Token indices sequence length is longer than the specified maximum sequence length for this model (XXX > 512). Running this sequence through the model will result in indexing errors - Githubissues

dvlab-research / LISA

Project Page for "LISA: Reasoning Segmentation via Large Language Model"

Apache License 2.0

1.61k stars 108 forks source link

Token indices sequence length is longer than the specified maximum sequence length for this model (XXX > 512). Running this sequence through the model will result in indexing errors #101

Open Ehsan-Yaghoubi opened 6 months ago

Ehsan-Yaghoubi commented 6 months ago

During training the model, I frequently receive the following warning:

Token indices sequence length is longer than the specified maximum sequence length for this model (590 > 512). Running this sequence through the model will result in indexing errors.

The number 590 varies in each warning, for example, 793, 743, ...

Questions: 1- This warning is harmless? 2- When using LLaVA-1.5v weights, it leads to an Assertion error: assert cur_len == total_len. It is line 135 of the dataset.py. Any suggestions to resolve the error?

XenoZLH commented 2 months ago

I also come across this warning, is there any solution for it now?