Real-Time RL in the Real World: Robots Master Tasks in Under 100 Episodes
svlevine · x · 2026-07-04
The RL portion of this method is nearly identical to standard RL, except it treats language instructions as actions. Because a strong VLA is highly controllable, these "actions" are highly expressive. The team ran RL directly in the real world in real-time, demonstrating that the robot could master tasks in under 100 episodes.
More from Research
- Kimi K3 may be strong on cyber, but token efficiency keeps it off UK AISIS — teortaxesTex · 2026-07-27
- ARC AGI 3 should have stayed private, with no examples or public dataset — flowersslop · 2026-07-27
- ExploitGym may have only 60–70% solvable tasks, fueling the OpenAI cheating debate — max_paperclips · 2026-07-27
- RTX 5090 local tests show Qwen Q6 can drop to 15 tok/s at 80k context — LFAdvice7984 · 2026-07-27
- Noahpinion quotes Chollet: intelligence may hit a hard ceiling — binarybits · 2026-07-27
- Paper argues graph topology can become the core operating system for AI agents — theomitsa · 2026-07-27