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
Is your feature request related to a problem? Please describe.
Training code should be refactored.
Describe the solution you'd like
Additional context
I read the code carefully, but PytorchLightning has no easy way to specify non-zero initial epochs and global_steps. (Admittedly, such a key exists at model save time, but it is not used.) It is very difficult to resume training from an existing model structure. However, global_steps and epochs are used everywhere, the only way may be to specify it with dirty code.
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
Training code should be refactored.
Describe the solution you'd like
Additional context
I read the code carefully, but PytorchLightning has no easy way to specify non-zero initial epochs and global_steps. (Admittedly, such a key exists at model save time, but it is not used.) It is very difficult to resume training from an existing model structure. However, global_steps and epochs are used everywhere, the only way may be to specify it with dirty code.
The text was updated successfully, but these errors were encountered: