nobody really understands how LLMs work and a big part of why is that gradient descent makes the weights appear out of nowhere. you "train" a neural network by carefully phrasing a wish called a…

Posters: QiaochuYuan. QiaochuYuan

3 likes recorded by the source corpus.