4 ms·
Classic, I've been doing the same, writing an alphazero for the transformation part. What seems _much_ harder is picking a decent set of transformations/concept
by bubblyworld 2y ago
Classic, I've been doing the same, writing an alphazero for the transformation part. What seems _much_ harder is picking a decent set of transformations/concepts to work with, or more generally automating that process. Maybe you're right that LLMs could help there!
- luke-stanley 2y agoReminds me of NVIDIA Eureka: https://github.com/eureka-research/Eureka https://github.com/eureka-research/Eureka
- bubblyworld 2y agoVery nice! Thanks for the link, that's great inspiration.
- nl 2y agoGreat link, thanks.