Find answers from the community

Updated 2 months ago

Token indices sequence length is longer

Token indices sequence length is longer than the specified maximum sequence length for this model (1215 > 1024). Running this sequence through the model will result in indexing errors. I updaetd the library to latest one and saw this error. WOnder where I could specify the length to be more than 1024.

5 comments

LLogan M

Are you using python3.8?

This error is very benign, it's due to using a transformers tokenizer for python 3.8

LLogan M

At least, the last time I tracked this down that was the cause lol

LLogan M

It's using the GPT2 tokenizer, which is why it gives that warning (but we aren't using a GPT2 model, so it doesn't matter)

MMilkman

Oh, no wonder the chat engine was working alright.

MMilkman

THAnks!

Add a reply