AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Learning to Interpret Weight Differences in Language Models

Avichal Goel, Yoon Kim, Nir N Shavit, Tony T. Wang

ICLR 2026 · Large language models › Retrieval and knowledge · approach: Knowledge editing in LLMs · 11 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use DIT adapter instead of Base QuestionQwen3-4B · Hidden topic reporting (SEP trigger)similarity score: 1.21 -> 4.76Improved
use DIT adapter instead of 20 QuestionsQwen3-4B · Hidden topic reporting (SEP trigger)similarity score: 1.38 -> 4.76Improved
evaluate DIT on OOD zero-width space trigger at start (ablation)DIT adapter · Hidden topic reportingsimilarity score: 4.76 -> 4.66No clear effect
evaluate DIT on OOD zero-width space trigger at random position (ablation)DIT adapter · Hidden topic reportingsimilarity score: 4.76 -> 4.58No clear effect
add 20 questions probing without DIT on OOD random-position trigger (ablation)20 Questions · Hidden topic reportingsimilarity score: 1.38 -> 2.31Improved
train on topic weight diffs, test on news summary weight diffs (Qwen3-4B) (ablation)Topic Adapter · Cross-behavior generalizationsimilarity score: 4.22 -> 1.25Hurt
train on news weight diffs, test on hidden topic weight diffs (Qwen3-4B) (ablation)News Adapter · Cross-behavior generalizationsimilarity score: 4.76 -> 1.22Hurt
train on topic weight diffs, test on news summary weight diffs (Gemma3-4B) (ablation)Topic Adapter · Cross-behavior generalizationsimilarity score: 4.56 -> 1.08Hurt
train on news weight diffs, test on hidden topic weight diffs (Gemma3-4B) (ablation)News Adapter · Cross-behavior generalizationsimilarity score: 4.82 -> 1.16Hurt
train DIT to predict the SEP trigger instead of hidden topicDIT adapter · Trigger recoverycorrect triggers: 0 out of 100Hurt
train DIT adapter on rank-1 LoRA diffs, test on higher rank and full-parameter diffs (ablation)DIT adapter · Hidden topic reportingsimilarity scoreMixed
increase DIT training data amount (ablation)DIT adapter · Hidden topic reportingsimilarity scoreImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy