Adaption Launches AutoScientist to Automate AI Model Fine-Tuning
On Wednesday, Adaption launched AutoScientist, a product designed to automate the fine-tuning of AI models, enabling them to rapidly acquire specific capabilities. The system works by co-optimizing both the training data and the model itself to boost learning efficiency.
"AutoScientist co-optimizes both the data and the model, and learns the best way to basically learn any capability." — Sara Hooker, CEO of Adaption
Key Details
- Built on a proven foundation: AutoScientist is powered by Adaption’s existing data platform, Adaptive Data, which specializes in creating high-quality datasets.
- Continuous improvement loop: The system aims to turn continuously improving datasets into continuously improving AI models.
- Adaptable stack: Hooker noted that Adaption’s approach makes the entire stack "completely adaptable" to optimize for specific tasks.
- Impressive early results: Adaption claims that AutoScientist has more than doubled win-rates across different models. However, these results have not yet been benchmarked against standard tests like SWE-Bench or ARC-AGI.
Availability
Adaption is offering AutoScientist free for the first 30 days after its release.
Looking Ahead
According to Hooker, the potential impact is significant:
"The same way that code generation unlocked a lot of tasks, this is going to unlock a lot of innovation at the frontier of different fields."