Haokui Zhang

Not All Attention Heads Are What You Need: Refining CLIP's Image Representation with Attention Ablation

2025 · arXiv (Cornell University)

This paper investigates the role of attention heads in CLIP's image encoder. Building on interpretability studies, we conduct an exhaustive analysis and find that certain heads, distributed across layers, are detrimental to the resulting representations. …

NN-Former: Rethinking Graph Structure in Neural Architecture Representation

2025 · arXiv (Cornell University)

The growing use of deep learning necessitates efficient network design and deployment, making neural predictors vital for estimating attributes such as accuracy and latency. Recently, Graph Neural Networks (GNNs) and transformers have shown promising performance …