There's been some work (e.g RASP - https://arxiv.org/abs/2106.06981) on taking l... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		leogao 48 days ago \| parent \| context \| favorite \| on: Weight-sparse transformers have interpretable circ... There's been some work (e.g RASP - https://arxiv.org/abs/2106.06981) on taking logical computations and compiling them into transformer weights.

astrange 47 days ago [–]

Sakana AI is also working on merging different transformer models together to combine skills.

https://sakana.ai/evolutionary-model-merge/

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