A random forest trained on 21 zero-cost proxies and network metadata predicts NATS-Bench test accuracy with Kendall correlations above 0.9 on held-out splits.
Title resolution pending
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
-
GreenFactory: Ensembling Zero-Cost Proxies to Estimate Performance of Neural Networks
A random forest trained on 21 zero-cost proxies and network metadata predicts NATS-Bench test accuracy with Kendall correlations above 0.9 on held-out splits.