Fetching the paper…

Reading the bibliography…

2024

BatchTopK Sparse Autoencoders

Bussmann, Bart, Leask, Patrick, Nanda, Neel

Original
PDF
Post on X

Understand

Sparse autoencoders (SAEs) have emerged as a powerful tool for interpreting language model activations by decomposing them into sparse, interpretable features.

A popular approach is the TopK SAE, that uses a fixed number of the most active latents per sample to reconstruct the model activations.
We introduce BatchTopK SAEs, a training method that improves upon TopK SAEs by relaxing the top-k constraint to the batch-level, allowing for a variable number of latents to be active per sample.
As a result, BatchTopK adaptively allocates more or fewer latents depending on the sample, improving reconstruction without sacrificing average sparsity.

BatchTopK Sparse Autoencoders · Around

Built on

k-sparse autoencoders, 2014
Alireza Makhzani and Brendan Frey · 2014
Earlier work this paper cites.
Language models are unsupervised multitask learners
Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al · 2019
Earlier work this paper cites.
Towards monosemanticity: Decomposing language models with dictionary learning
Trenton Bricken, Adly Templeton, Joshua Batson, Brian Chen, Adam Jermyn, Tom Conerly, Nick Turner, Cem Anil, Carson Denison, Amanda Askell, et al · 2023
Earlier work this paper cites.

Similar

Sparse autoencoders find highly interpretable features in language models
Original
Hoagy Cunningham, Aidan Ewart, Logan Riggs, Robert Huben, and Lee Sharkey · 2023
Cited alongside, same era.
Scaling and evaluating sparse autoencoders, 2024
Leo Gao, Tom Dupré la Tour, Henk Tillman, Gabriel Goh, Rajan Troll, Alec Radford, Ilya Sutskever, Jan Leike, and Jeffrey Wu · 2024
Cited alongside, same era.
Improving dictionary learning with gated sparse autoencoders, 2024
Senthooran Rajamanoharan, Arthur Conmy, Lewis Smith, Tom Lieberum, Vikrant Varma, János Kramár, Rohin Shah, and Neel Nanda · 2024
Cited alongside, same era.

Then

Jumping ahead: Improving reconstruction fidelity with jumprelu sparse autoencoders, 2024
Senthooran Rajamanoharan, Tom Lieberum, Nicolas Sonnerat, Arthur Conmy, Vikrant Varma, János Kramár, and Neel Nanda · 2024
Closest in time.
Gemma 2: Improving open language models at a practical size
Original
Gemma Team, Morgane Riviere, Shreya Pathak, Pier Giuseppe Sessa, Cassidy Hardin, Surya Bhupatiraju, Léonard Hussenot, Thomas Mesnard, Bobak Shahriari, Alexandre Ramé, et al · 2024
Closest in time.
Scaling monosemanticity: Extracting interpretable features from claude 3 sonnet
Adly Templeton, Tom Conerly, and et al · 2024
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…