search

PKU-YuanGroup / MoE-LLaVA

Mixture-of-Experts for Large Vision-Language Models
https://arxiv.org/abs/2401.15947
Apache License 2.0
1.97k stars 125 forks source link

supports Chinese or multiple images? #5

Closed BaoyanWang closed 9 months ago

BaoyanWang commented 9 months ago

Thank you for your work! I'm wondering, does the model you've developed support Chinese and multiple images?

LinB203 commented 9 months ago

For Chinese language support: this issue depends a lot on LLM, qwen-1.8B supports Chinese, so MoE-LLaVA-Qwen certainly does. However, Phi2 does not support Chinese, so MoE-LLaVA-Phi2 do not support Chinese. Multi-image: Our code supports multi-image training, multi-video training, and even image-video training together, however we have not released this version for the time being.

lucasjinreal commented 9 months ago

Deos MoE-LLaVA-Qwen available?

LinB203 commented 9 months ago

Deos MoE-LLaVA-Qwen available?

Sure. https://github.com/PKU-YuanGroup/MoE-LLaVA?tab=readme-ov-file#-model-zoo