AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Mixing Mechanisms: How Language Models Retrieve Bound Entities In-Context

Yoav Gur-Arieh, Mor Geva, Atticus Geiger

ICLR 2026 · Large language models › Retrieval and knowledge · approach: Retrieval-augmented reasoning · 19 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
model positional mechanism as Gaussian instead of one-hot (ablation)causal model M · Music task, n=20Jensen-Shannon similarity: 0.85 -> 0.95Improved
use full three-mechanism model instead of positional-only prevailing viewentity retrieval model · Music task, n=20Jensen-Shannon similarity: 0.44 -> 0.95Improved
ablate lexical term from M when target is first in group (te=1) (ablation)causal model M · Music task, te=1Jensen-Shannon similarity: nearly no effectNo clear effect
ablate reflexive term from M when target is last in group (te=3) (ablation)causal model M · Music task, te=3Jensen-Shannon similarity: nearly no effectNo clear effect
block attention from last token to query entity token in layers 11-16 (ablation)gemma-2-2b-it · Binding retrieval accuracyaccuracy: 98% -> 37%Hurt
also block attention to the token following the query token (ablation)gemma-2-2b-it · Binding retrieval accuracyaccuracy: 37% -> 8%Hurt
block last position to query token and its following token's attention (ablation)gemma-2-2b-it · Binding retrieval accuracyaccuracy: accuracy drops to 6%Hurt
block query token attention to matching entity token (reflexive signal) (ablation)gemma-2-2b-it · Reflexive, te=1accuracy: reducing accuracy to 6%Hurt
block query token attention to answer entity token (lexical signal) (ablation)gemma-2-2b-it · Lexical, te=2accuracy: reduces accuracy only to 86%No clear effect
block query token attention to all entity tokens (lexical) (ablation)gemma-2-2b-it · Lexical, te=2accuracy: accuracy at 90%No clear effect
add filler-text padding between entity groups (up to 500 tokens)gemma-2-2b-it · Boxes taskmechanism effect shareMixed
increase number of entity groups npositional mechanism · Middle entity groups, n in [3,20]positional efficacyHurt

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