NeurIPS 2026 workshop calls for continual world models that keep learning after deployment
jonLorraine9 · x · 2026-08-19
The Continual World Models workshop at NeurIPS 2026 (Sydney, December) is calling for submissions on world models that keep learning after deployment — "a world model should change when the world does."
It runs two non-archival tracks: Main for substantive technical work (up to 8 pages) and Idea for 2-4 page papers on early directions, positions, new formulations, negative results, and lessons from failed experiments. Submissions close Aug 29 AoE via OpenReview.
Related event: NeurIPS 2026 Workshop on Continual World Models Calls for Papers(3 posts)→
More from Research
- A decade in the making: new paper tackles AI retrosynthesis limits — rbhar90 · 2026-08-19
- Apple's Internalized Visual Thinking speeds up video reasoning — apple · 2026-08-19
- MIT, Stanford and 12 institutions launch Public AI Observatory to measure real AI use — yuntiandeng · 2026-08-19
- Visualizing Different Models in Embedding Space with Pangram Image — AaronBergman18 · 2026-08-19
- Study Finds LLMs Infer Drug Class from Suffixes, Not Knowledge — allen_ai · 2026-08-19
- Dataset Release: 1M+ 19th-Century Public Domain Images with Masks — wjb_mattingly · 2026-08-19