Fine-tuning a generalist on a semantically coherent label subset, using the generalist's own output weights for those classes, gives a small accuracy gain on that subset without extra data.
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
-
Neural network task specialization via domain constraining
Fine-tuning a generalist on a semantically coherent label subset, using the generalist's own output weights for those classes, gives a small accuracy gain on that subset without extra data.