Skip to content
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

Add plugin manager's callback hooks to training flow #2006

Merged

Conversation

chiragjn
Copy link
Contributor

@chiragjn chiragjn commented Oct 30, 2024

Description

Thanks a lot for creating the plugins system! It has really helped me remove a lot of monkey patching to add my custom callbacks to the Trainer. Now that it exists, all this PR is doing is invoking the callback adding hooks during the training flow allowing plugins to add their own callbacks

I have also made the plugins attribute a OrderedDict so that a plugin is only registered once no matter how many times .register(...) is called, preventing errors

How has this been tested?

Using it to run my own plugin for in house Experiment Tracking framework (E.g. https://github.com/truefoundry/llm-finetune/blob/cj_axolotl_update_oct2024/plugins/axolotl_truefoundry/axolotl_truefoundry/__init__.py)

Future Work

Add other methods of the plugin manager in the flow

@winglian winglian merged commit d4dbfa0 into axolotl-ai-cloud:main Oct 31, 2024
12 of 13 checks passed
bursteratom pushed a commit that referenced this pull request Nov 18, 2024
* Add plugin manager's callback hooks to training flow

* Use .values() instead of .items()
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants