From RuntimeWire · RuntimeWire · · 40s
Sakana AI Tests Agent Team Research for Self-Improvement
Sakana AI uses agents' research to train a model that self-improves.
In brief
Sakana AI uses agents' research to train a model that self-improves. Sakana AI tests if agent teams' collective research can train a model for self-improvement. Originally reported by RuntimeWire.
Sakana AI
Sakana AI co-founder and CEO David Ha has made self-evolving AI one of the Tokyo lab's research interests.
MASS Project
A new project from Sakana AI and UC Berkeley tests using a model to organize research agents and train on strongest team results.
Early Results
MASS tests whether agent coordination can become training data for a model's next generation with early research gains but unresolved long-term reliability and costs.
The one thing
Sakana AI tests if agent teams' collective research can train a model for self-improvement.
