MSME: A Multi-Stage Multi-Expert Framework for Zero-Shot Stance Detection

UW-BioNLP presented their methods for extracting chemotherapy timelines from clinical notes at the ChemoTimelines 2025 shared task, focusing on strategies like chain-of-thought thinking and supervised fine-tuning. Their best-performing model, fine-tuned Qwen3-14B, achieved a score of 0.678 on the test set leaderboard.

The Natural Language Actor-Critic (NLAC) algorithm has been introduced to enhance the training of large language model (LLM) agents, which interact with environments over extended periods. This method addresses challenges in learning from sparse rewards and aims to stabilize training through a generative LLM critic that evaluates actions in natural language space.

ExPairT-LLM has been introduced as an exact learning algorithm for code selection, addressing the challenges in code generation by large language models (LLMs). It utilizes pairwise membership and equivalence queries to enhance the accuracy of selecting the correct program from multiple outputs generated by LLMs, significantly improving success rates compared to existing algorithms.

Astra has been introduced as a pioneering multi-agent system designed for optimizing GPU kernel performance, addressing a long-standing challenge in high-performance computing and machine learning. This system leverages existing CUDA implementations from SGLang, a framework widely used for serving large language models (LLMs), marking a shift from traditional manual tuning methods.

CryptoBench has been introduced as the first expert-curated, dynamic benchmark aimed at evaluating the capabilities of Large Language Model (LLM) agents specifically in the cryptocurrency sector, addressing challenges such as time sensitivity and the need for data synthesis from specialized sources.

A new framework called ThinkDeeper has been introduced to enhance the visual grounding capabilities of autonomous vehicles by utilizing a Spatial-Aware World Model (SA-WM). This model enables vehicles to interpret natural-language commands more effectively by reasoning about future spatial states and disambiguating context-dependent instructions.

UniMo has been introduced as an innovative autoregressive model that simultaneously generates and understands 2D human videos and 3D human motions, marking a significant advancement in the integration of these two modalities. This model addresses the challenges posed by the structural and distributional differences between 2D and 3D data, which have largely remained unexplored in existing methodologies.

A recent study highlights the challenges faced by vision-language models (VLMs) in factual recall, identifying a two-hop problem that involves forming entity representations from visual inputs and recalling associated knowledge. The research benchmarks 14 VLMs, revealing that 11 of them show a decline in factual recall performance compared to their large language model (LLM) counterparts.