All issues

Search results

Question about internal signal localization

Status: Open.

#213 In ai-safety-foundation/sparse_autoencoder;

· luolearning opened on Apr 5, 2026

demo code running uses up system RAM in colab

Status: Open.

#212 In ai-safety-foundation/sparse_autoencoder;

· Tekikun opened on Apr 3, 2025

ImportError: cannot import name 'params_t' from 'torch.optim.optimizer'

Status: Open.

#210 In ai-safety-foundation/sparse_autoencoder;

· seansica opened on Jul 7, 2024

demo pre encoder bias device error

Status: Open.

#209 In ai-safety-foundation/sparse_autoencoder;

· qwenzo opened on Apr 6, 2024

[W/ Potential Severe Consequences] Running sweep() on different directory structure results in bugs.

Status: Open.

#208 In ai-safety-foundation/sparse_autoencoder;

· HuFY-dev opened on Mar 30, 2024

New torch version breaks optimizer code

Status: Open.

#202 In ai-safety-foundation/sparse_autoencoder;

· chefexperte opened on Feb 23, 2024

Training takes too long

Status: Open.

#198 In ai-safety-foundation/sparse_autoencoder;

· BiEchi opened on Feb 1, 2024

sum(latents) vs mean(hidden)

Status: Open.

#184 In ai-safety-foundation/sparse_autoencoder;

· wassname opened on Jan 11, 2024

Perhaps consider replace geometric_median_dataset with nn.BatchNorm1d(x, affine=False)

Status: Open.

#179 In ai-safety-foundation/sparse_autoencoder;

· wassname opened on Jan 6, 2024

Add automated interpretability metrics

enhancementNew feature or requestNew feature or request

help wantedExtra attention is neededExtra attention is needed

Status: Open.

#104 In ai-safety-foundation/sparse_autoencoder;

· alan-cooney opened on Nov 23, 2023

Log feature density histograms to weights and biases

Status: Open.

#88 In ai-safety-foundation/sparse_autoencoder;

· jbloomAus opened on Nov 21, 2023