Interpretability

Steering Safely or Off a Cliff? Rethinking Specificity and Robustness in Inference-Time Interventions EACL 2026 Stop Probing, Start Coding: Why Linear Probes and Sparse Autoencoders Fail at Compositional Generalization UAI 2026 Structural Graph Probing of Vision-Language Models CVPR 2026 SurgXBench: Explainable Vision-Language Model Benchmark for Surgery WACV 2026 Sycophancy Hides Linearly in the Attention Heads EACL 2026 Sycophantic Anchors: Localizing and Quantifying User Agreement in Reasoning Models ACL 2026 Synthesizing Visual Concepts as Vision-Language Programs CVPR 2026 Taming Object Hallucinations with Verified Atomic Confidence Estimation EACL 2026 Taming the Phantom: Token-Asymmetric Filtering for Hallucination Mitigation in Large Vision-Language Models AAAI 2026 Tandem Training for Language Models EACL 2026 Targeted Pathway Inference for Biological Knowledge Bases via Graph Learning and Explanation AAAI 2026 Targeted Syntactic Evaluation of Language Models on Georgian Case Alignment EACL 2026 Taxation Perspectives from Large Language Models: A Case Study on Additional Tax Penalties EACL 2026 TaxonRL: Reinforcement Learning with Intermediate Rewards for Interpretable Fine-Grained Visual Reasoning WACV 2026 Teaching and Critiquing Conceptualization and Operationalization in NLP EACL 2026 TensorLens: End-to-End Transformer Analysis via High-Order Attention Tensors ACL 2026 Test of Time: Rethinking Temporal Signal of Benchmark Contamination ACL 2026 Test-Time Adaptation through Semantically-guided Feature Decomposition for Few-shot Chest X-ray Diagnosis WACV 2026 Test Time Adaptation Using Adaptive Quantile Recalibration WACV 2026 Test-time Prompt Intervention AAAI 2026 Textual Steering Vectors Can Improve Visual Understanding in Multimodal Large Language Models ACL 2026 The Art of Saying "Maybe": A Conformal Lens for Uncertainty Benchmarking in VLMs EACL 2026 The Curse of Verbalization: How Presentation Order Constrains LLM Reasoning EACL 2026 The Digital Dunning-Kruger Effect: Decoupling Hallucinations via Geometric Hidden-state Observation for Semantic Truthfulness ACL 2026 The Dog the Cat Chased Stumped the Model: Measuring When Language Models Abandon Structure for Shortcuts EACL 2026