Any plans for the llama3.2 models on sambanova??
It’s gonna be crazy running the 90b for high level vision tasks with sambanova’s speed
Will have some updates to announce soon. What cool project do you have in mind?
It’s my side project called gaurika!
It’s an android app that has all the features i could think of like web grounding, single turn cot, multi turn cot (o1 at home kinda vibe… Really works well with 405b), ability to delete messages, redo them… its also Fully client side so no server processing and everything stays on device
It does have more features but these are the main highlights
I’m also now planning to add support for attaching documents and images (cz of 3.2 models being multimodal) and the option to use TTS and stt for a voice like experience!
That’s outstanding and seems exceptionally insightful.
Thanks 
, just waiting for two things from sambanova now!
- the full context length of llama models
- 3.2 models
But seriously this platform is straight up insanely good