DeepSeek-VL2
MoE vision-language, now easier to access
Description
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.
