OpenGVLab / Ask-Anything

[CVPR2024 Highlight][VideoChatGPT] ChatGPT with video understanding! And many more supported LMs such as miniGPT4, StableLM, and MOSS.
https://vchat.opengvlab.com/
MIT License
2.85k stars 230 forks source link

Question about attention mask #161

Open ikodoh opened 2 months ago

ikodoh commented 2 months ago

According to the code, it seems that causal masking is applied also in the visual queries in Stage 2 and 3. Is there any reason for this implementation?