BrucecarlL
5 hours ago
I'm not that impressed with this version. When using it, it often acts as if it has no visual capability and refuses to recognize images unless I remind it.
DeepSeek-V4-Flash-Vision-Exp
20 pointsposted 17 hours ago
by amulyabaral5 Comments
syntaxing
17 hours ago
I’m honestly surprised this is better benchmark wise than the text only model. I figured the addition of vision would take away from some of the text capabilities.
mcbuilder
17 hours ago
One of the early results from multimodal training is that it kinda works like cross training. Training vision helps with text tasks and visa versa.
Llamamoe
14 hours ago
I believe that multimodal training increases the robustness of latent representations regardless of which modality is being processed.
nateb2022
13 hours ago
Tangentially this makes me wonder how large Opus really is. Perhaps Opus is a lot smaller than most of the 1T+ assumptions, just a lot more post-training/ finetuning on a 300-400B sized MoE model.