SWE-bench Verified
Microsoft's Agent Lightning v1.0 Trains Agents With Their Real Harnesses
Microsoft Research Asia has open-sourced Agent Lightning v1.0, a rebuilt reinforcement learning framework of about 3,500 lines of code that lets the same harness used in…
Apple's SCLATE Gives Continual-Learning Agents One Shared Scheduler
Apple researchers have introduced SCLATE, an execution substrate for training and evaluating continual-learning agents, which compresses month-long scenarios into hours…