AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Shared Lexical Task Representations Explain Behavioral Variability In LLMs

Zhuonan Yang, Xiaochen Li, Francisco Piedrahita Velez, Eric Todd, David Bau, Michael Littman, Stephen Bach, Ellie Pavlick

ICML 2026 · Large language models › Language, safety and interpretability · approach: Task vectors and steering · 0 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
patch averaged lexical task head activations (scaled up) into zero-shot prompts (ablation)lexical task heads · 17 simple tasks, Llama-3.1-8B-Instructtask accuracy: recovers at least 10% to 90% of task performance across promptsImproved
patch activations from a related control task instead of the target task (ablation)lexical task heads · zero-shot prompts, Llama-3.1-8B-Instructtask accuracyNo clear effect
scale up average lexical task head activation from correct prompts in incorrect prompts (ablation)lexical task heads · incorrect prompts (baseline accuracy 0)accuracyImproved
strengthen lexical task head output (ablation)retrieval heads · country-capital style tasksretrieval heads retrieving correct answerImproved
increase number of in-context shotsexample-based prompts · 17 taskslexical task head count and output norm: 14 out of 17 tasks positively correlated with accuracyImproved
add misleading ambiguous demonstration pair to example promptexample-based prompts · Product-Producer tasktarget task head activationHurt
activate lexical task heads of both composing tasks (ablation)two-hop compositional tasks · compositional tasksperformance recovered: 12%-50%Improved
activate Python lexical task heads with a too strong scaling factor (ablation)code steering · HumanEval-X JavaScript promptsgeneration qualityHurt

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