Are models like GPT-3 ever used as a basis for transfer learning like the BERT/e... | Hacker News

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

wyldfire on July 4, 2021 | parent | context | favorite | on: GPT-J-6B – A 6 billion parameter, autoregressive t...

Are models like GPT-3 ever used as a basis for transfer learning like the BERT/etc ones?

minimaxir on July 4, 2021 [–]

OpenAI has allowed certain customers to finetune GPT-3 for specific datasets.

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