Stable Kullback-Leibler in evals.py #484
Closed
Conversation
This file contains hidden or 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
* fix: fix abnormally large llamascope L0. (jbloomAus#483) * make llamascope (base model) compatible with sae.fold_activation_norm_scaling_factor * make llamascope (base model) compatible with sae.fold_activation_norm_scaling_factor * 5.10.3 Automatically generated by python-semantic-release * chore: Remove unused files (jbloomAus#481) * removes make_hf_repo.sh * removes eval_metrics_resid_mid_oai.csv * removes check_open_ai_sae_metrics.ipynb * removes content dir * removes unused files in docs dir * wip * yes * renaming into singular_fisher * fmt * removing breakpoint * generalisation * removing debug log * Update training_sae.py --------- Co-authored-by: Zhengfu He <74953051+Hzfinfdu@users.noreply.github.com> Co-authored-by: github-actions <github-actions@github.com> Co-authored-by: Anthony Duong <42191920+anthonyduong9@users.noreply.github.com>
|
@rufimelo99 Hey there! Thanks for the PR. It looks like there are other changes in this (a PR for fisherSAE) - did you intend for that to be in this change? |
5 tasks
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Outdated suggestions cannot be applied.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.
Description
Hey ✋
I've been starting to look into the library and dive into it. I'm currently looking into the evaluation metrics, and I've noticed the KL divergence is being calculated manually, which sometimes leads to "nans" and "infs".
This ends up being a simple PR.
Fixes # (issue)
Type of change
Please delete options that are not relevant.
Checklist:
You have tested formatting, typing and tests
make check-cito check format and linting. (you can runmake formatto format code if needed.)Please links to wandb dashboards with a control and test group.