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689 results for “keyword:"mechanistic interpretability"” in ICLR 2027. 142 withdrawn or desk rejected listed apart.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]Where Features Are Written and Read: Localizing Native Transformer Channels for Residual Directions▼4.3k2.00
[PDF]Reliable Neuron Identification: A Statistical Learning Framework for Neuron Explanations▲3952.67
[PDF]Can We Read the Mind of an Audio LLM? A Verbalizable, Multilingual Middle-Layer Workspace▲432.75
[PDF]Committed by the First Token: Measuring Recovery from Induced Answer Commitment in Chain-of-Thought Reasoning▼4.3k2.00
[PDF]Fork Where the Model Changes Its Mind: Belief-Shift Branching for Tree-Structured Reinforcement Learning▼1.2k3.00
[PDF]Not All Thinking is Created Equal: Latent Reasoning Discovers a Recurrent Search Algorithm for Depth Generalization▲6.1k2.00
[PDF]ORION: EXACT CANONICAL-STATE SUPERVISION FOR SYSTEMATIC GENERALIZATION IN TRANSFORMERS▼1.2k1.67
[PDF]More Than Correct Answers: Testing, Tracing, and Predicting Multi-Hop Execution in Transformers▲6.1k2.00
[PDF]Shared Weights, Selected Computations: How Looped Transformers Route What Each Loop Does▼4.3k2.00
[PDF]SAE-Locked LoRA: Edit-Specific Sparse-Feature Gradient Routing for Low-Interference Knowledge Editing▼4.5k2.67
[PDF]WorldModelLens: From Attention to Rollouts - A framework for World Model Interpretability▼4.3k2.00
[PDF]What the Nonnegative Lens Misses: Energy-Matched Causal Tests of the J-Space in Open Language Models▲11.00
[PDF]MoE Routing Is Not Token-Local: How Expert Influence Propagates Across Tokens & Layers▼4.5k2.67
[PDF]Written as a Record, Read as an Address: What a Forward Pass Leaves in an Operation's KV Cache▲3.0k2.33
[PDF]QuTab: A Single Mutable-State Transformer for Mechanistically Interpretable Tabular In-Context Learning▼1.2k3.00
[PDF]From Input to Output: A Flexible Agent for Dual-End Interpretation of Sparse Autoencoder Features▲9.2k2.67
Withdrawn or desk rejected (142)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