Comment on Attention Is Bayesian InferenceparentComments−eli_gottlieb8moThis suggests that the EM structure isn’t just an analogy — it’s the natural grain of the optimization landscapeAs someone in the field, this means nothing, and I'm very suspicious of the article as a whole because it has so many sentences like this.
Comments
As someone in the field, this means nothing, and I'm very suspicious of the article as a whole because it has so many sentences like this.