Weak Ties, Strong Signals: Efficient Training Data Detection in Diffusion LLMs via Independent Token Sampling
arXiv:2609.22145v1 Announce Type: new Abstract: Diffusion large language models (dLLMs) offer a compelling alternative to autoregressive models, yet they may expose sensitive training data during denoising. Detecting such usage is challenging because dLLMs lack the efficient one-pass probability de…
Read the full story at arXiv cs.LG ↗
Timeline · 1 report
- 2026-09-22 04:00 · arXiv cs.LG
Weak Ties, Strong Signals: Efficient Training Data Detection in Diffusion LLMs via Independent Token Sampling