      "eval_rouge1": 0.0935,
+      "eval_rouge2": 0.0195,
+      "eval_rougeL": 0.0761,
+      "eval_rougeLsum": 0.0763,
+      "eval_runtime": 12.3892,
+      "eval_samples_per_second": 9.282,
+      "eval_steps_per_second": 1.614,
+      "step": 462
+    },
+    {
+      "epoch": 24.99,
+      "eval_gen_len": 17.6087,
+      "eval_loss": 17.844209671020508,
+      "eval_rouge1": 0.0936,
+      "eval_rouge2": 0.0225,
+      "eval_rougeL": 0.0756,
+      "eval_rougeLsum": 0.0758,
+      "eval_runtime": 12.2522,
+      "eval_samples_per_second": 9.386,
+      "eval_steps_per_second": 1.632,
+      "step": 481
+    },
+    {
+      "epoch": 25.97,
+      "grad_norm": 4.415005683898926,
+      "learning_rate": 9.474736842105265e-06,
+      "loss": 19.6383,
+      "step": 500
+    },
+    {
+      "epoch": 25.97,
+      "eval_gen_len": 17.7478,
+      "eval_loss": 17.39573097229004,
+      "eval_rouge1": 0.0967,
+      "eval_rouge2": 0.0221,
+      "eval_rougeL": 0.0765,
+      "eval_rougeLsum": 0.0764,
+      "eval_runtime": 12.5376,
+      "eval_samples_per_second": 9.172,
+      "eval_steps_per_second": 1.595,
+      "step": 500
+    },
+    {
+      "epoch": 26.96,
+      "eval_gen_len": 17.7478,
+      "eval_loss": 16.90059471130371,
+      "eval_rouge1": 0.0983,
+      "eval_rouge2": 0.0198,
+      "eval_rougeL": 0.0786,
+      "eval_rougeLsum": 0.0785,
+      "eval_runtime": 12.1506,
+      "eval_samples_per_second": 9.465,
+      "eval_steps_per_second": 1.646,
+      "step": 519
+    },
+    {
+      "epoch": 28.0,
+      "eval_gen_len": 17.3304,
+      "eval_loss": 16.37845802307129,
+      "eval_rouge1": 0.0916,
+      "eval_rouge2": 0.0162,
+      "eval_rougeL": 0.0723,
+      "eval_rougeLsum": 0.0724,
+      "eval_runtime": 12.9559,
+      "eval_samples_per_second": 8.876,
+      "eval_steps_per_second": 1.544,
+      "step": 539
+    },
+    {
+      "epoch": 28.99,
+      "eval_gen_len": 17.0522,
+      "eval_loss": 15.880877494812012,
+      "eval_rouge1": 0.0898,
+      "eval_rouge2": 0.0212,
+      "eval_rougeL": 0.0716,
+      "eval_rougeLsum": 0.0711,
+      "eval_runtime": 12.3917,
+      "eval_samples_per_second": 9.28,
+      "eval_steps_per_second": 1.614,
+      "step": 558
+    },
+    {
+      "epoch": 29.97,
+      "eval_gen_len": 17.6087,
+      "eval_loss": 15.38318920135498,
+      "eval_rouge1": 0.09,
+      "eval_rouge2": 0.0179,
+      "eval_rougeL": 0.0717,
+      "eval_rougeLsum": 0.0717,
+      "eval_runtime": 12.4102,
+      "eval_samples_per_second": 9.267,
+      "eval_steps_per_second": 1.612,
+      "step": 577
+    },
+    {
+      "epoch": 30.96,
+      "eval_gen_len": 17.6087,
+      "eval_loss": 14.880407333374023,
+      "eval_rouge1": 0.0921,
+      "eval_rouge2": 0.02,
+      "eval_rougeL": 0.0768,
+      "eval_rougeLsum": 0.0765,
+      "eval_runtime": 12.1352,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 1.648,
+      "step": 596
+    },
+    {
+      "epoch": 32.0,
+      "eval_gen_len": 17.7478,
+      "eval_loss": 14.299490928649902,
+      "eval_rouge1": 0.0856,
+      "eval_rouge2": 0.0197,
+      "eval_rougeL": 0.0713,
+      "eval_rougeLsum": 0.071,
+      "eval_runtime": 12.7436,
+      "eval_samples_per_second": 9.024,
+      "eval_steps_per_second": 1.569,
+      "step": 616
+    },
+    {
+      "epoch": 32.99,
+      "eval_gen_len": 17.0522,
+      "eval_loss": 13.684185981750488,
+      "eval_rouge1": 0.0796,
+      "eval_rouge2": 0.0153,
+      "eval_rougeL": 0.0627,
+      "eval_rougeLsum": 0.0626,
+      "eval_runtime": 12.5048,
+      "eval_samples_per_second": 9.196,
+      "eval_steps_per_second": 1.599,
+      "step": 635
+    },
+    {
+      "epoch": 33.97,
+      "eval_gen_len": 17.3304,
+      "eval_loss": 13.098108291625977,
+      "eval_rouge1": 0.0772,
+      "eval_rouge2": 0.0098,
+      "eval_rougeL": 0.0646,
+      "eval_rougeLsum": 0.0643,
+      "eval_runtime": 12.1222,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 1.65,
+      "step": 654
+    },
+    {
+      "epoch": 34.96,
+      "eval_gen_len": 17.4696,
+      "eval_loss": 12.545892715454102,
+      "eval_rouge1": 0.0751,
+      "eval_rouge2": 0.0107,
+      "eval_rougeL": 0.0622,
+      "eval_rougeLsum": 0.0622,
+      "eval_runtime": 12.1398,
+      "eval_samples_per_second": 9.473,
+      "eval_steps_per_second": 1.647,
+      "step": 673
+    },
+    {
+      "epoch": 36.0,
+      "eval_gen_len": 17.3304,
+      "eval_loss": 12.004250526428223,
+      "eval_rouge1": 0.0704,
+      "eval_rouge2": 0.0081,
+      "eval_rougeL": 0.0588,
+      "eval_rougeLsum": 0.0588,
+      "eval_runtime": 12.5092,
+      "eval_samples_per_second": 9.193,
+      "eval_steps_per_second": 1.599,
+      "step": 693
+    },
+    {
+      "epoch": 36.99,
+      "eval_gen_len": 17.887,
+      "eval_loss": 11.514721870422363,
+      "eval_rouge1": 0.0672,
+      "eval_rouge2": 0.0079,
+      "eval_rougeL": 0.0577,
+      "eval_rougeLsum": 0.0577,
+      "eval_runtime": 12.4809,
+      "eval_samples_per_second": 9.214,
+      "eval_steps_per_second": 1.602,
+      "step": 712
+    },
+    {
+      "epoch": 37.97,
+      "eval_gen_len": 17.6087,
+      "eval_loss": 11.038866996765137,
+      "eval_rouge1": 0.0513,
+      "eval_rouge2": 0.0045,
+      "eval_rougeL": 0.046,
+      "eval_rougeLsum": 0.0457,
+      "eval_runtime": 12.6414,
+      "eval_samples_per_second": 9.097,
+      "eval_steps_per_second": 1.582,
+      "step": 731
+    },
+    {
+      "epoch": 38.96,
+      "eval_gen_len": 17.1913,
+      "eval_loss": 10.570833206176758,
+      "eval_rouge1": 0.0468,
+      "eval_rouge2": 0.0029,
+      "eval_rougeL": 0.0404,
+      "eval_rougeLsum": 0.0403,
+      "eval_runtime": 12.7674,
+      "eval_samples_per_second": 9.007,
+      "eval_steps_per_second": 1.566,
+      "step": 750
+    },
+    {
+      "epoch": 40.0,
+      "eval_gen_len": 17.3565,
+      "eval_loss": 10.076028823852539,
+      "eval_rouge1": 0.0352,
+      "eval_rouge2": 0.0037,
+      "eval_rougeL": 0.0305,
+      "eval_rougeLsum": 0.0304,
+      "eval_runtime": 12.8987,
+      "eval_samples_per_second": 8.916,
+      "eval_steps_per_second": 1.551,
+      "step": 770
+    },
+    {
+      "epoch": 40.99,
+      "eval_gen_len": 17.4957,
+      "eval_loss": 9.603371620178223,
+      "eval_rouge1": 0.024,
+      "eval_rouge2": 0.0024,
+      "eval_rougeL": 0.0218,
+      "eval_rougeLsum": 0.0218,
+      "eval_runtime": 12.2083,
+      "eval_samples_per_second": 9.42,
+      "eval_steps_per_second": 1.638,
+      "step": 789
+    },
+    {
+      "epoch": 41.97,
+      "eval_gen_len": 18.3043,
+      "eval_loss": 9.131211280822754,
+      "eval_rouge1": 0.0193,
+      "eval_rouge2": 0.0022,
+      "eval_rougeL": 0.0184,
+      "eval_rougeLsum": 0.0185,
+      "eval_runtime": 12.1389,
+      "eval_samples_per_second": 9.474,
+      "eval_steps_per_second": 1.648,
+      "step": 808
+    },
+    {
+      "epoch": 42.96,
+      "eval_gen_len": 18.7217,
+      "eval_loss": 8.668445587158203,
+      "eval_rouge1": 0.0116,
+      "eval_rouge2": 0.0013,
+      "eval_rougeL": 0.0109,
+      "eval_rougeLsum": 0.0108,
+      "eval_runtime": 12.4712,
+      "eval_samples_per_second": 9.221,
+      "eval_steps_per_second": 1.604,
+      "step": 827
+    },
+    {
+      "epoch": 44.0,
+      "eval_gen_len": 18.8609,
+      "eval_loss": 8.1836576461792,
+      "eval_rouge1": 0.0031,
+      "eval_rouge2": 0.0004,
+      "eval_rougeL": 0.0032,
+      "eval_rougeLsum": 0.0032,
+      "eval_runtime": 12.1395,
+      "eval_samples_per_second": 9.473,
+      "eval_steps_per_second": 1.648,
+      "step": 847
+    },
+    {
+      "epoch": 44.99,
+      "eval_gen_len": 18.8609,
+      "eval_loss": 7.736245632171631,
+      "eval_rouge1": 0.0028,
+      "eval_rouge2": 0.0002,
+      "eval_rougeL": 0.0028,
+      "eval_rougeLsum": 0.0028,
+      "eval_runtime": 12.2043,
+      "eval_samples_per_second": 9.423,
+      "eval_steps_per_second": 1.639,
+      "step": 866
+    },
+    {
+      "epoch": 45.97,
+      "eval_gen_len": 18.8609,
+      "eval_loss": 7.298835277557373,
+      "eval_rouge1": 0.0018,
+      "eval_rouge2": 0.0004,
+      "eval_rougeL": 0.0018,
+      "eval_rougeLsum": 0.0018,
+      "eval_runtime": 12.1301,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 1.649,
+      "step": 885
+    },
+    {
+      "epoch": 46.96,
+      "eval_gen_len": 19.0,
+      "eval_loss": 6.873920917510986,
+      "eval_rouge1": 0.001,
+      "eval_rouge2": 0.0002,
+      "eval_rougeL": 0.001,
+      "eval_rougeLsum": 0.0009,
+      "eval_runtime": 12.1519,
+      "eval_samples_per_second": 9.464,
+      "eval_steps_per_second": 1.646,
+      "step": 904
+    },
+    {
+      "epoch": 48.0,
+      "eval_gen_len": 19.0,
+      "eval_loss": 6.440176486968994,
+      "eval_rouge1": 0.0001,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0001,
+      "eval_rougeLsum": 0.0001,
+      "eval_runtime": 12.3618,
+      "eval_samples_per_second": 9.303,
+      "eval_steps_per_second": 1.618,
+      "step": 924
+    },
+    {
+      "epoch": 48.99,
+      "eval_gen_len": 19.0,
+      "eval_loss": 6.049317359924316,
+      "eval_rouge1": 0.0006,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0006,
+      "eval_rougeLsum": 0.0005,
+      "eval_runtime": 12.167,
+      "eval_samples_per_second": 9.452,
+      "eval_steps_per_second": 1.644,
+      "step": 943
+    },
+    {
+      "epoch": 49.97,
+      "eval_gen_len": 19.0,
+      "eval_loss": 5.643195152282715,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 11.9138,
+      "eval_samples_per_second": 9.653,
+      "eval_steps_per_second": 1.679,
+      "step": 962
+    },
+    {
+      "epoch": 50.96,
+      "eval_gen_len": 19.0,
+      "eval_loss": 5.2581987380981445,
+      "eval_rouge1": 0.0007,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0008,
+      "eval_rougeLsum": 0.0007,
+      "eval_runtime": 12.5979,
+      "eval_samples_per_second": 9.129,
+      "eval_steps_per_second": 1.588,
+      "step": 981
+    },
+    {
+      "epoch": 51.95,
+      "grad_norm": 5.70858097076416,
+      "learning_rate": 8.949473684210527e-06,
+      "loss": 11.5478,
+      "step": 1000
+    },
+    {
+      "epoch": 52.0,
+      "eval_gen_len": 19.0,
+      "eval_loss": 4.877782344818115,
+      "eval_rouge1": 0.0006,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0006,
+      "eval_rougeLsum": 0.0006,
+      "eval_runtime": 12.0728,
+      "eval_samples_per_second": 9.526,
+      "eval_steps_per_second": 1.657,
+      "step": 1001
+    },
+    {
+      "epoch": 52.99,
+      "eval_gen_len": 19.0,
+      "eval_loss": 4.53688383102417,
+      "eval_rouge1": 0.0006,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0006,
+      "eval_rougeLsum": 0.0006,
+      "eval_runtime": 12.1263,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 1.649,
+      "step": 1020
+    },
+    {
+      "epoch": 53.97,
+      "eval_gen_len": 19.0,
+      "eval_loss": 4.222665309906006,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.5806,
+      "eval_samples_per_second": 9.141,
+      "eval_steps_per_second": 1.59,
+      "step": 1039
+    },
+    {
+      "epoch": 54.96,
+      "eval_gen_len": 19.0,
+      "eval_loss": 3.9297854900360107,
+      "eval_rouge1": 0.0008,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0008,
+      "eval_rougeLsum": 0.0007,
+      "eval_runtime": 12.7467,
+      "eval_samples_per_second": 9.022,
+      "eval_steps_per_second": 1.569,
+      "step": 1058
+    },
+    {
+      "epoch": 56.0,
+      "eval_gen_len": 19.0,
+      "eval_loss": 3.6505942344665527,
+      "eval_rouge1": 0.0005,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0005,
+      "eval_rougeLsum": 0.0005,
+      "eval_runtime": 12.6962,
+      "eval_samples_per_second": 9.058,
+      "eval_steps_per_second": 1.575,
+      "step": 1078
+    },
+    {
+      "epoch": 56.99,
+      "eval_gen_len": 19.0,
+      "eval_loss": 3.4100279808044434,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.5475,
+      "eval_samples_per_second": 9.165,
+      "eval_steps_per_second": 1.594,
+      "step": 1097
+    },
+    {
+      "epoch": 57.97,
+      "eval_gen_len": 19.0,
+      "eval_loss": 3.197094202041626,
+      "eval_rouge1": 0.0006,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0006,
+      "eval_rougeLsum": 0.0006,
+      "eval_runtime": 12.1721,
+      "eval_samples_per_second": 9.448,
+      "eval_steps_per_second": 1.643,
+      "step": 1116
+    },
+    {
+      "epoch": 58.96,
+      "eval_gen_len": 18.9913,
+      "eval_loss": 3.0094308853149414,
+      "eval_rouge1": 0.0006,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0006,
+      "eval_rougeLsum": 0.0006,
+      "eval_runtime": 12.2313,
+      "eval_samples_per_second": 9.402,
+      "eval_steps_per_second": 1.635,
+      "step": 1135
+    },
+    {
+      "epoch": 60.0,
+      "eval_gen_len": 18.9913,
+      "eval_loss": 2.841013193130493,
+      "eval_rouge1": 0.0008,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0008,
+      "eval_rougeLsum": 0.0008,
+      "eval_runtime": 12.5511,
+      "eval_samples_per_second": 9.163,
+      "eval_steps_per_second": 1.593,
+      "step": 1155
+    },
+    {
+      "epoch": 60.99,
+      "eval_gen_len": 18.9826,
+      "eval_loss": 2.698159694671631,
+      "eval_rouge1": 0.0007,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0007,
+      "eval_rougeLsum": 0.0007,
+      "eval_runtime": 12.0968,
+      "eval_samples_per_second": 9.507,
+      "eval_steps_per_second": 1.653,
+      "step": 1174
+    },
+    {
+      "epoch": 61.97,
+      "eval_gen_len": 18.6783,
+      "eval_loss": 2.5645217895507812,
+      "eval_rouge1": 0.0011,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0011,
+      "eval_rougeLsum": 0.0011,
+      "eval_runtime": 12.1821,
+      "eval_samples_per_second": 9.44,
+      "eval_steps_per_second": 1.642,
+      "step": 1193
+    },
+    {
+      "epoch": 62.96,
+      "eval_gen_len": 17.3565,
+      "eval_loss": 2.445538282394409,
+      "eval_rouge1": 0.0007,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0007,
+      "eval_rougeLsum": 0.0007,
+      "eval_runtime": 12.1339,
+      "eval_samples_per_second": 9.478,
+      "eval_steps_per_second": 1.648,
+      "step": 1212
+    },
+    {
+      "epoch": 64.0,
+      "eval_gen_len": 13.5043,
+      "eval_loss": 2.338679075241089,
+      "eval_rouge1": 0.0011,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.001,
+      "eval_rougeLsum": 0.001,
+      "eval_runtime": 12.4085,
+      "eval_samples_per_second": 9.268,
+      "eval_steps_per_second": 1.612,
+      "step": 1232
+    },
+    {
+      "epoch": 64.99,
+      "eval_gen_len": 9.4348,
+      "eval_loss": 2.2483484745025635,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.3896,
+      "eval_samples_per_second": 9.282,
+      "eval_steps_per_second": 1.614,
+      "step": 1251
+    },
+    {
+      "epoch": 65.97,
+      "eval_gen_len": 6.9652,
+      "eval_loss": 2.1728155612945557,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4531,
+      "eval_samples_per_second": 9.235,
+      "eval_steps_per_second": 1.606,
+      "step": 1270
+    },
+    {
+      "epoch": 66.96,
+      "eval_gen_len": 6.2957,
+      "eval_loss": 2.1103546619415283,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2925,
+      "eval_samples_per_second": 9.355,
+      "eval_steps_per_second": 1.627,
+      "step": 1289
+    },
+    {
+      "epoch": 68.0,
+      "eval_gen_len": 6.0,
+      "eval_loss": 2.0531256198883057,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.1269,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 1.649,
+      "step": 1309
+    },
+    {
+      "epoch": 68.99,
+      "eval_gen_len": 5.7043,
+      "eval_loss": 2.006763219833374,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.0803,
+      "eval_samples_per_second": 9.52,
+      "eval_steps_per_second": 1.656,
+      "step": 1328
+    },
+    {
+      "epoch": 69.97,
+      "eval_gen_len": 5.6609,
+      "eval_loss": 1.9675697088241577,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 13.197,
+      "eval_samples_per_second": 8.714,
+      "eval_steps_per_second": 1.515,
+      "step": 1347
+    },
+    {
+      "epoch": 70.96,
+      "eval_gen_len": 5.6,
+      "eval_loss": 1.9337714910507202,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.696,
+      "eval_samples_per_second": 9.058,
+      "eval_steps_per_second": 1.575,
+      "step": 1366
+    },
+    {
+      "epoch": 72.0,
+      "eval_gen_len": 5.6174,
+      "eval_loss": 1.9011404514312744,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5629,
+      "eval_samples_per_second": 9.154,
+      "eval_steps_per_second": 1.592,
+      "step": 1386
+    },
+    {
+      "epoch": 72.99,
+      "eval_gen_len": 5.6435,
+      "eval_loss": 1.8734184503555298,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.9654,
+      "eval_samples_per_second": 8.87,
+      "eval_steps_per_second": 1.543,
+      "step": 1405
+    },
+    {
+      "epoch": 73.97,
+      "eval_gen_len": 5.7739,
+      "eval_loss": 1.84665846824646,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.8257,
+      "eval_samples_per_second": 8.966,
+      "eval_steps_per_second": 1.559,
+      "step": 1424
+    },
+    {
+      "epoch": 74.96,
+      "eval_gen_len": 5.7478,
+      "eval_loss": 1.8196372985839844,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.89,
+      "eval_samples_per_second": 8.922,
+      "eval_steps_per_second": 1.552,
+      "step": 1443
+    },
+    {
+      "epoch": 76.0,
+      "eval_gen_len": 5.7217,
+      "eval_loss": 1.797453761100769,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4957,
+      "eval_samples_per_second": 9.203,
+      "eval_steps_per_second": 1.601,
+      "step": 1463
+    },
+    {
+      "epoch": 76.99,
+      "eval_gen_len": 5.8174,
+      "eval_loss": 1.7788159847259521,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4326,
+      "eval_samples_per_second": 9.25,
+      "eval_steps_per_second": 1.609,
+      "step": 1482
+    },
+    {
+      "epoch": 77.92,
+      "grad_norm": 2.0204899311065674,
+      "learning_rate": 8.42421052631579e-06,
+      "loss": 3.2357,
+      "step": 1500
+    },
+    {
+      "epoch": 77.97,
+      "eval_gen_len": 5.8957,
+      "eval_loss": 1.76212739944458,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5086,
+      "eval_samples_per_second": 9.194,
+      "eval_steps_per_second": 1.599,
+      "step": 1501
+    },
+    {
+      "epoch": 78.96,
+      "eval_gen_len": 5.8957,
+      "eval_loss": 1.744727373123169,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3492,
+      "eval_samples_per_second": 9.312,
+      "eval_steps_per_second": 1.62,
+      "step": 1520
+    },
+    {
+      "epoch": 80.0,
+      "eval_gen_len": 5.9391,
+      "eval_loss": 1.7277677059173584,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.126,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 1.649,
+      "step": 1540
+    },
+    {
+      "epoch": 80.99,
+      "eval_gen_len": 5.8435,
+      "eval_loss": 1.7146191596984863,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5132,
+      "eval_samples_per_second": 9.19,
+      "eval_steps_per_second": 1.598,
+      "step": 1559
+    },
+    {
+      "epoch": 81.97,
+      "eval_gen_len": 5.513,
+      "eval_loss": 1.7026437520980835,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7453,
+      "eval_samples_per_second": 9.023,
+      "eval_steps_per_second": 1.569,
+      "step": 1578
+    },
+    {
+      "epoch": 82.96,
+      "eval_gen_len": 5.5652,
+      "eval_loss": 1.68914794921875,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.98,
+      "eval_samples_per_second": 8.86,
+      "eval_steps_per_second": 1.541,
+      "step": 1597
+    },
+    {
+      "epoch": 84.0,
+      "eval_gen_len": 5.3304,
+      "eval_loss": 1.6754295825958252,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.127,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 1.649,
+      "step": 1617
+    },
+    {
+      "epoch": 84.99,
+      "eval_gen_len": 5.6435,
+      "eval_loss": 1.6632497310638428,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2091,
+      "eval_samples_per_second": 9.419,
+      "eval_steps_per_second": 1.638,
+      "step": 1636
+    },
+    {
+      "epoch": 85.97,
+      "eval_gen_len": 5.9652,
+      "eval_loss": 1.652411699295044,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3822,
+      "eval_samples_per_second": 9.288,
+      "eval_steps_per_second": 1.615,
+      "step": 1655
+    },
+    {
+      "epoch": 86.96,
+      "eval_gen_len": 5.9478,
+      "eval_loss": 1.642953872680664,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.0847,
+      "eval_samples_per_second": 9.516,
+      "eval_steps_per_second": 1.655,
+      "step": 1674
+    },
+    {
+      "epoch": 88.0,
+      "eval_gen_len": 5.4696,
+      "eval_loss": 1.6336156129837036,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5368,
+      "eval_samples_per_second": 9.173,
+      "eval_steps_per_second": 1.595,
+      "step": 1694
+    },
+    {
+      "epoch": 88.99,
+      "eval_gen_len": 5.4,
+      "eval_loss": 1.6246790885925293,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4452,
+      "eval_samples_per_second": 9.241,
+      "eval_steps_per_second": 1.607,
+      "step": 1713
+    },
+    {
+      "epoch": 89.97,
+      "eval_gen_len": 5.7739,
+      "eval_loss": 1.615963339805603,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.0491,
+      "eval_samples_per_second": 9.544,
+      "eval_steps_per_second": 1.66,
+      "step": 1732
+    },
+    {
+      "epoch": 90.96,
+      "eval_gen_len": 6.2348,
+      "eval_loss": 1.606810450553894,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2296,
+      "eval_samples_per_second": 9.403,
+      "eval_steps_per_second": 1.635,
+      "step": 1751
+    },
+    {
+      "epoch": 92.0,
+      "eval_gen_len": 6.1652,
+      "eval_loss": 1.59696626663208,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.4411,
+      "eval_samples_per_second": 9.244,
+      "eval_steps_per_second": 1.608,
+      "step": 1771
+    },
+    {
+      "epoch": 92.99,
+      "eval_gen_len": 6.3739,
+      "eval_loss": 1.5894649028778076,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 11.8551,
+      "eval_samples_per_second": 9.7,
+      "eval_steps_per_second": 1.687,
+      "step": 1790
+    },
+    {
+      "epoch": 93.97,
+      "eval_gen_len": 6.5043,
+      "eval_loss": 1.5818349123001099,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2434,
+      "eval_samples_per_second": 9.393,
+      "eval_steps_per_second": 1.634,
+      "step": 1809
+    },
+    {
+      "epoch": 94.96,
+      "eval_gen_len": 6.3565,
+      "eval_loss": 1.5746902227401733,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3859,
+      "eval_samples_per_second": 9.285,
+      "eval_steps_per_second": 1.615,
+      "step": 1828
+    },
+    {
+      "epoch": 96.0,
+      "eval_gen_len": 6.8087,
+      "eval_loss": 1.567280650138855,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.1358,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 1.648,
+      "step": 1848
+    },
+    {
+      "epoch": 96.99,
+      "eval_gen_len": 6.8,
+      "eval_loss": 1.5616425275802612,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3339,
+      "eval_samples_per_second": 9.324,
+      "eval_steps_per_second": 1.622,
+      "step": 1867
+    },
+    {
+      "epoch": 97.97,
+      "eval_gen_len": 6.6522,
+      "eval_loss": 1.5548292398452759,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.1749,
+      "eval_samples_per_second": 9.446,
+      "eval_steps_per_second": 1.643,
+      "step": 1886
+    },
+    {
+      "epoch": 98.96,
+      "eval_gen_len": 6.5913,
+      "eval_loss": 1.5485645532608032,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 11.995,
+      "eval_samples_per_second": 9.587,
+      "eval_steps_per_second": 1.667,
+      "step": 1905
+    },
+    {
+      "epoch": 100.0,
+      "eval_gen_len": 6.4522,
+      "eval_loss": 1.5418448448181152,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 11.7913,
+      "eval_samples_per_second": 9.753,
+      "eval_steps_per_second": 1.696,
+      "step": 1925
+    },
+    {
+      "epoch": 100.99,
+      "eval_gen_len": 5.6957,
+      "eval_loss": 1.5365816354751587,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5711,
+      "eval_samples_per_second": 9.148,
+      "eval_steps_per_second": 1.591,
+      "step": 1944
+    },
+    {
+      "epoch": 101.97,
+      "eval_gen_len": 5.5739,
+      "eval_loss": 1.5312349796295166,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7436,
+      "eval_samples_per_second": 9.024,
+      "eval_steps_per_second": 1.569,
+      "step": 1963
+    },
+    {
+      "epoch": 102.96,
+      "eval_gen_len": 5.4174,
+      "eval_loss": 1.5244060754776,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3775,
+      "eval_samples_per_second": 9.291,
+      "eval_steps_per_second": 1.616,
+      "step": 1982
+    },
+    {
+      "epoch": 103.9,
+      "grad_norm": 1.870866298675537,
+      "learning_rate": 7.898947368421053e-06,
+      "loss": 1.8779,
+      "step": 2000
+    },
+    {
+      "epoch": 104.0,
+      "eval_gen_len": 5.3565,
+      "eval_loss": 1.5186233520507812,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.0793,
+      "eval_samples_per_second": 9.52,
+      "eval_steps_per_second": 1.656,
+      "step": 2002
+    },
+    {
+      "epoch": 104.99,
+      "eval_gen_len": 5.6174,
+      "eval_loss": 1.5112248659133911,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 11.9289,
+      "eval_samples_per_second": 9.64,
+      "eval_steps_per_second": 1.677,
+      "step": 2021
+    },
+    {
+      "epoch": 105.97,
+      "eval_gen_len": 5.9217,
+      "eval_loss": 1.5045664310455322,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 11.9549,
+      "eval_samples_per_second": 9.619,
+      "eval_steps_per_second": 1.673,
+      "step": 2040
+    },
+    {
+      "epoch": 106.96,
+      "eval_gen_len": 5.9913,
+      "eval_loss": 1.4977103471755981,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5176,
+      "eval_samples_per_second": 9.187,
+      "eval_steps_per_second": 1.598,
+      "step": 2059
+    },
+    {
+      "epoch": 108.0,
+      "eval_gen_len": 5.4957,
+      "eval_loss": 1.491757869720459,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.36,
+      "eval_samples_per_second": 9.304,
+      "eval_steps_per_second": 1.618,
+      "step": 2079
+    },
+    {
+      "epoch": 108.99,
+      "eval_gen_len": 6.0348,
+      "eval_loss": 1.486743688583374,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5097,
+      "eval_samples_per_second": 9.193,
+      "eval_steps_per_second": 1.599,
+      "step": 2098
+    },
+    {
+      "epoch": 109.97,
+      "eval_gen_len": 6.3304,
+      "eval_loss": 1.480473279953003,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 11.9075,
+      "eval_samples_per_second": 9.658,
+      "eval_steps_per_second": 1.68,
+      "step": 2117
+    },
+    {
+      "epoch": 110.96,
+      "eval_gen_len": 6.2,
+      "eval_loss": 1.4745731353759766,
+      "eval_rouge1": 0.0003,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0003,
+      "eval_rougeLsum": 0.0003,
+      "eval_runtime": 12.2105,
+      "eval_samples_per_second": 9.418,
+      "eval_steps_per_second": 1.638,
+      "step": 2136
+    },
+    {
+      "epoch": 112.0,
+      "eval_gen_len": 5.9826,
+      "eval_loss": 1.468475341796875,
+      "eval_rouge1": 0.0009,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0008,
+      "eval_rougeLsum": 0.0008,
+      "eval_runtime": 12.6855,
+      "eval_samples_per_second": 9.065,
+      "eval_steps_per_second": 1.577,
+      "step": 2156
+    },
+    {
+      "epoch": 112.99,
+      "eval_gen_len": 5.8261,
+      "eval_loss": 1.4624364376068115,
+      "eval_rouge1": 0.0009,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0008,
+      "eval_rougeLsum": 0.0008,
+      "eval_runtime": 12.3065,
+      "eval_samples_per_second": 9.345,
+      "eval_steps_per_second": 1.625,
+      "step": 2175
+    },
+    {
+      "epoch": 113.97,
+      "eval_gen_len": 5.487,
+      "eval_loss": 1.4564381837844849,
+      "eval_rouge1": 0.0009,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0008,
+      "eval_rougeLsum": 0.0008,
+      "eval_runtime": 13.0224,
+      "eval_samples_per_second": 8.831,
+      "eval_steps_per_second": 1.536,
+      "step": 2194
+    },
+    {
+      "epoch": 114.96,
+      "eval_gen_len": 5.1565,
+      "eval_loss": 1.4514414072036743,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4022,
+      "eval_samples_per_second": 9.273,
+      "eval_steps_per_second": 1.613,
+      "step": 2213
+    },
+    {
+      "epoch": 116.0,
+      "eval_gen_len": 5.4957,
+      "eval_loss": 1.442409873008728,
+      "eval_rouge1": 0.001,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0009,
+      "eval_rougeLsum": 0.0009,
+      "eval_runtime": 12.3722,
+      "eval_samples_per_second": 9.295,
+      "eval_steps_per_second": 1.617,
+      "step": 2233
+    },
+    {
+      "epoch": 116.99,
+      "eval_gen_len": 5.7391,
+      "eval_loss": 1.4344819784164429,
+      "eval_rouge1": 0.0017,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0017,
+      "eval_rougeLsum": 0.0018,
+      "eval_runtime": 12.5554,
+      "eval_samples_per_second": 9.159,
+      "eval_steps_per_second": 1.593,
+      "step": 2252
+    },
+    {
+      "epoch": 117.97,
+      "eval_gen_len": 6.0435,
+      "eval_loss": 1.4248623847961426,
+      "eval_rouge1": 0.0021,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.002,
+      "eval_rougeLsum": 0.0021,
+      "eval_runtime": 12.1087,
+      "eval_samples_per_second": 9.497,
+      "eval_steps_per_second": 1.652,
+      "step": 2271
+    },
+    {
+      "epoch": 118.96,
+      "eval_gen_len": 6.4783,
+      "eval_loss": 1.4156382083892822,
+      "eval_rouge1": 0.0033,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0032,
+      "eval_rougeLsum": 0.0033,
+      "eval_runtime": 12.0374,
+      "eval_samples_per_second": 9.554,
+      "eval_steps_per_second": 1.661,
+      "step": 2290
+    },
+    {
+      "epoch": 120.0,
+      "eval_gen_len": 6.3043,
+      "eval_loss": 1.408909559249878,
+      "eval_rouge1": 0.0038,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0037,
+      "eval_rougeLsum": 0.0038,
+      "eval_runtime": 12.5996,
+      "eval_samples_per_second": 9.127,
+      "eval_steps_per_second": 1.587,
+      "step": 2310
+    },
+    {
+      "epoch": 120.99,
+      "eval_gen_len": 6.1043,
+      "eval_loss": 1.4028282165527344,
+      "eval_rouge1": 0.0043,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0042,
+      "eval_rougeLsum": 0.0043,
+      "eval_runtime": 12.384,
+      "eval_samples_per_second": 9.286,
+      "eval_steps_per_second": 1.615,
+      "step": 2329
+    },
+    {
+      "epoch": 121.97,
+      "eval_gen_len": 5.9478,
+      "eval_loss": 1.3989005088806152,
+      "eval_rouge1": 0.0036,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0036,
+      "eval_rougeLsum": 0.0037,
+      "eval_runtime": 12.3007,
+      "eval_samples_per_second": 9.349,
+      "eval_steps_per_second": 1.626,
+      "step": 2348
+    },
+    {
+      "epoch": 122.96,
+      "eval_gen_len": 5.4348,
+      "eval_loss": 1.3940106630325317,
+      "eval_rouge1": 0.0029,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0026,
+      "eval_rougeLsum": 0.0026,
+      "eval_runtime": 13.1431,
+      "eval_samples_per_second": 8.75,
+      "eval_steps_per_second": 1.522,
+      "step": 2367
+    },
+    {
+      "epoch": 124.0,
+      "eval_gen_len": 5.3913,
+      "eval_loss": 1.387468695640564,
+      "eval_rouge1": 0.0036,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0034,
+      "eval_rougeLsum": 0.0035,
+      "eval_runtime": 12.0602,
+      "eval_samples_per_second": 9.535,
+      "eval_steps_per_second": 1.658,
+      "step": 2387
+    },
+    {
+      "epoch": 124.99,
+      "eval_gen_len": 5.4174,
+      "eval_loss": 1.3833892345428467,
+      "eval_rouge1": 0.0031,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0031,
+      "eval_rougeLsum": 0.0032,
+      "eval_runtime": 12.5404,
+      "eval_samples_per_second": 9.17,
+      "eval_steps_per_second": 1.595,
+      "step": 2406
+    },
+    {
+      "epoch": 125.97,
+      "eval_gen_len": 5.8,
+      "eval_loss": 1.3742746114730835,
+      "eval_rouge1": 0.0034,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0034,
+      "eval_rougeLsum": 0.0034,
+      "eval_runtime": 12.5374,
+      "eval_samples_per_second": 9.173,
+      "eval_steps_per_second": 1.595,
+      "step": 2425
+    },
+    {
+      "epoch": 126.96,
+      "eval_gen_len": 6.2348,
+      "eval_loss": 1.3673855066299438,
+      "eval_rouge1": 0.0054,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0051,
+      "eval_rougeLsum": 0.0052,
+      "eval_runtime": 12.3787,
+      "eval_samples_per_second": 9.29,
+      "eval_steps_per_second": 1.616,
+      "step": 2444
+    },
+    {
+      "epoch": 128.0,
+      "eval_gen_len": 6.3739,
+      "eval_loss": 1.3610302209854126,
+      "eval_rouge1": 0.0051,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0051,
+      "eval_rougeLsum": 0.0053,
+      "eval_runtime": 12.0722,
+      "eval_samples_per_second": 9.526,
+      "eval_steps_per_second": 1.657,
+      "step": 2464
+    },
+    {
+      "epoch": 128.99,
+      "eval_gen_len": 7.1565,
+      "eval_loss": 1.351613163948059,
+      "eval_rouge1": 0.0062,
+      "eval_rouge2": 0.0002,
+      "eval_rougeL": 0.0056,
+      "eval_rougeLsum": 0.0057,
+      "eval_runtime": 11.971,
+      "eval_samples_per_second": 9.607,
+      "eval_steps_per_second": 1.671,
+      "step": 2483
+    },
+    {
+      "epoch": 129.87,
+      "grad_norm": 2.394576072692871,
+      "learning_rate": 7.3726315789473694e-06,
+      "loss": 1.6063,
+      "step": 2500
+    },
+    {
+      "epoch": 129.97,
+      "eval_gen_len": 7.4522,
+      "eval_loss": 1.3424580097198486,
+      "eval_rouge1": 0.0055,
+      "eval_rouge2": 0.0002,
+      "eval_rougeL": 0.0053,
+      "eval_rougeLsum": 0.0053,
+      "eval_runtime": 12.0884,
+      "eval_samples_per_second": 9.513,
+      "eval_steps_per_second": 1.654,
+      "step": 2502
+    },
+    {
+      "epoch": 130.96,
+      "eval_gen_len": 7.4609,
+      "eval_loss": 1.334855556488037,
+      "eval_rouge1": 0.0044,
+      "eval_rouge2": 0.0002,
+      "eval_rougeL": 0.0042,
+      "eval_rougeLsum": 0.0041,
+      "eval_runtime": 11.9536,
+      "eval_samples_per_second": 9.621,
+      "eval_steps_per_second": 1.673,
+      "step": 2521
+    },
+    {
+      "epoch": 132.0,
+      "eval_gen_len": 7.4522,
+      "eval_loss": 1.3267827033996582,
+      "eval_rouge1": 0.0048,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0046,
+      "eval_rougeLsum": 0.0046,
+      "eval_runtime": 12.6532,
+      "eval_samples_per_second": 9.089,
+      "eval_steps_per_second": 1.581,
+      "step": 2541
+    },
+    {
+      "epoch": 132.99,
+      "eval_gen_len": 7.8522,
+      "eval_loss": 1.3201897144317627,
+      "eval_rouge1": 0.0096,
+      "eval_rouge2": 0.0002,
+      "eval_rougeL": 0.0088,
+      "eval_rougeLsum": 0.0089,
+      "eval_runtime": 12.2562,
+      "eval_samples_per_second": 9.383,
+      "eval_steps_per_second": 1.632,
+      "step": 2560
+    },
+    {
+      "epoch": 133.97,
+      "eval_gen_len": 7.5304,
+      "eval_loss": 1.3138891458511353,
+      "eval_rouge1": 0.0074,
+      "eval_rouge2": 0.0002,
+      "eval_rougeL": 0.0075,
+      "eval_rougeLsum": 0.0075,
+      "eval_runtime": 11.9989,
+      "eval_samples_per_second": 9.584,
+      "eval_steps_per_second": 1.667,
+      "step": 2579
+    },
+    {
+      "epoch": 134.96,
+      "eval_gen_len": 7.2348,
+      "eval_loss": 1.3059097528457642,
+      "eval_rouge1": 0.005,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.005,
+      "eval_rougeLsum": 0.0051,
+      "eval_runtime": 11.9785,
+      "eval_samples_per_second": 9.6,
+      "eval_steps_per_second": 1.67,
+      "step": 2598
+    },
+    {
+      "epoch": 136.0,
+      "eval_gen_len": 7.1304,
+      "eval_loss": 1.298433780670166,
+      "eval_rouge1": 0.005,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0046,
+      "eval_rougeLsum": 0.0047,
+      "eval_runtime": 12.1248,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 1.65,
+      "step": 2618
+    },
+    {
+      "epoch": 136.99,
+      "eval_gen_len": 8.0261,
+      "eval_loss": 1.29219651222229,
+      "eval_rouge1": 0.0072,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0069,
+      "eval_rougeLsum": 0.0069,
+      "eval_runtime": 12.5125,
+      "eval_samples_per_second": 9.191,
+      "eval_steps_per_second": 1.598,
+      "step": 2637
+    },
+    {
+      "epoch": 137.97,
+      "eval_gen_len": 8.4087,
+      "eval_loss": 1.2833046913146973,
+      "eval_rouge1": 0.0108,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0097,
+      "eval_rougeLsum": 0.0099,
+      "eval_runtime": 12.0943,
+      "eval_samples_per_second": 9.509,
+      "eval_steps_per_second": 1.654,
+      "step": 2656
+    },
+    {
+      "epoch": 138.96,
+      "eval_gen_len": 8.3739,
+      "eval_loss": 1.278290033340454,
+      "eval_rouge1": 0.0111,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0094,
+      "eval_rougeLsum": 0.0095,
+      "eval_runtime": 12.8059,
+      "eval_samples_per_second": 8.98,
+      "eval_steps_per_second": 1.562,
+      "step": 2675
+    },
+    {
+      "epoch": 140.0,
+      "eval_gen_len": 8.5043,
+      "eval_loss": 1.2764371633529663,
+      "eval_rouge1": 0.0114,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0104,
+      "eval_rougeLsum": 0.0108,
+      "eval_runtime": 12.1122,
+      "eval_samples_per_second": 9.495,
+      "eval_steps_per_second": 1.651,
+      "step": 2695
+    },
+    {
+      "epoch": 140.99,
+      "eval_gen_len": 8.6261,
+      "eval_loss": 1.2698535919189453,
+      "eval_rouge1": 0.0139,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0126,
+      "eval_rougeLsum": 0.0128,
+      "eval_runtime": 12.5216,
+      "eval_samples_per_second": 9.184,
+      "eval_steps_per_second": 1.597,
+      "step": 2714
+    },
+    {
+      "epoch": 141.97,
+      "eval_gen_len": 8.2435,
+      "eval_loss": 1.2616974115371704,
+      "eval_rouge1": 0.0135,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0119,
+      "eval_rougeLsum": 0.012,
+      "eval_runtime": 12.2763,
+      "eval_samples_per_second": 9.368,
+      "eval_steps_per_second": 1.629,
+      "step": 2733
+    },
+    {
+      "epoch": 142.96,
+      "eval_gen_len": 8.713,
+      "eval_loss": 1.2539962530136108,
+      "eval_rouge1": 0.0144,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0123,
+      "eval_rougeLsum": 0.0125,
+      "eval_runtime": 12.1565,
+      "eval_samples_per_second": 9.46,
+      "eval_steps_per_second": 1.645,
+      "step": 2752
+    },
+    {
+      "epoch": 144.0,
+      "eval_gen_len": 8.9826,
+      "eval_loss": 1.2482250928878784,
+      "eval_rouge1": 0.0153,
+      "eval_rouge2": 0.0002,
+      "eval_rougeL": 0.0137,
+      "eval_rougeLsum": 0.0137,
+      "eval_runtime": 12.3974,
+      "eval_samples_per_second": 9.276,
+      "eval_steps_per_second": 1.613,
+      "step": 2772
+    },
+    {
+      "epoch": 144.99,
+      "eval_gen_len": 8.9391,
+      "eval_loss": 1.2442501783370972,
+      "eval_rouge1": 0.0139,
+      "eval_rouge2": 0.0006,
+      "eval_rougeL": 0.013,
+      "eval_rougeLsum": 0.0129,
+      "eval_runtime": 12.5443,
+      "eval_samples_per_second": 9.168,
+      "eval_steps_per_second": 1.594,
+      "step": 2791
+    },
+    {
+      "epoch": 145.97,
+      "eval_gen_len": 9.3565,
+      "eval_loss": 1.2381587028503418,
+      "eval_rouge1": 0.0187,
+      "eval_rouge2": 0.0011,
+      "eval_rougeL": 0.0151,
+      "eval_rougeLsum": 0.0151,
+      "eval_runtime": 12.381,
+      "eval_samples_per_second": 9.288,
+      "eval_steps_per_second": 1.615,
+      "step": 2810
+    },
+    {
+      "epoch": 146.96,
+      "eval_gen_len": 9.513,
+      "eval_loss": 1.2287580966949463,
+      "eval_rouge1": 0.0202,
+      "eval_rouge2": 0.0008,
+      "eval_rougeL": 0.0172,
+      "eval_rougeLsum": 0.0172,
+      "eval_runtime": 12.6591,
+      "eval_samples_per_second": 9.084,
+      "eval_steps_per_second": 1.58,
+      "step": 2829
+    },
+    {
+      "epoch": 148.0,
+      "eval_gen_len": 8.5565,
+      "eval_loss": 1.2264941930770874,
+      "eval_rouge1": 0.0147,
+      "eval_rouge2": 0.0004,
+      "eval_rougeL": 0.0126,
+      "eval_rougeLsum": 0.0124,
+      "eval_runtime": 11.9708,
+      "eval_samples_per_second": 9.607,
+      "eval_steps_per_second": 1.671,
+      "step": 2849
+    },
+    {
+      "epoch": 148.99,
+      "eval_gen_len": 8.7652,
+      "eval_loss": 1.222589135169983,
+      "eval_rouge1": 0.0153,
+      "eval_rouge2": 0.0005,
+      "eval_rougeL": 0.0129,
+      "eval_rougeLsum": 0.0129,
+      "eval_runtime": 12.5442,
+      "eval_samples_per_second": 9.168,
+      "eval_steps_per_second": 1.594,
+      "step": 2868
+    },
+    {
+      "epoch": 149.97,
+      "eval_gen_len": 8.8435,
+      "eval_loss": 1.2170130014419556,
+      "eval_rouge1": 0.0147,
+      "eval_rouge2": 0.0008,
+      "eval_rougeL": 0.0128,
+      "eval_rougeLsum": 0.0127,
+      "eval_runtime": 12.6121,
+      "eval_samples_per_second": 9.118,
+      "eval_steps_per_second": 1.586,
+      "step": 2887
+    },
+    {
+      "epoch": 150.96,
+      "eval_gen_len": 9.4174,
+      "eval_loss": 1.208147406578064,
+      "eval_rouge1": 0.0181,
+      "eval_rouge2": 0.001,
+      "eval_rougeL": 0.0162,
+      "eval_rougeLsum": 0.0162,
+      "eval_runtime": 13.01,
+      "eval_samples_per_second": 8.839,
+      "eval_steps_per_second": 1.537,
+      "step": 2906
+    },
+    {
+      "epoch": 152.0,
+      "eval_gen_len": 9.7739,
+      "eval_loss": 1.2039202451705933,
+      "eval_rouge1": 0.0216,
+      "eval_rouge2": 0.0013,
+      "eval_rougeL": 0.019,
+      "eval_rougeLsum": 0.0191,
+      "eval_runtime": 12.7191,
+      "eval_samples_per_second": 9.042,
+      "eval_steps_per_second": 1.572,
+      "step": 2926
+    },
+    {
+      "epoch": 152.99,
+      "eval_gen_len": 9.5652,
+      "eval_loss": 1.200941801071167,
+      "eval_rouge1": 0.02,
+      "eval_rouge2": 0.0011,
+      "eval_rougeL": 0.0176,
+      "eval_rougeLsum": 0.0178,
+      "eval_runtime": 12.6301,
+      "eval_samples_per_second": 9.105,
+      "eval_steps_per_second": 1.584,
+      "step": 2945
+    },
+    {
+      "epoch": 153.97,
+      "eval_gen_len": 9.4609,
+      "eval_loss": 1.195379376411438,
+      "eval_rouge1": 0.0156,
+      "eval_rouge2": 0.0008,
+      "eval_rougeL": 0.0131,
+      "eval_rougeLsum": 0.0133,
+      "eval_runtime": 12.3176,
+      "eval_samples_per_second": 9.336,
+      "eval_steps_per_second": 1.624,
+      "step": 2964
+    },
+    {
+      "epoch": 154.96,
+      "eval_gen_len": 9.6522,
+      "eval_loss": 1.1899113655090332,
+      "eval_rouge1": 0.0181,
+      "eval_rouge2": 0.001,
+      "eval_rougeL": 0.0156,
+      "eval_rougeLsum": 0.0157,
+      "eval_runtime": 12.5731,
+      "eval_samples_per_second": 9.147,
+      "eval_steps_per_second": 1.591,
+      "step": 2983
+    },
+    {
+      "epoch": 155.84,
+      "grad_norm": 2.1833202838897705,
+      "learning_rate": 6.846315789473684e-06,
+      "loss": 1.4271,
+      "step": 3000
+    },
+    {
+      "epoch": 156.0,
+      "eval_gen_len": 9.4696,
+      "eval_loss": 1.1842440366744995,
+      "eval_rouge1": 0.0203,
+      "eval_rouge2": 0.0008,
+      "eval_rougeL": 0.0174,
+      "eval_rougeLsum": 0.0174,
+      "eval_runtime": 12.2741,
+      "eval_samples_per_second": 9.369,
+      "eval_steps_per_second": 1.629,
+      "step": 3003
+    },
+    {
+      "epoch": 156.99,
+      "eval_gen_len": 9.8174,
+      "eval_loss": 1.1782081127166748,
+      "eval_rouge1": 0.0187,
+      "eval_rouge2": 0.0007,
+      "eval_rougeL": 0.0163,
+      "eval_rougeLsum": 0.0165,
+      "eval_runtime": 13.1906,
+      "eval_samples_per_second": 8.718,
+      "eval_steps_per_second": 1.516,
+      "step": 3022
+    },
+    {
+      "epoch": 157.97,
+      "eval_gen_len": 9.9304,
+      "eval_loss": 1.173979640007019,
+      "eval_rouge1": 0.0206,
+      "eval_rouge2": 0.0005,
+      "eval_rougeL": 0.018,
+      "eval_rougeLsum": 0.0183,
+      "eval_runtime": 12.6288,
+      "eval_samples_per_second": 9.106,
+      "eval_steps_per_second": 1.584,
+      "step": 3041
+    },
+    {
+      "epoch": 158.96,
+      "eval_gen_len": 10.0087,
+      "eval_loss": 1.1698901653289795,
+      "eval_rouge1": 0.0198,
+      "eval_rouge2": 0.0005,
+      "eval_rougeL": 0.0177,
+      "eval_rougeLsum": 0.018,
+      "eval_runtime": 12.5926,
+      "eval_samples_per_second": 9.132,
+      "eval_steps_per_second": 1.588,
+      "step": 3060
+    },
+    {
+      "epoch": 160.0,
+      "eval_gen_len": 10.2174,
+      "eval_loss": 1.1631128787994385,
+      "eval_rouge1": 0.0214,
+      "eval_rouge2": 0.0004,
+      "eval_rougeL": 0.0189,
+      "eval_rougeLsum": 0.0191,
+      "eval_runtime": 12.2102,
+      "eval_samples_per_second": 9.418,
+      "eval_steps_per_second": 1.638,
+      "step": 3080
+    },
+    {
+      "epoch": 160.99,
+      "eval_gen_len": 10.1304,
+      "eval_loss": 1.1569976806640625,
+      "eval_rouge1": 0.0221,
+      "eval_rouge2": 0.0009,
+      "eval_rougeL": 0.0185,
+      "eval_rougeLsum": 0.0187,
+      "eval_runtime": 12.421,
+      "eval_samples_per_second": 9.259,
+      "eval_steps_per_second": 1.61,
+      "step": 3099
+    },
+    {
+      "epoch": 161.97,
+      "eval_gen_len": 10.0609,
+      "eval_loss": 1.1523972749710083,
+      "eval_rouge1": 0.0202,
+      "eval_rouge2": 0.0009,
+      "eval_rougeL": 0.0169,
+      "eval_rougeLsum": 0.0171,
+      "eval_runtime": 12.3412,
+      "eval_samples_per_second": 9.318,
+      "eval_steps_per_second": 1.621,
+      "step": 3118
+    },
+    {
+      "epoch": 162.96,
+      "eval_gen_len": 9.8609,
+      "eval_loss": 1.1472958326339722,
+      "eval_rouge1": 0.0202,
+      "eval_rouge2": 0.0006,
+      "eval_rougeL": 0.0173,
+      "eval_rougeLsum": 0.0174,
+      "eval_runtime": 12.2339,
+      "eval_samples_per_second": 9.4,
+      "eval_steps_per_second": 1.635,
+      "step": 3137
+    },
+    {
+      "epoch": 164.0,
+      "eval_gen_len": 10.3913,
+      "eval_loss": 1.1415693759918213,
+      "eval_rouge1": 0.0218,
+      "eval_rouge2": 0.0011,
+      "eval_rougeL": 0.0184,
+      "eval_rougeLsum": 0.0183,
+      "eval_runtime": 12.6611,
+      "eval_samples_per_second": 9.083,
+      "eval_steps_per_second": 1.58,
+      "step": 3157
+    },
+    {
+      "epoch": 164.99,
+      "eval_gen_len": 9.713,
+      "eval_loss": 1.135535478591919,
+      "eval_rouge1": 0.0174,
+      "eval_rouge2": 0.0005,
+      "eval_rougeL": 0.0148,
+      "eval_rougeLsum": 0.0146,
+      "eval_runtime": 12.6693,
+      "eval_samples_per_second": 9.077,
+      "eval_steps_per_second": 1.579,
+      "step": 3176
+    },
+    {
+      "epoch": 165.97,
+      "eval_gen_len": 10.113,
+      "eval_loss": 1.1300948858261108,
+      "eval_rouge1": 0.0185,
+      "eval_rouge2": 0.0007,
+      "eval_rougeL": 0.0151,
+      "eval_rougeLsum": 0.0151,
+      "eval_runtime": 12.211,
+      "eval_samples_per_second": 9.418,
+      "eval_steps_per_second": 1.638,
+      "step": 3195
+    },
+    {
+      "epoch": 166.96,
+      "eval_gen_len": 10.1043,
+      "eval_loss": 1.125083088874817,
+      "eval_rouge1": 0.0205,
+      "eval_rouge2": 0.0008,
+      "eval_rougeL": 0.0165,
+      "eval_rougeLsum": 0.0164,
+      "eval_runtime": 12.1409,
+      "eval_samples_per_second": 9.472,
+      "eval_steps_per_second": 1.647,
+      "step": 3214
+    },
+    {
+      "epoch": 168.0,
+      "eval_gen_len": 10.2348,
+      "eval_loss": 1.1202179193496704,
+      "eval_rouge1": 0.0195,
+      "eval_rouge2": 0.0008,
+      "eval_rougeL": 0.0159,
+      "eval_rougeLsum": 0.0159,
+      "eval_runtime": 12.1317,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 1.649,
+      "step": 3234
+    },
+    {
+      "epoch": 168.99,
+      "eval_gen_len": 10.8957,
+      "eval_loss": 1.114139199256897,
+      "eval_rouge1": 0.0243,
+      "eval_rouge2": 0.0018,
+      "eval_rougeL": 0.0198,
+      "eval_rougeLsum": 0.0198,
+      "eval_runtime": 12.443,
+      "eval_samples_per_second": 9.242,
+      "eval_steps_per_second": 1.607,
+      "step": 3253
+    },
+    {
+      "epoch": 169.97,
+      "eval_gen_len": 11.0174,
+      "eval_loss": 1.1090463399887085,
+      "eval_rouge1": 0.0202,
+      "eval_rouge2": 0.0013,
+      "eval_rougeL": 0.0161,
+      "eval_rougeLsum": 0.0163,
+      "eval_runtime": 12.144,
+      "eval_samples_per_second": 9.47,
+      "eval_steps_per_second": 1.647,
+      "step": 3272
+    },
+    {
+      "epoch": 170.96,
+      "eval_gen_len": 11.313,
+      "eval_loss": 1.1036903858184814,
+      "eval_rouge1": 0.0223,
+      "eval_rouge2": 0.0015,
+      "eval_rougeL": 0.0186,
+      "eval_rougeLsum": 0.0186,
+      "eval_runtime": 12.3793,
+      "eval_samples_per_second": 9.29,
+      "eval_steps_per_second": 1.616,
+      "step": 3291
+    },
+    {
+      "epoch": 172.0,
+      "eval_gen_len": 11.3739,
+      "eval_loss": 1.0987364053726196,
+      "eval_rouge1": 0.0212,
+      "eval_rouge2": 0.0013,
+      "eval_rougeL": 0.0178,
+      "eval_rougeLsum": 0.0179,
+      "eval_runtime": 12.3964,
+      "eval_samples_per_second": 9.277,
+      "eval_steps_per_second": 1.613,
+      "step": 3311
+    },
+    {
+      "epoch": 172.99,
+      "eval_gen_len": 11.2522,
+      "eval_loss": 1.0937457084655762,
+      "eval_rouge1": 0.0219,
+      "eval_rouge2": 0.0015,
+      "eval_rougeL": 0.0182,
+      "eval_rougeLsum": 0.018,
+      "eval_runtime": 12.5831,
+      "eval_samples_per_second": 9.139,
+      "eval_steps_per_second": 1.589,
+      "step": 3330
+    },
+    {
+      "epoch": 173.97,
+      "eval_gen_len": 11.2174,
+      "eval_loss": 1.090100646018982,
+      "eval_rouge1": 0.0199,
+      "eval_rouge2": 0.0013,
+      "eval_rougeL": 0.0162,
+      "eval_rougeLsum": 0.0163,
+      "eval_runtime": 12.7101,
+      "eval_samples_per_second": 9.048,
+      "eval_steps_per_second": 1.574,
+      "step": 3349
+    },
+    {
+      "epoch": 174.96,
+      "eval_gen_len": 11.2174,
+      "eval_loss": 1.0861694812774658,
+      "eval_rouge1": 0.018,
+      "eval_rouge2": 0.0011,
+      "eval_rougeL": 0.0149,
+      "eval_rougeLsum": 0.0149,
+      "eval_runtime": 12.2255,
+      "eval_samples_per_second": 9.407,
+      "eval_steps_per_second": 1.636,
+      "step": 3368
+    },
+    {
+      "epoch": 176.0,
+      "eval_gen_len": 11.3304,
+      "eval_loss": 1.080249309539795,
+      "eval_rouge1": 0.0181,
+      "eval_rouge2": 0.0013,
+      "eval_rougeL": 0.0154,
+      "eval_rougeLsum": 0.0154,
+      "eval_runtime": 12.2755,
+      "eval_samples_per_second": 9.368,
+      "eval_steps_per_second": 1.629,
+      "step": 3388
+    },
+    {
+      "epoch": 176.99,
+      "eval_gen_len": 10.9739,
+      "eval_loss": 1.0751179456710815,
+      "eval_rouge1": 0.0147,
+      "eval_rouge2": 0.0012,
+      "eval_rougeL": 0.0124,
+      "eval_rougeLsum": 0.0125,
+      "eval_runtime": 12.3848,
+      "eval_samples_per_second": 9.286,
+      "eval_steps_per_second": 1.615,
+      "step": 3407
+    },
+    {
+      "epoch": 177.97,
+      "eval_gen_len": 10.8087,
+      "eval_loss": 1.069909930229187,
+      "eval_rouge1": 0.0149,
+      "eval_rouge2": 0.001,
+      "eval_rougeL": 0.0123,
+      "eval_rougeLsum": 0.0124,
+      "eval_runtime": 12.6526,
+      "eval_samples_per_second": 9.089,
+      "eval_steps_per_second": 1.581,
+      "step": 3426
+    },
+    {
+      "epoch": 178.96,
+      "eval_gen_len": 10.7217,
+      "eval_loss": 1.0651546716690063,
+      "eval_rouge1": 0.0134,
+      "eval_rouge2": 0.0011,
+      "eval_rougeL": 0.0122,
+      "eval_rougeLsum": 0.0122,
+      "eval_runtime": 13.0395,
+      "eval_samples_per_second": 8.819,
+      "eval_steps_per_second": 1.534,
+      "step": 3445
+    },
+    {
+      "epoch": 180.0,
+      "eval_gen_len": 10.6174,
+      "eval_loss": 1.060491681098938,
+      "eval_rouge1": 0.0121,
+      "eval_rouge2": 0.001,
+      "eval_rougeL": 0.0101,
+      "eval_rougeLsum": 0.0101,
+      "eval_runtime": 12.3246,
+      "eval_samples_per_second": 9.331,
+      "eval_steps_per_second": 1.623,
+      "step": 3465
+    },
+    {
+      "epoch": 180.99,
+      "eval_gen_len": 10.6435,
+      "eval_loss": 1.0562814474105835,
+      "eval_rouge1": 0.0131,
+      "eval_rouge2": 0.0007,
+      "eval_rougeL": 0.0113,
+      "eval_rougeLsum": 0.0111,
+      "eval_runtime": 13.1955,
+      "eval_samples_per_second": 8.715,
+      "eval_steps_per_second": 1.516,
+      "step": 3484
+    },
+    {
+      "epoch": 181.82,
+      "grad_norm": 0.9931882619857788,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 1.265,
+      "step": 3500
+    },
+    {
+      "epoch": 181.97,
+      "eval_gen_len": 10.3913,
+      "eval_loss": 1.0519821643829346,
+      "eval_rouge1": 0.0147,
+      "eval_rouge2": 0.0019,
+      "eval_rougeL": 0.0129,
+      "eval_rougeLsum": 0.0125,
+      "eval_runtime": 12.3478,
+      "eval_samples_per_second": 9.313,
+      "eval_steps_per_second": 1.62,
+      "step": 3503
+    },
+    {
+      "epoch": 182.96,
+      "eval_gen_len": 10.9826,
+      "eval_loss": 1.047600507736206,
+      "eval_rouge1": 0.0171,
+      "eval_rouge2": 0.0018,
+      "eval_rougeL": 0.0148,
+      "eval_rougeLsum": 0.0148,
+      "eval_runtime": 12.7548,
+      "eval_samples_per_second": 9.016,
+      "eval_steps_per_second": 1.568,
+      "step": 3522
+    },
+    {
+      "epoch": 184.0,
+      "eval_gen_len": 10.9478,
+      "eval_loss": 1.0428956747055054,
+      "eval_rouge1": 0.019,
+      "eval_rouge2": 0.0026,
+      "eval_rougeL": 0.0173,
+      "eval_rougeLsum": 0.0174,
+      "eval_runtime": 12.668,
+      "eval_samples_per_second": 9.078,
+      "eval_steps_per_second": 1.579,
+      "step": 3542
+    },
+    {
+      "epoch": 184.99,
+      "eval_gen_len": 10.6348,
+      "eval_loss": 1.0391294956207275,
+      "eval_rouge1": 0.0192,
+      "eval_rouge2": 0.0019,
+      "eval_rougeL": 0.016,
+      "eval_rougeLsum": 0.0162,
+      "eval_runtime": 12.2592,
+      "eval_samples_per_second": 9.381,
+      "eval_steps_per_second": 1.631,
+      "step": 3561
+    },
+    {
+      "epoch": 185.97,
+      "eval_gen_len": 10.6,
+      "eval_loss": 1.0354028940200806,
+      "eval_rouge1": 0.0192,
+      "eval_rouge2": 0.0021,
+      "eval_rougeL": 0.0154,
+      "eval_rougeLsum": 0.0155,
+      "eval_runtime": 12.3279,
+      "eval_samples_per_second": 9.328,
+      "eval_steps_per_second": 1.622,
+      "step": 3580
+    },
+    {
+      "epoch": 186.96,
+      "eval_gen_len": 10.2261,
+      "eval_loss": 1.0318480730056763,
+      "eval_rouge1": 0.0193,
+      "eval_rouge2": 0.003,
+      "eval_rougeL": 0.0162,
+      "eval_rougeLsum": 0.0163,
+      "eval_runtime": 12.7454,
+      "eval_samples_per_second": 9.023,
+      "eval_steps_per_second": 1.569,
+      "step": 3599
+    },
+    {
+      "epoch": 188.0,
+      "eval_gen_len": 10.6261,
+      "eval_loss": 1.0279144048690796,
+      "eval_rouge1": 0.0245,
+      "eval_rouge2": 0.0032,
+      "eval_rougeL": 0.0201,
+      "eval_rougeLsum": 0.02,
+      "eval_runtime": 12.5706,
+      "eval_samples_per_second": 9.148,
+      "eval_steps_per_second": 1.591,
+      "step": 3619
+    },
+    {
+      "epoch": 188.99,
+      "eval_gen_len": 10.5913,
+      "eval_loss": 1.0238802433013916,
+      "eval_rouge1": 0.025,
+      "eval_rouge2": 0.0029,
+      "eval_rougeL": 0.0206,
+      "eval_rougeLsum": 0.0207,
+      "eval_runtime": 12.4275,
+      "eval_samples_per_second": 9.254,
+      "eval_steps_per_second": 1.609,
+      "step": 3638
+    },
+    {
+      "epoch": 189.97,
+      "eval_gen_len": 10.2261,
+      "eval_loss": 1.0197361707687378,
+      "eval_rouge1": 0.0249,
+      "eval_rouge2": 0.0029,
+      "eval_rougeL": 0.0198,
+      "eval_rougeLsum": 0.0199,
+      "eval_runtime": 12.6121,
+      "eval_samples_per_second": 9.118,
+      "eval_steps_per_second": 1.586,
+      "step": 3657
+    },
+    {
+      "epoch": 190.96,
+      "eval_gen_len": 10.1391,
+      "eval_loss": 1.0159963369369507,
+      "eval_rouge1": 0.0245,
+      "eval_rouge2": 0.003,
+      "eval_rougeL": 0.019,
+      "eval_rougeLsum": 0.0191,
+      "eval_runtime": 12.81,
+      "eval_samples_per_second": 8.977,
+      "eval_steps_per_second": 1.561,
+      "step": 3676
+    },
+    {
+      "epoch": 192.0,
+      "eval_gen_len": 10.2435,
+      "eval_loss": 1.0119863748550415,
+      "eval_rouge1": 0.0243,
+      "eval_rouge2": 0.0027,
+      "eval_rougeL": 0.019,
+      "eval_rougeLsum": 0.019,
+      "eval_runtime": 12.53,
+      "eval_samples_per_second": 9.178,
+      "eval_steps_per_second": 1.596,
+      "step": 3696
+    },
+    {
+      "epoch": 192.99,
+      "eval_gen_len": 10.3826,
+      "eval_loss": 1.008431315422058,
+      "eval_rouge1": 0.0247,
+      "eval_rouge2": 0.0029,
+      "eval_rougeL": 0.0194,
+      "eval_rougeLsum": 0.0193,
+      "eval_runtime": 12.6196,
+      "eval_samples_per_second": 9.113,
+      "eval_steps_per_second": 1.585,
+      "step": 3715
+    },
+    {
+      "epoch": 193.97,
+      "eval_gen_len": 10.6696,
+      "eval_loss": 1.0049232244491577,
+      "eval_rouge1": 0.0239,
+      "eval_rouge2": 0.0027,
+      "eval_rougeL": 0.0186,
+      "eval_rougeLsum": 0.0185,
+      "eval_runtime": 12.5612,
+      "eval_samples_per_second": 9.155,
+      "eval_steps_per_second": 1.592,
+      "step": 3734
+    },
+    {
+      "epoch": 194.96,
+      "eval_gen_len": 11.1043,
+      "eval_loss": 1.0015385150909424,
+      "eval_rouge1": 0.0248,
+      "eval_rouge2": 0.0029,
+      "eval_rougeL": 0.0195,
+      "eval_rougeLsum": 0.0195,
+      "eval_runtime": 11.9572,
+      "eval_samples_per_second": 9.618,
+      "eval_steps_per_second": 1.673,
+      "step": 3753
+    },
+    {
+      "epoch": 196.0,
+      "eval_gen_len": 10.8609,
+      "eval_loss": 0.9973338842391968,
+      "eval_rouge1": 0.0233,
+      "eval_rouge2": 0.0026,
+      "eval_rougeL": 0.0189,
+      "eval_rougeLsum": 0.0189,
+      "eval_runtime": 11.9829,
+      "eval_samples_per_second": 9.597,
+      "eval_steps_per_second": 1.669,
+      "step": 3773
+    },
+    {
+      "epoch": 196.99,
+      "eval_gen_len": 10.6783,
+      "eval_loss": 0.9933551549911499,
+      "eval_rouge1": 0.0209,
+      "eval_rouge2": 0.0028,
+      "eval_rougeL": 0.0172,
+      "eval_rougeLsum": 0.0172,
+      "eval_runtime": 12.3195,
+      "eval_samples_per_second": 9.335,
+      "eval_steps_per_second": 1.623,
+      "step": 3792
+    },
+    {
+      "epoch": 197.97,
+      "eval_gen_len": 10.9043,
+      "eval_loss": 0.9898021817207336,
+      "eval_rouge1": 0.0224,
+      "eval_rouge2": 0.0028,
+      "eval_rougeL": 0.0183,
+      "eval_rougeLsum": 0.0182,
+      "eval_runtime": 12.6246,
+      "eval_samples_per_second": 9.109,
+      "eval_steps_per_second": 1.584,
+      "step": 3811
+    },
+    {
+      "epoch": 198.96,
+      "eval_gen_len": 11.2435,
+      "eval_loss": 0.9868430495262146,
+      "eval_rouge1": 0.0223,
+      "eval_rouge2": 0.0034,
+      "eval_rougeL": 0.0186,
+      "eval_rougeLsum": 0.0186,
+      "eval_runtime": 12.4605,
+      "eval_samples_per_second": 9.229,
+      "eval_steps_per_second": 1.605,
+      "step": 3830
+    },
+    {
+      "epoch": 200.0,
+      "eval_gen_len": 11.1565,
+      "eval_loss": 0.9835883975028992,
+      "eval_rouge1": 0.0212,
+      "eval_rouge2": 0.0033,
+      "eval_rougeL": 0.0182,
+      "eval_rougeLsum": 0.018,
+      "eval_runtime": 12.0605,
+      "eval_samples_per_second": 9.535,
+      "eval_steps_per_second": 1.658,
+      "step": 3850
+    },
+    {
+      "epoch": 200.99,
+      "eval_gen_len": 11.2087,
+      "eval_loss": 0.9812867641448975,
+      "eval_rouge1": 0.0202,
+      "eval_rouge2": 0.003,
+      "eval_rougeL": 0.0164,
+      "eval_rougeLsum": 0.0164,
+      "eval_runtime": 12.6958,
+      "eval_samples_per_second": 9.058,
+      "eval_steps_per_second": 1.575,
+      "step": 3869
+    },
+    {
+      "epoch": 201.97,
+      "eval_gen_len": 11.2783,
+      "eval_loss": 0.9780998229980469,
+      "eval_rouge1": 0.0192,
+      "eval_rouge2": 0.0032,
+      "eval_rougeL": 0.0158,
+      "eval_rougeLsum": 0.0158,
+      "eval_runtime": 12.2767,
+      "eval_samples_per_second": 9.367,
+      "eval_steps_per_second": 1.629,
+      "step": 3888
+    },
+    {
+      "epoch": 202.96,
+      "eval_gen_len": 11.113,
+      "eval_loss": 0.9748227596282959,
+      "eval_rouge1": 0.0174,
+      "eval_rouge2": 0.0028,
+      "eval_rougeL": 0.0144,
+      "eval_rougeLsum": 0.0144,
+      "eval_runtime": 12.2251,
+      "eval_samples_per_second": 9.407,
+      "eval_steps_per_second": 1.636,
+      "step": 3907
+    },
+    {
+      "epoch": 204.0,
+      "eval_gen_len": 11.3304,
+      "eval_loss": 0.9713881015777588,
+      "eval_rouge1": 0.0187,
+      "eval_rouge2": 0.0026,
+      "eval_rougeL": 0.0157,
+      "eval_rougeLsum": 0.0157,
+      "eval_runtime": 12.1776,
+      "eval_samples_per_second": 9.444,
+      "eval_steps_per_second": 1.642,
+      "step": 3927
+    },
+    {
+      "epoch": 204.99,
+      "eval_gen_len": 11.5043,
+      "eval_loss": 0.968216598033905,
+      "eval_rouge1": 0.0199,
+      "eval_rouge2": 0.0026,
+      "eval_rougeL": 0.0164,
+      "eval_rougeLsum": 0.0166,
+      "eval_runtime": 12.4259,
+      "eval_samples_per_second": 9.255,
+      "eval_steps_per_second": 1.61,
+      "step": 3946
+    },
+    {
+      "epoch": 205.97,
+      "eval_gen_len": 11.4261,
+      "eval_loss": 0.9647319912910461,
+      "eval_rouge1": 0.0184,
+      "eval_rouge2": 0.0025,
+      "eval_rougeL": 0.0154,
+      "eval_rougeLsum": 0.0154,
+      "eval_runtime": 12.6085,
+      "eval_samples_per_second": 9.121,
+      "eval_steps_per_second": 1.586,
+      "step": 3965
+    },
+    {
+      "epoch": 206.96,
+      "eval_gen_len": 11.6087,
+      "eval_loss": 0.9613582491874695,
+      "eval_rouge1": 0.0172,
+      "eval_rouge2": 0.0018,
+      "eval_rougeL": 0.0146,
+      "eval_rougeLsum": 0.0145,
+      "eval_runtime": 12.5542,
+      "eval_samples_per_second": 9.16,
+      "eval_steps_per_second": 1.593,
+      "step": 3984
+    },
+    {
+      "epoch": 207.79,
+      "grad_norm": 0.6749991178512573,
+      "learning_rate": 5.793684210526316e-06,
+      "loss": 1.119,
+      "step": 4000
+    },
+    {
+      "epoch": 208.0,
+      "eval_gen_len": 11.8087,
+      "eval_loss": 0.9580429196357727,
+      "eval_rouge1": 0.0206,
+      "eval_rouge2": 0.0023,
+      "eval_rougeL": 0.0168,
+      "eval_rougeLsum": 0.0167,
+      "eval_runtime": 12.4965,
+      "eval_samples_per_second": 9.203,
+      "eval_steps_per_second": 1.6,
+      "step": 4004
+    },
+    {
+      "epoch": 208.99,
+      "eval_gen_len": 12.0957,
+      "eval_loss": 0.9548400640487671,
+      "eval_rouge1": 0.0233,
+      "eval_rouge2": 0.0023,
+      "eval_rougeL": 0.019,
+      "eval_rougeLsum": 0.019,
+      "eval_runtime": 13.061,
+      "eval_samples_per_second": 8.805,
+      "eval_steps_per_second": 1.531,
+      "step": 4023
+    },
+    {
+      "epoch": 209.97,
+      "eval_gen_len": 11.9826,
+      "eval_loss": 0.9517626166343689,
+      "eval_rouge1": 0.0214,
+      "eval_rouge2": 0.0021,
+      "eval_rougeL": 0.0181,
+      "eval_rougeLsum": 0.018,
+      "eval_runtime": 12.6476,
+      "eval_samples_per_second": 9.093,
+      "eval_steps_per_second": 1.581,
+      "step": 4042
+    },
+    {
+      "epoch": 210.96,
+      "eval_gen_len": 11.9304,
+      "eval_loss": 0.9485481381416321,
+      "eval_rouge1": 0.0208,
+      "eval_rouge2": 0.0018,
+      "eval_rougeL": 0.0171,
+      "eval_rougeLsum": 0.0172,
+      "eval_runtime": 12.4352,
+      "eval_samples_per_second": 9.248,
+      "eval_steps_per_second": 1.608,
+      "step": 4061
+    },
+    {
+      "epoch": 212.0,
+      "eval_gen_len": 11.7826,
+      "eval_loss": 0.9455087184906006,
+      "eval_rouge1": 0.0184,
+      "eval_rouge2": 0.0016,
+      "eval_rougeL": 0.0152,
+      "eval_rougeLsum": 0.015,
+      "eval_runtime": 12.868,
+      "eval_samples_per_second": 8.937,
+      "eval_steps_per_second": 1.554,
+      "step": 4081
+    },
+    {
+      "epoch": 212.99,
+      "eval_gen_len": 11.7565,
+      "eval_loss": 0.9424554109573364,
+      "eval_rouge1": 0.0186,
+      "eval_rouge2": 0.0028,
+      "eval_rougeL": 0.0153,
+      "eval_rougeLsum": 0.0153,
+      "eval_runtime": 12.3925,
+      "eval_samples_per_second": 9.28,
+      "eval_steps_per_second": 1.614,
+      "step": 4100
+    },
+    {
+      "epoch": 213.97,
+      "eval_gen_len": 11.3913,
+      "eval_loss": 0.939349889755249,
+      "eval_rouge1": 0.0165,
+      "eval_rouge2": 0.002,
+      "eval_rougeL": 0.0131,
+      "eval_rougeLsum": 0.0131,
+      "eval_runtime": 12.3014,
+      "eval_samples_per_second": 9.349,
+      "eval_steps_per_second": 1.626,
+      "step": 4119
+    },
+    {
+      "epoch": 214.96,
+      "eval_gen_len": 11.4522,
+      "eval_loss": 0.9365057349205017,
+      "eval_rouge1": 0.0177,
+      "eval_rouge2": 0.0022,
+      "eval_rougeL": 0.0143,
+      "eval_rougeLsum": 0.0143,
+      "eval_runtime": 12.8423,
+      "eval_samples_per_second": 8.955,
+      "eval_steps_per_second": 1.557,
+      "step": 4138
+    },
+    {
+      "epoch": 216.0,
+      "eval_gen_len": 11.7391,
+      "eval_loss": 0.9332289099693298,
+      "eval_rouge1": 0.0213,
+      "eval_rouge2": 0.0028,
+      "eval_rougeL": 0.0177,
+      "eval_rougeLsum": 0.0173,
+      "eval_runtime": 12.4944,
+      "eval_samples_per_second": 9.204,
+      "eval_steps_per_second": 1.601,
+      "step": 4158
+    },
+    {
+      "epoch": 216.99,
+      "eval_gen_len": 11.6522,
+      "eval_loss": 0.9310381412506104,
+      "eval_rouge1": 0.0197,
+      "eval_rouge2": 0.0028,
+      "eval_rougeL": 0.0159,
+      "eval_rougeLsum": 0.0157,
+      "eval_runtime": 12.5982,
+      "eval_samples_per_second": 9.128,
+      "eval_steps_per_second": 1.588,
+      "step": 4177
+    },
+    {
+      "epoch": 217.97,
+      "eval_gen_len": 11.687,
+      "eval_loss": 0.9279318451881409,
+      "eval_rouge1": 0.0203,
+      "eval_rouge2": 0.0029,
+      "eval_rougeL": 0.0168,
+      "eval_rougeLsum": 0.0165,
+      "eval_runtime": 12.2551,
+      "eval_samples_per_second": 9.384,
+      "eval_steps_per_second": 1.632,
+      "step": 4196
+    },
+    {
+      "epoch": 218.96,
+      "eval_gen_len": 11.7043,
+      "eval_loss": 0.9249460697174072,
+      "eval_rouge1": 0.0228,
+      "eval_rouge2": 0.0032,
+      "eval_rougeL": 0.0191,
+      "eval_rougeLsum": 0.019,
+      "eval_runtime": 12.661,
+      "eval_samples_per_second": 9.083,
+      "eval_steps_per_second": 1.58,
+      "step": 4215
+    },
+    {
+      "epoch": 220.0,
+      "eval_gen_len": 11.2783,
+      "eval_loss": 0.9218883514404297,
+      "eval_rouge1": 0.0219,
+      "eval_rouge2": 0.0032,
+      "eval_rougeL": 0.0182,
+      "eval_rougeLsum": 0.018,
+      "eval_runtime": 12.5398,
+      "eval_samples_per_second": 9.171,
+      "eval_steps_per_second": 1.595,
+      "step": 4235
+    },
+    {
+      "epoch": 220.99,
+      "eval_gen_len": 11.0087,
+      "eval_loss": 0.9194144010543823,
+      "eval_rouge1": 0.0203,
+      "eval_rouge2": 0.0029,
+      "eval_rougeL": 0.0171,
+      "eval_rougeLsum": 0.0167,
+      "eval_runtime": 12.4516,
+      "eval_samples_per_second": 9.236,
+      "eval_steps_per_second": 1.606,
+      "step": 4254
+    },
+    {
+      "epoch": 221.97,
+      "eval_gen_len": 10.8174,
+      "eval_loss": 0.9165053963661194,
+      "eval_rouge1": 0.0197,
+      "eval_rouge2": 0.0021,
+      "eval_rougeL": 0.0164,
+      "eval_rougeLsum": 0.0161,
+      "eval_runtime": 12.4796,
+      "eval_samples_per_second": 9.215,
+      "eval_steps_per_second": 1.603,
+      "step": 4273
+    },
+    {
+      "epoch": 222.96,
+      "eval_gen_len": 10.9652,
+      "eval_loss": 0.9133741855621338,
+      "eval_rouge1": 0.0226,
+      "eval_rouge2": 0.0027,
+      "eval_rougeL": 0.0185,
+      "eval_rougeLsum": 0.0182,
+      "eval_runtime": 12.2854,
+      "eval_samples_per_second": 9.361,
+      "eval_steps_per_second": 1.628,
+      "step": 4292
+    },
+    {
+      "epoch": 224.0,
+      "eval_gen_len": 10.7565,
+      "eval_loss": 0.9105393886566162,
+      "eval_rouge1": 0.0245,
+      "eval_rouge2": 0.0032,
+      "eval_rougeL": 0.0199,
+      "eval_rougeLsum": 0.0197,
+      "eval_runtime": 12.0278,
+      "eval_samples_per_second": 9.561,
+      "eval_steps_per_second": 1.663,
+      "step": 4312
+    },
+    {
+      "epoch": 224.99,
+      "eval_gen_len": 10.1391,
+      "eval_loss": 0.907561719417572,
+      "eval_rouge1": 0.0198,
+      "eval_rouge2": 0.0025,
+      "eval_rougeL": 0.0163,
+      "eval_rougeLsum": 0.0161,
+      "eval_runtime": 12.2338,
+      "eval_samples_per_second": 9.4,
+      "eval_steps_per_second": 1.635,
+      "step": 4331
+    },
+    {
+      "epoch": 225.97,
+      "eval_gen_len": 9.8522,
+      "eval_loss": 0.9046717286109924,
+      "eval_rouge1": 0.0171,
+      "eval_rouge2": 0.0029,
+      "eval_rougeL": 0.0145,
+      "eval_rougeLsum": 0.0141,
+      "eval_runtime": 12.5334,
+      "eval_samples_per_second": 9.175,
+      "eval_steps_per_second": 1.596,
+      "step": 4350
+    },
+    {
+      "epoch": 226.96,
+      "eval_gen_len": 9.6,
+      "eval_loss": 0.9021281599998474,
+      "eval_rouge1": 0.0167,
+      "eval_rouge2": 0.0025,
+      "eval_rougeL": 0.0145,
+      "eval_rougeLsum": 0.0141,
+      "eval_runtime": 12.1542,
+      "eval_samples_per_second": 9.462,
+      "eval_steps_per_second": 1.646,
+      "step": 4369
+    },
+    {
+      "epoch": 228.0,
+      "eval_gen_len": 9.6261,
+      "eval_loss": 0.8991298675537109,
+      "eval_rouge1": 0.0181,
+      "eval_rouge2": 0.0019,
+      "eval_rougeL": 0.0153,
+      "eval_rougeLsum": 0.0148,
+      "eval_runtime": 12.1167,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 1.651,
+      "step": 4389
+    },
+    {
+      "epoch": 228.99,
+      "eval_gen_len": 9.687,
+      "eval_loss": 0.8962268829345703,
+      "eval_rouge1": 0.0217,
+      "eval_rouge2": 0.0027,
+      "eval_rougeL": 0.0176,
+      "eval_rougeLsum": 0.0172,
+      "eval_runtime": 12.1668,
+      "eval_samples_per_second": 9.452,
+      "eval_steps_per_second": 1.644,
+      "step": 4408
+    },
+    {
+      "epoch": 229.97,
+      "eval_gen_len": 9.2435,
+      "eval_loss": 0.8939462304115295,
+      "eval_rouge1": 0.0223,
+      "eval_rouge2": 0.0029,
+      "eval_rougeL": 0.0178,
+      "eval_rougeLsum": 0.0175,
+      "eval_runtime": 12.2423,
+      "eval_samples_per_second": 9.394,
+      "eval_steps_per_second": 1.634,
+      "step": 4427
+    },
+    {
+      "epoch": 230.96,
+      "eval_gen_len": 9.1304,
+      "eval_loss": 0.8907042145729065,
+      "eval_rouge1": 0.0216,
+      "eval_rouge2": 0.0029,
+      "eval_rougeL": 0.0179,
+      "eval_rougeLsum": 0.0175,
+      "eval_runtime": 12.2251,
+      "eval_samples_per_second": 9.407,
+      "eval_steps_per_second": 1.636,
+      "step": 4446
+    },
+    {
+      "epoch": 232.0,
+      "eval_gen_len": 8.9652,
+      "eval_loss": 0.8877010345458984,
+      "eval_rouge1": 0.0211,
+      "eval_rouge2": 0.0025,
+      "eval_rougeL": 0.0169,
+      "eval_rougeLsum": 0.0166,
+      "eval_runtime": 12.123,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 1.65,
+      "step": 4466
+    },
+    {
+      "epoch": 232.99,
+      "eval_gen_len": 8.7739,
+      "eval_loss": 0.8858217597007751,
+      "eval_rouge1": 0.0209,
+      "eval_rouge2": 0.0027,
+      "eval_rougeL": 0.0168,
+      "eval_rougeLsum": 0.0164,
+      "eval_runtime": 12.3317,
+      "eval_samples_per_second": 9.326,
+      "eval_steps_per_second": 1.622,
+      "step": 4485
+    },
+    {
+      "epoch": 233.77,
+      "grad_norm": 0.45308393239974976,
+      "learning_rate": 5.267368421052632e-06,
+      "loss": 1.0189,
+      "step": 4500
+    },
+    {
+      "epoch": 233.97,
+      "eval_gen_len": 8.6087,
+      "eval_loss": 0.8837451934814453,
+      "eval_rouge1": 0.0221,
+      "eval_rouge2": 0.0032,
+      "eval_rougeL": 0.0177,
+      "eval_rougeLsum": 0.0173,
+      "eval_runtime": 13.1885,
+      "eval_samples_per_second": 8.72,
+      "eval_steps_per_second": 1.516,
+      "step": 4504
+    },
+    {
+      "epoch": 234.96,
+      "eval_gen_len": 8.487,
+      "eval_loss": 0.8812865614891052,
+      "eval_rouge1": 0.0224,
+      "eval_rouge2": 0.003,
+      "eval_rougeL": 0.0175,
+      "eval_rougeLsum": 0.0172,
+      "eval_runtime": 12.2108,
+      "eval_samples_per_second": 9.418,
+      "eval_steps_per_second": 1.638,
+      "step": 4523
+    },
+    {
+      "epoch": 236.0,
+      "eval_gen_len": 8.2957,
+      "eval_loss": 0.8780920505523682,
+      "eval_rouge1": 0.0225,
+      "eval_rouge2": 0.0028,
+      "eval_rougeL": 0.0171,
+      "eval_rougeLsum": 0.0168,
+      "eval_runtime": 12.2449,
+      "eval_samples_per_second": 9.392,
+      "eval_steps_per_second": 1.633,
+      "step": 4543
+    },
+    {
+      "epoch": 236.99,
+      "eval_gen_len": 7.9304,
+      "eval_loss": 0.8753093481063843,
+      "eval_rouge1": 0.0215,
+      "eval_rouge2": 0.0027,
+      "eval_rougeL": 0.016,
+      "eval_rougeLsum": 0.0158,
+      "eval_runtime": 12.4484,
+      "eval_samples_per_second": 9.238,
+      "eval_steps_per_second": 1.607,
+      "step": 4562
+    },
+    {
+      "epoch": 237.97,
+      "eval_gen_len": 7.8174,
+      "eval_loss": 0.8730840086936951,
+      "eval_rouge1": 0.0211,
+      "eval_rouge2": 0.0027,
+      "eval_rougeL": 0.016,
+      "eval_rougeLsum": 0.0156,
+      "eval_runtime": 12.5923,
+      "eval_samples_per_second": 9.133,
+      "eval_steps_per_second": 1.588,
+      "step": 4581
+    },
+    {
+      "epoch": 238.96,
+      "eval_gen_len": 7.687,
+      "eval_loss": 0.8703946471214294,
+      "eval_rouge1": 0.0209,
+      "eval_rouge2": 0.0027,
+      "eval_rougeL": 0.0158,
+      "eval_rougeLsum": 0.0154,
+      "eval_runtime": 12.8465,
+      "eval_samples_per_second": 8.952,
+      "eval_steps_per_second": 1.557,
+      "step": 4600
+    },
+    {
+      "epoch": 240.0,
+      "eval_gen_len": 7.3652,
+      "eval_loss": 0.8674846887588501,
+      "eval_rouge1": 0.0211,
+      "eval_rouge2": 0.0027,
+      "eval_rougeL": 0.0158,
+      "eval_rougeLsum": 0.0154,
+      "eval_runtime": 12.4238,
+      "eval_samples_per_second": 9.256,
+      "eval_steps_per_second": 1.61,
+      "step": 4620
+    },
+    {
+      "epoch": 240.99,
+      "eval_gen_len": 7.2609,
+      "eval_loss": 0.8647277355194092,
+      "eval_rouge1": 0.0204,
+      "eval_rouge2": 0.0022,
+      "eval_rougeL": 0.0147,
+      "eval_rougeLsum": 0.0143,
+      "eval_runtime": 12.6703,
+      "eval_samples_per_second": 9.076,
+      "eval_steps_per_second": 1.578,
+      "step": 4639
+    },
+    {
+      "epoch": 241.97,
+      "eval_gen_len": 7.0609,
+      "eval_loss": 0.8625157475471497,
+      "eval_rouge1": 0.0206,
+      "eval_rouge2": 0.0023,
+      "eval_rougeL": 0.0152,
+      "eval_rougeLsum": 0.0149,
+      "eval_runtime": 12.5819,
+      "eval_samples_per_second": 9.14,
+      "eval_steps_per_second": 1.59,
+      "step": 4658
+    },
+    {
+      "epoch": 242.96,
+      "eval_gen_len": 6.5652,
+      "eval_loss": 0.8605428338050842,
+      "eval_rouge1": 0.0182,
+      "eval_rouge2": 0.0017,
+      "eval_rougeL": 0.0133,
+      "eval_rougeLsum": 0.0131,
+      "eval_runtime": 13.1768,
+      "eval_samples_per_second": 8.727,
+      "eval_steps_per_second": 1.518,
+      "step": 4677
+    },
+    {
+      "epoch": 244.0,
+      "eval_gen_len": 6.4261,
+      "eval_loss": 0.8578657507896423,
+      "eval_rouge1": 0.0177,
+      "eval_rouge2": 0.0021,
+      "eval_rougeL": 0.0134,
+      "eval_rougeLsum": 0.0131,
+      "eval_runtime": 12.6947,
+      "eval_samples_per_second": 9.059,
+      "eval_steps_per_second": 1.575,
+      "step": 4697
+    },
+    {
+      "epoch": 244.99,
+      "eval_gen_len": 6.2783,
+      "eval_loss": 0.8557173609733582,
+      "eval_rouge1": 0.0177,
+      "eval_rouge2": 0.0021,
+      "eval_rougeL": 0.0134,
+      "eval_rougeLsum": 0.013,
+      "eval_runtime": 12.5391,
+      "eval_samples_per_second": 9.171,
+      "eval_steps_per_second": 1.595,
+      "step": 4716
+    },
+    {
+      "epoch": 245.97,
+      "eval_gen_len": 6.2435,
+      "eval_loss": 0.8529919981956482,
+      "eval_rouge1": 0.0169,
+      "eval_rouge2": 0.0014,
+      "eval_rougeL": 0.0131,
+      "eval_rougeLsum": 0.0127,
+      "eval_runtime": 12.7454,
+      "eval_samples_per_second": 9.023,
+      "eval_steps_per_second": 1.569,
+      "step": 4735
+    },
+    {
+      "epoch": 246.96,
+      "eval_gen_len": 6.1565,
+      "eval_loss": 0.850603461265564,
+      "eval_rouge1": 0.0191,
+      "eval_rouge2": 0.0019,
+      "eval_rougeL": 0.0145,
+      "eval_rougeLsum": 0.0141,
+      "eval_runtime": 12.284,
+      "eval_samples_per_second": 9.362,
+      "eval_steps_per_second": 1.628,
+      "step": 4754
+    },
+    {
+      "epoch": 248.0,
+      "eval_gen_len": 5.9478,
+      "eval_loss": 0.8480112552642822,
+      "eval_rouge1": 0.0186,
+      "eval_rouge2": 0.0015,
+      "eval_rougeL": 0.0146,
+      "eval_rougeLsum": 0.0142,
+      "eval_runtime": 12.3791,
+      "eval_samples_per_second": 9.29,
+      "eval_steps_per_second": 1.616,
+      "step": 4774
+    },
+    {
+      "epoch": 248.99,
+      "eval_gen_len": 5.7043,
+      "eval_loss": 0.8458153009414673,
+      "eval_rouge1": 0.0173,
+      "eval_rouge2": 0.0013,
+      "eval_rougeL": 0.0137,
+      "eval_rougeLsum": 0.0131,
+      "eval_runtime": 12.4132,
+      "eval_samples_per_second": 9.264,
+      "eval_steps_per_second": 1.611,
+      "step": 4793
+    },
+    {
+      "epoch": 249.97,
+      "eval_gen_len": 5.7478,
+      "eval_loss": 0.8430487513542175,
+      "eval_rouge1": 0.0169,
+      "eval_rouge2": 0.0015,
+      "eval_rougeL": 0.0136,
+      "eval_rougeLsum": 0.0133,
+      "eval_runtime": 12.8141,
+      "eval_samples_per_second": 8.974,
+      "eval_steps_per_second": 1.561,
+      "step": 4812
+    },
+    {
+      "epoch": 250.96,
+      "eval_gen_len": 5.3739,
+      "eval_loss": 0.841323971748352,
+      "eval_rouge1": 0.0152,
+      "eval_rouge2": 0.0016,
+      "eval_rougeL": 0.0124,
+      "eval_rougeLsum": 0.0121,
+      "eval_runtime": 12.5346,
+      "eval_samples_per_second": 9.175,
+      "eval_steps_per_second": 1.596,
+      "step": 4831
+    },
+    {
+      "epoch": 252.0,
+      "eval_gen_len": 5.1565,
+      "eval_loss": 0.838948130607605,
+      "eval_rouge1": 0.0149,
+      "eval_rouge2": 0.0011,
+      "eval_rougeL": 0.0124,
+      "eval_rougeLsum": 0.012,
+      "eval_runtime": 13.0143,
+      "eval_samples_per_second": 8.836,
+      "eval_steps_per_second": 1.537,
+      "step": 4851
+    },
+    {
+      "epoch": 252.99,
+      "eval_gen_len": 4.9739,
+      "eval_loss": 0.8368021249771118,
+      "eval_rouge1": 0.0148,
+      "eval_rouge2": 0.0011,
+      "eval_rougeL": 0.0123,
+      "eval_rougeLsum": 0.0119,
+      "eval_runtime": 12.4217,
+      "eval_samples_per_second": 9.258,
+      "eval_steps_per_second": 1.61,
+      "step": 4870
+    },
+    {
+      "epoch": 253.97,
+      "eval_gen_len": 4.9652,
+      "eval_loss": 0.8342902660369873,
+      "eval_rouge1": 0.0158,
+      "eval_rouge2": 0.0011,
+      "eval_rougeL": 0.013,
+      "eval_rougeLsum": 0.0127,
+      "eval_runtime": 12.2067,
+      "eval_samples_per_second": 9.421,
+      "eval_steps_per_second": 1.638,
+      "step": 4889
+    },
+    {
+      "epoch": 254.96,
+      "eval_gen_len": 4.6522,
+      "eval_loss": 0.8321281671524048,
+      "eval_rouge1": 0.0145,
+      "eval_rouge2": 0.0009,
+      "eval_rougeL": 0.012,
+      "eval_rougeLsum": 0.0117,
+      "eval_runtime": 12.8126,
+      "eval_samples_per_second": 8.976,
+      "eval_steps_per_second": 1.561,
+      "step": 4908
+    },
+    {
+      "epoch": 256.0,
+      "eval_gen_len": 4.2522,
+      "eval_loss": 0.8296378254890442,
+      "eval_rouge1": 0.0139,
+      "eval_rouge2": 0.0009,
+      "eval_rougeL": 0.0113,
+      "eval_rougeLsum": 0.0112,
+      "eval_runtime": 12.3844,
+      "eval_samples_per_second": 9.286,
+      "eval_steps_per_second": 1.615,
+      "step": 4928
+    },
+    {
+      "epoch": 256.99,
+      "eval_gen_len": 4.1826,
+      "eval_loss": 0.8276596069335938,
+      "eval_rouge1": 0.0143,
+      "eval_rouge2": 0.0009,
+      "eval_rougeL": 0.0118,
+      "eval_rougeLsum": 0.0117,
+      "eval_runtime": 12.4227,
+      "eval_samples_per_second": 9.257,
+      "eval_steps_per_second": 1.61,
+      "step": 4947
+    },
+    {
+      "epoch": 257.97,
+      "eval_gen_len": 3.6261,
+      "eval_loss": 0.8265025019645691,
+      "eval_rouge1": 0.0127,
+      "eval_rouge2": 0.0007,
+      "eval_rougeL": 0.0103,
+      "eval_rougeLsum": 0.01,
+      "eval_runtime": 12.2122,
+      "eval_samples_per_second": 9.417,
+      "eval_steps_per_second": 1.638,
+      "step": 4966
+    },
+    {
+      "epoch": 258.96,
+      "eval_gen_len": 3.2609,
+      "eval_loss": 0.8242406845092773,
+      "eval_rouge1": 0.0122,
+      "eval_rouge2": 0.0013,
+      "eval_rougeL": 0.0102,
+      "eval_rougeLsum": 0.0101,
+      "eval_runtime": 12.2099,
+      "eval_samples_per_second": 9.419,
+      "eval_steps_per_second": 1.638,
+      "step": 4985
+    },
+    {
+      "epoch": 259.74,
+      "grad_norm": 0.4785182774066925,
+      "learning_rate": 4.741052631578948e-06,
+      "loss": 0.9442,
+      "step": 5000
+    },
+    {
+      "epoch": 260.0,
+      "eval_gen_len": 2.8,
+      "eval_loss": 0.8225219249725342,
+      "eval_rouge1": 0.0097,
+      "eval_rouge2": 0.0011,
+      "eval_rougeL": 0.0082,
+      "eval_rougeLsum": 0.0082,
+      "eval_runtime": 12.9652,
+      "eval_samples_per_second": 8.87,
+      "eval_steps_per_second": 1.543,
+      "step": 5005
+    },
+    {
+      "epoch": 260.99,
+      "eval_gen_len": 2.5652,
+      "eval_loss": 0.8206771612167358,
+      "eval_rouge1": 0.0087,
+      "eval_rouge2": 0.0011,
+      "eval_rougeL": 0.0071,
+      "eval_rougeLsum": 0.0069,
+      "eval_runtime": 12.2186,
+      "eval_samples_per_second": 9.412,
+      "eval_steps_per_second": 1.637,
+      "step": 5024
+    },
+    {
+      "epoch": 261.97,
+      "eval_gen_len": 2.2348,
+      "eval_loss": 0.818169891834259,
+      "eval_rouge1": 0.0072,
+      "eval_rouge2": 0.0005,
+      "eval_rougeL": 0.0059,
+      "eval_rougeLsum": 0.0058,
+      "eval_runtime": 12.5714,
+      "eval_samples_per_second": 9.148,
+      "eval_steps_per_second": 1.591,
+      "step": 5043
+    },
+    {
+      "epoch": 262.96,
+      "eval_gen_len": 2.2,
+      "eval_loss": 0.8162385821342468,
+      "eval_rouge1": 0.0062,
+      "eval_rouge2": 0.0002,
+      "eval_rougeL": 0.0051,
+      "eval_rougeLsum": 0.0051,
+      "eval_runtime": 12.5747,
+      "eval_samples_per_second": 9.145,
+      "eval_steps_per_second": 1.59,
+      "step": 5062
+    },
+    {
+      "epoch": 264.0,
+      "eval_gen_len": 2.2087,
+      "eval_loss": 0.8145304918289185,
+      "eval_rouge1": 0.0068,
+      "eval_rouge2": 0.0005,
+      "eval_rougeL": 0.0056,
+      "eval_rougeLsum": 0.0057,
+      "eval_runtime": 12.6293,
+      "eval_samples_per_second": 9.106,
+      "eval_steps_per_second": 1.584,
+      "step": 5082
+    },
+    {
+      "epoch": 264.99,
+      "eval_gen_len": 2.3304,
+      "eval_loss": 0.8127499222755432,
+      "eval_rouge1": 0.0086,
+      "eval_rouge2": 0.002,
+      "eval_rougeL": 0.0076,
+      "eval_rougeLsum": 0.0075,
+      "eval_runtime": 13.2312,
+      "eval_samples_per_second": 8.692,
+      "eval_steps_per_second": 1.512,
+      "step": 5101
+    },
+    {
+      "epoch": 265.97,
+      "eval_gen_len": 1.8957,
+      "eval_loss": 0.811177670955658,
+      "eval_rouge1": 0.0057,
+      "eval_rouge2": 0.001,
+      "eval_rougeL": 0.0052,
+      "eval_rougeLsum": 0.0051,
+      "eval_runtime": 12.5731,
+      "eval_samples_per_second": 9.147,
+      "eval_steps_per_second": 1.591,
+      "step": 5120
+    },
+    {
+      "epoch": 266.96,
+      "eval_gen_len": 1.513,
+      "eval_loss": 0.8090675473213196,
+      "eval_rouge1": 0.0042,
+      "eval_rouge2": 0.0007,
+      "eval_rougeL": 0.004,
+      "eval_rougeLsum": 0.004,
+      "eval_runtime": 12.391,
+      "eval_samples_per_second": 9.281,
+      "eval_steps_per_second": 1.614,
+      "step": 5139
+    },
+    {
+      "epoch": 268.0,
+      "eval_gen_len": 1.2435,
+      "eval_loss": 0.8073368668556213,
+      "eval_rouge1": 0.0031,
+      "eval_rouge2": 0.0006,
+      "eval_rougeL": 0.0029,
+      "eval_rougeLsum": 0.0029,
+      "eval_runtime": 12.3681,
+      "eval_samples_per_second": 9.298,
+      "eval_steps_per_second": 1.617,
+      "step": 5159
+    },
+    {
+      "epoch": 268.99,
+      "eval_gen_len": 1.0348,
+      "eval_loss": 0.8059150576591492,
+      "eval_rouge1": 0.0031,
+      "eval_rouge2": 0.0006,
+      "eval_rougeL": 0.0029,
+      "eval_rougeLsum": 0.0029,
+      "eval_runtime": 12.3463,
+      "eval_samples_per_second": 9.315,
+      "eval_steps_per_second": 1.62,
+      "step": 5178
+    },
+    {
+      "epoch": 269.97,
+      "eval_gen_len": 0.6348,
+      "eval_loss": 0.8042454123497009,
+      "eval_rouge1": 0.0007,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0007,
+      "eval_rougeLsum": 0.0007,
+      "eval_runtime": 12.4194,
+      "eval_samples_per_second": 9.26,
+      "eval_steps_per_second": 1.61,
+      "step": 5197
+    },
+    {
+      "epoch": 270.96,
+      "eval_gen_len": 0.7304,
+      "eval_loss": 0.8023030161857605,
+      "eval_rouge1": 0.0012,
+      "eval_rouge2": 0.0003,
+      "eval_rougeL": 0.0011,
+      "eval_rougeLsum": 0.001,
+      "eval_runtime": 12.5901,
+      "eval_samples_per_second": 9.134,
+      "eval_steps_per_second": 1.589,
+      "step": 5216
+    },
+    {
+      "epoch": 272.0,
+      "eval_gen_len": 0.8,
+      "eval_loss": 0.8001125454902649,
+      "eval_rouge1": 0.0012,
+      "eval_rouge2": 0.0003,
+      "eval_rougeL": 0.0011,
+      "eval_rougeLsum": 0.001,
+      "eval_runtime": 12.5732,
+      "eval_samples_per_second": 9.146,
+      "eval_steps_per_second": 1.591,
+      "step": 5236
+    },
+    {
+      "epoch": 272.99,
+      "eval_gen_len": 0.6348,
+      "eval_loss": 0.7986020445823669,
+      "eval_rouge1": 0.0012,
+      "eval_rouge2": 0.0003,
+      "eval_rougeL": 0.0011,
+      "eval_rougeLsum": 0.001,
+      "eval_runtime": 12.389,
+      "eval_samples_per_second": 9.282,
+      "eval_steps_per_second": 1.614,
+      "step": 5255
+    },
+    {
+      "epoch": 273.97,
+      "eval_gen_len": 0.7478,
+      "eval_loss": 0.7969604730606079,
+      "eval_rouge1": 0.0012,
+      "eval_rouge2": 0.0003,
+      "eval_rougeL": 0.0011,
+      "eval_rougeLsum": 0.001,
+      "eval_runtime": 12.3419,
+      "eval_samples_per_second": 9.318,
+      "eval_steps_per_second": 1.62,
+      "step": 5274
+    },
+    {
+      "epoch": 274.96,
+      "eval_gen_len": 0.5826,
+      "eval_loss": 0.795600175857544,
+      "eval_rouge1": 0.0004,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0004,
+      "eval_rougeLsum": 0.0004,
+      "eval_runtime": 12.3627,
+      "eval_samples_per_second": 9.302,
+      "eval_steps_per_second": 1.618,
+      "step": 5293
+    },
+    {
+      "epoch": 276.0,
+      "eval_gen_len": 0.4,
+      "eval_loss": 0.7938172817230225,
+      "eval_rouge1": 0.0004,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0004,
+      "eval_rougeLsum": 0.0004,
+      "eval_runtime": 12.3706,
+      "eval_samples_per_second": 9.296,
+      "eval_steps_per_second": 1.617,
+      "step": 5313
+    },
+    {
+      "epoch": 276.99,
+      "eval_gen_len": 0.2261,
+      "eval_loss": 0.7923696041107178,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.4258,
+      "eval_samples_per_second": 9.255,
+      "eval_steps_per_second": 1.61,
+      "step": 5332
+    },
+    {
+      "epoch": 277.97,
+      "eval_gen_len": 0.2261,
+      "eval_loss": 0.7907570600509644,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.5909,
+      "eval_samples_per_second": 9.134,
+      "eval_steps_per_second": 1.588,
+      "step": 5351
+    },
+    {
+      "epoch": 278.96,
+      "eval_gen_len": 0.2,
+      "eval_loss": 0.7891109585762024,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 13.0018,
+      "eval_samples_per_second": 8.845,
+      "eval_steps_per_second": 1.538,
+      "step": 5370
+    },
+    {
+      "epoch": 280.0,
+      "eval_gen_len": 0.1826,
+      "eval_loss": 0.787673830986023,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.0698,
+      "eval_samples_per_second": 9.528,
+      "eval_steps_per_second": 1.657,
+      "step": 5390
+    },
+    {
+      "epoch": 280.99,
+      "eval_gen_len": 0.1739,
+      "eval_loss": 0.785959005355835,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.4254,
+      "eval_samples_per_second": 9.255,
+      "eval_steps_per_second": 1.61,
+      "step": 5409
+    },
+    {
+      "epoch": 281.97,
+      "eval_gen_len": 0.1739,
+      "eval_loss": 0.7843196988105774,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.6986,
+      "eval_samples_per_second": 9.056,
+      "eval_steps_per_second": 1.575,
+      "step": 5428
+    },
+    {
+      "epoch": 282.96,
+      "eval_gen_len": 0.1739,
+      "eval_loss": 0.7826663851737976,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.2827,
+      "eval_samples_per_second": 9.363,
+      "eval_steps_per_second": 1.628,
+      "step": 5447
+    },
+    {
+      "epoch": 284.0,
+      "eval_gen_len": 0.1739,
+      "eval_loss": 0.7811480760574341,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.6013,
+      "eval_samples_per_second": 9.126,
+      "eval_steps_per_second": 1.587,
+      "step": 5467
+    },
+    {
+      "epoch": 284.99,
+      "eval_gen_len": 0.1652,
+      "eval_loss": 0.7799001932144165,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.3727,
+      "eval_samples_per_second": 9.295,
+      "eval_steps_per_second": 1.616,
+      "step": 5486
+    },
+    {
+      "epoch": 285.71,
+      "grad_norm": 0.36042362451553345,
+      "learning_rate": 4.214736842105263e-06,
+      "loss": 0.8855,
+      "step": 5500
+    },
+    {
+      "epoch": 285.97,
+      "eval_gen_len": 0.1652,
+      "eval_loss": 0.7784348726272583,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.4298,
+      "eval_samples_per_second": 9.252,
+      "eval_steps_per_second": 1.609,
+      "step": 5505
+    },
+    {
+      "epoch": 286.96,
+      "eval_gen_len": 0.1652,
+      "eval_loss": 0.7772350311279297,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.5935,
+      "eval_samples_per_second": 9.132,
+      "eval_steps_per_second": 1.588,
+      "step": 5524
+    },
+    {
+      "epoch": 288.0,
+      "eval_gen_len": 0.1652,
+      "eval_loss": 0.775896430015564,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.6907,
+      "eval_samples_per_second": 9.062,
+      "eval_steps_per_second": 1.576,
+      "step": 5544
+    },
+    {
+      "epoch": 288.99,
+      "eval_gen_len": 0.1652,
+      "eval_loss": 0.7743993401527405,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.7195,
+      "eval_samples_per_second": 9.041,
+      "eval_steps_per_second": 1.572,
+      "step": 5563
+    },
+    {
+      "epoch": 289.97,
+      "eval_gen_len": 0.1652,
+      "eval_loss": 0.7728458046913147,
+      "eval_rouge1": 0.0002,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0002,
+      "eval_rougeLsum": 0.0002,
+      "eval_runtime": 12.71,
+      "eval_samples_per_second": 9.048,
+      "eval_steps_per_second": 1.574,
+      "step": 5582
+    },
+    {
+      "epoch": 290.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7715795636177063,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6969,
+      "eval_samples_per_second": 9.057,
+      "eval_steps_per_second": 1.575,
+      "step": 5601
+    },
+    {
+      "epoch": 292.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7701930999755859,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.902,
+      "eval_samples_per_second": 8.913,
+      "eval_steps_per_second": 1.55,
+      "step": 5621
+    },
+    {
+      "epoch": 292.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7691376209259033,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3513,
+      "eval_samples_per_second": 9.311,
+      "eval_steps_per_second": 1.619,
+      "step": 5640
+    },
+    {
+      "epoch": 293.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7679579257965088,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4173,
+      "eval_samples_per_second": 9.261,
+      "eval_steps_per_second": 1.611,
+      "step": 5659
+    },
+    {
+      "epoch": 294.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7667289972305298,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6438,
+      "eval_samples_per_second": 9.095,
+      "eval_steps_per_second": 1.582,
+      "step": 5678
+    },
+    {
+      "epoch": 296.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7650233507156372,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.1233,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 1.65,
+      "step": 5698
+    },
+    {
+      "epoch": 296.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7638988494873047,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.1018,
+      "eval_samples_per_second": 9.503,
+      "eval_steps_per_second": 1.653,
+      "step": 5717
+    },
+    {
+      "epoch": 297.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7627271413803101,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.0736,
+      "eval_samples_per_second": 9.525,
+      "eval_steps_per_second": 1.657,
+      "step": 5736
+    },
+    {
+      "epoch": 298.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.761401891708374,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4179,
+      "eval_samples_per_second": 9.261,
+      "eval_steps_per_second": 1.611,
+      "step": 5755
+    },
+    {
+      "epoch": 300.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7603045105934143,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.1939,
+      "eval_samples_per_second": 9.431,
+      "eval_steps_per_second": 1.64,
+      "step": 5775
+    },
+    {
+      "epoch": 300.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7593241333961487,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4302,
+      "eval_samples_per_second": 9.252,
+      "eval_steps_per_second": 1.609,
+      "step": 5794
+    },
+    {
+      "epoch": 301.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7581080198287964,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7272,
+      "eval_samples_per_second": 9.036,
+      "eval_steps_per_second": 1.571,
+      "step": 5813
+    },
+    {
+      "epoch": 302.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7565290927886963,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7621,
+      "eval_samples_per_second": 9.011,
+      "eval_steps_per_second": 1.567,
+      "step": 5832
+    },
+    {
+      "epoch": 304.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7556654810905457,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4563,
+      "eval_samples_per_second": 9.232,
+      "eval_steps_per_second": 1.606,
+      "step": 5852
+    },
+    {
+      "epoch": 304.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.754369854927063,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.207,
+      "eval_samples_per_second": 9.421,
+      "eval_steps_per_second": 1.638,
+      "step": 5871
+    },
+    {
+      "epoch": 305.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7534385919570923,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6305,
+      "eval_samples_per_second": 9.105,
+      "eval_steps_per_second": 1.583,
+      "step": 5890
+    },
+    {
+      "epoch": 306.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7526547908782959,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4513,
+      "eval_samples_per_second": 9.236,
+      "eval_steps_per_second": 1.606,
+      "step": 5909
+    },
+    {
+      "epoch": 308.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7513379454612732,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.349,
+      "eval_samples_per_second": 9.312,
+      "eval_steps_per_second": 1.62,
+      "step": 5929
+    },
+    {
+      "epoch": 308.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7506363987922668,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.0976,
+      "eval_samples_per_second": 9.506,
+      "eval_steps_per_second": 1.653,
+      "step": 5948
+    },
+    {
+      "epoch": 309.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7496155500411987,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3015,
+      "eval_samples_per_second": 9.348,
+      "eval_steps_per_second": 1.626,
+      "step": 5967
+    },
+    {
+      "epoch": 310.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7488384246826172,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3367,
+      "eval_samples_per_second": 9.322,
+      "eval_steps_per_second": 1.621,
+      "step": 5986
+    },
+    {
+      "epoch": 311.69,
+      "grad_norm": 0.3260189890861511,
+      "learning_rate": 3.6884210526315794e-06,
+      "loss": 0.8402,
+      "step": 6000
+    },
+    {
+      "epoch": 312.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7474696040153503,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3992,
+      "eval_samples_per_second": 9.275,
+      "eval_steps_per_second": 1.613,
+      "step": 6006
+    },
+    {
+      "epoch": 312.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7464930415153503,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 11.9232,
+      "eval_samples_per_second": 9.645,
+      "eval_steps_per_second": 1.677,
+      "step": 6025
+    },
+    {
+      "epoch": 313.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7456102967262268,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6801,
+      "eval_samples_per_second": 9.069,
+      "eval_steps_per_second": 1.577,
+      "step": 6044
+    },
+    {
+      "epoch": 314.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7446662783622742,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3568,
+      "eval_samples_per_second": 9.307,
+      "eval_steps_per_second": 1.619,
+      "step": 6063
+    },
+    {
+      "epoch": 316.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7433856725692749,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 11.987,
+      "eval_samples_per_second": 9.594,
+      "eval_steps_per_second": 1.668,
+      "step": 6083
+    },
+    {
+      "epoch": 316.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7426111698150635,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6966,
+      "eval_samples_per_second": 9.058,
+      "eval_steps_per_second": 1.575,
+      "step": 6102
+    },
+    {
+      "epoch": 317.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7413787841796875,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2025,
+      "eval_samples_per_second": 9.424,
+      "eval_steps_per_second": 1.639,
+      "step": 6121
+    },
+    {
+      "epoch": 318.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7404425144195557,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.9943,
+      "eval_samples_per_second": 8.85,
+      "eval_steps_per_second": 1.539,
+      "step": 6140
+    },
+    {
+      "epoch": 320.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7396877408027649,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3803,
+      "eval_samples_per_second": 9.289,
+      "eval_steps_per_second": 1.615,
+      "step": 6160
+    },
+    {
+      "epoch": 320.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.739030122756958,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 13.3182,
+      "eval_samples_per_second": 8.635,
+      "eval_steps_per_second": 1.502,
+      "step": 6179
+    },
+    {
+      "epoch": 321.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7381538152694702,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7464,
+      "eval_samples_per_second": 9.022,
+      "eval_steps_per_second": 1.569,
+      "step": 6198
+    },
+    {
+      "epoch": 322.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7372981309890747,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.1988,
+      "eval_samples_per_second": 9.427,
+      "eval_steps_per_second": 1.64,
+      "step": 6217
+    },
+    {
+      "epoch": 324.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7361249327659607,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 13.3472,
+      "eval_samples_per_second": 8.616,
+      "eval_steps_per_second": 1.498,
+      "step": 6237
+    },
+    {
+      "epoch": 324.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.735177218914032,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2208,
+      "eval_samples_per_second": 9.41,
+      "eval_steps_per_second": 1.637,
+      "step": 6256
+    },
+    {
+      "epoch": 325.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7344561219215393,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.9594,
+      "eval_samples_per_second": 8.874,
+      "eval_steps_per_second": 1.543,
+      "step": 6275
+    },
+    {
+      "epoch": 326.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7334731817245483,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 13.1094,
+      "eval_samples_per_second": 8.772,
+      "eval_steps_per_second": 1.526,
+      "step": 6294
+    },
+    {
+      "epoch": 328.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7326551079750061,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7729,
+      "eval_samples_per_second": 9.003,
+      "eval_steps_per_second": 1.566,
+      "step": 6314
+    },
+    {
+      "epoch": 328.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7316291332244873,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.415,
+      "eval_samples_per_second": 9.263,
+      "eval_steps_per_second": 1.611,
+      "step": 6333
+    },
+    {
+      "epoch": 329.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7311994433403015,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.9044,
+      "eval_samples_per_second": 8.912,
+      "eval_steps_per_second": 1.55,
+      "step": 6352
+    },
+    {
+      "epoch": 330.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7306154370307922,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 13.0335,
+      "eval_samples_per_second": 8.823,
+      "eval_steps_per_second": 1.535,
+      "step": 6371
+    },
+    {
+      "epoch": 332.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7298057675361633,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5384,
+      "eval_samples_per_second": 9.172,
+      "eval_steps_per_second": 1.595,
+      "step": 6391
+    },
+    {
+      "epoch": 332.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7290323972702026,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7202,
+      "eval_samples_per_second": 9.041,
+      "eval_steps_per_second": 1.572,
+      "step": 6410
+    },
+    {
+      "epoch": 333.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7283275127410889,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5197,
+      "eval_samples_per_second": 9.186,
+      "eval_steps_per_second": 1.597,
+      "step": 6429
+    },
+    {
+      "epoch": 334.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7273982763290405,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6808,
+      "eval_samples_per_second": 9.069,
+      "eval_steps_per_second": 1.577,
+      "step": 6448
+    },
+    {
+      "epoch": 336.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7265883684158325,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 13.0488,
+      "eval_samples_per_second": 8.813,
+      "eval_steps_per_second": 1.533,
+      "step": 6468
+    },
+    {
+      "epoch": 336.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7261592745780945,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6573,
+      "eval_samples_per_second": 9.086,
+      "eval_steps_per_second": 1.58,
+      "step": 6487
+    },
+    {
+      "epoch": 337.66,
+      "grad_norm": 0.288989782333374,
+      "learning_rate": 3.1621052631578953e-06,
+      "loss": 0.8058,
+      "step": 6500
+    },
+    {
+      "epoch": 337.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7252629995346069,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4295,
+      "eval_samples_per_second": 9.252,
+      "eval_steps_per_second": 1.609,
+      "step": 6506
+    },
+    {
+      "epoch": 338.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7245468497276306,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7865,
+      "eval_samples_per_second": 8.994,
+      "eval_steps_per_second": 1.564,
+      "step": 6525
+    },
+    {
+      "epoch": 340.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7239726185798645,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.9112,
+      "eval_samples_per_second": 8.907,
+      "eval_steps_per_second": 1.549,
+      "step": 6545
+    },
+    {
+      "epoch": 340.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7231466770172119,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5487,
+      "eval_samples_per_second": 9.164,
+      "eval_steps_per_second": 1.594,
+      "step": 6564
+    },
+    {
+      "epoch": 341.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7223904728889465,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 13.3347,
+      "eval_samples_per_second": 8.624,
+      "eval_steps_per_second": 1.5,
+      "step": 6583
+    },
+    {
+      "epoch": 342.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7218188643455505,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 13.0031,
+      "eval_samples_per_second": 8.844,
+      "eval_steps_per_second": 1.538,
+      "step": 6602
+    },
+    {
+      "epoch": 344.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7209810614585876,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4608,
+      "eval_samples_per_second": 9.229,
+      "eval_steps_per_second": 1.605,
+      "step": 6622
+    },
+    {
+      "epoch": 344.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7203324437141418,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5304,
+      "eval_samples_per_second": 9.178,
+      "eval_steps_per_second": 1.596,
+      "step": 6641
+    },
+    {
+      "epoch": 345.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7196723818778992,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6257,
+      "eval_samples_per_second": 9.108,
+      "eval_steps_per_second": 1.584,
+      "step": 6660
+    },
+    {
+      "epoch": 346.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7190775275230408,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4134,
+      "eval_samples_per_second": 9.264,
+      "eval_steps_per_second": 1.611,
+      "step": 6679
+    },
+    {
+      "epoch": 348.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7185074090957642,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.166,
+      "eval_samples_per_second": 9.453,
+      "eval_steps_per_second": 1.644,
+      "step": 6699
+    },
+    {
+      "epoch": 348.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7180371880531311,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.559,
+      "eval_samples_per_second": 9.157,
+      "eval_steps_per_second": 1.592,
+      "step": 6718
+    },
+    {
+      "epoch": 349.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.717097818851471,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3,
+      "eval_samples_per_second": 9.35,
+      "eval_steps_per_second": 1.626,
+      "step": 6737
+    },
+    {
+      "epoch": 350.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7164217829704285,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3583,
+      "eval_samples_per_second": 9.306,
+      "eval_steps_per_second": 1.618,
+      "step": 6756
+    },
+    {
+      "epoch": 352.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7158520817756653,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 13.0047,
+      "eval_samples_per_second": 8.843,
+      "eval_steps_per_second": 1.538,
+      "step": 6776
+    },
+    {
+      "epoch": 352.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7151947021484375,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6741,
+      "eval_samples_per_second": 9.074,
+      "eval_steps_per_second": 1.578,
+      "step": 6795
+    },
+    {
+      "epoch": 353.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7145124077796936,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6036,
+      "eval_samples_per_second": 9.124,
+      "eval_steps_per_second": 1.587,
+      "step": 6814
+    },
+    {
+      "epoch": 354.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7140352725982666,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2129,
+      "eval_samples_per_second": 9.416,
+      "eval_steps_per_second": 1.638,
+      "step": 6833
+    },
+    {
+      "epoch": 356.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7135369777679443,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4624,
+      "eval_samples_per_second": 9.228,
+      "eval_steps_per_second": 1.605,
+      "step": 6853
+    },
+    {
+      "epoch": 356.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7128369808197021,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.422,
+      "eval_samples_per_second": 9.258,
+      "eval_steps_per_second": 1.61,
+      "step": 6872
+    },
+    {
+      "epoch": 357.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7122591137886047,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2268,
+      "eval_samples_per_second": 9.406,
+      "eval_steps_per_second": 1.636,
+      "step": 6891
+    },
+    {
+      "epoch": 358.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7116859555244446,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.866,
+      "eval_samples_per_second": 8.938,
+      "eval_steps_per_second": 1.554,
+      "step": 6910
+    },
+    {
+      "epoch": 360.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7112235426902771,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4942,
+      "eval_samples_per_second": 9.204,
+      "eval_steps_per_second": 1.601,
+      "step": 6930
+    },
+    {
+      "epoch": 360.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7106695771217346,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4042,
+      "eval_samples_per_second": 9.271,
+      "eval_steps_per_second": 1.612,
+      "step": 6949
+    },
+    {
+      "epoch": 361.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.710101306438446,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2879,
+      "eval_samples_per_second": 9.359,
+      "eval_steps_per_second": 1.628,
+      "step": 6968
+    },
+    {
+      "epoch": 362.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7094107270240784,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.269,
+      "eval_samples_per_second": 9.373,
+      "eval_steps_per_second": 1.63,
+      "step": 6987
+    },
+    {
+      "epoch": 363.64,
+      "grad_norm": 0.44062402844429016,
+      "learning_rate": 2.635789473684211e-06,
+      "loss": 0.7798,
+      "step": 7000
+    },
+    {
+      "epoch": 364.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.708891749382019,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.9085,
+      "eval_samples_per_second": 8.909,
+      "eval_steps_per_second": 1.549,
+      "step": 7007
+    },
+    {
+      "epoch": 364.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7082711458206177,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.1554,
+      "eval_samples_per_second": 9.461,
+      "eval_steps_per_second": 1.645,
+      "step": 7026
+    },
+    {
+      "epoch": 365.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7078844904899597,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6546,
+      "eval_samples_per_second": 9.088,
+      "eval_steps_per_second": 1.58,
+      "step": 7045
+    },
+    {
+      "epoch": 366.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7073128819465637,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7172,
+      "eval_samples_per_second": 9.043,
+      "eval_steps_per_second": 1.573,
+      "step": 7064
+    },
+    {
+      "epoch": 368.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7066096067428589,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.1125,
+      "eval_samples_per_second": 9.494,
+      "eval_steps_per_second": 1.651,
+      "step": 7084
+    },
+    {
+      "epoch": 368.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7057228088378906,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5605,
+      "eval_samples_per_second": 9.156,
+      "eval_steps_per_second": 1.592,
+      "step": 7103
+    },
+    {
+      "epoch": 369.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7048721313476562,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3178,
+      "eval_samples_per_second": 9.336,
+      "eval_steps_per_second": 1.624,
+      "step": 7122
+    },
+    {
+      "epoch": 370.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7041941285133362,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2739,
+      "eval_samples_per_second": 9.369,
+      "eval_steps_per_second": 1.629,
+      "step": 7141
+    },
+    {
+      "epoch": 372.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7035704255104065,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3207,
+      "eval_samples_per_second": 9.334,
+      "eval_steps_per_second": 1.623,
+      "step": 7161
+    },
+    {
+      "epoch": 372.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7029441595077515,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.094,
+      "eval_samples_per_second": 9.509,
+      "eval_steps_per_second": 1.654,
+      "step": 7180
+    },
+    {
+      "epoch": 373.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7022525668144226,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7042,
+      "eval_samples_per_second": 9.052,
+      "eval_steps_per_second": 1.574,
+      "step": 7199
+    },
+    {
+      "epoch": 374.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7016597986221313,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3663,
+      "eval_samples_per_second": 9.299,
+      "eval_steps_per_second": 1.617,
+      "step": 7218
+    },
+    {
+      "epoch": 376.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7011125683784485,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.2533,
+      "eval_samples_per_second": 9.385,
+      "eval_steps_per_second": 1.632,
+      "step": 7238
+    },
+    {
+      "epoch": 376.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7006986737251282,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4459,
+      "eval_samples_per_second": 9.24,
+      "eval_steps_per_second": 1.607,
+      "step": 7257
+    },
+    {
+      "epoch": 377.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.7000675201416016,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5813,
+      "eval_samples_per_second": 9.141,
+      "eval_steps_per_second": 1.59,
+      "step": 7276
+    },
+    {
+      "epoch": 378.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6994682550430298,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7191,
+      "eval_samples_per_second": 9.041,
+      "eval_steps_per_second": 1.572,
+      "step": 7295
+    },
+    {
+      "epoch": 380.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6987762451171875,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7423,
+      "eval_samples_per_second": 9.025,
+      "eval_steps_per_second": 1.57,
+      "step": 7315
+    },
+    {
+      "epoch": 380.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6981701254844666,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7919,
+      "eval_samples_per_second": 8.99,
+      "eval_steps_per_second": 1.563,
+      "step": 7334
+    },
+    {
+      "epoch": 381.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6976540088653564,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.6842,
+      "eval_samples_per_second": 9.066,
+      "eval_steps_per_second": 1.577,
+      "step": 7353
+    },
+    {
+      "epoch": 382.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6971992254257202,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7985,
+      "eval_samples_per_second": 8.985,
+      "eval_steps_per_second": 1.563,
+      "step": 7372
+    },
+    {
+      "epoch": 384.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6967973113059998,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5355,
+      "eval_samples_per_second": 9.174,
+      "eval_steps_per_second": 1.595,
+      "step": 7392
+    },
+    {
+      "epoch": 384.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6962406039237976,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.7996,
+      "eval_samples_per_second": 8.985,
+      "eval_steps_per_second": 1.563,
+      "step": 7411
+    },
+    {
+      "epoch": 385.97,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6957660913467407,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.4031,
+      "eval_samples_per_second": 9.272,
+      "eval_steps_per_second": 1.612,
+      "step": 7430
+    },
+    {
+      "epoch": 386.96,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6953439116477966,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 13.2572,
+      "eval_samples_per_second": 8.675,
+      "eval_steps_per_second": 1.509,
+      "step": 7449
+    },
+    {
+      "epoch": 388.0,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6948480606079102,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.5809,
+      "eval_samples_per_second": 9.141,
+      "eval_steps_per_second": 1.59,
+      "step": 7469
+    },
+    {
+      "epoch": 388.99,
+      "eval_gen_len": 0.0,
+      "eval_loss": 0.6944136023521423,
+      "eval_rouge1": 0.0,
+      "eval_rouge2": 0.0,
+      "eval_rougeL": 0.0,
+      "eval_rougeLsum": 0.0,
+      "eval_runtime": 12.3151,
+      "eval_samples_per_second": 9.338,
+      "eval_steps_per_second": 1.624,
+      "step": 7488
+    },
+    {
+      "epoch": 389.61,
+      "grad_norm": 0.2672542333602905,
+      "learning_rate": 2.1094736842105264e-06,
+      "loss": 0.7599,
+      "step": 7500
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 9500,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 500,
+  "save_steps": 500,
+  "total_flos": 2.1853261061318246e+17,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}