Batch upload: checkpoint-1000 checkpoint-10000 checkpoint-100000 checkpoint-101000 checkpoint-102000 checkpoint-103000 checkpoint-104000 checkpoint-105000 checkpoint-106000 checkpoint-107000 checkpoint-108000 checkpoint-108168 checkpoint-11000 checkpoint-12000 checkpoint-13000 checkpoint-14000 checkpoint-15000 checkpoint-16000 checkpoint-17000 checkpoint-18000 checkpoint-19000 checkpoint-2000 checkpoint-20000 checkpoint-21000 checkpoint-22000 checkpoint-23000 checkpoint-24000 checkpoint-25000 checkpoint-26000 checkpoint-27000 checkpoint-28000 checkpoint-29000 checkpoint-3000 checkpoint-30000 checkpoint-31000 checkpoint-32000 checkpoint-33000 checkpoint-34000 checkpoint-35000 checkpoint-36000 checkpoint-37000 checkpoint-38000 checkpoint-39000 checkpoint-4000 checkpoint-40000 checkpoint-41000 checkpoint-42000 checkpoint-43000 checkpoint-44000 checkpoint-45000
e1ac6ff
verified
| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.02773453886128917, | |
| "eval_steps": 1000, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0013867269430644586, | |
| "grad_norm": 1.8933687210083008, | |
| "learning_rate": 2.957486136783734e-06, | |
| "loss": 1.2241, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.002773453886128917, | |
| "grad_norm": 0.7502820491790771, | |
| "learning_rate": 6.038200862600124e-06, | |
| "loss": 1.0267, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.004160180829193376, | |
| "grad_norm": 0.5821689963340759, | |
| "learning_rate": 9.118915588416513e-06, | |
| "loss": 0.8167, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.005546907772257834, | |
| "grad_norm": 0.5138927698135376, | |
| "learning_rate": 1.2199630314232902e-05, | |
| "loss": 0.6408, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.006933634715322293, | |
| "grad_norm": 0.619263768196106, | |
| "learning_rate": 1.5280345040049293e-05, | |
| "loss": 0.5468, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.008320361658386751, | |
| "grad_norm": 0.5078439712524414, | |
| "learning_rate": 1.836105976586568e-05, | |
| "loss": 0.4952, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.00970708860145121, | |
| "grad_norm": 0.5653749108314514, | |
| "learning_rate": 2.144177449168207e-05, | |
| "loss": 0.4388, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.011093815544515669, | |
| "grad_norm": 0.6189213991165161, | |
| "learning_rate": 2.452248921749846e-05, | |
| "loss": 0.4232, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.012480542487580126, | |
| "grad_norm": 0.6082913875579834, | |
| "learning_rate": 2.760320394331485e-05, | |
| "loss": 0.401, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.013867269430644586, | |
| "grad_norm": 0.6956301331520081, | |
| "learning_rate": 3.068391866913124e-05, | |
| "loss": 0.3895, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.015253996373709043, | |
| "grad_norm": 0.7030412554740906, | |
| "learning_rate": 3.3764633394947633e-05, | |
| "loss": 0.3676, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.016640723316773503, | |
| "grad_norm": 0.6779190897941589, | |
| "learning_rate": 3.684534812076402e-05, | |
| "loss": 0.3653, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.01802745025983796, | |
| "grad_norm": 0.8930213451385498, | |
| "learning_rate": 3.992606284658041e-05, | |
| "loss": 0.3645, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.01941417720290242, | |
| "grad_norm": 0.6423994302749634, | |
| "learning_rate": 4.30067775723968e-05, | |
| "loss": 0.3514, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.02080090414596688, | |
| "grad_norm": 0.7728660106658936, | |
| "learning_rate": 4.608749229821319e-05, | |
| "loss": 0.3468, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.022187631089031337, | |
| "grad_norm": 0.7561061978340149, | |
| "learning_rate": 4.916820702402958e-05, | |
| "loss": 0.3499, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.023574358032095795, | |
| "grad_norm": 0.6163890957832336, | |
| "learning_rate": 5.224892174984597e-05, | |
| "loss": 0.3417, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.024961084975160253, | |
| "grad_norm": 0.7334563732147217, | |
| "learning_rate": 5.532963647566236e-05, | |
| "loss": 0.3299, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.026347811918224714, | |
| "grad_norm": 0.655237078666687, | |
| "learning_rate": 5.841035120147874e-05, | |
| "loss": 0.3306, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.02773453886128917, | |
| "grad_norm": 0.8147113919258118, | |
| "learning_rate": 6.149106592729513e-05, | |
| "loss": 0.3281, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02773453886128917, | |
| "eval_loss": 0.32194069027900696, | |
| "eval_runtime": 501.2457, | |
| "eval_samples_per_second": 5.7, | |
| "eval_steps_per_second": 5.7, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 108168, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.802385752064e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |