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689 results for “keyword:"mechanistic interpretability"” in ICLR 2027. 142 withdrawn or desk rejected listed apart.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]The Right Source Is Not Enough: Understanding History Interference in Speech-LLM Memory▼1.2k1.67
[PDF]ECHO: Emergent Common-Mode Hidden Offsets Drive Vocabulary Collapse in Preference Optimization▲3.0k2.33
[PDF]KV Compression Cannot Double as a Jailbreak Defense: The Harmfulness-then-Refusal Trajectory Guides Token Deletion▼5.8k2.33
[PDF]Does the Workspace Do the Work? A Causal Protocol for Verifying Global-Workspace Computation in a Language Model▲5103.00
[PDF]Don't Judge an LLM Only by Its Activations: Discovering Suppressed Safety Features via Counterfactual Activation Potential▼4.3k2.00
[PDF]Attention Sinks as Knowledge-Conflict Arbiters: Training-Free Gating-Preserving Routing for Contextual Faithfulness▲3952.67
[PDF]The Signal Survives, the Threshold Slips: Small LLM Agents After Quantization and Fine-Tuning▼1.2k1.67
[PDF]Same Task, Different Answer: Persona Conditioning Modulates Shared Task Mechanisms in Language Models▼4.5k2.67
[PDF]Where Do LLMs Decide to Break the Rules? Mechanistic Localization of Prompt Injection Compliance▲5103.00
[PDF]When Does a Probe Identify a Representation? Semantic Counterfactual Validation in Language Models▼4.3k2.00
[PDF]MoE Mirage: Selection Noise and Routing-Weight Geometry in Token-Level Counterfactuals▼1.2k3.00
[PDF]Identifying Token and Task-Relevant Representational Subspaces in Large Language Models▼4.3k2.00
[PDF]What Behavior Cloning Gives and Takes: Action Learning and Visual Reorganization in VLAs▼1361.33
[PDF]The Verbalization Gap: Stress-Testing Activation Monitoring with Natural Language Autoencoders▼1361.33
[PDF]Interpreting and Steering Trajectory Entrainment Behavior in Goal-Oriented Conversations▼1.2k1.67
[PDF]Safety Fragility Is a Routing Problem: The Attention Bottleneck Between Harm Detection and Refusal▲1.9k3.67
Withdrawn or desk rejected (142)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