Jannis Schönleber
@joennlae.bsky.social
Founder @44ai
Co-Founder @viRACE
I build actual stuff with AI.
Ex: PhD Student @ETH, Researcher @Huawei
Open-Source Contributor
https://github.com/joennlae
Co-Founder @viRACE
I build actual stuff with AI.
Ex: PhD Student @ETH, Researcher @Huawei
Open-Source Contributor
https://github.com/joennlae
That explains a lot.
I think they are all more than aware that AGI will not be achieved with this model generation. A new communication mechanism is needed (better than attention or convolution). But on that front, there are physics limitations (memory bandwidth, scalability).
I think they are all more than aware that AGI will not be achieved with this model generation. A new communication mechanism is needed (better than attention or convolution). But on that front, there are physics limitations (memory bandwidth, scalability).
December 5, 2024 at 8:31 AM
That explains a lot.
I think they are all more than aware that AGI will not be achieved with this model generation. A new communication mechanism is needed (better than attention or convolution). But on that front, there are physics limitations (memory bandwidth, scalability).
I think they are all more than aware that AGI will not be achieved with this model generation. A new communication mechanism is needed (better than attention or convolution). But on that front, there are physics limitations (memory bandwidth, scalability).