用自举分组法分离图像全局与局部信息，提升压缩与生成效率。
Balancing Image Compression and Generation with Bootstrapped Tokenization
- 将图像信息按粒度拆分为全局和局部令牌，避免冗余混合。
- 生成器仅需全局令牌，计算量减少约40%，重建与生成质量更优。
- 可扩展性强，64令牌下达到1.56的gFID新纪录，适合高效图像生成场景。
尽管图像标记化取得进展，但标准方法在每个标记中混合不同粒度的信息，导致标记间仍存在冗余，且干扰生成器训练。本文提出SelfBootTok，通过自举学习将信息清晰分解为全局与局部标记组。模型仅从全局标记预测局部细节，将视觉细节负担从生成器转移至标记器。结果表明，生成器仅需全局标记即可运行，计算量降低约40%，同时实现更优的重建与生成效果。该范式具备良好可扩展性：通过更多数据或参数自监督学习局部表示，仅用64个标记即达成1.56的gFID新纪录。
原文摘要 · Abstract (English)
Despite progress in image tokenization, standard methods encode redundant information by mixing all granularities within each token, thus redundancy persists between tokens. The mix of information of different granularity also complicates the training of generators. This paper introduces SelfBootTok, a method that resolves this by cleanly decomposing information into global and local token groups. Through self-bootstrapped learning, the model predicts local details exclusively from global tokens, shifting the burden of visual details from the generator to the tokenizer. Consequently, our generator is far more efficient, requiring only global tokens and reducing computation by approximately 40%, while delivering superior reconstruction and generation. Moreover, this paradigm scales elegantly: by leveraging more data or parameters to self-supervise local representation learning, SelfBootTok achieves a new state-of-the-art gFID score of 1.56 using only 64 tokens.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。