Announcement
ML4 is out, with a claim it can recursively self-improve
A post announcing ML4 says it is especially strong at general agentic tasks, including day-to-day knowledge work.
TLDR
A post announcing ML4 says it is especially strong at general agentic tasks, described as day-to-day knowledge work. Its author also claims ML4 has passed a threshold to make environments and recursively self-improve, and invites people to work on those efforts.
Combined views
2.2K
2 Sources, first seen ago
