CollabSkill Paper Accepted at COLM, Open-Sources Trajectories and Rating Code
EchoShao8899 · x · 2026-08-21
The author announces CollabSkill was accepted at COLM (scores 8/7/9/8), releasing collected trajectories and the official rating implementation. The work brings human-agent collaboration skill measurement into Co-Gym, studying which agents collaborate best with humans, inter-human variability, and AI literacy. Fun facts: it's the first paper of their PhD where no reviewer asked why humans matter in AI research; and the dataset—uploaded 8/1 and not yet linked from the project page—already has 1700+ downloads, suggesting bots/auto research agents are scraping Hugging Face at scale.
Related event: CollabSkill Paper Accepted to COLM, Dataset Open-Sourced(2 posts)→
More from Research
- Researcher publishes 8 months of primary-source records on frontier model behavior — rayanpal_ · 2026-08-21
- Solo dev trains 250M model SHADOW: 60 MB total, 400 tok/s on laptop CPU, 100M-token retrieval archive — Final-Data-1410 · 2026-08-21
- Hydra-0: a generalist world model that represents robot actions as action flow — mangahomanga · 2026-08-21
- Does Claude perform better in 'claudish'? Researchers call for empirical measures — voooooogel · 2026-08-21
- Sakana's DiffusionBlocks trains networks block-by-block, cutting memory up to 4x — z_latent · 2026-08-21
- Re-evaluation: memory-based self-improving agents ride on noise and task order — dair_ai · 2026-08-21