NeurIPS 2026 reviews are up: reviews, scores, rebuttals and meta-reviews for 7,877 papers. See the highest rated
acceptodds: the papers of ICLR 2027
689 results for “keyword:"mechanistic interpretability"” in ICLR 2027. 142 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]Utility-Preserving Relabeling Reveals Semantic Sensitivity in Language-Model Decision Making▼1.2k1.67
[PDF]TRACE: Causal Tracing and Editing Visual Evidence Retrieval to Mitigate Vision-Language Model Hallucinations▼13k2.00
[PDF]A Linear Name-Origin Prior in the Recall Heads: How Language Models Fabricate and Misrecall Facts About People▲5103.00
[PDF]Which State Should Prediction Read? A Mechanistic Analysis of State-Prediction Separation▲5.4k3.00
[PDF]Right on Average, Wrong per Respondent: A Mechanistic Diagnosis of LLM-Simulated Choice▼1.2k1.67
[PDF]Bridging Singular Learning Theory and Mechanistic Interpretability at the Billion-Parameter Scale▼1.2k1.67
[PDF]TS-Edit: A Model Editing Framework for Adversarial Defense in Time-Series Foundation Models▼1361.33
[PDF]Evidence States Survive Where Answers Fail: Decodability and Causal Access in Language Models▼4.3k2.00
[PDF]Anchored but Not Protected: When Refusal Erodes During Fine-Tuning, and Whether Holding One Direction Stops It▲13k2.33
[PDF]Spectral Signatures And Emergent Equivariance In Looped Transformers For In-Context Dynamical-System Prediction▼4.3k2.00
[PDF]Deterministic Addresses Constrain Reads, Not Ordinary Writes: Deletion Boundaries in Conditional-Memory Language Models▲6.1k2.00
[PDF]The Circuit Basis of Metacognitive Failure in Language Models: Conflict and Hallucination▲9.2k2.67
[PDF]INTERVENTION IS NOT IDENTIFICATION: ACTIVATION EDITS CAN PRODUCE THE PREDICTED EFFECT FOR THE WRONG REASON▼1361.33
[PDF]Stable MLP Pools with Task-Dependent Communication Geometry in Context-Memory Arbitration▼4.5k2.25
[PDF]Path-Space: Estimating Transformer Component Effects and Their Interactions from Shared Backward Passes▼5.8k2.33
[PDF]Out of Sight, Still in Mind? Separating Internal and Answer-Level Effects of Retained Key-Value History▼1.2k1.67
Withdrawn or desk rejected (142)
[PDF]Why Benign Post-Training Induces Safety Gate Collapse: Changes in Harmful Activation Geometry Relative to the Inherited Refusal GateWithdrawn1.67
[PDF]Concept Subspaces Compute Beyond the Logit Lens: A Weights-Only Test for Locating Representations Upstream of ReadoutWithdrawn