DeepSeek-VL2
MoE vision-language, now easier to access
About DeepSeek-VL2
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.
Gallery
Comments
No comments yet. Be the first!
Product Details
- Launched: 3 months ago
- Votes: 478
- Featured
Maker
John Doe
Creator of DeepSeek-VL2
Topics
DeepSeek-VL2
vision-language models
MoE
mixture of experts
multimodal understanding
open-source
Hugging Face
demo
VLMs
Related Products
