| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.020380434782608696, | |
| "eval_steps": 5, | |
| "global_step": 150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0006793478260869565, | |
| "eval_loss": 4.499350547790527, | |
| "eval_runtime": 20.1523, | |
| "eval_samples_per_second": 975.424, | |
| "eval_steps_per_second": 7.642, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.001358695652173913, | |
| "eval_loss": 4.498117923736572, | |
| "eval_runtime": 20.4966, | |
| "eval_samples_per_second": 959.037, | |
| "eval_steps_per_second": 7.513, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0020380434782608695, | |
| "eval_loss": 4.495996475219727, | |
| "eval_runtime": 21.0485, | |
| "eval_samples_per_second": 933.891, | |
| "eval_steps_per_second": 7.316, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.002717391304347826, | |
| "eval_loss": 4.492975234985352, | |
| "eval_runtime": 21.3506, | |
| "eval_samples_per_second": 920.677, | |
| "eval_steps_per_second": 7.213, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0033967391304347825, | |
| "eval_loss": 4.489028453826904, | |
| "eval_runtime": 21.5145, | |
| "eval_samples_per_second": 913.661, | |
| "eval_steps_per_second": 7.158, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.004076086956521739, | |
| "eval_loss": 4.484206199645996, | |
| "eval_runtime": 21.5836, | |
| "eval_samples_per_second": 910.738, | |
| "eval_steps_per_second": 7.135, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.004755434782608696, | |
| "eval_loss": 4.478429317474365, | |
| "eval_runtime": 21.7892, | |
| "eval_samples_per_second": 902.143, | |
| "eval_steps_per_second": 7.068, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.005434782608695652, | |
| "eval_loss": 4.471565246582031, | |
| "eval_runtime": 21.9727, | |
| "eval_samples_per_second": 894.608, | |
| "eval_steps_per_second": 7.009, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.006114130434782609, | |
| "eval_loss": 4.463589668273926, | |
| "eval_runtime": 21.6519, | |
| "eval_samples_per_second": 907.863, | |
| "eval_steps_per_second": 7.113, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.006793478260869565, | |
| "eval_loss": 4.454347610473633, | |
| "eval_runtime": 21.7784, | |
| "eval_samples_per_second": 902.593, | |
| "eval_steps_per_second": 7.071, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.007472826086956522, | |
| "eval_loss": 4.443789958953857, | |
| "eval_runtime": 21.9874, | |
| "eval_samples_per_second": 894.012, | |
| "eval_steps_per_second": 7.004, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.008152173913043478, | |
| "eval_loss": 4.432095527648926, | |
| "eval_runtime": 21.6944, | |
| "eval_samples_per_second": 906.087, | |
| "eval_steps_per_second": 7.099, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.008831521739130434, | |
| "eval_loss": 4.419074535369873, | |
| "eval_runtime": 21.7766, | |
| "eval_samples_per_second": 902.668, | |
| "eval_steps_per_second": 7.072, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.009510869565217392, | |
| "eval_loss": 4.404222011566162, | |
| "eval_runtime": 21.9472, | |
| "eval_samples_per_second": 895.649, | |
| "eval_steps_per_second": 7.017, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.010190217391304348, | |
| "eval_loss": 4.387468338012695, | |
| "eval_runtime": 22.1085, | |
| "eval_samples_per_second": 889.115, | |
| "eval_steps_per_second": 6.966, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.010869565217391304, | |
| "eval_loss": 4.368589878082275, | |
| "eval_runtime": 21.8154, | |
| "eval_samples_per_second": 901.063, | |
| "eval_steps_per_second": 7.059, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.01154891304347826, | |
| "eval_loss": 4.347426891326904, | |
| "eval_runtime": 21.8937, | |
| "eval_samples_per_second": 897.84, | |
| "eval_steps_per_second": 7.034, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.012228260869565218, | |
| "eval_loss": 4.32356595993042, | |
| "eval_runtime": 21.9031, | |
| "eval_samples_per_second": 897.453, | |
| "eval_steps_per_second": 7.031, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.012907608695652174, | |
| "eval_loss": 4.296773433685303, | |
| "eval_runtime": 22.045, | |
| "eval_samples_per_second": 891.674, | |
| "eval_steps_per_second": 6.986, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.01358695652173913, | |
| "grad_norm": 11.771966934204102, | |
| "learning_rate": 4.528985507246377e-07, | |
| "loss": 4.4995, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01358695652173913, | |
| "eval_loss": 4.266582489013672, | |
| "eval_runtime": 21.7748, | |
| "eval_samples_per_second": 902.742, | |
| "eval_steps_per_second": 7.072, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.014266304347826086, | |
| "eval_loss": 4.232626914978027, | |
| "eval_runtime": 21.7463, | |
| "eval_samples_per_second": 903.924, | |
| "eval_steps_per_second": 7.082, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.014945652173913044, | |
| "eval_loss": 4.194711208343506, | |
| "eval_runtime": 22.0767, | |
| "eval_samples_per_second": 890.396, | |
| "eval_steps_per_second": 6.976, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.015625, | |
| "eval_loss": 4.151618957519531, | |
| "eval_runtime": 21.7284, | |
| "eval_samples_per_second": 904.667, | |
| "eval_steps_per_second": 7.087, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.016304347826086956, | |
| "eval_loss": 4.102933883666992, | |
| "eval_runtime": 21.8375, | |
| "eval_samples_per_second": 900.149, | |
| "eval_steps_per_second": 7.052, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.016983695652173912, | |
| "eval_loss": 4.047597408294678, | |
| "eval_runtime": 21.8792, | |
| "eval_samples_per_second": 898.434, | |
| "eval_steps_per_second": 7.039, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.017663043478260868, | |
| "eval_loss": 3.9850316047668457, | |
| "eval_runtime": 22.0074, | |
| "eval_samples_per_second": 893.2, | |
| "eval_steps_per_second": 6.998, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.018342391304347828, | |
| "eval_loss": 3.9161784648895264, | |
| "eval_runtime": 21.9382, | |
| "eval_samples_per_second": 896.016, | |
| "eval_steps_per_second": 7.02, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.019021739130434784, | |
| "eval_loss": 3.839725971221924, | |
| "eval_runtime": 21.8144, | |
| "eval_samples_per_second": 901.1, | |
| "eval_steps_per_second": 7.06, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.01970108695652174, | |
| "eval_loss": 3.7521822452545166, | |
| "eval_runtime": 21.6841, | |
| "eval_samples_per_second": 906.516, | |
| "eval_steps_per_second": 7.102, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.020380434782608696, | |
| "eval_loss": 3.652061700820923, | |
| "eval_runtime": 21.8254, | |
| "eval_samples_per_second": 900.648, | |
| "eval_steps_per_second": 7.056, | |
| "step": 150 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 22080, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 5, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 128, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |