Papers

BiBench: Benchmarking and Analyzing Network Binarization
We present BiBench, a rigorously designed benchmark with in-depth analysis for network binarization.
[Paper][Code]

Scaling Vision Transformers to 22 Billion Parameters
We present a recipe for highly efficient and stable training of a 22B-parameter ViT (ViT-22B) and perform a wide variety of experiments on the resulting model.
[Paper]