-
Notifications
You must be signed in to change notification settings - Fork 1.5k
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
[DPO] Refactor eval logging of dpo trainer #954
Conversation
Thanks for working on this! Just ping me whenever you would like to run the CI. |
@lvwerra, I completed this part of the refactor, can you run the CI? @kashif as part of this, I removed logging all the |
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. |
@mnoukhov are you also logging sample outputs as well here? |
ok makes sense! |
so this PR LGTM! feed free to merge |
@mnoukhov so folks really want the additional metrics to be logged during training which isn't possible at the moment, so lets revert this for now and continue this refactoring to figure that out. |
I don't think this change prevents additional metrics from being logged. I can create a followup PR that addresses those comments |
Ok so let's try to add the previous behavior if possible... I couldn't figure it out at first glance... I think the trainer's training loop doesn't seem to have a way to do it... can you kindly confirm? |
Basically what #1046 wants is to log metrics during training, which is not how AFAIK the problem with trying to put If you want metrics for the training set, we can either add keep this |
* first attempts at refactor of dpo trainer * removed extra stuff in prediction step * import fixes * label names * all working --------- Co-authored-by: Leandro von Werra <[email protected]>
Address part 1 of #952 by removing
store_metrics
and switching to more standardcompute_metrics
lettingTrainer
handle logging