This new MoE AI model looks promising with 1T parameters (32B activated)
The results speaks for themselves.
This new MoE AI model looks promising with 1T parameters (32B activated)
The results speaks for themselves.
@hello1 Our ML teams are aware of it . There is no present decision to productize this on our cloud. If our product managers decide to do so; I will revert with a follow up post .
-Coby
As an aside I have opened up an Idea to the product team for their consideration.
-Coby
thanks - this does look like a very valid model for real world use
@david.keane thank you for joining the community. I will definitely append all additional feedback to the product team.
-Coby
Thank you for sharing. It does look ver promising!
I’d love to know if you have tried it for any of your agentic use-cases and whether you have seen any significant improvement on your results.
Not yet, as it is significantly larger than my setup which I can use. With 128GB ram my model running capabilities are limited
ah ok! Thanks for letting us know about this model!
Looking forward to put this model through the token ripper!
Thank you for joining the community .
Coby
Yes, this would be an amazing model to have! Thank you for considering adding it.