Training doesn't start. There are no errors.

[seoeaa]

Training doesn't start. There are no errors.

[INFO|training_args.py:1838] 2024-01-21 06:21:04,624 >> PyTorch: setting up devices [INFO|trainer.py:568] 2024-01-21 06:21:04,630 >> Using cpu_amp half precision backend [INFO|trainer.py:1706] 2024-01-21 06:21:04,775 >> ***** Running training ***** [INFO|trainer.py:1707] 2024-01-21 06:21:04,775 >> Num examples = 1,000 [INFO|trainer.py:1708] 2024-01-21 06:21:04,775 >> Num Epochs = 3 [INFO|trainer.py:1709] 2024-01-21 06:21:04,775 >> Instantaneous batch size per device = 4 [INFO|trainer.py:1712] 2024-01-21 06:21:04,775 >> Total train batch size (w. parallel, distributed & accumulation) = 16 [INFO|trainer.py:1713] 2024-01-21 06:21:04,775 >> Gradient Accumulation steps = 4 [INFO|trainer.py:1714] 2024-01-21 06:21:04,775 >> Total optimization steps = 186 [INFO|trainer.py:1715] 2024-01-21 06:21:04,776 >> Number of trainable parameters = 4,194,304 D:\Programm\AI\LLaMA-Factory\myenv\lib\site-packages\torch\utils\checkpoint.py:429: UserWarning: torch.utils.checkpoint: please pass in use_reentrant=True or use_reentrant=False explicitly. The default value of use_reentrant will be updated to be False in the future. To maintain current behavior, pass use_reentrant=True. It is recommended that you use use_reentrant=False. Refer to docs for more details on the differences between the two variants. warnings.warn(

[hiyouga]

Using cpu_amp half precision backend

It seems that the CUDA environment was not configured properly