-
Notifications
You must be signed in to change notification settings - Fork 3.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[3/3] Update lightning callbacks to Stateful
, deprecations for old on_save/load_checkpoint
signatures
#11887
[3/3] Update lightning callbacks to Stateful
, deprecations for old on_save/load_checkpoint
signatures
#11887
Conversation
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Once you have a proof of concept working, I'd strongly recommend splitting this into a few different PRs:
- Add state_dict/load_state_dict to the base Callback
- Add support for this inside of the checkpoint connector/trainer
- Change relevant callback implementations to use state_dict/load_state_dict instead
This will be much easier to review for merging
Stateful
CallbacksStateful
Callbacks
for more information, see https://pre-commit.ci
edit: was missing
|
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
Head branch was pushed to by a user without write access
What does this PR do?
Note this currently includes the changes in part [2/n] (PR #12232) until that can be merged - review here for parts 3 and 4 only.
part of / Fixes #11429: introduce Stateful Callbacks (see Appendix code changes outlined in issue)
This part:
Stateful
on_save/load_checkpoint
signaturesDoes your PR introduce any breaking changes? If yes, please list them.
see CHANGELOG "changed" section: removed lightning
Callback.on_save/load_checkpoint
methods in favor ofstate_dict/load_state_dict
for the callbacks listed under Labels: early stopping, finetuning, model checkpoint, pruning, timer.Before submitting
PR review
Anyone in the community is welcome to review the PR.
Before you start reviewing make sure you have read Review guidelines. In short, see the following bullet-list:
Did you have fun?
Make sure you had fun coding 🙃