[KTO] prevent nans from appearing in metrics #1386
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…e batch_size losses
add reference to paper Co-authored-by: lewtun <lewis.c.tunstall@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
kashif
approved these changes
Mar 1, 2024
kashif
added a commit
to fe1ixxu/trl
that referenced
this pull request
Mar 15, 2024
* add warning for imbalanced data * update documentation * update script commands to be same as in dpo * use batch_size KL examples and batch_size target examples to calculate batch_size losses * fix deepspeed issue * speed up forward with no_grad for KL * add some removed metrics * Update trl/trainer/kto_trainer.py * Update trl/trainer/kto_trainer.py * Update trl/trainer/kto_trainer.py add reference to paper Co-authored-by: lewtun <lewis.c.tunstall@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * add more detailed comments * convert assert to ValueError * Update kto_trainer.py * precommit formatting * remove nans in metrics by gathering across machines * fix formatting --------- Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> Co-authored-by: lewtun <lewis.c.tunstall@gmail.com>
lapp0
pushed a commit
to lapp0/trl
that referenced
this pull request
May 10, 2024
* add warning for imbalanced data * update documentation * update script commands to be same as in dpo * use batch_size KL examples and batch_size target examples to calculate batch_size losses * fix deepspeed issue * speed up forward with no_grad for KL * add some removed metrics * Update trl/trainer/kto_trainer.py * Update trl/trainer/kto_trainer.py * Update trl/trainer/kto_trainer.py add reference to paper Co-authored-by: lewtun <lewis.c.tunstall@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * Update trl/trainer/kto_trainer.py Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> * add more detailed comments * convert assert to ValueError * Update kto_trainer.py * precommit formatting * remove nans in metrics by gathering across machines * fix formatting --------- Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com> Co-authored-by: lewtun <lewis.c.tunstall@gmail.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR addresses the issue of NaNs appearing in KTO metric logs (even when training works). See here: #1342 (comment)
This was fixed by doing an all_gather over lists of [nan] instead of empty lists when a particular metric has no entries for a given microbatch (for some reason, accelerate, unlike pytorch, cannot all_gather over empty tensors).
cc @kashif