Diffusion large language models (dLLMs) offer a promising paradigm for parallel text generation, but in practice they face an accuracy-parallelism trade-off, where increasing tokens per forward (TPF) often degrades generation quality. Existing acceleration methods often gain speed at the cost of accuracy. To address this limitation, we propose TAD, a Temporal-Aware trajectory self-Distillation framework. During data construction, we condition a teacher model on both the prompt and the ground-truth response to generate decoding trajectories, recording the intermediate masked states throughout the process. Based on how many decoding steps remain before each masked token is revealed, we partition masked positions into near and distant subsets. For near tokens, we train the student with a hard cross-entropy loss using the teacher trajectory tokens as labels, encouraging confident predictions for tokens that are about to be decoded. For distant tokens, we apply a soft KL divergence loss between the teacher and student token distributions, providing softer supervision and preserving future planning knowledge. This temporal-aware partition naturally gives rise to two deployment configurations: a Quality model that prioritizes accuracy and a Speed model that favors more aggressive acceleration. Experiments show that TAD consistently improves the accuracy-parallelism trade-off. On LLaDA, it raises average accuracy from 46.2\% to 51.6\% with the Quality model and average AUP from 46.2 to 257.1 with the Speed model. Our code is available at: https://github.com/BHmingyang/TAD
d3LLM: Ultra-Fast Diffusion LLM using Pseudo-Trajectory Distillation
University of California, San Diego(加州大学圣地亚哥分校);School of Artificial Intelligence, Nanjing University(南京大学人工智能学院);National Key Laboratory for Novel Software Technology, Nanjing University(南京大学新型软件技术国家实验室);Shanghai Jiao Tong University(上海交通大学)
Beyond Trajectory Imitation: Strategy-Guided Policy Optimization for LLM Reasoning
School of Computer Science and Engineering, Sun Yat-sen University, China(中山大学计算机科学与工程学院);Meituan, Inc., China(美团);Shenzhen Loop Area Institute, China(深圳环区研究所);Meta AI, USA(Meta AI)
From Segments to Scenes: Temporal Understanding for Agentic Autonomous Driving via Vision-Language Models
University of California, Berkeley(加州大学伯克利分校);University of Cambridge(剑桥大学);University of Toronto(多伦多大学);ETH Zurich(苏黎世联邦理工学院);University of Washington(华盛顿大学);University of Southern California(南加州大学)
Reasoning Compression with Mixed-Policy Distillation
Technical University of Munich(慕尼黑技术大学);GESIS – Leibniz Institute for the Social Sciences(莱布尼茨社会科学研究所);Northeastern University(东北大学);LMU Munich(慕尼黑大学);Siemens AG(西门子股份公司);University of Cambridge(剑桥大学);University of Oxford(牛津大学);Mina AI
SEED: Self-Evolving On-Policy Distillation for Agentic Reinforcement Learning
Tsinghua University(清华大学);Zhejiang University(浙江大学);The Chinese University of Hong Kong(香港中文大学);Nanyang Technological University(南洋理工大学);Tongji University(同济大学)
Look Ahead Before You Distill: Future Trajectory Validation of Teacher Guidance for Agentic On-Policy Distillation
Meituan LongCat Interaction(美团龙猫交互);Shanghai Jiao Tong University(上海交通大学);Fudan University(复旦大学);Peking University(北京大学);Nanjing University(南京大学);University of Chinese Academy of Sciences(中国科学院大学);Jilin University(吉林大学);University of Science and Technology of China(中国科学技术大学);The Hong Kong Polytechnic University(香港理工大学)
Prefix Teach, Suffix Fade: Local Teachability Collapse in Strong-to-Weak On-Policy Distillation
College of Computer Science and Technology, Zhejiang University(浙江大学计算机科学与技术学院);Meituan LongCat Team, China(美团LongCat团队,中国);College of Computer Science and Technology, Jilin University(吉林大学计算机科学与技术学院)
Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);Harbin Institute of Technology(哈尔滨工业大学);Fudan University(复旦大学);The Chinese University of Hong Kong(香港中文大学)