🧑‍🔬 Tom Lieberum

Affiliations

• Google DeepMind (United Kingdom) · London, United Kingdom — current · 2024 on 1 paper

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine.

Topics

Interpretability2

Works by year

2020s

Papers

The most cited first (OpenAlex and our library).

1. Progress measures for grokking via mechanistic interpretability (2023 · 207 citations)
   Deep networks sometimes gain new abilities abruptly, which makes their behavior hard to anticipate. This work asks whether such jumps hide smooth underlying progress, and whether interpretability can reveal it. The test case is grokking: small transformers learning addition modulo a prime first memorize the training…
2. Does Circuit Analysis Interpretability Scale? Evidence from Multiple Choice Capabilities in Chinchilla (2023 · 51 citations)
   Circuit analysis has mostly been demonstrated on small networks, such as two-layer attention-only transformers or the 117M GPT-2.
3. Improving Dictionary Learning with Gated Sparse Autoencoders (2024 · 39 citations)
   Sparse autoencoders (SAEs) are used to break language model activations into sparse combinations of interpretable directions.

Also in our library

3 more works of theirs are in our library without a page here yet; they will appear as we publish them.

Co-authors here

6 co-authors with a page, the most papers together first.

• Neel Nanda ×3
• Janos Kramar ×2
• Rohin Shah ×2
• Geoffrey Irving
• Jacob Steinhardt
• Vladimir Mikulik

Cited by

Papers of our library cite their works 394 times (of the 140 884 papers whose bibliographies we read); 14 citing papers with a page here →.

This page in Markdown → https://md.research.elamur.ai/free/en/authors/tom-lieberum/