margaretwood
@margaretwood
Where is all of this headed? Jiaming wants to build a personalized multimodal model that can interact with humans in the real world — even if it means extending the context window to trillions of tokens.
0 reply
0 recast
0 reaction