用扩散模型并行恢复嵌入对应文本，无需编码器即可高效还原。
Embedding Inversion via Conditional Masked Diffusion Language Models
- 通过条件掩码扩散模型并行去噪恢复全部词元
- 32词元序列下仅需8次前向传播即完成还原
- 无需编码器、无需迭代修正，适配多种嵌入模型
我们将嵌入反演问题建模为条件掩码扩散过程，通过迭代去噪并行恢复所有词元，而非传统的自回归逐个生成。掩码扩散语言模型通过自适应层归一化条件于目标嵌入，推理时仅需8次前向传播，且无需访问目标编码器。在三种嵌入模型的32词元序列上，该方法实现了无需编码器、无需迭代修正或架构特异性对齐的并行词元恢复。源代码与在线演示已公开于 https://github.com/jina-ai/embedding-inversion-demo。
原文摘要 · Abstract (English)
We frame embedding inversion as conditional masked diffusion, recovering all tokens in parallel through iterative denoising rather than sequential autoregressive generation. A masked diffusion language model is conditioned on the target embedding via adaptive layer normalization, requiring only 8 forward passes with no access to the target encoder at inference time. On 32-token sequences across three embedding models, the method achieves token recovery through parallel denoising without requiring encoder access, iterative correction, or architecture-specific alignment. Source code and live demo are available at https://github.com/jina-ai/embedding-inversion-demo.
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。