Graphs follow the wrong number (steps instead of epochs) #234
Labels
No Label
bug
duplicate
enhancement
help wanted
insufficient info
invalid
news
not a bug
question
wontfix
No Milestone
No project
No Assignees
2 Participants
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: mrq/ai-voice-cloning#234
Loading…
Reference in New Issue
Block a user
No description provided.
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
As you can see in the file, I'm doing a training of 500 epochs, but each epoch is like 18 iterations so its 9000 steps in total.
The graph is showing the current number of steps (133), but the graph axis stops at 500 (the number of epochs) instead of 9000.
Meaning if the training ends when the graph gets to the end, it would only have done 500 steps, instead of 500 epochs..(?)
The backend is dropping the schedule at the correct times it seems (by epochs) so it seems it's just the graphs are wrong, but I think when I've done this before it stops at the '500 steps' mark, even though thats only 500 out of 9000. So something thinks the steps are epochs.
Oops, somehow I had that change make its way upstream. For my VALL-E training, I needed it to show by iteration step rather than epoch count, and my local change on my training machine somehow made its way to the repo.
Should be fixed in commit
74bd0f0cdc
, but you can set yourX Max
to 9000 and it should re-scale. I think a default of 0 will check your training settings and set itself to500
, thinking it's 500 epochs and not 9000 steps.No problem, I did think it seemed weird!
How's the vall-e experience going? Is it proving to be better?
edit: I updated and now I'm getting -
AttributeError: module 'numpy' has no attribute 'complex'.
np.complex
was a deprecated alias for the builtincomplex
. To avoid this error in existing code, usecomplex
by itself. Doing this will not modify any behavior and is safe. If you specifically wanted the numpy scalar type, usenp.complex128
here.The aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:
https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations. Did you mean: 'complex_'?
Press any key to continue . . .
You'll need to do something like (after activating the venv):
I could have sworn I had the version frozen for this repo, but I can't for the life of me find which repo ended up having it.