Why Multi-Step Tool-Use Reinforcement Learning Collapses and How Supervisory Signals Fix It Paper • 2606.26027 • Published 4 days ago • 15
Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application Paper • 2606.12191 • Published 18 days ago • 70
Think While Watching: Online Streaming Segment-Level Memory for Multi-Turn Video Reasoning in Multimodal Large Language Models Paper • 2603.11896 • Published Mar 12 • 10
MMR-Life: Piecing Together Real-life Scenes for Multimodal Multi-image Reasoning Paper • 2603.02024 • Published Mar 2 • 47
