|
|
|
|
|
|
by matusp
1174 days ago
|
|
|
Shameless self-promotion, I have recently written a blog about this. ChatGPT actually is usually a little bit worse than older models for these classical NLP tasks. Of course the older models are not zero-shot. https://www.opensamizdat.com/posts/chatgpt_survey/ |
|
|