pipeline for text generation seems to be giving garbage outputs
#12
by
lunahr
- opened
I don't think it's a chat completion, that's a text completion and it doesn't consider the chat history
Please check as it kept repeating the final few tokens until it hit the token limit
Hi.
Could you please share some more details?
Oh wait I figured my model is just corrupt.
Oops.
lunahr
changed discussion status to
closed