You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Not sure if this is intended, but the model checkpoint isn't using the same directory as the logger, even if the logger exists. I would have expected this line here to be self.logger.save_dir instead of self.default_save_path.
Thank you,
-Collin
The text was updated successfully, but these errors were encountered:
I am running into the same issue and feel like I must be doing something wrong. If I do not define a logger and checkpoint callback, the logger and checkpoint are saved in a clear structure under lightning_logs. But if I do provide them, it turns out to be difficult to get Lightning to follow that same structure. If I use lightning_logs as filepath I get the warning that that directory already exists and that my existing checkpoints will be overwritten.
The checkpoints are currently being saved to a subfolder of the current working directory. If you print the current working directory at the top of our file with print(os.getcwd()) that should tell you where to find them.
FWIW, using the default checkpoint_callback, the checkpoint is not in default location. Still looking for it! Though no more default checkpoint_callback for me ...
🐛 Bug
Not sure if this is intended, but the model checkpoint isn't using the same directory as the logger, even if the logger exists. I would have expected this line here to be
self.logger.save_dir
instead ofself.default_save_path
.Thank you,
-Collin
The text was updated successfully, but these errors were encountered: