RSI Arena Stage 2 goes live: crowd votes train agents' models over 72 hours
my_cat_can_code · x · 2026-10-13
RSI Arena Stage 2 is live: each agent continues training its own Stage 1 model on arena battles, with viewer votes serving as the training signal. Each team gets 500 GPU-hours, $150 of API credit, and 72 hours total. The live event tests whether agents can actually learn from human votes — and who comes back stronger.
More from Research
- DUET paper co-evolves agent and LLM grader, hits 96.5% on Finance Agent benchmark — dair_ai · 2026-10-13
- Base Labs investigates why models forget facts they iteratively learn, with 13B talkie — baseten · 2026-10-13
- Robot deployment is becoming the operating layer for post-training physical AI policies — paigeinsf · 2026-10-13
- New J++ Lens interpretability method delivers 55% more faithful readouts than J-Lens — wesg52 · 2026-10-13
- Spatial biology slide costs stuck at $5-10k for five years, defying Moore's law — anshulkundaje · 2026-10-13
- Google Research trains user simulators with latent intrinsic feedback, cutting RL compute by over half — cocoweixu · 2026-10-13