Yuhan Wang

Standard pass rates hide severe technical debt: coding agents duplicate logic in over half of multi-turn task chains by turn 5 while progressively abandoning repository exploration.

Tiling multi-shot video chunks across a 2D spatial grid instead of stretching them along a single temporal axis yields 6× more narrative shots under the exact same token budget while decisively outperforming prior consistency baselines.

PaperGym achieves a remarkable 73.48 on ResearchQA, outperforming larger models and redefining how AI can generate and evaluate research plans.

MM-Spectrum achieves substantial performance gains in molecular structure elucidation by intelligently routing multimodal data, revealing the power of tailored expert mechanisms in handling spectral heterogeneity.

A unified policy can achieve near-perfect performance across diverse robotic embodiments without requiring extensive fine-tuning for specific tasks.

HERO enables robots to autonomously evolve manipulation skills from scratch, drastically reducing reliance on human demonstrations.

Transforming diverse visual tasks into a unified RGB format enables a single model to excel across multiple domains without task-specific tuning.

Achieving 100% success on complex robotic tasks with just one demonstration could revolutionize how we approach real-world robotic learning.

Current AI models miss critical tumor detections in underrepresented demographics, revealing a hidden bias that could compromise patient outcomes.