Deepseek-VL2
MoE Vision-Language, Now Easier to Access
589 followers
DeepSeek-VL2 are open-source vision-language models with strong multimodal understanding, powered by an efficient MoE architecture. Easily test them out with the new Hugging Face demo.
