| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 218.1818181818182, | |
| "eval_steps": 500, | |
| "global_step": 6000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.98, | |
| "eval_gen_len": 14.9909, | |
| "eval_loss": 21.427785873413086, | |
| "eval_rouge1": 0.0931, | |
| "eval_rouge2": 0.021, | |
| "eval_rougeL": 0.0719, | |
| "eval_rougeLsum": 0.0715, | |
| "eval_runtime": 11.4039, | |
| "eval_samples_per_second": 9.646, | |
| "eval_steps_per_second": 1.228, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_gen_len": 14.9909, | |
| "eval_loss": 21.134571075439453, | |
| "eval_rouge1": 0.0948, | |
| "eval_rouge2": 0.0238, | |
| "eval_rougeL": 0.073, | |
| "eval_rougeLsum": 0.0724, | |
| "eval_runtime": 9.5984, | |
| "eval_samples_per_second": 11.46, | |
| "eval_steps_per_second": 1.459, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "eval_gen_len": 14.9909, | |
| "eval_loss": 20.702985763549805, | |
| "eval_rouge1": 0.0917, | |
| "eval_rouge2": 0.0203, | |
| "eval_rougeL": 0.0707, | |
| "eval_rougeLsum": 0.0703, | |
| "eval_runtime": 9.5997, | |
| "eval_samples_per_second": 11.459, | |
| "eval_steps_per_second": 1.458, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_gen_len": 15.3455, | |
| "eval_loss": 20.22304344177246, | |
| "eval_rouge1": 0.0991, | |
| "eval_rouge2": 0.0246, | |
| "eval_rougeL": 0.0781, | |
| "eval_rougeLsum": 0.0778, | |
| "eval_runtime": 9.5876, | |
| "eval_samples_per_second": 11.473, | |
| "eval_steps_per_second": 1.46, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "eval_gen_len": 15.4818, | |
| "eval_loss": 19.888574600219727, | |
| "eval_rouge1": 0.1057, | |
| "eval_rouge2": 0.0297, | |
| "eval_rougeL": 0.0845, | |
| "eval_rougeLsum": 0.0845, | |
| "eval_runtime": 9.549, | |
| "eval_samples_per_second": 11.519, | |
| "eval_steps_per_second": 1.466, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_gen_len": 15.7182, | |
| "eval_loss": 19.539011001586914, | |
| "eval_rouge1": 0.1034, | |
| "eval_rouge2": 0.0276, | |
| "eval_rougeL": 0.084, | |
| "eval_rougeLsum": 0.0839, | |
| "eval_runtime": 9.551, | |
| "eval_samples_per_second": 11.517, | |
| "eval_steps_per_second": 1.466, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "eval_gen_len": 15.8727, | |
| "eval_loss": 19.162174224853516, | |
| "eval_rouge1": 0.1023, | |
| "eval_rouge2": 0.0303, | |
| "eval_rougeL": 0.0837, | |
| "eval_rougeLsum": 0.0835, | |
| "eval_runtime": 9.5511, | |
| "eval_samples_per_second": 11.517, | |
| "eval_steps_per_second": 1.466, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_gen_len": 15.5091, | |
| "eval_loss": 18.730688095092773, | |
| "eval_rouge1": 0.1034, | |
| "eval_rouge2": 0.0342, | |
| "eval_rougeL": 0.0832, | |
| "eval_rougeLsum": 0.083, | |
| "eval_runtime": 9.5901, | |
| "eval_samples_per_second": 11.47, | |
| "eval_steps_per_second": 1.46, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "eval_gen_len": 15.8, | |
| "eval_loss": 18.190088272094727, | |
| "eval_rouge1": 0.0969, | |
| "eval_rouge2": 0.0344, | |
| "eval_rougeL": 0.0818, | |
| "eval_rougeLsum": 0.0815, | |
| "eval_runtime": 9.5799, | |
| "eval_samples_per_second": 11.482, | |
| "eval_steps_per_second": 1.461, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_gen_len": 15.9455, | |
| "eval_loss": 17.473072052001953, | |
| "eval_rouge1": 0.1041, | |
| "eval_rouge2": 0.0337, | |
| "eval_rougeL": 0.0857, | |
| "eval_rougeLsum": 0.0853, | |
| "eval_runtime": 9.5524, | |
| "eval_samples_per_second": 11.515, | |
| "eval_steps_per_second": 1.466, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "eval_gen_len": 18.0182, | |
| "eval_loss": 16.60153579711914, | |
| "eval_rouge1": 0.1001, | |
| "eval_rouge2": 0.029, | |
| "eval_rougeL": 0.0828, | |
| "eval_rougeLsum": 0.0828, | |
| "eval_runtime": 9.5623, | |
| "eval_samples_per_second": 11.503, | |
| "eval_steps_per_second": 1.464, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_gen_len": 16.9636, | |
| "eval_loss": 15.59052848815918, | |
| "eval_rouge1": 0.0875, | |
| "eval_rouge2": 0.0222, | |
| "eval_rougeL": 0.0724, | |
| "eval_rougeLsum": 0.0724, | |
| "eval_runtime": 9.5436, | |
| "eval_samples_per_second": 11.526, | |
| "eval_steps_per_second": 1.467, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 12.98, | |
| "eval_gen_len": 15.1455, | |
| "eval_loss": 14.362702369689941, | |
| "eval_rouge1": 0.0657, | |
| "eval_rouge2": 0.0153, | |
| "eval_rougeL": 0.0545, | |
| "eval_rougeLsum": 0.0543, | |
| "eval_runtime": 9.5681, | |
| "eval_samples_per_second": 11.497, | |
| "eval_steps_per_second": 1.463, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_gen_len": 13.2273, | |
| "eval_loss": 12.995957374572754, | |
| "eval_rouge1": 0.0449, | |
| "eval_rouge2": 0.0083, | |
| "eval_rougeL": 0.0363, | |
| "eval_rougeLsum": 0.0361, | |
| "eval_runtime": 9.506, | |
| "eval_samples_per_second": 11.572, | |
| "eval_steps_per_second": 1.473, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 14.98, | |
| "eval_gen_len": 13.1818, | |
| "eval_loss": 11.857050895690918, | |
| "eval_rouge1": 0.0337, | |
| "eval_rouge2": 0.0071, | |
| "eval_rougeL": 0.0303, | |
| "eval_rougeLsum": 0.0304, | |
| "eval_runtime": 9.5174, | |
| "eval_samples_per_second": 11.558, | |
| "eval_steps_per_second": 1.471, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_gen_len": 12.0545, | |
| "eval_loss": 10.728998184204102, | |
| "eval_rouge1": 0.0137, | |
| "eval_rouge2": 0.002, | |
| "eval_rougeL": 0.0114, | |
| "eval_rougeLsum": 0.0112, | |
| "eval_runtime": 9.5263, | |
| "eval_samples_per_second": 11.547, | |
| "eval_steps_per_second": 1.47, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 16.98, | |
| "eval_gen_len": 14.0818, | |
| "eval_loss": 9.675826072692871, | |
| "eval_rouge1": 0.0051, | |
| "eval_rouge2": 0.0007, | |
| "eval_rougeL": 0.005, | |
| "eval_rougeLsum": 0.0049, | |
| "eval_runtime": 9.4916, | |
| "eval_samples_per_second": 11.589, | |
| "eval_steps_per_second": 1.475, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_gen_len": 17.2545, | |
| "eval_loss": 8.581353187561035, | |
| "eval_rouge1": 0.0013, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0013, | |
| "eval_rougeLsum": 0.0013, | |
| "eval_runtime": 9.4979, | |
| "eval_samples_per_second": 11.581, | |
| "eval_steps_per_second": 1.474, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 18.18, | |
| "grad_norm": 5.896080017089844, | |
| "learning_rate": 1.8360493827160496e-05, | |
| "loss": 16.7039, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 18.98, | |
| "eval_gen_len": 17.5455, | |
| "eval_loss": 7.5404815673828125, | |
| "eval_rouge1": 0.0015, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0014, | |
| "eval_rougeLsum": 0.0014, | |
| "eval_runtime": 9.5287, | |
| "eval_samples_per_second": 11.544, | |
| "eval_steps_per_second": 1.469, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_gen_len": 18.5636, | |
| "eval_loss": 6.503509044647217, | |
| "eval_rouge1": 0.0006, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0006, | |
| "eval_rougeLsum": 0.0006, | |
| "eval_runtime": 9.5264, | |
| "eval_samples_per_second": 11.547, | |
| "eval_steps_per_second": 1.47, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 20.98, | |
| "eval_gen_len": 19.0, | |
| "eval_loss": 5.559322834014893, | |
| "eval_rouge1": 0.0008, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0007, | |
| "eval_rougeLsum": 0.0007, | |
| "eval_runtime": 9.5196, | |
| "eval_samples_per_second": 11.555, | |
| "eval_steps_per_second": 1.471, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_gen_len": 19.0, | |
| "eval_loss": 4.715544700622559, | |
| "eval_rouge1": 0.0002, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0002, | |
| "eval_rougeLsum": 0.0002, | |
| "eval_runtime": 9.5225, | |
| "eval_samples_per_second": 11.552, | |
| "eval_steps_per_second": 1.47, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 22.98, | |
| "eval_gen_len": 19.0, | |
| "eval_loss": 4.022518634796143, | |
| "eval_rouge1": 0.0007, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0007, | |
| "eval_rougeLsum": 0.0007, | |
| "eval_runtime": 9.5406, | |
| "eval_samples_per_second": 11.53, | |
| "eval_steps_per_second": 1.467, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_gen_len": 19.0, | |
| "eval_loss": 3.401372194290161, | |
| "eval_rouge1": 0.0006, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0006, | |
| "eval_rougeLsum": 0.0006, | |
| "eval_runtime": 9.5371, | |
| "eval_samples_per_second": 11.534, | |
| "eval_steps_per_second": 1.468, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 24.98, | |
| "eval_gen_len": 19.0, | |
| "eval_loss": 2.971496105194092, | |
| "eval_rouge1": 0.0005, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0005, | |
| "eval_rougeLsum": 0.0005, | |
| "eval_runtime": 9.5401, | |
| "eval_samples_per_second": 11.53, | |
| "eval_steps_per_second": 1.467, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_gen_len": 18.7273, | |
| "eval_loss": 2.6538097858428955, | |
| "eval_rouge1": 0.0008, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0008, | |
| "eval_rougeLsum": 0.0008, | |
| "eval_runtime": 9.5288, | |
| "eval_samples_per_second": 11.544, | |
| "eval_steps_per_second": 1.469, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 26.98, | |
| "eval_gen_len": 14.5, | |
| "eval_loss": 2.4105889797210693, | |
| "eval_rouge1": 0.0003, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0003, | |
| "eval_rougeLsum": 0.0003, | |
| "eval_runtime": 9.478, | |
| "eval_samples_per_second": 11.606, | |
| "eval_steps_per_second": 1.477, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_gen_len": 8.2636, | |
| "eval_loss": 2.2297565937042236, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4449, | |
| "eval_samples_per_second": 11.646, | |
| "eval_steps_per_second": 1.482, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 28.98, | |
| "eval_gen_len": 7.1455, | |
| "eval_loss": 2.1042511463165283, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4539, | |
| "eval_samples_per_second": 11.635, | |
| "eval_steps_per_second": 1.481, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_gen_len": 6.5545, | |
| "eval_loss": 2.011505126953125, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4645, | |
| "eval_samples_per_second": 11.622, | |
| "eval_steps_per_second": 1.479, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 30.98, | |
| "eval_gen_len": 6.0273, | |
| "eval_loss": 1.927019715309143, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.5048, | |
| "eval_samples_per_second": 11.573, | |
| "eval_steps_per_second": 1.473, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_gen_len": 5.4091, | |
| "eval_loss": 1.8653310537338257, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.49, | |
| "eval_samples_per_second": 11.591, | |
| "eval_steps_per_second": 1.475, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 32.98, | |
| "eval_gen_len": 5.1727, | |
| "eval_loss": 1.8196451663970947, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4647, | |
| "eval_samples_per_second": 11.622, | |
| "eval_steps_per_second": 1.479, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_gen_len": 5.1909, | |
| "eval_loss": 1.7811836004257202, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4798, | |
| "eval_samples_per_second": 11.604, | |
| "eval_steps_per_second": 1.477, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 34.98, | |
| "eval_gen_len": 5.3182, | |
| "eval_loss": 1.7490955591201782, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4662, | |
| "eval_samples_per_second": 11.62, | |
| "eval_steps_per_second": 1.479, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_gen_len": 5.3091, | |
| "eval_loss": 1.721943974494934, | |
| "eval_rouge1": 0.0003, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0003, | |
| "eval_rougeLsum": 0.0003, | |
| "eval_runtime": 9.4582, | |
| "eval_samples_per_second": 11.63, | |
| "eval_steps_per_second": 1.48, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 36.36, | |
| "grad_norm": 3.1814663410186768, | |
| "learning_rate": 1.6720987654320987e-05, | |
| "loss": 3.9957, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 36.98, | |
| "eval_gen_len": 5.3273, | |
| "eval_loss": 1.69921875, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.5006, | |
| "eval_samples_per_second": 11.578, | |
| "eval_steps_per_second": 1.474, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_gen_len": 5.2091, | |
| "eval_loss": 1.678276777267456, | |
| "eval_rouge1": 0.0003, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0003, | |
| "eval_rougeLsum": 0.0003, | |
| "eval_runtime": 9.4859, | |
| "eval_samples_per_second": 11.596, | |
| "eval_steps_per_second": 1.476, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 38.98, | |
| "eval_gen_len": 5.9273, | |
| "eval_loss": 1.657922625541687, | |
| "eval_rouge1": 0.0006, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0006, | |
| "eval_rougeLsum": 0.0006, | |
| "eval_runtime": 9.4757, | |
| "eval_samples_per_second": 11.609, | |
| "eval_steps_per_second": 1.477, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_gen_len": 6.1909, | |
| "eval_loss": 1.6388959884643555, | |
| "eval_rouge1": 0.0007, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0007, | |
| "eval_rougeLsum": 0.0007, | |
| "eval_runtime": 9.4468, | |
| "eval_samples_per_second": 11.644, | |
| "eval_steps_per_second": 1.482, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 40.98, | |
| "eval_gen_len": 5.6818, | |
| "eval_loss": 1.622721552848816, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4505, | |
| "eval_samples_per_second": 11.64, | |
| "eval_steps_per_second": 1.481, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_gen_len": 5.0818, | |
| "eval_loss": 1.6065285205841064, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4897, | |
| "eval_samples_per_second": 11.592, | |
| "eval_steps_per_second": 1.475, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 42.98, | |
| "eval_gen_len": 5.5091, | |
| "eval_loss": 1.5892502069473267, | |
| "eval_rouge1": 0.0002, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0002, | |
| "eval_rougeLsum": 0.0002, | |
| "eval_runtime": 9.4662, | |
| "eval_samples_per_second": 11.62, | |
| "eval_steps_per_second": 1.479, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_gen_len": 6.0636, | |
| "eval_loss": 1.568816900253296, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4716, | |
| "eval_samples_per_second": 11.614, | |
| "eval_steps_per_second": 1.478, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 44.98, | |
| "eval_gen_len": 6.6273, | |
| "eval_loss": 1.5522329807281494, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4809, | |
| "eval_samples_per_second": 11.602, | |
| "eval_steps_per_second": 1.477, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_gen_len": 6.5364, | |
| "eval_loss": 1.539686918258667, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4493, | |
| "eval_samples_per_second": 11.641, | |
| "eval_steps_per_second": 1.482, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 46.98, | |
| "eval_gen_len": 6.3909, | |
| "eval_loss": 1.527640700340271, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.441, | |
| "eval_samples_per_second": 11.651, | |
| "eval_steps_per_second": 1.483, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_gen_len": 7.2818, | |
| "eval_loss": 1.5062768459320068, | |
| "eval_rouge1": 0.0, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0, | |
| "eval_rougeLsum": 0.0, | |
| "eval_runtime": 9.4248, | |
| "eval_samples_per_second": 11.671, | |
| "eval_steps_per_second": 1.485, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 48.98, | |
| "eval_gen_len": 7.9273, | |
| "eval_loss": 1.4878034591674805, | |
| "eval_rouge1": 0.0002, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0002, | |
| "eval_rougeLsum": 0.0002, | |
| "eval_runtime": 9.5189, | |
| "eval_samples_per_second": 11.556, | |
| "eval_steps_per_second": 1.471, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_gen_len": 7.5636, | |
| "eval_loss": 1.4774630069732666, | |
| "eval_rouge1": 0.0006, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0006, | |
| "eval_rougeLsum": 0.0006, | |
| "eval_runtime": 9.4679, | |
| "eval_samples_per_second": 11.618, | |
| "eval_steps_per_second": 1.479, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 50.98, | |
| "eval_gen_len": 8.2273, | |
| "eval_loss": 1.4622873067855835, | |
| "eval_rouge1": 0.0007, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0007, | |
| "eval_rougeLsum": 0.0007, | |
| "eval_runtime": 9.4582, | |
| "eval_samples_per_second": 11.63, | |
| "eval_steps_per_second": 1.48, | |
| "step": 1402 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_gen_len": 8.5636, | |
| "eval_loss": 1.4518091678619385, | |
| "eval_rouge1": 0.0008, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0008, | |
| "eval_rougeLsum": 0.0008, | |
| "eval_runtime": 9.4604, | |
| "eval_samples_per_second": 11.627, | |
| "eval_steps_per_second": 1.48, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 52.98, | |
| "eval_gen_len": 8.2545, | |
| "eval_loss": 1.4443649053573608, | |
| "eval_rouge1": 0.0005, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0005, | |
| "eval_rougeLsum": 0.0005, | |
| "eval_runtime": 9.4485, | |
| "eval_samples_per_second": 11.642, | |
| "eval_steps_per_second": 1.482, | |
| "step": 1457 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_gen_len": 7.9545, | |
| "eval_loss": 1.4318201541900635, | |
| "eval_rouge1": 0.0005, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0003, | |
| "eval_rougeLsum": 0.0003, | |
| "eval_runtime": 9.5193, | |
| "eval_samples_per_second": 11.555, | |
| "eval_steps_per_second": 1.471, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 54.55, | |
| "grad_norm": 4.463714599609375, | |
| "learning_rate": 1.5078189300411523e-05, | |
| "loss": 1.9182, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 54.98, | |
| "eval_gen_len": 8.8273, | |
| "eval_loss": 1.4121521711349487, | |
| "eval_rouge1": 0.0009, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0009, | |
| "eval_rougeLsum": 0.0009, | |
| "eval_runtime": 9.4998, | |
| "eval_samples_per_second": 11.579, | |
| "eval_steps_per_second": 1.474, | |
| "step": 1512 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_gen_len": 7.8455, | |
| "eval_loss": 1.400160789489746, | |
| "eval_rouge1": 0.0012, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0013, | |
| "eval_rougeLsum": 0.0012, | |
| "eval_runtime": 9.4886, | |
| "eval_samples_per_second": 11.593, | |
| "eval_steps_per_second": 1.475, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 56.98, | |
| "eval_gen_len": 7.7909, | |
| "eval_loss": 1.3821996450424194, | |
| "eval_rouge1": 0.0003, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0003, | |
| "eval_rougeLsum": 0.0003, | |
| "eval_runtime": 9.4712, | |
| "eval_samples_per_second": 11.614, | |
| "eval_steps_per_second": 1.478, | |
| "step": 1567 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_gen_len": 8.5545, | |
| "eval_loss": 1.370047688484192, | |
| "eval_rouge1": 0.001, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.001, | |
| "eval_rougeLsum": 0.001, | |
| "eval_runtime": 9.4659, | |
| "eval_samples_per_second": 11.621, | |
| "eval_steps_per_second": 1.479, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 58.98, | |
| "eval_gen_len": 8.4636, | |
| "eval_loss": 1.3583662509918213, | |
| "eval_rouge1": 0.0006, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0006, | |
| "eval_rougeLsum": 0.0006, | |
| "eval_runtime": 9.472, | |
| "eval_samples_per_second": 11.613, | |
| "eval_steps_per_second": 1.478, | |
| "step": 1622 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_gen_len": 8.0909, | |
| "eval_loss": 1.3374855518341064, | |
| "eval_rouge1": 0.0002, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0002, | |
| "eval_rougeLsum": 0.0002, | |
| "eval_runtime": 9.4676, | |
| "eval_samples_per_second": 11.619, | |
| "eval_steps_per_second": 1.479, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 60.98, | |
| "eval_gen_len": 8.7, | |
| "eval_loss": 1.3220137357711792, | |
| "eval_rouge1": 0.0015, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0016, | |
| "eval_rougeLsum": 0.0016, | |
| "eval_runtime": 9.504, | |
| "eval_samples_per_second": 11.574, | |
| "eval_steps_per_second": 1.473, | |
| "step": 1677 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_gen_len": 8.5818, | |
| "eval_loss": 1.3142321109771729, | |
| "eval_rouge1": 0.0002, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0002, | |
| "eval_rougeLsum": 0.0002, | |
| "eval_runtime": 9.4996, | |
| "eval_samples_per_second": 11.579, | |
| "eval_steps_per_second": 1.474, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 62.98, | |
| "eval_gen_len": 8.7727, | |
| "eval_loss": 1.3096009492874146, | |
| "eval_rouge1": 0.0016, | |
| "eval_rouge2": 0.0, | |
| "eval_rougeL": 0.0017, | |
| "eval_rougeLsum": 0.0017, | |
| "eval_runtime": 9.463, | |
| "eval_samples_per_second": 11.624, | |
| "eval_steps_per_second": 1.479, | |
| "step": 1732 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_gen_len": 9.0455, | |
| "eval_loss": 1.2836058139801025, | |
| "eval_rouge1": 0.0013, | |
| "eval_rouge2": 0.0003, | |
| "eval_rougeL": 0.0012, | |
| "eval_rougeLsum": 0.0013, | |
| "eval_runtime": 9.4725, | |
| "eval_samples_per_second": 11.613, | |
| "eval_steps_per_second": 1.478, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 64.98, | |
| "eval_gen_len": 9.0, | |
| "eval_loss": 1.2642889022827148, | |
| "eval_rouge1": 0.002, | |
| "eval_rouge2": 0.0003, | |
| "eval_rougeL": 0.002, | |
| "eval_rougeLsum": 0.002, | |
| "eval_runtime": 9.6032, | |
| "eval_samples_per_second": 11.454, | |
| "eval_steps_per_second": 1.458, | |
| "step": 1787 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_gen_len": 8.8182, | |
| "eval_loss": 1.2471901178359985, | |
| "eval_rouge1": 0.0052, | |
| "eval_rouge2": 0.001, | |
| "eval_rougeL": 0.0052, | |
| "eval_rougeLsum": 0.0052, | |
| "eval_runtime": 9.4575, | |
| "eval_samples_per_second": 11.631, | |
| "eval_steps_per_second": 1.48, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 66.98, | |
| "eval_gen_len": 9.6636, | |
| "eval_loss": 1.226246953010559, | |
| "eval_rouge1": 0.0086, | |
| "eval_rouge2": 0.0013, | |
| "eval_rougeL": 0.0088, | |
| "eval_rougeLsum": 0.0088, | |
| "eval_runtime": 9.4953, | |
| "eval_samples_per_second": 11.585, | |
| "eval_steps_per_second": 1.474, | |
| "step": 1842 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_gen_len": 9.5273, | |
| "eval_loss": 1.2092421054840088, | |
| "eval_rouge1": 0.0107, | |
| "eval_rouge2": 0.0019, | |
| "eval_rougeL": 0.009, | |
| "eval_rougeLsum": 0.0088, | |
| "eval_runtime": 9.5273, | |
| "eval_samples_per_second": 11.546, | |
| "eval_steps_per_second": 1.469, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 68.98, | |
| "eval_gen_len": 9.8091, | |
| "eval_loss": 1.1953155994415283, | |
| "eval_rouge1": 0.0123, | |
| "eval_rouge2": 0.0032, | |
| "eval_rougeL": 0.012, | |
| "eval_rougeLsum": 0.0118, | |
| "eval_runtime": 9.4864, | |
| "eval_samples_per_second": 11.596, | |
| "eval_steps_per_second": 1.476, | |
| "step": 1897 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_gen_len": 9.3636, | |
| "eval_loss": 1.1815446615219116, | |
| "eval_rouge1": 0.0104, | |
| "eval_rouge2": 0.0033, | |
| "eval_rougeL": 0.0101, | |
| "eval_rougeLsum": 0.0098, | |
| "eval_runtime": 9.4807, | |
| "eval_samples_per_second": 11.603, | |
| "eval_steps_per_second": 1.477, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 70.98, | |
| "eval_gen_len": 9.1091, | |
| "eval_loss": 1.161879301071167, | |
| "eval_rouge1": 0.0064, | |
| "eval_rouge2": 0.0008, | |
| "eval_rougeL": 0.0056, | |
| "eval_rougeLsum": 0.0055, | |
| "eval_runtime": 9.4717, | |
| "eval_samples_per_second": 11.613, | |
| "eval_steps_per_second": 1.478, | |
| "step": 1952 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_gen_len": 9.3273, | |
| "eval_loss": 1.139123797416687, | |
| "eval_rouge1": 0.0105, | |
| "eval_rouge2": 0.002, | |
| "eval_rougeL": 0.0099, | |
| "eval_rougeLsum": 0.0098, | |
| "eval_runtime": 9.4501, | |
| "eval_samples_per_second": 11.64, | |
| "eval_steps_per_second": 1.481, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 72.73, | |
| "grad_norm": 3.556974172592163, | |
| "learning_rate": 1.34320987654321e-05, | |
| "loss": 1.6026, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 72.98, | |
| "eval_gen_len": 9.7364, | |
| "eval_loss": 1.1243102550506592, | |
| "eval_rouge1": 0.0108, | |
| "eval_rouge2": 0.0019, | |
| "eval_rougeL": 0.0104, | |
| "eval_rougeLsum": 0.0101, | |
| "eval_runtime": 9.4726, | |
| "eval_samples_per_second": 11.612, | |
| "eval_steps_per_second": 1.478, | |
| "step": 2007 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_gen_len": 10.6182, | |
| "eval_loss": 1.1061749458312988, | |
| "eval_rouge1": 0.0204, | |
| "eval_rouge2": 0.0031, | |
| "eval_rougeL": 0.0185, | |
| "eval_rougeLsum": 0.0183, | |
| "eval_runtime": 9.4719, | |
| "eval_samples_per_second": 11.613, | |
| "eval_steps_per_second": 1.478, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 74.98, | |
| "eval_gen_len": 10.1545, | |
| "eval_loss": 1.0930193662643433, | |
| "eval_rouge1": 0.0159, | |
| "eval_rouge2": 0.0017, | |
| "eval_rougeL": 0.0145, | |
| "eval_rougeLsum": 0.0143, | |
| "eval_runtime": 9.4961, | |
| "eval_samples_per_second": 11.584, | |
| "eval_steps_per_second": 1.474, | |
| "step": 2062 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_gen_len": 10.7364, | |
| "eval_loss": 1.077184796333313, | |
| "eval_rouge1": 0.0151, | |
| "eval_rouge2": 0.0023, | |
| "eval_rougeL": 0.0133, | |
| "eval_rougeLsum": 0.0133, | |
| "eval_runtime": 9.5109, | |
| "eval_samples_per_second": 11.566, | |
| "eval_steps_per_second": 1.472, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 76.98, | |
| "eval_gen_len": 10.7, | |
| "eval_loss": 1.0578068494796753, | |
| "eval_rouge1": 0.0195, | |
| "eval_rouge2": 0.0044, | |
| "eval_rougeL": 0.0178, | |
| "eval_rougeLsum": 0.0178, | |
| "eval_runtime": 9.4786, | |
| "eval_samples_per_second": 11.605, | |
| "eval_steps_per_second": 1.477, | |
| "step": 2117 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_gen_len": 11.0636, | |
| "eval_loss": 1.0393445491790771, | |
| "eval_rouge1": 0.0237, | |
| "eval_rouge2": 0.0048, | |
| "eval_rougeL": 0.0214, | |
| "eval_rougeLsum": 0.0211, | |
| "eval_runtime": 9.4415, | |
| "eval_samples_per_second": 11.651, | |
| "eval_steps_per_second": 1.483, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 78.98, | |
| "eval_gen_len": 10.1455, | |
| "eval_loss": 1.0262919664382935, | |
| "eval_rouge1": 0.0121, | |
| "eval_rouge2": 0.0014, | |
| "eval_rougeL": 0.0113, | |
| "eval_rougeLsum": 0.0112, | |
| "eval_runtime": 9.4285, | |
| "eval_samples_per_second": 11.667, | |
| "eval_steps_per_second": 1.485, | |
| "step": 2172 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_gen_len": 11.3818, | |
| "eval_loss": 1.0064616203308105, | |
| "eval_rouge1": 0.0273, | |
| "eval_rouge2": 0.0048, | |
| "eval_rougeL": 0.0238, | |
| "eval_rougeLsum": 0.0235, | |
| "eval_runtime": 9.448, | |
| "eval_samples_per_second": 11.643, | |
| "eval_steps_per_second": 1.482, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 80.98, | |
| "eval_gen_len": 10.9545, | |
| "eval_loss": 0.98997563123703, | |
| "eval_rouge1": 0.0228, | |
| "eval_rouge2": 0.0042, | |
| "eval_rougeL": 0.0197, | |
| "eval_rougeLsum": 0.0196, | |
| "eval_runtime": 9.4798, | |
| "eval_samples_per_second": 11.604, | |
| "eval_steps_per_second": 1.477, | |
| "step": 2227 | |
| }, | |
| { | |
| "epoch": 82.0, | |
| "eval_gen_len": 10.9909, | |
| "eval_loss": 0.9675103425979614, | |
| "eval_rouge1": 0.024, | |
| "eval_rouge2": 0.0046, | |
| "eval_rougeL": 0.0204, | |
| "eval_rougeLsum": 0.0202, | |
| "eval_runtime": 9.4448, | |
| "eval_samples_per_second": 11.647, | |
| "eval_steps_per_second": 1.482, | |
| "step": 2255 | |
| }, | |
| { | |
| "epoch": 82.98, | |
| "eval_gen_len": 10.7182, | |
| "eval_loss": 0.9506540298461914, | |
| "eval_rouge1": 0.0244, | |
| "eval_rouge2": 0.0051, | |
| "eval_rougeL": 0.0211, | |
| "eval_rougeLsum": 0.0211, | |
| "eval_runtime": 9.5012, | |
| "eval_samples_per_second": 11.578, | |
| "eval_steps_per_second": 1.474, | |
| "step": 2282 | |
| }, | |
| { | |
| "epoch": 84.0, | |
| "eval_gen_len": 10.8636, | |
| "eval_loss": 0.9340613484382629, | |
| "eval_rouge1": 0.0249, | |
| "eval_rouge2": 0.0058, | |
| "eval_rougeL": 0.022, | |
| "eval_rougeLsum": 0.0223, | |
| "eval_runtime": 9.447, | |
| "eval_samples_per_second": 11.644, | |
| "eval_steps_per_second": 1.482, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 84.98, | |
| "eval_gen_len": 10.0909, | |
| "eval_loss": 0.9161014556884766, | |
| "eval_rouge1": 0.0243, | |
| "eval_rouge2": 0.0077, | |
| "eval_rougeL": 0.0224, | |
| "eval_rougeLsum": 0.0226, | |
| "eval_runtime": 9.4492, | |
| "eval_samples_per_second": 11.641, | |
| "eval_steps_per_second": 1.482, | |
| "step": 2337 | |
| }, | |
| { | |
| "epoch": 86.0, | |
| "eval_gen_len": 9.7182, | |
| "eval_loss": 0.8942736983299255, | |
| "eval_rouge1": 0.0176, | |
| "eval_rouge2": 0.0035, | |
| "eval_rougeL": 0.0152, | |
| "eval_rougeLsum": 0.0153, | |
| "eval_runtime": 9.4727, | |
| "eval_samples_per_second": 11.612, | |
| "eval_steps_per_second": 1.478, | |
| "step": 2365 | |
| }, | |
| { | |
| "epoch": 86.98, | |
| "eval_gen_len": 10.0, | |
| "eval_loss": 0.8758471608161926, | |
| "eval_rouge1": 0.0239, | |
| "eval_rouge2": 0.0093, | |
| "eval_rougeL": 0.0214, | |
| "eval_rougeLsum": 0.0215, | |
| "eval_runtime": 9.4864, | |
| "eval_samples_per_second": 11.596, | |
| "eval_steps_per_second": 1.476, | |
| "step": 2392 | |
| }, | |
| { | |
| "epoch": 88.0, | |
| "eval_gen_len": 10.2273, | |
| "eval_loss": 0.854742169380188, | |
| "eval_rouge1": 0.0254, | |
| "eval_rouge2": 0.0116, | |
| "eval_rougeL": 0.0237, | |
| "eval_rougeLsum": 0.0238, | |
| "eval_runtime": 9.5043, | |
| "eval_samples_per_second": 11.574, | |
| "eval_steps_per_second": 1.473, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 88.98, | |
| "eval_gen_len": 10.2545, | |
| "eval_loss": 0.8352662324905396, | |
| "eval_rouge1": 0.0196, | |
| "eval_rouge2": 0.007, | |
| "eval_rougeL": 0.0183, | |
| "eval_rougeLsum": 0.0182, | |
| "eval_runtime": 9.455, | |
| "eval_samples_per_second": 11.634, | |
| "eval_steps_per_second": 1.481, | |
| "step": 2447 | |
| }, | |
| { | |
| "epoch": 90.0, | |
| "eval_gen_len": 9.2364, | |
| "eval_loss": 0.8150736093521118, | |
| "eval_rouge1": 0.0104, | |
| "eval_rouge2": 0.0032, | |
| "eval_rougeL": 0.0095, | |
| "eval_rougeLsum": 0.0098, | |
| "eval_runtime": 9.4416, | |
| "eval_samples_per_second": 11.651, | |
| "eval_steps_per_second": 1.483, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 90.91, | |
| "grad_norm": 1.338108777999878, | |
| "learning_rate": 1.1786008230452676e-05, | |
| "loss": 1.2934, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 90.98, | |
| "eval_gen_len": 9.3455, | |
| "eval_loss": 0.7920636534690857, | |
| "eval_rouge1": 0.01, | |
| "eval_rouge2": 0.0036, | |
| "eval_rougeL": 0.0095, | |
| "eval_rougeLsum": 0.0096, | |
| "eval_runtime": 9.4783, | |
| "eval_samples_per_second": 11.605, | |
| "eval_steps_per_second": 1.477, | |
| "step": 2502 | |
| }, | |
| { | |
| "epoch": 92.0, | |
| "eval_gen_len": 9.2545, | |
| "eval_loss": 0.7697137594223022, | |
| "eval_rouge1": 0.012, | |
| "eval_rouge2": 0.0051, | |
| "eval_rougeL": 0.0111, | |
| "eval_rougeLsum": 0.011, | |
| "eval_runtime": 9.5826, | |
| "eval_samples_per_second": 11.479, | |
| "eval_steps_per_second": 1.461, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 92.98, | |
| "eval_gen_len": 8.8455, | |
| "eval_loss": 0.7492441534996033, | |
| "eval_rouge1": 0.0106, | |
| "eval_rouge2": 0.0048, | |
| "eval_rougeL": 0.0098, | |
| "eval_rougeLsum": 0.0098, | |
| "eval_runtime": 9.4826, | |
| "eval_samples_per_second": 11.6, | |
| "eval_steps_per_second": 1.476, | |
| "step": 2557 | |
| }, | |
| { | |
| "epoch": 94.0, | |
| "eval_gen_len": 9.4, | |
| "eval_loss": 0.7301111221313477, | |
| "eval_rouge1": 0.0112, | |
| "eval_rouge2": 0.0053, | |
| "eval_rougeL": 0.0095, | |
| "eval_rougeLsum": 0.0095, | |
| "eval_runtime": 9.4682, | |
| "eval_samples_per_second": 11.618, | |
| "eval_steps_per_second": 1.479, | |
| "step": 2585 | |
| }, | |
| { | |
| "epoch": 94.98, | |
| "eval_gen_len": 8.9, | |
| "eval_loss": 0.7125746607780457, | |
| "eval_rouge1": 0.0069, | |
| "eval_rouge2": 0.0026, | |
| "eval_rougeL": 0.0063, | |
| "eval_rougeLsum": 0.0064, | |
| "eval_runtime": 9.4699, | |
| "eval_samples_per_second": 11.616, | |
| "eval_steps_per_second": 1.478, | |
| "step": 2612 | |
| }, | |
| { | |
| "epoch": 96.0, | |
| "eval_gen_len": 9.2, | |
| "eval_loss": 0.6931119561195374, | |
| "eval_rouge1": 0.0107, | |
| "eval_rouge2": 0.0053, | |
| "eval_rougeL": 0.0088, | |
| "eval_rougeLsum": 0.0089, | |
| "eval_runtime": 9.4599, | |
| "eval_samples_per_second": 11.628, | |
| "eval_steps_per_second": 1.48, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 96.98, | |
| "eval_gen_len": 8.8909, | |
| "eval_loss": 0.6750566363334656, | |
| "eval_rouge1": 0.0063, | |
| "eval_rouge2": 0.0029, | |
| "eval_rougeL": 0.0049, | |
| "eval_rougeLsum": 0.0048, | |
| "eval_runtime": 9.4656, | |
| "eval_samples_per_second": 11.621, | |
| "eval_steps_per_second": 1.479, | |
| "step": 2667 | |
| }, | |
| { | |
| "epoch": 98.0, | |
| "eval_gen_len": 9.3182, | |
| "eval_loss": 0.6581041216850281, | |
| "eval_rouge1": 0.0091, | |
| "eval_rouge2": 0.005, | |
| "eval_rougeL": 0.0069, | |
| "eval_rougeLsum": 0.0071, | |
| "eval_runtime": 9.4802, | |
| "eval_samples_per_second": 11.603, | |
| "eval_steps_per_second": 1.477, | |
| "step": 2695 | |
| }, | |
| { | |
| "epoch": 98.98, | |
| "eval_gen_len": 9.1091, | |
| "eval_loss": 0.6413628458976746, | |
| "eval_rouge1": 0.0038, | |
| "eval_rouge2": 0.0027, | |
| "eval_rougeL": 0.0026, | |
| "eval_rougeLsum": 0.0034, | |
| "eval_runtime": 9.4864, | |
| "eval_samples_per_second": 11.595, | |
| "eval_steps_per_second": 1.476, | |
| "step": 2722 | |
| }, | |
| { | |
| "epoch": 100.0, | |
| "eval_gen_len": 9.5273, | |
| "eval_loss": 0.6237995028495789, | |
| "eval_rouge1": 0.0125, | |
| "eval_rouge2": 0.0088, | |
| "eval_rougeL": 0.0109, | |
| "eval_rougeLsum": 0.0109, | |
| "eval_runtime": 9.479, | |
| "eval_samples_per_second": 11.605, | |
| "eval_steps_per_second": 1.477, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 100.98, | |
| "eval_gen_len": 9.3727, | |
| "eval_loss": 0.6053850054740906, | |
| "eval_rouge1": 0.0053, | |
| "eval_rouge2": 0.0027, | |
| "eval_rougeL": 0.0042, | |
| "eval_rougeLsum": 0.0045, | |
| "eval_runtime": 9.4604, | |
| "eval_samples_per_second": 11.627, | |
| "eval_steps_per_second": 1.48, | |
| "step": 2777 | |
| }, | |
| { | |
| "epoch": 102.0, | |
| "eval_gen_len": 9.7091, | |
| "eval_loss": 0.5907317996025085, | |
| "eval_rouge1": 0.0115, | |
| "eval_rouge2": 0.0096, | |
| "eval_rougeL": 0.0108, | |
| "eval_rougeLsum": 0.0109, | |
| "eval_runtime": 9.4533, | |
| "eval_samples_per_second": 11.636, | |
| "eval_steps_per_second": 1.481, | |
| "step": 2805 | |
| }, | |
| { | |
| "epoch": 102.98, | |
| "eval_gen_len": 9.5364, | |
| "eval_loss": 0.5752558708190918, | |
| "eval_rouge1": 0.0055, | |
| "eval_rouge2": 0.0045, | |
| "eval_rougeL": 0.005, | |
| "eval_rougeLsum": 0.0053, | |
| "eval_runtime": 9.4454, | |
| "eval_samples_per_second": 11.646, | |
| "eval_steps_per_second": 1.482, | |
| "step": 2832 | |
| }, | |
| { | |
| "epoch": 104.0, | |
| "eval_gen_len": 9.7273, | |
| "eval_loss": 0.562364399433136, | |
| "eval_rouge1": 0.01, | |
| "eval_rouge2": 0.0057, | |
| "eval_rougeL": 0.0091, | |
| "eval_rougeLsum": 0.0094, | |
| "eval_runtime": 9.4478, | |
| "eval_samples_per_second": 11.643, | |
| "eval_steps_per_second": 1.482, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 104.98, | |
| "eval_gen_len": 9.6909, | |
| "eval_loss": 0.5496523380279541, | |
| "eval_rouge1": 0.0078, | |
| "eval_rouge2": 0.0038, | |
| "eval_rougeL": 0.0066, | |
| "eval_rougeLsum": 0.0069, | |
| "eval_runtime": 9.4963, | |
| "eval_samples_per_second": 11.583, | |
| "eval_steps_per_second": 1.474, | |
| "step": 2887 | |
| }, | |
| { | |
| "epoch": 106.0, | |
| "eval_gen_len": 9.6, | |
| "eval_loss": 0.5380507111549377, | |
| "eval_rouge1": 0.0077, | |
| "eval_rouge2": 0.0041, | |
| "eval_rougeL": 0.0068, | |
| "eval_rougeLsum": 0.0071, | |
| "eval_runtime": 9.4581, | |
| "eval_samples_per_second": 11.63, | |
| "eval_steps_per_second": 1.48, | |
| "step": 2915 | |
| }, | |
| { | |
| "epoch": 106.98, | |
| "eval_gen_len": 9.2909, | |
| "eval_loss": 0.5269507765769958, | |
| "eval_rouge1": 0.0109, | |
| "eval_rouge2": 0.0068, | |
| "eval_rougeL": 0.0101, | |
| "eval_rougeLsum": 0.0103, | |
| "eval_runtime": 9.4431, | |
| "eval_samples_per_second": 11.649, | |
| "eval_steps_per_second": 1.483, | |
| "step": 2942 | |
| }, | |
| { | |
| "epoch": 108.0, | |
| "eval_gen_len": 8.9636, | |
| "eval_loss": 0.5166643857955933, | |
| "eval_rouge1": 0.0095, | |
| "eval_rouge2": 0.004, | |
| "eval_rougeL": 0.008, | |
| "eval_rougeLsum": 0.0079, | |
| "eval_runtime": 9.4573, | |
| "eval_samples_per_second": 11.631, | |
| "eval_steps_per_second": 1.48, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 108.98, | |
| "eval_gen_len": 9.3818, | |
| "eval_loss": 0.5079358816146851, | |
| "eval_rouge1": 0.0078, | |
| "eval_rouge2": 0.0035, | |
| "eval_rougeL": 0.0055, | |
| "eval_rougeLsum": 0.0059, | |
| "eval_runtime": 9.4478, | |
| "eval_samples_per_second": 11.643, | |
| "eval_steps_per_second": 1.482, | |
| "step": 2997 | |
| }, | |
| { | |
| "epoch": 109.09, | |
| "grad_norm": 2.720813512802124, | |
| "learning_rate": 1.0139917695473251e-05, | |
| "loss": 0.9194, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 110.0, | |
| "eval_gen_len": 9.2909, | |
| "eval_loss": 0.5007773041725159, | |
| "eval_rouge1": 0.0044, | |
| "eval_rouge2": 0.0019, | |
| "eval_rougeL": 0.0037, | |
| "eval_rougeLsum": 0.0033, | |
| "eval_runtime": 9.4749, | |
| "eval_samples_per_second": 11.61, | |
| "eval_steps_per_second": 1.478, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 110.98, | |
| "eval_gen_len": 9.2909, | |
| "eval_loss": 0.49183493852615356, | |
| "eval_rouge1": 0.0108, | |
| "eval_rouge2": 0.0069, | |
| "eval_rougeL": 0.0087, | |
| "eval_rougeLsum": 0.0088, | |
| "eval_runtime": 9.5004, | |
| "eval_samples_per_second": 11.578, | |
| "eval_steps_per_second": 1.474, | |
| "step": 3052 | |
| }, | |
| { | |
| "epoch": 112.0, | |
| "eval_gen_len": 9.2455, | |
| "eval_loss": 0.48342087864875793, | |
| "eval_rouge1": 0.0112, | |
| "eval_rouge2": 0.008, | |
| "eval_rougeL": 0.0092, | |
| "eval_rougeLsum": 0.009, | |
| "eval_runtime": 9.5074, | |
| "eval_samples_per_second": 11.57, | |
| "eval_steps_per_second": 1.473, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 112.98, | |
| "eval_gen_len": 8.9364, | |
| "eval_loss": 0.47708794474601746, | |
| "eval_rouge1": 0.0057, | |
| "eval_rouge2": 0.0031, | |
| "eval_rougeL": 0.0042, | |
| "eval_rougeLsum": 0.004, | |
| "eval_runtime": 9.5044, | |
| "eval_samples_per_second": 11.574, | |
| "eval_steps_per_second": 1.473, | |
| "step": 3107 | |
| }, | |
| { | |
| "epoch": 114.0, | |
| "eval_gen_len": 9.1364, | |
| "eval_loss": 0.47102922201156616, | |
| "eval_rouge1": 0.0122, | |
| "eval_rouge2": 0.0065, | |
| "eval_rougeL": 0.0089, | |
| "eval_rougeLsum": 0.0087, | |
| "eval_runtime": 9.4603, | |
| "eval_samples_per_second": 11.628, | |
| "eval_steps_per_second": 1.48, | |
| "step": 3135 | |
| }, | |
| { | |
| "epoch": 114.98, | |
| "eval_gen_len": 9.2182, | |
| "eval_loss": 0.4654105007648468, | |
| "eval_rouge1": 0.0102, | |
| "eval_rouge2": 0.0063, | |
| "eval_rougeL": 0.0085, | |
| "eval_rougeLsum": 0.0083, | |
| "eval_runtime": 9.4614, | |
| "eval_samples_per_second": 11.626, | |
| "eval_steps_per_second": 1.48, | |
| "step": 3162 | |
| }, | |
| { | |
| "epoch": 116.0, | |
| "eval_gen_len": 8.9455, | |
| "eval_loss": 0.4600731432437897, | |
| "eval_rouge1": 0.0119, | |
| "eval_rouge2": 0.0092, | |
| "eval_rougeL": 0.0104, | |
| "eval_rougeLsum": 0.0102, | |
| "eval_runtime": 9.461, | |
| "eval_samples_per_second": 11.627, | |
| "eval_steps_per_second": 1.48, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 116.98, | |
| "eval_gen_len": 8.4182, | |
| "eval_loss": 0.4526377320289612, | |
| "eval_rouge1": 0.0107, | |
| "eval_rouge2": 0.0082, | |
| "eval_rougeL": 0.0099, | |
| "eval_rougeLsum": 0.0097, | |
| "eval_runtime": 9.4731, | |
| "eval_samples_per_second": 11.612, | |
| "eval_steps_per_second": 1.478, | |
| "step": 3217 | |
| }, | |
| { | |
| "epoch": 118.0, | |
| "eval_gen_len": 8.4273, | |
| "eval_loss": 0.4467811584472656, | |
| "eval_rouge1": 0.0052, | |
| "eval_rouge2": 0.0044, | |
| "eval_rougeL": 0.0045, | |
| "eval_rougeLsum": 0.0045, | |
| "eval_runtime": 9.4604, | |
| "eval_samples_per_second": 11.627, | |
| "eval_steps_per_second": 1.48, | |
| "step": 3245 | |
| }, | |
| { | |
| "epoch": 118.98, | |
| "eval_gen_len": 8.3, | |
| "eval_loss": 0.4426099956035614, | |
| "eval_rouge1": 0.0054, | |
| "eval_rouge2": 0.0041, | |
| "eval_rougeL": 0.005, | |
| "eval_rougeLsum": 0.0052, | |
| "eval_runtime": 9.4891, | |
| "eval_samples_per_second": 11.592, | |
| "eval_steps_per_second": 1.475, | |
| "step": 3272 | |
| }, | |
| { | |
| "epoch": 120.0, | |
| "eval_gen_len": 8.2727, | |
| "eval_loss": 0.43666210770606995, | |
| "eval_rouge1": 0.0107, | |
| "eval_rouge2": 0.0079, | |
| "eval_rougeL": 0.0102, | |
| "eval_rougeLsum": 0.0101, | |
| "eval_runtime": 9.5096, | |
| "eval_samples_per_second": 11.567, | |
| "eval_steps_per_second": 1.472, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 120.98, | |
| "eval_gen_len": 8.7182, | |
| "eval_loss": 0.4338292181491852, | |
| "eval_rouge1": 0.0142, | |
| "eval_rouge2": 0.0102, | |
| "eval_rougeL": 0.0134, | |
| "eval_rougeLsum": 0.0131, | |
| "eval_runtime": 9.4732, | |
| "eval_samples_per_second": 11.612, | |
| "eval_steps_per_second": 1.478, | |
| "step": 3327 | |
| }, | |
| { | |
| "epoch": 122.0, | |
| "eval_gen_len": 7.5727, | |
| "eval_loss": 0.4293038547039032, | |
| "eval_rouge1": 0.0045, | |
| "eval_rouge2": 0.0035, | |
| "eval_rougeL": 0.0039, | |
| "eval_rougeLsum": 0.0039, | |
| "eval_runtime": 9.458, | |
| "eval_samples_per_second": 11.63, | |
| "eval_steps_per_second": 1.48, | |
| "step": 3355 | |
| }, | |
| { | |
| "epoch": 122.98, | |
| "eval_gen_len": 7.8818, | |
| "eval_loss": 0.4247772991657257, | |
| "eval_rouge1": 0.0082, | |
| "eval_rouge2": 0.0056, | |
| "eval_rougeL": 0.0078, | |
| "eval_rougeLsum": 0.0076, | |
| "eval_runtime": 9.4976, | |
| "eval_samples_per_second": 11.582, | |
| "eval_steps_per_second": 1.474, | |
| "step": 3382 | |
| }, | |
| { | |
| "epoch": 124.0, | |
| "eval_gen_len": 7.4273, | |
| "eval_loss": 0.4226304888725281, | |
| "eval_rouge1": 0.0047, | |
| "eval_rouge2": 0.0039, | |
| "eval_rougeL": 0.0047, | |
| "eval_rougeLsum": 0.0047, | |
| "eval_runtime": 9.4775, | |
| "eval_samples_per_second": 11.606, | |
| "eval_steps_per_second": 1.477, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 124.98, | |
| "eval_gen_len": 7.7091, | |
| "eval_loss": 0.4187226891517639, | |
| "eval_rouge1": 0.0096, | |
| "eval_rouge2": 0.0065, | |
| "eval_rougeL": 0.0097, | |
| "eval_rougeLsum": 0.0095, | |
| "eval_runtime": 9.4596, | |
| "eval_samples_per_second": 11.628, | |
| "eval_steps_per_second": 1.48, | |
| "step": 3437 | |
| }, | |
| { | |
| "epoch": 126.0, | |
| "eval_gen_len": 7.1364, | |
| "eval_loss": 0.4152156412601471, | |
| "eval_rouge1": 0.0026, | |
| "eval_rouge2": 0.0024, | |
| "eval_rougeL": 0.0026, | |
| "eval_rougeLsum": 0.0026, | |
| "eval_runtime": 9.4787, | |
| "eval_samples_per_second": 11.605, | |
| "eval_steps_per_second": 1.477, | |
| "step": 3465 | |
| }, | |
| { | |
| "epoch": 126.98, | |
| "eval_gen_len": 6.8909, | |
| "eval_loss": 0.4114760458469391, | |
| "eval_rouge1": 0.0026, | |
| "eval_rouge2": 0.0024, | |
| "eval_rougeL": 0.0026, | |
| "eval_rougeLsum": 0.0026, | |
| "eval_runtime": 9.4927, | |
| "eval_samples_per_second": 11.588, | |
| "eval_steps_per_second": 1.475, | |
| "step": 3492 | |
| }, | |
| { | |
| "epoch": 127.27, | |
| "grad_norm": 5.880601406097412, | |
| "learning_rate": 8.493827160493828e-06, | |
| "loss": 0.6369, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 128.0, | |
| "eval_gen_len": 6.7182, | |
| "eval_loss": 0.4087616503238678, | |
| "eval_rouge1": 0.0051, | |
| "eval_rouge2": 0.0048, | |
| "eval_rougeL": 0.0051, | |
| "eval_rougeLsum": 0.0051, | |
| "eval_runtime": 9.5116, | |
| "eval_samples_per_second": 11.565, | |
| "eval_steps_per_second": 1.472, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 128.98, | |
| "eval_gen_len": 7.3091, | |
| "eval_loss": 0.4050390422344208, | |
| "eval_rouge1": 0.0113, | |
| "eval_rouge2": 0.0097, | |
| "eval_rougeL": 0.0115, | |
| "eval_rougeLsum": 0.0115, | |
| "eval_runtime": 9.5251, | |
| "eval_samples_per_second": 11.548, | |
| "eval_steps_per_second": 1.47, | |
| "step": 3547 | |
| }, | |
| { | |
| "epoch": 130.0, | |
| "eval_gen_len": 7.2727, | |
| "eval_loss": 0.40342459082603455, | |
| "eval_rouge1": 0.0097, | |
| "eval_rouge2": 0.0086, | |
| "eval_rougeL": 0.0098, | |
| "eval_rougeLsum": 0.0099, | |
| "eval_runtime": 9.5061, | |
| "eval_samples_per_second": 11.572, | |
| "eval_steps_per_second": 1.473, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 130.98, | |
| "eval_gen_len": 6.9455, | |
| "eval_loss": 0.39917439222335815, | |
| "eval_rouge1": 0.0096, | |
| "eval_rouge2": 0.0081, | |
| "eval_rougeL": 0.0097, | |
| "eval_rougeLsum": 0.0097, | |
| "eval_runtime": 9.5045, | |
| "eval_samples_per_second": 11.573, | |
| "eval_steps_per_second": 1.473, | |
| "step": 3602 | |
| }, | |
| { | |
| "epoch": 132.0, | |
| "eval_gen_len": 6.6, | |
| "eval_loss": 0.3954027593135834, | |
| "eval_rouge1": 0.0053, | |
| "eval_rouge2": 0.0056, | |
| "eval_rougeL": 0.0053, | |
| "eval_rougeLsum": 0.0061, | |
| "eval_runtime": 9.4882, | |
| "eval_samples_per_second": 11.593, | |
| "eval_steps_per_second": 1.476, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 132.98, | |
| "eval_gen_len": 6.6727, | |
| "eval_loss": 0.3916667103767395, | |
| "eval_rouge1": 0.0061, | |
| "eval_rouge2": 0.0056, | |
| "eval_rougeL": 0.006, | |
| "eval_rougeLsum": 0.0061, | |
| "eval_runtime": 9.4921, | |
| "eval_samples_per_second": 11.589, | |
| "eval_steps_per_second": 1.475, | |
| "step": 3657 | |
| }, | |
| { | |
| "epoch": 134.0, | |
| "eval_gen_len": 6.9455, | |
| "eval_loss": 0.3922131359577179, | |
| "eval_rouge1": 0.0084, | |
| "eval_rouge2": 0.0072, | |
| "eval_rougeL": 0.0082, | |
| "eval_rougeLsum": 0.0086, | |
| "eval_runtime": 9.4995, | |
| "eval_samples_per_second": 11.58, | |
| "eval_steps_per_second": 1.474, | |
| "step": 3685 | |
| }, | |
| { | |
| "epoch": 134.98, | |
| "eval_gen_len": 6.3727, | |
| "eval_loss": 0.38674217462539673, | |
| "eval_rouge1": 0.0052, | |
| "eval_rouge2": 0.003, | |
| "eval_rougeL": 0.0046, | |
| "eval_rougeLsum": 0.0046, | |
| "eval_runtime": 9.4561, | |
| "eval_samples_per_second": 11.633, | |
| "eval_steps_per_second": 1.481, | |
| "step": 3712 | |
| }, | |
| { | |
| "epoch": 136.0, | |
| "eval_gen_len": 6.6273, | |
| "eval_loss": 0.38475027680397034, | |
| "eval_rouge1": 0.009, | |
| "eval_rouge2": 0.0061, | |
| "eval_rougeL": 0.0083, | |
| "eval_rougeLsum": 0.0086, | |
| "eval_runtime": 9.4487, | |
| "eval_samples_per_second": 11.642, | |
| "eval_steps_per_second": 1.482, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 136.98, | |
| "eval_gen_len": 6.4455, | |
| "eval_loss": 0.37981557846069336, | |
| "eval_rouge1": 0.0123, | |
| "eval_rouge2": 0.0095, | |
| "eval_rougeL": 0.0119, | |
| "eval_rougeLsum": 0.0121, | |
| "eval_runtime": 9.4838, | |
| "eval_samples_per_second": 11.599, | |
| "eval_steps_per_second": 1.476, | |
| "step": 3767 | |
| }, | |
| { | |
| "epoch": 138.0, | |
| "eval_gen_len": 6.6909, | |
| "eval_loss": 0.3788022994995117, | |
| "eval_rouge1": 0.0138, | |
| "eval_rouge2": 0.01, | |
| "eval_rougeL": 0.0132, | |
| "eval_rougeLsum": 0.0133, | |
| "eval_runtime": 9.4736, | |
| "eval_samples_per_second": 11.611, | |
| "eval_steps_per_second": 1.478, | |
| "step": 3795 | |
| }, | |
| { | |
| "epoch": 138.98, | |
| "eval_gen_len": 6.3909, | |
| "eval_loss": 0.3755718171596527, | |
| "eval_rouge1": 0.0119, | |
| "eval_rouge2": 0.0085, | |
| "eval_rougeL": 0.0116, | |
| "eval_rougeLsum": 0.0116, | |
| "eval_runtime": 9.4922, | |
| "eval_samples_per_second": 11.588, | |
| "eval_steps_per_second": 1.475, | |
| "step": 3822 | |
| }, | |
| { | |
| "epoch": 140.0, | |
| "eval_gen_len": 6.6545, | |
| "eval_loss": 0.3745150864124298, | |
| "eval_rouge1": 0.0135, | |
| "eval_rouge2": 0.0095, | |
| "eval_rougeL": 0.013, | |
| "eval_rougeLsum": 0.013, | |
| "eval_runtime": 9.5082, | |
| "eval_samples_per_second": 11.569, | |
| "eval_steps_per_second": 1.472, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 140.98, | |
| "eval_gen_len": 6.9636, | |
| "eval_loss": 0.37223342061042786, | |
| "eval_rouge1": 0.0175, | |
| "eval_rouge2": 0.0123, | |
| "eval_rougeL": 0.0171, | |
| "eval_rougeLsum": 0.0168, | |
| "eval_runtime": 9.506, | |
| "eval_samples_per_second": 11.572, | |
| "eval_steps_per_second": 1.473, | |
| "step": 3877 | |
| }, | |
| { | |
| "epoch": 142.0, | |
| "eval_gen_len": 7.0727, | |
| "eval_loss": 0.36922305822372437, | |
| "eval_rouge1": 0.0188, | |
| "eval_rouge2": 0.0127, | |
| "eval_rougeL": 0.0183, | |
| "eval_rougeLsum": 0.018, | |
| "eval_runtime": 9.4985, | |
| "eval_samples_per_second": 11.581, | |
| "eval_steps_per_second": 1.474, | |
| "step": 3905 | |
| }, | |
| { | |
| "epoch": 142.98, | |
| "eval_gen_len": 7.2727, | |
| "eval_loss": 0.3674834668636322, | |
| "eval_rouge1": 0.0201, | |
| "eval_rouge2": 0.0136, | |
| "eval_rougeL": 0.0197, | |
| "eval_rougeLsum": 0.0194, | |
| "eval_runtime": 9.4928, | |
| "eval_samples_per_second": 11.588, | |
| "eval_steps_per_second": 1.475, | |
| "step": 3932 | |
| }, | |
| { | |
| "epoch": 144.0, | |
| "eval_gen_len": 7.1818, | |
| "eval_loss": 0.36526089906692505, | |
| "eval_rouge1": 0.0215, | |
| "eval_rouge2": 0.0139, | |
| "eval_rougeL": 0.0211, | |
| "eval_rougeLsum": 0.0208, | |
| "eval_runtime": 9.4987, | |
| "eval_samples_per_second": 11.581, | |
| "eval_steps_per_second": 1.474, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 144.98, | |
| "eval_gen_len": 7.2, | |
| "eval_loss": 0.36316850781440735, | |
| "eval_rouge1": 0.0209, | |
| "eval_rouge2": 0.0128, | |
| "eval_rougeL": 0.0201, | |
| "eval_rougeLsum": 0.0202, | |
| "eval_runtime": 9.4792, | |
| "eval_samples_per_second": 11.604, | |
| "eval_steps_per_second": 1.477, | |
| "step": 3987 | |
| }, | |
| { | |
| "epoch": 145.45, | |
| "grad_norm": 0.6945245265960693, | |
| "learning_rate": 6.847736625514404e-06, | |
| "loss": 0.5099, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 146.0, | |
| "eval_gen_len": 7.1364, | |
| "eval_loss": 0.3603822886943817, | |
| "eval_rouge1": 0.022, | |
| "eval_rouge2": 0.0145, | |
| "eval_rougeL": 0.0213, | |
| "eval_rougeLsum": 0.0212, | |
| "eval_runtime": 9.5107, | |
| "eval_samples_per_second": 11.566, | |
| "eval_steps_per_second": 1.472, | |
| "step": 4015 | |
| }, | |
| { | |
| "epoch": 146.98, | |
| "eval_gen_len": 7.1182, | |
| "eval_loss": 0.35853010416030884, | |
| "eval_rouge1": 0.022, | |
| "eval_rouge2": 0.0145, | |
| "eval_rougeL": 0.0213, | |
| "eval_rougeLsum": 0.0212, | |
| "eval_runtime": 9.5074, | |
| "eval_samples_per_second": 11.57, | |
| "eval_steps_per_second": 1.473, | |
| "step": 4042 | |
| }, | |
| { | |
| "epoch": 148.0, | |
| "eval_gen_len": 7.7364, | |
| "eval_loss": 0.35745835304260254, | |
| "eval_rouge1": 0.0283, | |
| "eval_rouge2": 0.018, | |
| "eval_rougeL": 0.0269, | |
| "eval_rougeLsum": 0.0269, | |
| "eval_runtime": 9.482, | |
| "eval_samples_per_second": 11.601, | |
| "eval_steps_per_second": 1.476, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 148.98, | |
| "eval_gen_len": 8.3364, | |
| "eval_loss": 0.3559305965900421, | |
| "eval_rouge1": 0.0374, | |
| "eval_rouge2": 0.0236, | |
| "eval_rougeL": 0.0357, | |
| "eval_rougeLsum": 0.0359, | |
| "eval_runtime": 9.4877, | |
| "eval_samples_per_second": 11.594, | |
| "eval_steps_per_second": 1.476, | |
| "step": 4097 | |
| }, | |
| { | |
| "epoch": 150.0, | |
| "eval_gen_len": 8.9091, | |
| "eval_loss": 0.3528214395046234, | |
| "eval_rouge1": 0.0384, | |
| "eval_rouge2": 0.0236, | |
| "eval_rougeL": 0.0366, | |
| "eval_rougeLsum": 0.0369, | |
| "eval_runtime": 9.4747, | |
| "eval_samples_per_second": 11.61, | |
| "eval_steps_per_second": 1.478, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 150.98, | |
| "eval_gen_len": 9.4455, | |
| "eval_loss": 0.35084155201911926, | |
| "eval_rouge1": 0.0416, | |
| "eval_rouge2": 0.0254, | |
| "eval_rougeL": 0.0399, | |
| "eval_rougeLsum": 0.0399, | |
| "eval_runtime": 9.4685, | |
| "eval_samples_per_second": 11.618, | |
| "eval_steps_per_second": 1.479, | |
| "step": 4152 | |
| }, | |
| { | |
| "epoch": 152.0, | |
| "eval_gen_len": 9.6091, | |
| "eval_loss": 0.34896430373191833, | |
| "eval_rouge1": 0.0439, | |
| "eval_rouge2": 0.0257, | |
| "eval_rougeL": 0.0413, | |
| "eval_rougeLsum": 0.0415, | |
| "eval_runtime": 9.5218, | |
| "eval_samples_per_second": 11.552, | |
| "eval_steps_per_second": 1.47, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 152.98, | |
| "eval_gen_len": 9.7636, | |
| "eval_loss": 0.34783267974853516, | |
| "eval_rouge1": 0.0479, | |
| "eval_rouge2": 0.0297, | |
| "eval_rougeL": 0.045, | |
| "eval_rougeLsum": 0.0454, | |
| "eval_runtime": 9.5136, | |
| "eval_samples_per_second": 11.562, | |
| "eval_steps_per_second": 1.472, | |
| "step": 4207 | |
| }, | |
| { | |
| "epoch": 154.0, | |
| "eval_gen_len": 10.1909, | |
| "eval_loss": 0.3452661335468292, | |
| "eval_rouge1": 0.0495, | |
| "eval_rouge2": 0.0291, | |
| "eval_rougeL": 0.0464, | |
| "eval_rougeLsum": 0.0464, | |
| "eval_runtime": 9.5097, | |
| "eval_samples_per_second": 11.567, | |
| "eval_steps_per_second": 1.472, | |
| "step": 4235 | |
| }, | |
| { | |
| "epoch": 154.98, | |
| "eval_gen_len": 10.6, | |
| "eval_loss": 0.34583330154418945, | |
| "eval_rouge1": 0.0576, | |
| "eval_rouge2": 0.035, | |
| "eval_rougeL": 0.055, | |
| "eval_rougeLsum": 0.0551, | |
| "eval_runtime": 9.4851, | |
| "eval_samples_per_second": 11.597, | |
| "eval_steps_per_second": 1.476, | |
| "step": 4262 | |
| }, | |
| { | |
| "epoch": 156.0, | |
| "eval_gen_len": 10.2909, | |
| "eval_loss": 0.3417557179927826, | |
| "eval_rouge1": 0.0533, | |
| "eval_rouge2": 0.0314, | |
| "eval_rougeL": 0.0506, | |
| "eval_rougeLsum": 0.0507, | |
| "eval_runtime": 9.474, | |
| "eval_samples_per_second": 11.611, | |
| "eval_steps_per_second": 1.478, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 156.98, | |
| "eval_gen_len": 10.9364, | |
| "eval_loss": 0.3396497666835785, | |
| "eval_rouge1": 0.0591, | |
| "eval_rouge2": 0.0351, | |
| "eval_rougeL": 0.0561, | |
| "eval_rougeLsum": 0.0561, | |
| "eval_runtime": 9.496, | |
| "eval_samples_per_second": 11.584, | |
| "eval_steps_per_second": 1.474, | |
| "step": 4317 | |
| }, | |
| { | |
| "epoch": 158.0, | |
| "eval_gen_len": 11.0364, | |
| "eval_loss": 0.3386593759059906, | |
| "eval_rouge1": 0.0633, | |
| "eval_rouge2": 0.0387, | |
| "eval_rougeL": 0.0605, | |
| "eval_rougeLsum": 0.0601, | |
| "eval_runtime": 9.4912, | |
| "eval_samples_per_second": 11.59, | |
| "eval_steps_per_second": 1.475, | |
| "step": 4345 | |
| }, | |
| { | |
| "epoch": 158.98, | |
| "eval_gen_len": 11.4455, | |
| "eval_loss": 0.3368191123008728, | |
| "eval_rouge1": 0.0614, | |
| "eval_rouge2": 0.0371, | |
| "eval_rougeL": 0.0593, | |
| "eval_rougeLsum": 0.0583, | |
| "eval_runtime": 9.5012, | |
| "eval_samples_per_second": 11.578, | |
| "eval_steps_per_second": 1.474, | |
| "step": 4372 | |
| }, | |
| { | |
| "epoch": 160.0, | |
| "eval_gen_len": 11.4545, | |
| "eval_loss": 0.33700016140937805, | |
| "eval_rouge1": 0.0702, | |
| "eval_rouge2": 0.0444, | |
| "eval_rougeL": 0.0672, | |
| "eval_rougeLsum": 0.0671, | |
| "eval_runtime": 9.4911, | |
| "eval_samples_per_second": 11.59, | |
| "eval_steps_per_second": 1.475, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 160.98, | |
| "eval_gen_len": 11.4182, | |
| "eval_loss": 0.3347805440425873, | |
| "eval_rouge1": 0.0702, | |
| "eval_rouge2": 0.0444, | |
| "eval_rougeL": 0.0672, | |
| "eval_rougeLsum": 0.0671, | |
| "eval_runtime": 9.4829, | |
| "eval_samples_per_second": 11.6, | |
| "eval_steps_per_second": 1.476, | |
| "step": 4427 | |
| }, | |
| { | |
| "epoch": 162.0, | |
| "eval_gen_len": 11.2818, | |
| "eval_loss": 0.332653284072876, | |
| "eval_rouge1": 0.0691, | |
| "eval_rouge2": 0.0438, | |
| "eval_rougeL": 0.0661, | |
| "eval_rougeLsum": 0.0656, | |
| "eval_runtime": 9.5355, | |
| "eval_samples_per_second": 11.536, | |
| "eval_steps_per_second": 1.468, | |
| "step": 4455 | |
| }, | |
| { | |
| "epoch": 162.98, | |
| "eval_gen_len": 11.3545, | |
| "eval_loss": 0.3307643234729767, | |
| "eval_rouge1": 0.0648, | |
| "eval_rouge2": 0.0405, | |
| "eval_rougeL": 0.0614, | |
| "eval_rougeLsum": 0.0609, | |
| "eval_runtime": 9.4851, | |
| "eval_samples_per_second": 11.597, | |
| "eval_steps_per_second": 1.476, | |
| "step": 4482 | |
| }, | |
| { | |
| "epoch": 163.64, | |
| "grad_norm": 0.6090702414512634, | |
| "learning_rate": 5.20164609053498e-06, | |
| "loss": 0.4471, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 164.0, | |
| "eval_gen_len": 11.7909, | |
| "eval_loss": 0.32992881536483765, | |
| "eval_rouge1": 0.0711, | |
| "eval_rouge2": 0.0441, | |
| "eval_rougeL": 0.0677, | |
| "eval_rougeLsum": 0.0667, | |
| "eval_runtime": 9.5142, | |
| "eval_samples_per_second": 11.562, | |
| "eval_steps_per_second": 1.471, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 164.98, | |
| "eval_gen_len": 12.0273, | |
| "eval_loss": 0.3291892111301422, | |
| "eval_rouge1": 0.0749, | |
| "eval_rouge2": 0.0476, | |
| "eval_rougeL": 0.0709, | |
| "eval_rougeLsum": 0.0701, | |
| "eval_runtime": 9.5232, | |
| "eval_samples_per_second": 11.551, | |
| "eval_steps_per_second": 1.47, | |
| "step": 4537 | |
| }, | |
| { | |
| "epoch": 166.0, | |
| "eval_gen_len": 12.4364, | |
| "eval_loss": 0.3260752558708191, | |
| "eval_rouge1": 0.078, | |
| "eval_rouge2": 0.0484, | |
| "eval_rougeL": 0.0728, | |
| "eval_rougeLsum": 0.0724, | |
| "eval_runtime": 9.5359, | |
| "eval_samples_per_second": 11.535, | |
| "eval_steps_per_second": 1.468, | |
| "step": 4565 | |
| }, | |
| { | |
| "epoch": 166.98, | |
| "eval_gen_len": 12.6636, | |
| "eval_loss": 0.32543399930000305, | |
| "eval_rouge1": 0.0865, | |
| "eval_rouge2": 0.0555, | |
| "eval_rougeL": 0.0812, | |
| "eval_rougeLsum": 0.0806, | |
| "eval_runtime": 9.5501, | |
| "eval_samples_per_second": 11.518, | |
| "eval_steps_per_second": 1.466, | |
| "step": 4592 | |
| }, | |
| { | |
| "epoch": 168.0, | |
| "eval_gen_len": 12.8455, | |
| "eval_loss": 0.3237576484680176, | |
| "eval_rouge1": 0.081, | |
| "eval_rouge2": 0.0501, | |
| "eval_rougeL": 0.0763, | |
| "eval_rougeLsum": 0.0754, | |
| "eval_runtime": 9.5386, | |
| "eval_samples_per_second": 11.532, | |
| "eval_steps_per_second": 1.468, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 168.98, | |
| "eval_gen_len": 13.1273, | |
| "eval_loss": 0.32304662466049194, | |
| "eval_rouge1": 0.0899, | |
| "eval_rouge2": 0.0584, | |
| "eval_rougeL": 0.0846, | |
| "eval_rougeLsum": 0.0836, | |
| "eval_runtime": 9.4963, | |
| "eval_samples_per_second": 11.583, | |
| "eval_steps_per_second": 1.474, | |
| "step": 4647 | |
| }, | |
| { | |
| "epoch": 170.0, | |
| "eval_gen_len": 13.5364, | |
| "eval_loss": 0.3217833638191223, | |
| "eval_rouge1": 0.0936, | |
| "eval_rouge2": 0.0579, | |
| "eval_rougeL": 0.0864, | |
| "eval_rougeLsum": 0.0854, | |
| "eval_runtime": 9.5121, | |
| "eval_samples_per_second": 11.564, | |
| "eval_steps_per_second": 1.472, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 170.98, | |
| "eval_gen_len": 13.9455, | |
| "eval_loss": 0.3209010660648346, | |
| "eval_rouge1": 0.0976, | |
| "eval_rouge2": 0.063, | |
| "eval_rougeL": 0.0914, | |
| "eval_rougeLsum": 0.0904, | |
| "eval_runtime": 9.507, | |
| "eval_samples_per_second": 11.57, | |
| "eval_steps_per_second": 1.473, | |
| "step": 4702 | |
| }, | |
| { | |
| "epoch": 172.0, | |
| "eval_gen_len": 14.2545, | |
| "eval_loss": 0.31976738572120667, | |
| "eval_rouge1": 0.1024, | |
| "eval_rouge2": 0.0663, | |
| "eval_rougeL": 0.0959, | |
| "eval_rougeLsum": 0.0946, | |
| "eval_runtime": 9.527, | |
| "eval_samples_per_second": 11.546, | |
| "eval_steps_per_second": 1.47, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 172.98, | |
| "eval_gen_len": 13.7909, | |
| "eval_loss": 0.31930533051490784, | |
| "eval_rouge1": 0.0943, | |
| "eval_rouge2": 0.0596, | |
| "eval_rougeL": 0.0883, | |
| "eval_rougeLsum": 0.0873, | |
| "eval_runtime": 9.5147, | |
| "eval_samples_per_second": 11.561, | |
| "eval_steps_per_second": 1.471, | |
| "step": 4757 | |
| }, | |
| { | |
| "epoch": 174.0, | |
| "eval_gen_len": 13.9182, | |
| "eval_loss": 0.3177041709423065, | |
| "eval_rouge1": 0.0964, | |
| "eval_rouge2": 0.0605, | |
| "eval_rougeL": 0.0905, | |
| "eval_rougeLsum": 0.0896, | |
| "eval_runtime": 9.4816, | |
| "eval_samples_per_second": 11.601, | |
| "eval_steps_per_second": 1.477, | |
| "step": 4785 | |
| }, | |
| { | |
| "epoch": 174.98, | |
| "eval_gen_len": 13.4636, | |
| "eval_loss": 0.3157893121242523, | |
| "eval_rouge1": 0.0978, | |
| "eval_rouge2": 0.0628, | |
| "eval_rougeL": 0.0918, | |
| "eval_rougeLsum": 0.0905, | |
| "eval_runtime": 9.4906, | |
| "eval_samples_per_second": 11.59, | |
| "eval_steps_per_second": 1.475, | |
| "step": 4812 | |
| }, | |
| { | |
| "epoch": 176.0, | |
| "eval_gen_len": 14.3273, | |
| "eval_loss": 0.3147883713245392, | |
| "eval_rouge1": 0.1021, | |
| "eval_rouge2": 0.0643, | |
| "eval_rougeL": 0.0951, | |
| "eval_rougeLsum": 0.0943, | |
| "eval_runtime": 9.5161, | |
| "eval_samples_per_second": 11.559, | |
| "eval_steps_per_second": 1.471, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 176.98, | |
| "eval_gen_len": 14.3727, | |
| "eval_loss": 0.31470227241516113, | |
| "eval_rouge1": 0.1043, | |
| "eval_rouge2": 0.0661, | |
| "eval_rougeL": 0.0982, | |
| "eval_rougeLsum": 0.0972, | |
| "eval_runtime": 9.5231, | |
| "eval_samples_per_second": 11.551, | |
| "eval_steps_per_second": 1.47, | |
| "step": 4867 | |
| }, | |
| { | |
| "epoch": 178.0, | |
| "eval_gen_len": 14.4455, | |
| "eval_loss": 0.31245002150535583, | |
| "eval_rouge1": 0.1068, | |
| "eval_rouge2": 0.0691, | |
| "eval_rougeL": 0.1016, | |
| "eval_rougeLsum": 0.1001, | |
| "eval_runtime": 9.5249, | |
| "eval_samples_per_second": 11.549, | |
| "eval_steps_per_second": 1.47, | |
| "step": 4895 | |
| }, | |
| { | |
| "epoch": 178.98, | |
| "eval_gen_len": 14.7091, | |
| "eval_loss": 0.3121263384819031, | |
| "eval_rouge1": 0.1073, | |
| "eval_rouge2": 0.0682, | |
| "eval_rougeL": 0.1012, | |
| "eval_rougeLsum": 0.1, | |
| "eval_runtime": 9.531, | |
| "eval_samples_per_second": 11.541, | |
| "eval_steps_per_second": 1.469, | |
| "step": 4922 | |
| }, | |
| { | |
| "epoch": 180.0, | |
| "eval_gen_len": 14.7182, | |
| "eval_loss": 0.310048907995224, | |
| "eval_rouge1": 0.1095, | |
| "eval_rouge2": 0.0698, | |
| "eval_rougeL": 0.1038, | |
| "eval_rougeLsum": 0.1022, | |
| "eval_runtime": 9.5031, | |
| "eval_samples_per_second": 11.575, | |
| "eval_steps_per_second": 1.473, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 180.98, | |
| "eval_gen_len": 14.5727, | |
| "eval_loss": 0.3098377585411072, | |
| "eval_rouge1": 0.1122, | |
| "eval_rouge2": 0.0717, | |
| "eval_rougeL": 0.1054, | |
| "eval_rougeLsum": 0.1048, | |
| "eval_runtime": 9.5178, | |
| "eval_samples_per_second": 11.557, | |
| "eval_steps_per_second": 1.471, | |
| "step": 4977 | |
| }, | |
| { | |
| "epoch": 181.82, | |
| "grad_norm": 0.6373523473739624, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.4093, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 182.0, | |
| "eval_gen_len": 14.8182, | |
| "eval_loss": 0.3088683784008026, | |
| "eval_rouge1": 0.1148, | |
| "eval_rouge2": 0.074, | |
| "eval_rougeL": 0.1082, | |
| "eval_rougeLsum": 0.1071, | |
| "eval_runtime": 9.5147, | |
| "eval_samples_per_second": 11.561, | |
| "eval_steps_per_second": 1.471, | |
| "step": 5005 | |
| }, | |
| { | |
| "epoch": 182.98, | |
| "eval_gen_len": 14.6727, | |
| "eval_loss": 0.30828168988227844, | |
| "eval_rouge1": 0.1088, | |
| "eval_rouge2": 0.0689, | |
| "eval_rougeL": 0.1021, | |
| "eval_rougeLsum": 0.1014, | |
| "eval_runtime": 9.5092, | |
| "eval_samples_per_second": 11.568, | |
| "eval_steps_per_second": 1.472, | |
| "step": 5032 | |
| }, | |
| { | |
| "epoch": 184.0, | |
| "eval_gen_len": 15.2182, | |
| "eval_loss": 0.30716758966445923, | |
| "eval_rouge1": 0.1167, | |
| "eval_rouge2": 0.0746, | |
| "eval_rougeL": 0.1098, | |
| "eval_rougeLsum": 0.1084, | |
| "eval_runtime": 9.5556, | |
| "eval_samples_per_second": 11.512, | |
| "eval_steps_per_second": 1.465, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 184.98, | |
| "eval_gen_len": 15.9364, | |
| "eval_loss": 0.3059370815753937, | |
| "eval_rouge1": 0.1233, | |
| "eval_rouge2": 0.08, | |
| "eval_rougeL": 0.1166, | |
| "eval_rougeLsum": 0.1157, | |
| "eval_runtime": 9.5537, | |
| "eval_samples_per_second": 11.514, | |
| "eval_steps_per_second": 1.465, | |
| "step": 5087 | |
| }, | |
| { | |
| "epoch": 186.0, | |
| "eval_gen_len": 15.1727, | |
| "eval_loss": 0.3056795597076416, | |
| "eval_rouge1": 0.1128, | |
| "eval_rouge2": 0.0707, | |
| "eval_rougeL": 0.1055, | |
| "eval_rougeLsum": 0.1049, | |
| "eval_runtime": 9.5621, | |
| "eval_samples_per_second": 11.504, | |
| "eval_steps_per_second": 1.464, | |
| "step": 5115 | |
| }, | |
| { | |
| "epoch": 186.98, | |
| "eval_gen_len": 15.1818, | |
| "eval_loss": 0.3043256402015686, | |
| "eval_rouge1": 0.1131, | |
| "eval_rouge2": 0.0707, | |
| "eval_rougeL": 0.1057, | |
| "eval_rougeLsum": 0.105, | |
| "eval_runtime": 9.5545, | |
| "eval_samples_per_second": 11.513, | |
| "eval_steps_per_second": 1.465, | |
| "step": 5142 | |
| }, | |
| { | |
| "epoch": 188.0, | |
| "eval_gen_len": 15.1727, | |
| "eval_loss": 0.30425599217414856, | |
| "eval_rouge1": 0.1125, | |
| "eval_rouge2": 0.0703, | |
| "eval_rougeL": 0.1052, | |
| "eval_rougeLsum": 0.1046, | |
| "eval_runtime": 9.5018, | |
| "eval_samples_per_second": 11.577, | |
| "eval_steps_per_second": 1.473, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 188.98, | |
| "eval_gen_len": 15.1636, | |
| "eval_loss": 0.30395984649658203, | |
| "eval_rouge1": 0.1128, | |
| "eval_rouge2": 0.0705, | |
| "eval_rougeL": 0.1054, | |
| "eval_rougeLsum": 0.1049, | |
| "eval_runtime": 9.5213, | |
| "eval_samples_per_second": 11.553, | |
| "eval_steps_per_second": 1.47, | |
| "step": 5197 | |
| }, | |
| { | |
| "epoch": 190.0, | |
| "eval_gen_len": 15.2455, | |
| "eval_loss": 0.30324217677116394, | |
| "eval_rouge1": 0.1136, | |
| "eval_rouge2": 0.0705, | |
| "eval_rougeL": 0.1061, | |
| "eval_rougeLsum": 0.1056, | |
| "eval_runtime": 9.5432, | |
| "eval_samples_per_second": 11.527, | |
| "eval_steps_per_second": 1.467, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 190.98, | |
| "eval_gen_len": 15.6182, | |
| "eval_loss": 0.30257585644721985, | |
| "eval_rouge1": 0.1149, | |
| "eval_rouge2": 0.071, | |
| "eval_rougeL": 0.1075, | |
| "eval_rougeLsum": 0.107, | |
| "eval_runtime": 9.5398, | |
| "eval_samples_per_second": 11.531, | |
| "eval_steps_per_second": 1.468, | |
| "step": 5252 | |
| }, | |
| { | |
| "epoch": 192.0, | |
| "eval_gen_len": 15.7545, | |
| "eval_loss": 0.301755428314209, | |
| "eval_rouge1": 0.118, | |
| "eval_rouge2": 0.0744, | |
| "eval_rougeL": 0.1114, | |
| "eval_rougeLsum": 0.1105, | |
| "eval_runtime": 9.5491, | |
| "eval_samples_per_second": 11.519, | |
| "eval_steps_per_second": 1.466, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 192.98, | |
| "eval_gen_len": 15.7545, | |
| "eval_loss": 0.30100175738334656, | |
| "eval_rouge1": 0.1186, | |
| "eval_rouge2": 0.0756, | |
| "eval_rougeL": 0.1122, | |
| "eval_rougeLsum": 0.1116, | |
| "eval_runtime": 9.5605, | |
| "eval_samples_per_second": 11.506, | |
| "eval_steps_per_second": 1.464, | |
| "step": 5307 | |
| }, | |
| { | |
| "epoch": 194.0, | |
| "eval_gen_len": 15.6727, | |
| "eval_loss": 0.3014240562915802, | |
| "eval_rouge1": 0.1169, | |
| "eval_rouge2": 0.0738, | |
| "eval_rougeL": 0.1106, | |
| "eval_rougeLsum": 0.1094, | |
| "eval_runtime": 9.5649, | |
| "eval_samples_per_second": 11.5, | |
| "eval_steps_per_second": 1.464, | |
| "step": 5335 | |
| }, | |
| { | |
| "epoch": 194.98, | |
| "eval_gen_len": 15.6364, | |
| "eval_loss": 0.3001127541065216, | |
| "eval_rouge1": 0.1161, | |
| "eval_rouge2": 0.0734, | |
| "eval_rougeL": 0.1104, | |
| "eval_rougeLsum": 0.1092, | |
| "eval_runtime": 9.558, | |
| "eval_samples_per_second": 11.509, | |
| "eval_steps_per_second": 1.465, | |
| "step": 5362 | |
| }, | |
| { | |
| "epoch": 196.0, | |
| "eval_gen_len": 15.7909, | |
| "eval_loss": 0.29918792843818665, | |
| "eval_rouge1": 0.1192, | |
| "eval_rouge2": 0.0752, | |
| "eval_rougeL": 0.1132, | |
| "eval_rougeLsum": 0.1125, | |
| "eval_runtime": 9.5635, | |
| "eval_samples_per_second": 11.502, | |
| "eval_steps_per_second": 1.464, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 196.98, | |
| "eval_gen_len": 15.6364, | |
| "eval_loss": 0.2991277277469635, | |
| "eval_rouge1": 0.1205, | |
| "eval_rouge2": 0.0772, | |
| "eval_rougeL": 0.1141, | |
| "eval_rougeLsum": 0.1133, | |
| "eval_runtime": 9.524, | |
| "eval_samples_per_second": 11.55, | |
| "eval_steps_per_second": 1.47, | |
| "step": 5417 | |
| }, | |
| { | |
| "epoch": 198.0, | |
| "eval_gen_len": 15.7909, | |
| "eval_loss": 0.2986967861652374, | |
| "eval_rouge1": 0.1202, | |
| "eval_rouge2": 0.0768, | |
| "eval_rougeL": 0.1143, | |
| "eval_rougeLsum": 0.1138, | |
| "eval_runtime": 9.5106, | |
| "eval_samples_per_second": 11.566, | |
| "eval_steps_per_second": 1.472, | |
| "step": 5445 | |
| }, | |
| { | |
| "epoch": 198.98, | |
| "eval_gen_len": 15.6182, | |
| "eval_loss": 0.2981720566749573, | |
| "eval_rouge1": 0.1217, | |
| "eval_rouge2": 0.0793, | |
| "eval_rougeL": 0.1153, | |
| "eval_rougeLsum": 0.1148, | |
| "eval_runtime": 9.5676, | |
| "eval_samples_per_second": 11.497, | |
| "eval_steps_per_second": 1.463, | |
| "step": 5472 | |
| }, | |
| { | |
| "epoch": 200.0, | |
| "grad_norm": 1.2200640439987183, | |
| "learning_rate": 1.909465020576132e-06, | |
| "loss": 0.3864, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 200.0, | |
| "eval_gen_len": 15.7818, | |
| "eval_loss": 0.29755449295043945, | |
| "eval_rouge1": 0.1218, | |
| "eval_rouge2": 0.079, | |
| "eval_rougeL": 0.1154, | |
| "eval_rougeLsum": 0.1147, | |
| "eval_runtime": 9.547, | |
| "eval_samples_per_second": 11.522, | |
| "eval_steps_per_second": 1.466, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 200.98, | |
| "eval_gen_len": 15.6182, | |
| "eval_loss": 0.29704758524894714, | |
| "eval_rouge1": 0.1219, | |
| "eval_rouge2": 0.0811, | |
| "eval_rougeL": 0.1163, | |
| "eval_rougeLsum": 0.116, | |
| "eval_runtime": 9.5552, | |
| "eval_samples_per_second": 11.512, | |
| "eval_steps_per_second": 1.465, | |
| "step": 5527 | |
| }, | |
| { | |
| "epoch": 202.0, | |
| "eval_gen_len": 15.5273, | |
| "eval_loss": 0.29719075560569763, | |
| "eval_rouge1": 0.122, | |
| "eval_rouge2": 0.0811, | |
| "eval_rougeL": 0.116, | |
| "eval_rougeLsum": 0.1157, | |
| "eval_runtime": 9.5815, | |
| "eval_samples_per_second": 11.48, | |
| "eval_steps_per_second": 1.461, | |
| "step": 5555 | |
| }, | |
| { | |
| "epoch": 202.98, | |
| "eval_gen_len": 15.6909, | |
| "eval_loss": 0.29677239060401917, | |
| "eval_rouge1": 0.1209, | |
| "eval_rouge2": 0.0803, | |
| "eval_rougeL": 0.1156, | |
| "eval_rougeLsum": 0.1149, | |
| "eval_runtime": 9.499, | |
| "eval_samples_per_second": 11.58, | |
| "eval_steps_per_second": 1.474, | |
| "step": 5582 | |
| }, | |
| { | |
| "epoch": 204.0, | |
| "eval_gen_len": 15.5273, | |
| "eval_loss": 0.2963531017303467, | |
| "eval_rouge1": 0.1251, | |
| "eval_rouge2": 0.0846, | |
| "eval_rougeL": 0.1207, | |
| "eval_rougeLsum": 0.1194, | |
| "eval_runtime": 9.5143, | |
| "eval_samples_per_second": 11.562, | |
| "eval_steps_per_second": 1.471, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 204.98, | |
| "eval_gen_len": 15.6909, | |
| "eval_loss": 0.29531627893447876, | |
| "eval_rouge1": 0.1239, | |
| "eval_rouge2": 0.0831, | |
| "eval_rougeL": 0.1193, | |
| "eval_rougeLsum": 0.1184, | |
| "eval_runtime": 9.5198, | |
| "eval_samples_per_second": 11.555, | |
| "eval_steps_per_second": 1.471, | |
| "step": 5637 | |
| }, | |
| { | |
| "epoch": 206.0, | |
| "eval_gen_len": 15.5273, | |
| "eval_loss": 0.29536357522010803, | |
| "eval_rouge1": 0.1236, | |
| "eval_rouge2": 0.0835, | |
| "eval_rougeL": 0.1192, | |
| "eval_rougeLsum": 0.1182, | |
| "eval_runtime": 9.5066, | |
| "eval_samples_per_second": 11.571, | |
| "eval_steps_per_second": 1.473, | |
| "step": 5665 | |
| }, | |
| { | |
| "epoch": 206.98, | |
| "eval_gen_len": 15.6727, | |
| "eval_loss": 0.2951861023902893, | |
| "eval_rouge1": 0.1236, | |
| "eval_rouge2": 0.0832, | |
| "eval_rougeL": 0.1191, | |
| "eval_rougeLsum": 0.1181, | |
| "eval_runtime": 9.5179, | |
| "eval_samples_per_second": 11.557, | |
| "eval_steps_per_second": 1.471, | |
| "step": 5692 | |
| }, | |
| { | |
| "epoch": 208.0, | |
| "eval_gen_len": 15.8273, | |
| "eval_loss": 0.29457393288612366, | |
| "eval_rouge1": 0.1262, | |
| "eval_rouge2": 0.0856, | |
| "eval_rougeL": 0.1223, | |
| "eval_rougeLsum": 0.121, | |
| "eval_runtime": 9.5235, | |
| "eval_samples_per_second": 11.55, | |
| "eval_steps_per_second": 1.47, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 208.98, | |
| "eval_gen_len": 15.8273, | |
| "eval_loss": 0.29461580514907837, | |
| "eval_rouge1": 0.1269, | |
| "eval_rouge2": 0.086, | |
| "eval_rougeL": 0.1227, | |
| "eval_rougeLsum": 0.1213, | |
| "eval_runtime": 9.5193, | |
| "eval_samples_per_second": 11.556, | |
| "eval_steps_per_second": 1.471, | |
| "step": 5747 | |
| }, | |
| { | |
| "epoch": 210.0, | |
| "eval_gen_len": 15.6727, | |
| "eval_loss": 0.2948046028614044, | |
| "eval_rouge1": 0.1261, | |
| "eval_rouge2": 0.0859, | |
| "eval_rougeL": 0.1221, | |
| "eval_rougeLsum": 0.1208, | |
| "eval_runtime": 9.5297, | |
| "eval_samples_per_second": 11.543, | |
| "eval_steps_per_second": 1.469, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 210.98, | |
| "eval_gen_len": 15.7636, | |
| "eval_loss": 0.29468077421188354, | |
| "eval_rouge1": 0.129, | |
| "eval_rouge2": 0.0888, | |
| "eval_rougeL": 0.1244, | |
| "eval_rougeLsum": 0.1235, | |
| "eval_runtime": 9.5285, | |
| "eval_samples_per_second": 11.544, | |
| "eval_steps_per_second": 1.469, | |
| "step": 5802 | |
| }, | |
| { | |
| "epoch": 212.0, | |
| "eval_gen_len": 15.9091, | |
| "eval_loss": 0.2943172752857208, | |
| "eval_rouge1": 0.1308, | |
| "eval_rouge2": 0.0909, | |
| "eval_rougeL": 0.1267, | |
| "eval_rougeLsum": 0.1254, | |
| "eval_runtime": 9.5333, | |
| "eval_samples_per_second": 11.538, | |
| "eval_steps_per_second": 1.469, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 212.98, | |
| "eval_gen_len": 15.9091, | |
| "eval_loss": 0.2938406467437744, | |
| "eval_rouge1": 0.1293, | |
| "eval_rouge2": 0.0888, | |
| "eval_rougeL": 0.1251, | |
| "eval_rougeLsum": 0.1236, | |
| "eval_runtime": 9.5369, | |
| "eval_samples_per_second": 11.534, | |
| "eval_steps_per_second": 1.468, | |
| "step": 5857 | |
| }, | |
| { | |
| "epoch": 214.0, | |
| "eval_gen_len": 15.9727, | |
| "eval_loss": 0.2933821678161621, | |
| "eval_rouge1": 0.1284, | |
| "eval_rouge2": 0.0877, | |
| "eval_rougeL": 0.1243, | |
| "eval_rougeLsum": 0.123, | |
| "eval_runtime": 9.5327, | |
| "eval_samples_per_second": 11.539, | |
| "eval_steps_per_second": 1.469, | |
| "step": 5885 | |
| }, | |
| { | |
| "epoch": 214.98, | |
| "eval_gen_len": 15.8727, | |
| "eval_loss": 0.2933785617351532, | |
| "eval_rouge1": 0.1262, | |
| "eval_rouge2": 0.0855, | |
| "eval_rougeL": 0.1221, | |
| "eval_rougeLsum": 0.1208, | |
| "eval_runtime": 9.542, | |
| "eval_samples_per_second": 11.528, | |
| "eval_steps_per_second": 1.467, | |
| "step": 5912 | |
| }, | |
| { | |
| "epoch": 216.0, | |
| "eval_gen_len": 15.8727, | |
| "eval_loss": 0.29337677359580994, | |
| "eval_rouge1": 0.1267, | |
| "eval_rouge2": 0.0863, | |
| "eval_rougeL": 0.1225, | |
| "eval_rougeLsum": 0.1214, | |
| "eval_runtime": 9.5387, | |
| "eval_samples_per_second": 11.532, | |
| "eval_steps_per_second": 1.468, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 216.98, | |
| "eval_gen_len": 15.9636, | |
| "eval_loss": 0.2933517396450043, | |
| "eval_rouge1": 0.1292, | |
| "eval_rouge2": 0.0894, | |
| "eval_rougeL": 0.1254, | |
| "eval_rougeLsum": 0.1241, | |
| "eval_runtime": 9.5629, | |
| "eval_samples_per_second": 11.503, | |
| "eval_steps_per_second": 1.464, | |
| "step": 5967 | |
| }, | |
| { | |
| "epoch": 218.0, | |
| "eval_gen_len": 15.9636, | |
| "eval_loss": 0.29335835576057434, | |
| "eval_rouge1": 0.1289, | |
| "eval_rouge2": 0.0889, | |
| "eval_rougeL": 0.1249, | |
| "eval_rougeLsum": 0.1234, | |
| "eval_runtime": 9.5531, | |
| "eval_samples_per_second": 11.515, | |
| "eval_steps_per_second": 1.465, | |
| "step": 5995 | |
| }, | |
| { | |
| "epoch": 218.18, | |
| "grad_norm": 0.6804682016372681, | |
| "learning_rate": 2.6337448559670784e-07, | |
| "loss": 0.3747, | |
| "step": 6000 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 6075, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 225, | |
| "save_steps": 500, | |
| "total_flos": 1.1611644167297434e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |