Hello, I’m Sk4Dl

A postgraduate student of Harbin Institude of Technology, Shenzhen. A deeping learning amateur.

Latest Posts

Finite Scalar Quantization: VQ-VAE Made Simple

Contributions FSQ can serve as a replacement for VQ in various architectures, for different datasets and tasks. There is a reduction of only $0.5 - 3\%$ in the respective met...

MaskBit: Embedding-free Image Generation via Bit Tokens

Contributions An empirical and systematic examination of VQGANs, leading to a modernized VQGAN. A novel embedding-free generation network operating directly on...