/usr/local/python3.10.17/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py:632: UserWarning: torch.utils.checkpoint: the use_reentrant parameter should be passed explicitly. In version 2.5 we will raise an exception if use_reentrant is not passed. use_reentrant=False is recommended, but if you need to preserve the current default behavior, you can pass use_reentrant=True. Refer to docs for more details on the differences between the two variants.
return fn(*args, **kwargs)
{'loss': 1.955, 'grad_norm': 0.41192638874053955, 'learning_rate': 0.0002, 'num_tokens': 12374.0, 'mean_token_accuracy': 0.5659834313392639, 'epoch': 0.0}
{'loss': 1.6544, 'grad_norm': 0.9751525521278381, 'learning_rate': 0.0002, 'num_tokens': 16766.0, 'mean_token_accuracy': 0.6307526516914368, 'epoch': 0.0}
{'loss': 1.8638, 'grad_norm': 0.3490130603313446, 'learning_rate': 0.0002, 'num_tokens': 27174.0, 'mean_token_accuracy': 0.5735858237743378, 'epoch': 0.01}
{'loss': 1.7149, 'grad_norm': 0.6998162269592285, 'learning_rate': 0.0002, 'num_tokens': 31047.0, 'mean_token_accuracy': 0.6218746590614319, 'epoch': 0.01}
1%|█▉ | 100/13001 [01:13<2:25:55, 1.47it/s]/usr/local/python3.10.17/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py:632: UserWarning: torch.utils.checkpoint: the use_reentrant parameter should be passed explicitly. In version 2.5 we will raise an exception if use_reentrant is not passed. use_reentrant=False is recommended, but if you need to preserve the current default behavior, you can pass use_reentrant=True. Refer to docs for more details on the differences between the two variants.
return fn(*args, **kwargs)
{'loss': 1.7628, 'grad_norm': 0.29718583822250366, 'learning_rate': 0.0002, 'num_tokens': 42117.0, 'mean_token_accuracy': 0.5874369502067566, 'epoch': 0.01}
{'loss': 1.6219, 'grad_norm': 0.5728892087936401, 'learning_rate': 0.0002, 'num_tokens': 46196.0, 'mean_token_accuracy': 0.6381562113761902, 'epoch': 0.01}
{'loss': 1.8255, 'grad_norm': 0.31880176067352295, 'learning_rate': 0.0002, 'num_tokens': 58459.0, 'mean_token_accuracy': 0.5713030004501343, 'epoch': 0.01}
{'loss': 1.5681, 'grad_norm': 0.8215921521186829, 'learning_rate': 0.0002, 'num_tokens': 62713.0, 'mean_token_accuracy': 0.6404920220375061, 'epoch': 0.02}
2%|███▊ | 200/13001 [02:25<2:22:47, 1.49it/s]/usr/local/python3.10.17/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py:632: UserWarning: torch.utils.checkpoint: the use_reentrant parameter should be passed explicitly. In version 2.5 we will raise an exception if use_reentrant is not passed. use_reentrant=False is recommended, but if you need to preserve the current default behavior, you can pass use_reentrant=True. Refer to docs for more details on the differences between the two variants.
return fn(*args, **kwargs)
{'loss': 1.8265, 'grad_norm': 0.31884294748306274, 'learning_rate': 0.0002, 'num_tokens': 74551.0, 'mean_token_accuracy': 0.5766114640235901, 'epoch': 0.02}
{'loss': 1.6236, 'grad_norm': 0.6962191462516785, 'learning_rate': 0.0002, 'num_tokens': 78901.0, 'mean_token_accuracy': 0.6332866501808166, 'epoch': 0.02}
{'loss': 1.7901, 'grad_norm': 0.31407737731933594, 'learning_rate': 0.0002, 'num_tokens': 91128.0, 'mean_token_accuracy': 0.5836749339103698, 'epoch': 0.02}
{'loss': 1.5906, 'grad_norm': 0.5636782646179199, 'learning_rate': 0.0002, 'num_tokens': 95364.0, 'mean_token_accuracy': 0.637250554561615, 'epoch': 0.02}
2%|█████▋ | 300/13001 [03:45<2:17:26, 1.54it/s]/usr/local/python3.10.17/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py:632: UserWarning: torch.utils.checkpoint: the use_reentrant parameter should be passed explicitly. In version 2.5 we will raise an exception if use_reentrant is not passed. use_reentrant=False is recommended, but if you need to preserve the current default behavior, you can pass use_reentrant=True. Refer to docs for more details on the differences between the two variants.
return fn(*args, **kwargs)
这个日志输出是什么意思???
最新发布