StableVQ: Practical Guidelines for Stable Vector-Quantized Tokenizer Training

A new paper, "StableVQ: Practical Guidelines for Stable Vector-Quantized Tokenizer Training," addresses training stability in vector-quantized visual tokenizers, which underpin autoregressive and masked image generation models. The work targets the stability challenge that persists even after shared-projection codebook methods substantially improved codebook utilization. The paper's stated contribution is a set of practical guidelines for stable VQ tokenizer training.

Vector Quantization VQ is fundamental to discrete visual tokenizers that power modern autoregressive and masked image generation models. While recent shared-projection codebook methods have substantially advanced codebook utilization, training stability remains a critical and underexplored challen