OpenClaw-RL
Learns from replies, tool output, and GUI state during use.
A research software framework for continuously improving agents from next-state signals, including reply feedback, tool outputs, and GUI state, without manually supplied reward labels.
Recent stories
0 linked stories
No linked stories yet.