| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 8.71313672922252, | |
| "eval_steps": 500, | |
| "global_step": 6500, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.6702412868632708, | |
| "grad_norm": 9.020133018493652, | |
| "learning_rate": 1.4299394279479472e-06, | |
| "loss": 0.6244, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7267851233482361, | |
| "eval_loss": 0.5142917633056641, | |
| "eval_runtime": 9.4466, | |
| "eval_samples_per_second": 315.773, | |
| "eval_steps_per_second": 19.795, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.3404825737265416, | |
| "grad_norm": 10.139765739440918, | |
| "learning_rate": 1.3272138943309972e-06, | |
| "loss": 0.5438, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.7405296564102173, | |
| "eval_loss": 0.49428224563598633, | |
| "eval_runtime": 9.4794, | |
| "eval_samples_per_second": 314.682, | |
| "eval_steps_per_second": 19.727, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 2.0107238605898123, | |
| "grad_norm": 7.425673961639404, | |
| "learning_rate": 1.2244883607140467e-06, | |
| "loss": 0.5108, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.680965147453083, | |
| "grad_norm": 8.044295310974121, | |
| "learning_rate": 1.1217628270970966e-06, | |
| "loss": 0.4787, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.7468990683555603, | |
| "eval_loss": 0.5006520748138428, | |
| "eval_runtime": 9.5789, | |
| "eval_samples_per_second": 311.415, | |
| "eval_steps_per_second": 19.522, | |
| "step": 2238 | |
| }, | |
| { | |
| "epoch": 3.351206434316354, | |
| "grad_norm": 19.568927764892578, | |
| "learning_rate": 1.0190372934801463e-06, | |
| "loss": 0.458, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.7522628307342529, | |
| "eval_loss": 0.49020281434059143, | |
| "eval_runtime": 9.4558, | |
| "eval_samples_per_second": 315.467, | |
| "eval_steps_per_second": 19.776, | |
| "step": 2984 | |
| }, | |
| { | |
| "epoch": 4.021447721179625, | |
| "grad_norm": 5.546102046966553, | |
| "learning_rate": 9.16311759863196e-07, | |
| "loss": 0.4303, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.6916890080428955, | |
| "grad_norm": 15.734197616577148, | |
| "learning_rate": 8.135862262462459e-07, | |
| "loss": 0.4115, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.7536037564277649, | |
| "eval_loss": 0.5176616907119751, | |
| "eval_runtime": 9.5583, | |
| "eval_samples_per_second": 312.084, | |
| "eval_steps_per_second": 19.564, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 5.361930294906166, | |
| "grad_norm": 15.1699857711792, | |
| "learning_rate": 7.108606926292956e-07, | |
| "loss": 0.401, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.755950391292572, | |
| "eval_loss": 0.5166484713554382, | |
| "eval_runtime": 9.6024, | |
| "eval_samples_per_second": 310.652, | |
| "eval_steps_per_second": 19.474, | |
| "step": 4476 | |
| }, | |
| { | |
| "epoch": 6.032171581769437, | |
| "grad_norm": 15.72288990020752, | |
| "learning_rate": 6.081351590123454e-07, | |
| "loss": 0.3829, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 6.702412868632708, | |
| "grad_norm": 9.840254783630371, | |
| "learning_rate": 5.054096253953951e-07, | |
| "loss": 0.3686, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.7596379518508911, | |
| "eval_loss": 0.5389177799224854, | |
| "eval_runtime": 9.5625, | |
| "eval_samples_per_second": 311.946, | |
| "eval_steps_per_second": 19.555, | |
| "step": 5222 | |
| }, | |
| { | |
| "epoch": 7.372654155495979, | |
| "grad_norm": 13.200855255126953, | |
| "learning_rate": 4.0268409177844495e-07, | |
| "loss": 0.358, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.755950391292572, | |
| "eval_loss": 0.562088131904602, | |
| "eval_runtime": 9.4712, | |
| "eval_samples_per_second": 314.955, | |
| "eval_steps_per_second": 19.744, | |
| "step": 5968 | |
| }, | |
| { | |
| "epoch": 8.04289544235925, | |
| "grad_norm": 15.2745361328125, | |
| "learning_rate": 2.9995855816149467e-07, | |
| "loss": 0.3402, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 8.71313672922252, | |
| "grad_norm": 24.888896942138672, | |
| "learning_rate": 1.9723302454454444e-07, | |
| "loss": 0.3335, | |
| "step": 6500 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 7460, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 1.6928810928603216e+16, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 1.5326649615648975e-06, | |
| "per_device_train_batch_size": 16 | |
| } | |
| } | |