GameVerse releases minute-scale gameplay dataset for long-horizon interactive world modeling
AlayaLab · hf · 2026-11-05
Alaya Lab, with Shanghai Innovation Institute, Wuhan University and Tsinghua, released GameVerse (arXiv, Oct 31 2026), a large-scale gameplay dataset targeting long-horizon interactive world modeling.
- Key feature: minute-long sequences with event-aligned annotations, moving past second-level clips
- Five-step curation pipeline: quality filter, content assessment, stability & UI detection, deduplication, scene transition detection
- Annotations include trajectory annotation (with pose filtering) and caption annotation generated with Kimi-K2.5 on 2-minute clips
- Project page, GitHub and datasets are open
More from Research
- Template-based multilingual safety pipeline shows English-only evals misjudge alignment — swiss-ai · 2026-11-05
- 'Grafting' paper: SFT on donor checkpoints + weight merging beats on-policy self-distillation — AdtRaghunathan · 2026-11-05
- Physical Intelligence's RLT uses RL tokens to make robot fine manipulation 3x faster — svlevine · 2026-11-05
- FrankenSim's physics-based flute synthesis gets graded against real recordings — doodlestein · 2026-11-05
- Randomized Algorithms Everywhere: Why Randomness Beats Determinism in Production — arpit_bhayani · 2026-11-05
- CloudOpsBench: open-source benchmark tests AI agents on 45 real cloud ops scenarios — DarkMG1 · 2026-11-05