Primer on Softmax
There is so much detail to softmax like numerical issues, why it is used with CE and implications of its inductive bias.
There is so much detail to softmax like numerical issues, why it is used with CE and implications of its inductive bias.
A fun and nuanced explainer of Word2Vec, the earliest successful word embedding algorithm.
I have recently applied to SERIMATS where I was tasked to answer the question Why is it surprising, from the perspective of classic machine learning, that ne...