DeepSeek-VL2

MoE vision-language, now easier to access

DeepSeek-VL2 media 1
DeepSeek-VL2 media 2
DeepSeek-VL2 media 3
Truy cập trang web

Mô tả

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.

Sản phẩm đề xuất