kohjingyu / gill

🐟 Code and models for the NeurIPS 2023 paper "Generating Images with Multimodal Language Models".
https://jykoh.com/gill
Apache License 2.0
430 stars 36 forks source link

Inference shape is not 8 #36

Open taemin6697 opened 9 months ago

taemin6697 commented 9 months ago

Thank you for the good code. However, the inference code appears as follows. The value of the first dimension of the actual raw_emb tensor is 0, not 8. image

taemin6697 commented 9 months ago

The transformers and tokenizer library versions are the same.