Skip to content
ml-recall

#infonce

1 question from Transfer Learning

Timed test

1 question · 1 min · score and explanations at the end

Take a test
Q 1 / 1Score 0 / 0

Question 1: Short answer

mediumAI-checkedTransfer Learning

A contrastive loss for an anchor xx with one positive y+y^+ and one negative y−y^- is −log⁡ef(x)⊤f(y+)ef(x)⊤f(y+)+ef(x)⊤f(y−),-\log \frac{e^{f(x)^\top f(y^+)}}{e^{f(x)^\top f(y^+)} + e^{f(x)^\top f(y^-)}}, where the encoder ff outputs unit vectors and there is no temperature. What is the smallest value this loss can take?

a number, 3 decimals