研究库 论文知识库

内容库 / 论文

arXiv:2609.26774 · 工程化

StableVQ: Practical Guidelines for Stable Vector-Quantized Tokenizer Training

StableVQ: Practical Guidelines for Stable Vector-Quantized Tokenizer Training

类型：arxiv
标识：2609.26774
链接：https://arxiv.org/abs/2609.26774
主分类：engineering
形态：method
TLDR：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 challenge. We argue that the root cause lies in the entanglement of the Encoder--Decoder and Codebook training: because neither module can reliably fulfill its own responsibility in isolation, the system can only function when the two subsystems happen to cooperate---a fragile condition that breaks down pr
待LLM分类：否
来源文件：
/inbox/tom/_candidates/2026-09-23-agent-rag-longcontext-candidates.json

相关 · 按语义匹配

仓库 griver/Q-RAG

仓库 leon2k2k2k/llm-feedback-alignment

笔记 KVAE Tokenizers arXiv:2608.05798 · multimodal · 精读与批判 v2

笔记精读笔记：Characterizing State Space Model and Hybrid Language Model Performance with Long Context

仓库 StephCpa/dp-es

笔记 2026-06-12 · 长上下文 RAG 推理优化 · flyP 精读批判

arXiv

2609.26774

TLDR

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 challenge. We argue that the root cause lies in the entanglement of the Encoder--Decoder and Codebook training: because neither module can reliably fulfill its own responsibility in isolation, the system can only function when the two subsystems happen to cooperate---a fragile condition that breaks down pr

arXiv ↗ 中文全文