Cheng-Yu Hsieh

6

Papers

20

Total Citations

1

Affiliations

Affiliations

University of Washington

Papers (6)

The Hard Positive Truth about Vision-Language Compositionality

RealEdit: Reddit Edits As a Large-scale Empirical Dataset for Image Transformations

NVILA: Efficient Frontier Visual Language Models

Perception Tokens Enhance Visual Reasoning in Multimodal Language Models

Synthetic Visual Genome

Outlier Weighed Layerwise Sparsity (OWL): A Missing Secret Sauce for Pruning LLMs to High Sparsity