Positive users express optimism about the Mythos model's 800B parameter claim being a breakthrough, while negative users call it a terrible experiment that's too impractical for real use.
Based on 3 visible X reactions from 9 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@teortaxesTex Might be some quote-unquote teacher model too! 800BA is definitely too psycho to use IRL in any possible way.
@teortaxesTex it would be best thing ever if Wenfeng picked up that number from you, which is, plausible
Might be GPT-4.5 or some such terrible experiment
some back of napkin calcs on the 800B active, if the model shared something similar to the rumored gpt4 architecture instead of the dsv3 sparse architecture, then that would put it around 10T-12T total params. no idea if its true or not but wenfeng is usually very well informed and the math checks out
> Mythos 800B active I think Wenfeng didn't literally mean it, but if he does… lmao
Positive users express optimism about the Mythos model's 800B parameter claim being a breakthrough, while negative users call it a terrible experiment that's too impractical for real use.
Based on 3 visible X reactions from 9 accounts; directional sample.
Ask a question below.
Published answers will appear here.