NeurIPS 2026 paper: RL-trained skills transfer poorly to composed reasoning tasks
StanfordAILab · x · 2026-11-17
A NeurIPS 2026 paper, "Compositional Reasoning in Language Models under Reinforcement Learning Post-Training," finds that training LLMs on individual RL-learned skills transfers poorly to composed tasks, while transfer in the other direction is much stronger — a caution for skill-by-skill RL post-training pipelines.
Related event: Study Finds RL Skills Struggle to Compose into Complex Reasoning(2 posts)→
More from Research
- Open Vision Banana: fully open recipe turns image generators into multi-task perception systems — NielsRogge · 2026-11-17
- WebArena-Pro debuts as largest browser-agent benchmark, top agents score 62.3% — xhluca · 2026-11-17
- OpenAI warns CoT monitorability will degrade as models scale, researchers push back — gleech · 2026-11-17
- New preprint unifies polygenic and monogenic risk with longitudinal EHRs for age-resolved disease prediction — anshulkundaje · 2026-11-17
- ICLR authors say AI-reviewed peer review feels like 'pleasing an agent, not a person' — murat_kocaoglu_ · 2026-11-17
- NVIDIA's LidarVerse simulates any LiDAR sensor in real time with one world model — NVIDIAAI · 2026-11-17