Logit-Contribution Scoring Identifies Non-Literal Retrieval Heads
In long-context use, large language models frequently synthesize answers from the meaning of a relevant context span rather than literally copy-pasting them. …
Human-LLM Dialogue Improves Diagnostic Accuracy in Emergency Care
Clinical decision-making in emergency medicine demands rapid, accurate diagnoses under uncertainty. Despite benchmark progress, evidence for LLMs as interactive aids in live …
SCOPE: Self-Play via Co-Evolving Policies for Open-Ended Tasks
Self-play can train language models without external supervision. However, existing methods require rule-checkable answers, leaving open-ended tasks dependent on curated prompts or …
Do Composed Image Retrieval Benchmarks Require Multimodal Composition?
Composed Image Retrieval (CIR) is a multimodal retrieval task where a query consists of a reference image and a textual modification, and …
Can Retrieval Heads See Images? Multimodal Retrieval Heads in Long-Context Vision-Language Models
Large vision-language models increasingly rely on long-context modeling to reason over documents, hour-level videos, and long-horizon agent trajectories, requiring them to locate …
Learning and Reasoning on Knowledge and Heterogeneous Graphs in the era of Graph Foundation and Large Language Models
Knowledge Graphs (KGs) and heterogeneous graphs (HGs) offer a principled way to represent multi-entity, multi-relational systems, while also revealing a persistent tension …
Rethinking the Harmonic Loss via Non-Euclidean Distance Layers
Cross-entropy loss has long been the standard choice for training deep neural networks, yet it suffers from interpretability limitations, unbounded weight growth, …
Analyzing LLM Instruction Optimization for Tabular Fact Verification
Instruction optimization provides a lightweight, model-agnostic approach to enhancing the reasoning performance of large language models (LLMs). This paper presents the first …
VLM-RobustBench: A Comprehensive Benchmark for Robustness of Vision-Language Models
Vision-language models (VLMs) achieve strong performance on standard, high-quality datasets, but we still do not fully understand how they perform under real-world …