| { | |
| "best_metric": 42.7479, | |
| "best_model_checkpoint": "opus_base_lsp_adapt_wce_prop_0.4_weight_1.25/checkpoint-80000", | |
| "epoch": 3.943217665615142, | |
| "global_step": 80000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9753918572555207e-05, | |
| "loss": 0.203, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_bleu": 35.9536, | |
| "eval_gen_len": 39.4238, | |
| "eval_loss": 0.10900890082120895, | |
| "eval_runtime": 184.2385, | |
| "eval_samples_per_second": 5.661, | |
| "eval_steps_per_second": 0.179, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.950746746845426e-05, | |
| "loss": 0.1702, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "eval_bleu": 37.6437, | |
| "eval_gen_len": 37.813, | |
| "eval_loss": 0.10653014481067657, | |
| "eval_runtime": 154.7637, | |
| "eval_samples_per_second": 6.739, | |
| "eval_steps_per_second": 0.213, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.9261016364353314e-05, | |
| "loss": 0.1634, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "eval_bleu": 41.2278, | |
| "eval_gen_len": 34.6328, | |
| "eval_loss": 0.10528820008039474, | |
| "eval_runtime": 161.838, | |
| "eval_samples_per_second": 6.445, | |
| "eval_steps_per_second": 0.204, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.9014565260252367e-05, | |
| "loss": 0.1581, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "eval_bleu": 41.3223, | |
| "eval_gen_len": 35.1716, | |
| "eval_loss": 0.10417595505714417, | |
| "eval_runtime": 134.2346, | |
| "eval_samples_per_second": 7.77, | |
| "eval_steps_per_second": 0.246, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.876811415615142e-05, | |
| "loss": 0.1558, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_bleu": 41.7183, | |
| "eval_gen_len": 35.8869, | |
| "eval_loss": 0.10362090915441513, | |
| "eval_runtime": 140.7284, | |
| "eval_samples_per_second": 7.411, | |
| "eval_steps_per_second": 0.234, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.8521786277602524e-05, | |
| "loss": 0.1444, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "eval_bleu": 41.7158, | |
| "eval_gen_len": 35.8408, | |
| "eval_loss": 0.10352081805467606, | |
| "eval_runtime": 141.9378, | |
| "eval_samples_per_second": 7.348, | |
| "eval_steps_per_second": 0.232, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.8275520011829655e-05, | |
| "loss": 0.1442, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "eval_bleu": 41.8593, | |
| "eval_gen_len": 34.3624, | |
| "eval_loss": 0.10272093862295151, | |
| "eval_runtime": 126.4713, | |
| "eval_samples_per_second": 8.247, | |
| "eval_steps_per_second": 0.261, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.8029130520504733e-05, | |
| "loss": 0.1437, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "eval_bleu": 41.6189, | |
| "eval_gen_len": 35.7239, | |
| "eval_loss": 0.1021599993109703, | |
| "eval_runtime": 129.5482, | |
| "eval_samples_per_second": 8.051, | |
| "eval_steps_per_second": 0.255, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.7782741029179815e-05, | |
| "loss": 0.1424, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "eval_bleu": 41.4617, | |
| "eval_gen_len": 36.1496, | |
| "eval_loss": 0.10178348422050476, | |
| "eval_runtime": 111.4531, | |
| "eval_samples_per_second": 9.358, | |
| "eval_steps_per_second": 0.296, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.7536413150630914e-05, | |
| "loss": 0.1416, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "eval_bleu": 42.2455, | |
| "eval_gen_len": 34.1246, | |
| "eval_loss": 0.10190726071596146, | |
| "eval_runtime": 113.6774, | |
| "eval_samples_per_second": 9.175, | |
| "eval_steps_per_second": 0.29, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.729008527208202e-05, | |
| "loss": 0.1342, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "eval_bleu": 39.6896, | |
| "eval_gen_len": 37.3634, | |
| "eval_loss": 0.10224150866270065, | |
| "eval_runtime": 175.0132, | |
| "eval_samples_per_second": 5.96, | |
| "eval_steps_per_second": 0.189, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.70436957807571e-05, | |
| "loss": 0.1334, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "eval_bleu": 42.2249, | |
| "eval_gen_len": 35.9415, | |
| "eval_loss": 0.10214059799909592, | |
| "eval_runtime": 118.9158, | |
| "eval_samples_per_second": 8.771, | |
| "eval_steps_per_second": 0.278, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1.6797367902208205e-05, | |
| "loss": 0.1323, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "eval_bleu": 42.292, | |
| "eval_gen_len": 34.9655, | |
| "eval_loss": 0.10183346271514893, | |
| "eval_runtime": 145.4935, | |
| "eval_samples_per_second": 7.169, | |
| "eval_steps_per_second": 0.227, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.6551040023659308e-05, | |
| "loss": 0.1326, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "eval_bleu": 41.9963, | |
| "eval_gen_len": 35.5781, | |
| "eval_loss": 0.10167970508337021, | |
| "eval_runtime": 188.6137, | |
| "eval_samples_per_second": 5.53, | |
| "eval_steps_per_second": 0.175, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 1.630471214511041e-05, | |
| "loss": 0.1329, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "eval_bleu": 42.4612, | |
| "eval_gen_len": 34.8341, | |
| "eval_loss": 0.10102493315935135, | |
| "eval_runtime": 127.7158, | |
| "eval_samples_per_second": 8.167, | |
| "eval_steps_per_second": 0.258, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 1.6058384266561514e-05, | |
| "loss": 0.1267, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "eval_bleu": 41.9042, | |
| "eval_gen_len": 33.8245, | |
| "eval_loss": 0.10175066441297531, | |
| "eval_runtime": 141.7885, | |
| "eval_samples_per_second": 7.356, | |
| "eval_steps_per_second": 0.233, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 1.581205638801262e-05, | |
| "loss": 0.1262, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "eval_bleu": 42.5921, | |
| "eval_gen_len": 34.6807, | |
| "eval_loss": 0.10201378166675568, | |
| "eval_runtime": 132.2812, | |
| "eval_samples_per_second": 7.885, | |
| "eval_steps_per_second": 0.249, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 1.55656668966877e-05, | |
| "loss": 0.125, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "eval_bleu": 42.0669, | |
| "eval_gen_len": 35.1496, | |
| "eval_loss": 0.10195770859718323, | |
| "eval_runtime": 90.9683, | |
| "eval_samples_per_second": 11.466, | |
| "eval_steps_per_second": 0.363, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 1.5319339018138805e-05, | |
| "loss": 0.1255, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "eval_bleu": 42.2988, | |
| "eval_gen_len": 35.1266, | |
| "eval_loss": 0.10172092169523239, | |
| "eval_runtime": 123.5502, | |
| "eval_samples_per_second": 8.442, | |
| "eval_steps_per_second": 0.267, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.5073011139589906e-05, | |
| "loss": 0.1262, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "eval_bleu": 42.7479, | |
| "eval_gen_len": 33.9904, | |
| "eval_loss": 0.10153283178806305, | |
| "eval_runtime": 148.6262, | |
| "eval_samples_per_second": 7.018, | |
| "eval_steps_per_second": 0.222, | |
| "step": 80000 | |
| } | |
| ], | |
| "max_steps": 324608, | |
| "num_train_epochs": 16, | |
| "total_flos": 1.3796012702564352e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |