3 ms·Toward Training Superintelligent Software Agents Through Self-Play SWE-RL (Meta)1 points by xhevahir 10mo ago