NeurIPS

NeurIPS — Neural Information Processing Systems

Surveys and per-paper notes for NeurIPS, organized by year.

Years

Year	Surveys	Notes
2025	VLA & Manipulation	~5,290 accepted (24.5%). VLA/manipulation sits between CoRL 2025 and ICLR 2026 in the lineage; Knowledge Insulation (PI spotlight) bridges π0.5 → π0.6.
2024	—	placeholder (HPT, Octo era)

Scope

NeurIPS is the broadest ML venue; VLA / manipulation work appears alongside LLMs, diffusion, and RL theory. In 2025 the VLA concentration was strongest in dual-system architectures, world-model-conditioned policies, RL post-training, and safety/robustness — the long tail that ICLR 2026 then expanded on.

← Back to Home