| { | |
| "best_metric": 1.0448113679885864, | |
| "best_model_checkpoint": "th_cl_5epochs_lora_pos_neg/checkpoint-96", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 160, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.3125, | |
| "grad_norm": 49.992950439453125, | |
| "learning_rate": 9.375e-05, | |
| "loss": 1.0996, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 29.154659271240234, | |
| "learning_rate": 8.75e-05, | |
| "loss": 1.1576, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.9375, | |
| "grad_norm": 39.784446716308594, | |
| "learning_rate": 8.125000000000001e-05, | |
| "loss": 1.1552, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.5111111111111111, | |
| "eval_balanced_accuracy": 0.5118577075098814, | |
| "eval_loss": 1.2024661302566528, | |
| "eval_runtime": 84.971, | |
| "eval_samples_per_second": 0.53, | |
| "eval_steps_per_second": 0.071, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 44.496421813964844, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.7388, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.5625, | |
| "grad_norm": 39.3575325012207, | |
| "learning_rate": 6.875e-05, | |
| "loss": 0.5455, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.875, | |
| "grad_norm": 24.707012176513672, | |
| "learning_rate": 6.25e-05, | |
| "loss": 0.7688, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.4888888888888889, | |
| "eval_balanced_accuracy": 0.4939271255060729, | |
| "eval_loss": 1.1111503839492798, | |
| "eval_runtime": 84.9531, | |
| "eval_samples_per_second": 0.53, | |
| "eval_steps_per_second": 0.071, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 2.1875, | |
| "grad_norm": 10.749614715576172, | |
| "learning_rate": 5.6250000000000005e-05, | |
| "loss": 0.4859, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 24.55562973022461, | |
| "learning_rate": 5e-05, | |
| "loss": 0.41, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.8125, | |
| "grad_norm": 38.890281677246094, | |
| "learning_rate": 4.375e-05, | |
| "loss": 0.4098, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.4888888888888889, | |
| "eval_balanced_accuracy": 0.48814229249011853, | |
| "eval_loss": 1.0448113679885864, | |
| "eval_runtime": 85.1798, | |
| "eval_samples_per_second": 0.528, | |
| "eval_steps_per_second": 0.07, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 3.125, | |
| "grad_norm": 11.055780410766602, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 0.341, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 3.4375, | |
| "grad_norm": 18.11351203918457, | |
| "learning_rate": 3.125e-05, | |
| "loss": 0.2338, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 10.180174827575684, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.1714, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.4888888888888889, | |
| "eval_balanced_accuracy": 0.48814229249011853, | |
| "eval_loss": 1.0591027736663818, | |
| "eval_runtime": 84.9785, | |
| "eval_samples_per_second": 0.53, | |
| "eval_steps_per_second": 0.071, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 4.0625, | |
| "grad_norm": 6.068151473999023, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 0.157, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 4.375, | |
| "grad_norm": 2.7363812923431396, | |
| "learning_rate": 1.25e-05, | |
| "loss": 0.0971, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 4.6875, | |
| "grad_norm": 13.663867950439453, | |
| "learning_rate": 6.25e-06, | |
| "loss": 0.1185, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 20.521446228027344, | |
| "learning_rate": 0.0, | |
| "loss": 0.1392, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.4888888888888889, | |
| "eval_balanced_accuracy": 0.48814229249011853, | |
| "eval_loss": 1.0568149089813232, | |
| "eval_runtime": 85.0351, | |
| "eval_samples_per_second": 0.529, | |
| "eval_steps_per_second": 0.071, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 160, | |
| "total_flos": 2.696144222158848e+16, | |
| "train_loss": 0.5018329162150621, | |
| "train_runtime": 8068.1557, | |
| "train_samples_per_second": 0.156, | |
| "train_steps_per_second": 0.02 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 160, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "total_flos": 2.696144222158848e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |