这项研究解决了扩散语言模型在少步解码时的质量-长度权衡问题，做文本生成或摘要的开发者可以直接用DSL-LLaDA获得更好的零样本效果，值得关注。
离散掩码扩散语言模型在少步解码时面临长度与质量的权衡：固定步数下，要么生成短而高质量的输出，要么生成长但重复的文本。连续去噪通过在嵌入空间联合演化所有位置来规避此问题，但大规模从头构建此类模型仍是难题。本文证明，预训练的掩码DLM可轻量适配以支持连续嵌入空间去噪：从LLaDA-8B-Instruct出发，仅用1000步继续预训练（离散随机定位DSL），将二元掩码替换为连续逐token高斯噪声作为软掩码。适配后的模型支持连续推理，在嵌入空间联合演化所有位置，并在最后一步才做出硬token承诺。在低步数（≤16次前向传播）的零样本摘要任务中，DSL-LLaDA-SDE在所有四个基准上取得最佳ROUGE-1，并大幅避免了迭代去掩码的提前终止/重复权衡。该适配还带来选择性噪声状态鲁棒性：模型能纠正被破坏的token，同时保留干净的token。对照实验表明，使用相同计算量的标准掩码扩散训练无法产生这两种行为。
DSL-LLaDA: Scaling Continuous Denoising to 8B Masked Diffusion LMs
Discrete Masked diffusion language models generate text by iterative parallel decoding, but few-step decoding suffers from a tradeoff between length and quality: with a fixed step budget, standard methods can generate a short, high-quality output, or they can produce long but repetitive text. Continuous denoising can sidestep this tradeoff by evolving all positions jointly in embedding space, but building such a model from scratch at scale remains an open problem. We show that a pretrained masked DLM can instead be lightly adapted to support continuous embedding-space denoising. Starting from LLaDA-8B-Instruct, we continue-pretrain for only 1,000 steps with Discrete Stochastic Localization (DSL), replacing binary masking with continuous per-token Gaussian noise as a soft mask. The adapted model supports continuous inference that evolves all positions jointly in embedding space and defers hard token commitment to the final step. On zero-shot summarization at low step budgets (<=16 forward passes), DSL-LLaDA-SDE achieves the best ROUGE-1 on all four benchmarks and largely avoids the premature-termination / repetition tradeoff of iterative unmasking. The same adaptation also yields selective noisy-state robustness: the model corrects corrupted tokens while preserving clean ones. Control experiments using standard masked diffusion training with the same compute demonstrate neither behavior.