Project
Sakana AI Research Lab
Source-grounded study of Sakana AI's research programs — Continuous Thought Machines, AI Scientist, Darwin Gödel Machine — turned into concept notes, falsifiable reproduction gates, and inspectable atlases.
Status Active
Sakana AI Research Lab
Treat Sakana AI’s publication corpus as a coherent research lab to study: reconstruct mechanisms from primary sources, separate claims from evidence, and turn the strongest ideas into concept notes and falsifiable local protocols. This is the external study surface for the Recursive Self Improvement Lab — RSI Lab, AI Scientist, and Darwin Gödel Machine are organizational context; local publications only claim what sealed notebooks and audits support.
The first public artifact is a Continuous Thought Machines research atlas: source-grounded dissection plus a validation-only parity protocol whose local reproduction state remains disputed. No paper-parity result is claimed. Club B role: import external RSI methods into a verifier-first local evidence pipeline.
Reports
- Report
Continuous Thought Machines: A Research Atlas
A source-grounded dissection of Sakana AI's Continuous Thought Machines, paired with an inspectable parity protocol and an explicitly disputed local reproduction state.