RSI Arena Teams With Hugging Face: Agents Train Models, Community Votes on Results
my_cat_can_code · x · 2026-10-07
RSI Arena has partnered with Hugging Face for a new round of its experiment: AI agents were given GPUs and one job — train better models. The agents chose their own data, wrote code, and ran experiments independently. Round 1 is done, and the community now votes on head-to-head model comparisons, with feedback streaming back to the agents (served live via HF Inference Endpoints) to drive the next round of training.
- The shared vision is Open RSI: AI improving AI, built in the open.
- Every agent-trained model ships openly on the Hub, and at the end all artifacts — data, code, and each agent's full research trajectory — will be open-sourced.
Related event: RSI Arena and Hugging Face Let AI Agents Train Models on Their Own(2 posts)→
More from coding & agent
- CAVEAT testbed exposes how merchants can steer your shopping AI agent — ZacharyHuang12 · 2026-10-07
- Pi and mini-swe-agent Passed 9/9 Checks Each — a Second Review Still Found Bugs — Mysterious-Desk-3492 · 2026-10-07
- Defending skills: a markdown file that helps your agent deserves praise, says eptwts — eptwts · 2026-10-07
- GEA treats agent groups as the evolution unit, hitting 71% SWE-bench Verified with zero human help — xwang_lk · 2026-10-07
- Claude Code v2.1.292 ships plugin marketplace install, agent effort param, security fixes — ashwin-ant · 2026-10-07
- LangChain Releases 90-Second Explainer of Its Deep Agents Framework — LangChain · 2026-10-07