Direct Preference Optimization: Your Language Model Is a Reward Model

[rahimnathwani]

https://arxiv.org/abs/2305.18290

[rahimnathwani]

Previously mentioned by a few commenters:

https://news.ycombinator.com/item?id=36139342

https://news.ycombinator.com/item?id=36423421

https://news.ycombinator.com/item?id=38659012

Newly interesting because Andrew Ng raved about it in his latest newsletter: https://www.deeplearning.ai/the-batch/issue-231/

[amai → rahimnathwani]

see also https://hn.algolia.com/?q=Direct+Preference+Optimization