Welcome to The Neural Nook! In this inaugural blog post, we will be exploring Speculative Decoding, an algorithm for faster decoding in transformers.
Clear and crisp article written Shreya.
Can you check if this is correct?
"In this case, q(up)<p(up), and we calculate Z=p(up)/q(up)"
I think Z should be equal to q(up)/p(up)
Clear and crisp article written Shreya.
Can you check if this is correct?
"In this case, q(up)<p(up), and we calculate Z=p(up)/q(up)"
I think Z should be equal to q(up)/p(up)