Hacker News new | past | comments | ask | show | jobs | submit login
Learning Complex Goals with Iterated Amplification (openai.com)
41 points by stablemap 5 months ago | hide | past | web | favorite | 3 comments



Could this be used to build ML models themselves?


They use Iterated Amplification to train ML models on toy problems to validate the approach. So, yes.


I meant, could a toy problem be construction of a ML model for a given dataset?




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: