Smarter computer agents, better reasoning, and robot manipulation breakthroughs.
Today’s Papers
ScaleCUA: Scaling Open-Source Computer Use Agents with Cross-Platform Data
➡️ Large dataset across 6 OSs and 3 task domains; closed-loop pipeline of auto-agents + human curation; big benchmark gains for GUI agents.
FlowRL: Matching Reward Distributions for LLM Reasoning
➡️ Shifts RL objective from reward maximization to reward distribution matching; preserves diverse reasoning paths; strong math & code benchmark results.
RynnVLA-001: Using Human Demonstrations to Improve Robot Manipulation
➡️ Two-stage pretraining on 12M ego-centric videos + trajectory-aware modeling; adds ActionVAE for action compression; stronger transfer to real robot tasks.