fix: update mdl sae for new SAELens api#94
Merged
Conversation
sae-lens 6.28.0 renamed ActivationsStore.get_filtered_buffer(n_batches) to get_filtered_llm_batch() (single-batch). Bumps the floor pin and adds a small _get_filtered_buffer helper that loops the new API to preserve the original sample size at the three mdl call sites. The dropped in-place shuffle is harmless: all callers compute order-invariant aggregates (histograms, MSE, min/max). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Covers build_bins (num_bins, bin_precision, the min_pos-zero quirk, arg validation), quantize_features_to_bin_midpoints (correctness + out-of-range clamping), IdentityAE, and an integration test for the new _get_filtered_buffer helper that uses a real ActivationsStore on gpt2 + the gpt2-small-res-jb SAE (already-cached fixtures from tests/conftest.py) and asserts the exact concatenated shape. Suite runs in ~13s on cpu. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 fixes a bug in the MDL eval due to a refactor to the SAELens ActivationStore from Jan 2026 (decoderesearch/SAELens#614).