| { | |
| "best_metric": 0.7325565814971924, | |
| "best_model_checkpoint": "../../models/lora/codellama-13b-codtok-multitask_cot_reflection/checkpoint-200", | |
| "epoch": 25.0, | |
| "eval_steps": 200, | |
| "global_step": 400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.9999999999999997e-05, | |
| "loss": 2.0085, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.9999999999999995e-05, | |
| "loss": 1.9902, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 8.999999999999999e-05, | |
| "loss": 1.8388, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 0.00011999999999999999, | |
| "loss": 1.5146, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 0.00015, | |
| "loss": 1.1563, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 0.00017999999999999998, | |
| "loss": 0.9863, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 0.00020999999999999998, | |
| "loss": 0.9537, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.00023999999999999998, | |
| "loss": 0.912, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 0.00027, | |
| "loss": 0.8958, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 0.0003, | |
| "loss": 0.8613, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 0.0002921052631578947, | |
| "loss": 0.8499, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 0.0002842105263157894, | |
| "loss": 0.8511, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 0.0002763157894736842, | |
| "loss": 0.8283, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 0.0002684210526315789, | |
| "loss": 0.812, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "learning_rate": 0.0002605263157894737, | |
| "loss": 0.7994, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 0.00025263157894736836, | |
| "loss": 0.8173, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 10.62, | |
| "learning_rate": 0.00024473684210526314, | |
| "loss": 0.8028, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 11.25, | |
| "learning_rate": 0.00023684210526315788, | |
| "loss": 0.7976, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "learning_rate": 0.00022894736842105263, | |
| "loss": 0.783, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 0.00022105263157894733, | |
| "loss": 0.7765, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "eval_loss": 0.7325565814971924, | |
| "eval_runtime": 1.7436, | |
| "eval_samples_per_second": 12.044, | |
| "eval_steps_per_second": 1.721, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 13.12, | |
| "learning_rate": 0.00021315789473684208, | |
| "loss": 0.7681, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 13.75, | |
| "learning_rate": 0.00020526315789473683, | |
| "loss": 0.751, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 14.38, | |
| "learning_rate": 0.00019736842105263157, | |
| "loss": 0.7483, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 0.0001894736842105263, | |
| "loss": 0.7378, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 15.62, | |
| "learning_rate": 0.00018157894736842105, | |
| "loss": 0.751, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 16.25, | |
| "learning_rate": 0.0001736842105263158, | |
| "loss": 0.7225, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 16.88, | |
| "learning_rate": 0.00016578947368421052, | |
| "loss": 0.7144, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 17.5, | |
| "learning_rate": 0.00015789473684210524, | |
| "loss": 0.7131, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 18.12, | |
| "learning_rate": 0.00015, | |
| "loss": 0.691, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 0.0001421052631578947, | |
| "loss": 0.6649, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 19.38, | |
| "learning_rate": 0.00013421052631578946, | |
| "loss": 0.6819, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.00012631578947368418, | |
| "loss": 0.6695, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 20.62, | |
| "learning_rate": 0.00011842105263157894, | |
| "loss": 0.6846, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 21.25, | |
| "learning_rate": 0.00011052631578947366, | |
| "loss": 0.6421, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 21.88, | |
| "learning_rate": 0.00010263157894736841, | |
| "loss": 0.6467, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 22.5, | |
| "learning_rate": 9.473684210526315e-05, | |
| "loss": 0.6359, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 23.12, | |
| "learning_rate": 8.68421052631579e-05, | |
| "loss": 0.63, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 23.75, | |
| "learning_rate": 7.894736842105262e-05, | |
| "loss": 0.5977, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 24.38, | |
| "learning_rate": 7.105263157894735e-05, | |
| "loss": 0.611, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 6.315789473684209e-05, | |
| "loss": 0.6014, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_loss": 0.8105262517929077, | |
| "eval_runtime": 1.7389, | |
| "eval_samples_per_second": 12.076, | |
| "eval_steps_per_second": 1.725, | |
| "step": 400 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 480, | |
| "num_train_epochs": 30, | |
| "save_steps": 200, | |
| "total_flos": 1.0367112417376666e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |