New SWAP-Score metric evaluates neural networks without training

Researchers have introduced SWAP-Score, a novel zero-shot metric designed to evaluate neural networks without requiring training. This method measures a network's expressivity using sample-wise activation patterns and demonstrates strong predictive performance across various architectures, including CNNs and Transformers. SWAP-Score significantly outperforms existing metrics in computer vision and natural language processing tasks, showing high correlations with ground-truth performance and enabling faster neural architecture search. AI

影响 Enables faster and more accurate neural architecture search by reducing computational overhead in model evaluation.

排序理由 The cluster contains an academic paper introducing a new method for evaluating neural networks. [lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.LG 阅读 →

• CIFAR-10
• CNNs
• DARTS
• FlexiBERT
• GLUE
• ImageNet
• Sample-Wise Activation Patterns
• SWAP-Score
• Transformers

• 论文
• 其他

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →