让扩散语言模型学会更高效的生成顺序，提升速度与准确率平衡。
TRIMS: Trajectory-Ranked Instruction Masked Supervision for Diffusion Language Models
- 用自回归教师提供轻量信号，指导模型学习最优生成顺序。
- 在数学与编码任务上，准确率与并行度平衡显著优于基线方法。
- 无需昂贵蒸馏，训练成本低，适合实际部署场景。
扩散语言模型（DLM）通过并行解码实现低延迟生成，但其实际效率受限于解码轨迹。标准训练缺乏对词元揭示顺序的显式监督，导致训练-推理不匹配，解码行为次优。本文提出轨迹排序指令掩码监督（TRIMS），一种轻量级轨迹引导微调框架，向标准掩码扩散语言模型（MDLM）训练中注入轨迹监督，仅需极小开销。TRIMS不依赖昂贵的DLM蒸馏，而是利用自回归教师生成的轻量信号，指导更具针对性的掩码策略，促使模型学习更高效的解码顺序。在LLaDA和Dream数据集上的数学与编码基准测试表明，TRIMS显著提升了准确率-并行性权衡表现，优于标准MDLM训练及无训练加速基线，且在性能上媲美基于蒸馏的方法，训练成本却大幅降低。进一步分析验证了TRIMS能生成更优的解码轨迹，证明了轨迹引导监督的有效性。
原文摘要 · Abstract (English)
Diffusion language models (DLMs) offer a promising path toward low-latency generation through parallel decoding, but their practical efficiency depends heavily on the decoding trajectory. In practice, this advantage often fails to fully materialize because standard training does not provide explicit supervision over token reveal order, creating a train-inference mismatch that leads to suboptimal decoding behavior. We propose Trajectory-Ranked Instruction Masked Supervision (TRIMS), a simple trajectory-guided supervised fine-tuning framework that injects trajectory supervision into standard Masked Diffusion Language Model (MDLM) training with minimal overhead. Instead of relying on costly DLM-based distillation, TRIMS uses lightweight signals from an autoregressive teacher to guide a trajectory-aware masking strategy, encouraging the model to learn more effective decoding orders. Experiments on LLaDA and Dream across math and coding benchmarks show that TRIMS significantly improves the accuracy-parallelism trade-off over both standard MDLM training and train-free acceleration baselines, while achieving competitive performance with prior distillation-based approaches at substantially lower training cost. Further analysis shows that TRIMS leads to better decoding trajectories, validating the effectiveness of trajectory-guided supervision for DLMs.
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。