A logit-based distillation method that amplifies and decouples the teacher's Top-K outputs reports gains over several existing losses and can be added to other methods.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
TopKD: Top-scaled Knowledge Distillation
A logit-based distillation method that amplifies and decouples the teacher's Top-K outputs reports gains over several existing losses and can be added to other methods.