Rim
@RimShayakhmetov
Congrats to the @liquidai team, great job @ramin_m_h @xanamini @maximelabonne !
If you thought 100–500 ms was fast for a model call, d1-3B just reset the bar: 8 ms (!) per decision on a single RTX 4090.
Our current molecular property specialists are 4B models too, so we know
If you thought 100–500 ms was fast for a model call, d1-3B just reset the bar: 8 ms (!) per decision on a single RTX 4090.
Our current molecular property specialists are 4B models too, so we know
Liquid AI@liquidai · Oct 7Today we release Open d1: two open-weight multimodal models in our d1 decision model family.
> d1-3B: text + vision
> d1-omni-600M: text + image or text + audio
> Real-time decision making anywhere, from data centers such as @nvidia DGX to RTX workstations to Jetson at the edge.
> d1-3B: text + vision
> d1-omni-600M: text + image or text + audio
> Real-time decision making anywhere, from data centers such as @nvidia DGX to RTX workstations to Jetson at the edge.
Open quoted post →
0 10