An adversarial early-exit method for frozen-backbone vision language models that reuses the final classifier and reports 1.5x inference speedup with comparable accuracy.
Beem: Boosting performance of early exit dnns using multi-exit classifiers as experts
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
FREE: Fast and Robust Vision Language Models with Early Exits
An adversarial early-exit method for frozen-backbone vision language models that reuses the final classifier and reports 1.5x inference speedup with comparable accuracy.