auto_select_gpus is deprecated in Pytorch Lightning 2.0.2
Closed this issue · 1 comments
First off, thanks for the great contribution to the community!
I am getting the following error when running the training code:
Error executing job with overrides: []
Traceback (most recent call last):
File "D:\User\TEMOS\train.py", line 12, in _train
return train(cfg)
File "D:\User\TEMOS\train.py", line 66, in train
trainer = pl.Trainer(
File "D:\User\Anaconda37\envs\temos\lib\site-packages\pytorch_lightning\utilities\argparse.py", line 69, in insert_env_defaults
return fn(self, **kwargs)
TypeError: __init__() got an unexpected keyword argument 'auto_select_gpus'
Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.
It seems like the auto_select_gpus
argument is now deprecated in the latest version of Pytorch Lightning. See discussion here: Lightning-AI/pytorch-lightning#16147
Therefore it shouldn't be need on this line for the latest Pytorch Lightning:
TEMOS/configs/trainer/gpu.yaml
Line 1 in aedf871
EDIT: I verified that this isn't an issue in version 1.9.5
I am happy to make a PR if need be, otherwise a frozen requirements.txt that is compatible with the state of the code as it is would be also really helpful.
Thanks for you help!
Hello @simnyatsanga,
Thanks for your issue! I removed the "auto_select_gpus" from the configs.