hiracat

3 ms·

1. 1.▲
   by hiracat 2y ago

   You could, and I do think that the more ways they have to get input the more "alive" they will be, but my point in freezing the model in time and then just running inference on it still stands.
2. 2.▲
   by hiracat 2y ago

   I think its because animals are many interconnected systems or models which are made to create "intelligent" behavior, like one for speak one for movement and alike. LLMs are trained purely off text, so they are more similar to a
3. 3.▲
   by hiracat 2y ago

   That's an amazing description, I don't have aphantasia, but that made me understand what it could be like.