| { | |
| "best_metric": 0.9973708987236023, | |
| "best_model_checkpoint": "th_cl_28epochs_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": 11.923697471618652, | |
| "learning_rate": 9.375e-05, | |
| "loss": 0.2148, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 27.81431770324707, | |
| "learning_rate": 8.75e-05, | |
| "loss": 0.3845, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.9375, | |
| "grad_norm": 12.166078567504883, | |
| "learning_rate": 8.125000000000001e-05, | |
| "loss": 0.4669, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.5111111111111111, | |
| "eval_balanced_accuracy": 0.515, | |
| "eval_loss": 1.022578477859497, | |
| "eval_runtime": 84.4402, | |
| "eval_samples_per_second": 0.533, | |
| "eval_steps_per_second": 0.071, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 23.626405715942383, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.2006, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.5625, | |
| "grad_norm": 22.549772262573242, | |
| "learning_rate": 6.875e-05, | |
| "loss": 0.1906, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.875, | |
| "grad_norm": 21.862329483032227, | |
| "learning_rate": 6.25e-05, | |
| "loss": 0.2575, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.5333333333333333, | |
| "eval_balanced_accuracy": 0.5394736842105263, | |
| "eval_loss": 1.006407618522644, | |
| "eval_runtime": 84.7861, | |
| "eval_samples_per_second": 0.531, | |
| "eval_steps_per_second": 0.071, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 2.1875, | |
| "grad_norm": 8.992438316345215, | |
| "learning_rate": 5.6250000000000005e-05, | |
| "loss": 0.1647, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 11.991353988647461, | |
| "learning_rate": 5e-05, | |
| "loss": 0.157, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.8125, | |
| "grad_norm": 5.575663089752197, | |
| "learning_rate": 4.375e-05, | |
| "loss": 0.0998, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.5777777777777777, | |
| "eval_balanced_accuracy": 0.5743534482758621, | |
| "eval_loss": 0.9973708987236023, | |
| "eval_runtime": 84.5235, | |
| "eval_samples_per_second": 0.532, | |
| "eval_steps_per_second": 0.071, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 3.125, | |
| "grad_norm": 2.0587921142578125, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 0.0978, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 3.4375, | |
| "grad_norm": 24.552135467529297, | |
| "learning_rate": 3.125e-05, | |
| "loss": 0.0775, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 6.25897741317749, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.0339, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.5777777777777777, | |
| "eval_balanced_accuracy": 0.5740740740740741, | |
| "eval_loss": 1.1087063550949097, | |
| "eval_runtime": 86.0452, | |
| "eval_samples_per_second": 0.523, | |
| "eval_steps_per_second": 0.07, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 4.0625, | |
| "grad_norm": 0.7533628940582275, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 0.0369, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 4.375, | |
| "grad_norm": 1.4770331382751465, | |
| "learning_rate": 1.25e-05, | |
| "loss": 0.0142, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 4.6875, | |
| "grad_norm": 4.743549823760986, | |
| "learning_rate": 6.25e-06, | |
| "loss": 0.0246, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 9.260613441467285, | |
| "learning_rate": 0.0, | |
| "loss": 0.048, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.5111111111111111, | |
| "eval_balanced_accuracy": 0.5118577075098814, | |
| "eval_loss": 1.11983060836792, | |
| "eval_runtime": 85.4666, | |
| "eval_samples_per_second": 0.527, | |
| "eval_steps_per_second": 0.07, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 160, | |
| "total_flos": 2.696144222158848e+16, | |
| "train_loss": 0.15434152334928514, | |
| "train_runtime": 7802.4738, | |
| "train_samples_per_second": 0.161, | |
| "train_steps_per_second": 0.021 | |
| } | |
| ], | |
| "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 | |
| } | |