| { | |
| "best_metric": 1.008945345878601, | |
| "best_model_checkpoint": "th_cl_10epochs_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": 34.58167266845703, | |
| "learning_rate": 9.375e-05, | |
| "loss": 0.6976, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 27.326923370361328, | |
| "learning_rate": 8.75e-05, | |
| "loss": 0.8167, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.9375, | |
| "grad_norm": 39.94591522216797, | |
| "learning_rate": 8.125000000000001e-05, | |
| "loss": 0.9267, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4888888888888889, | |
| "eval_balanced_accuracy": 0.4910714285714286, | |
| "eval_loss": 1.0786933898925781, | |
| "eval_runtime": 84.4951, | |
| "eval_samples_per_second": 0.533, | |
| "eval_steps_per_second": 0.071, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 36.723445892333984, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.5409, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.5625, | |
| "grad_norm": 33.607635498046875, | |
| "learning_rate": 6.875e-05, | |
| "loss": 0.4148, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.875, | |
| "grad_norm": 24.300277709960938, | |
| "learning_rate": 6.25e-05, | |
| "loss": 0.583, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.4666666666666667, | |
| "eval_balanced_accuracy": 0.47, | |
| "eval_loss": 1.035575270652771, | |
| "eval_runtime": 84.4534, | |
| "eval_samples_per_second": 0.533, | |
| "eval_steps_per_second": 0.071, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 2.1875, | |
| "grad_norm": 8.202654838562012, | |
| "learning_rate": 5.6250000000000005e-05, | |
| "loss": 0.3811, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 25.63054084777832, | |
| "learning_rate": 5e-05, | |
| "loss": 0.3316, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.8125, | |
| "grad_norm": 23.79229736328125, | |
| "learning_rate": 4.375e-05, | |
| "loss": 0.302, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.5111111111111111, | |
| "eval_balanced_accuracy": 0.5060728744939271, | |
| "eval_loss": 1.008945345878601, | |
| "eval_runtime": 84.6786, | |
| "eval_samples_per_second": 0.531, | |
| "eval_steps_per_second": 0.071, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 3.125, | |
| "grad_norm": 7.110757827758789, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 0.2366, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 3.4375, | |
| "grad_norm": 22.723182678222656, | |
| "learning_rate": 3.125e-05, | |
| "loss": 0.1844, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 9.120621681213379, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.1245, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.4666666666666667, | |
| "eval_balanced_accuracy": 0.4642857142857143, | |
| "eval_loss": 1.0418320894241333, | |
| "eval_runtime": 84.6576, | |
| "eval_samples_per_second": 0.532, | |
| "eval_steps_per_second": 0.071, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 4.0625, | |
| "grad_norm": 4.578421115875244, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 0.1225, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 4.375, | |
| "grad_norm": 3.0875754356384277, | |
| "learning_rate": 1.25e-05, | |
| "loss": 0.0645, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 4.6875, | |
| "grad_norm": 10.612360000610352, | |
| "learning_rate": 6.25e-06, | |
| "loss": 0.0888, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 18.231754302978516, | |
| "learning_rate": 0.0, | |
| "loss": 0.1163, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.4888888888888889, | |
| "eval_balanced_accuracy": 0.4910714285714286, | |
| "eval_loss": 1.0460317134857178, | |
| "eval_runtime": 84.6266, | |
| "eval_samples_per_second": 0.532, | |
| "eval_steps_per_second": 0.071, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 160, | |
| "total_flos": 2.696144222158848e+16, | |
| "train_loss": 0.3707559358328581, | |
| "train_runtime": 7964.4497, | |
| "train_samples_per_second": 0.158, | |
| "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 | |
| } | |