wh (@nrehiew_) 's Twitter Profile
wh

@nrehiew_

eng primarily, ml mostly, research previously

ID: 1718879852827484160

calendar_today30-10-2023 06:37:59

2,2K Tweet

12,12K Followers

85 Following

wh (@nrehiew_) 's Twitter Profile Photo
wh
@nrehiew_

2 months ago

This result that "reasoning" features learnt by an SAEs can be transferred **as is** across MODELS and datasets is super cool and similar in spirit to Mistral's finding that there exists a low dim reasoning direction

This result that "reasoning" features learnt by an SAEs can be transferred **as is** across MODELS and datasets is super cool and similar in spirit to Mistral's finding that there exists a low dim reasoning direction