118 likesNew · 8d old
New 2B Decision Model Answers Text, Images, Video in Milliseconds
What is d1-2B?
d1-2B is a decision-making model that can look at text, pictures or video and return a calibrated answer without generating tokens. It also extracts spans, numbers, dates, locates objects and estimates quantities, all in a single forward pass.
Who it is for
Developers building routing, moderation, extraction or visual inspection pipelines can get fast, multi-modal decisions without the overhead of generation.
How to try it
Open the model page on Hugging Face and call the hosted inference API with your state and questions.