Skip to yearly menu bar Skip to main content

Workshop Fri, Dec 11, 2026 • 8:00 AM – 5:00 PM AEDT Parkside 1

Diffusion Language Models: Foundations, Efficiency, and Reasoning

Irina Belousova ⋅ Shansan Gong ⋅ Amin Karimi Monsefi ⋅ Pavlo Molchanov ⋅ Jinjie Ni ⋅ Michael Shieh ⋅ Yizhe Zhang ⋅ Yuchen Zhu ⋅ Amin Karimi Monsefi

Project Page [ OpenReview]

Abstract

Autoregressive (AR) generation has dominated language modeling for years, but a fundamentally different paradigm is gaining rapid traction: diffusion language models (DLMs). Rather than generating tokens left-to-right, diffusion language models corrupt sequences through forward noising over categorical spaces and learn to reverse this corruption, enabling parallel decoding, bidirectional context, and fine-grained controllable generation. In 2025–2026, this paradigm transitioned from theoretical curiosity to commercial reality: Inception Labs launched Mercury, the first commercial-scale diffusion LLM; academic labs released LLaDA and Dream; and academic work—SEDD, MDLM, FS-DFM, LaViDa—has shown diffusion language models can match or exceed AR baselines while achieving up to 10× inference speedups. This full-day workshop brings together researchers from academia and industry to consolidate theoretical understanding, benchmark competing approaches, and chart a roadmap for diffusion language models, catalyzing collaborations across the generative modeling, NLP, and systems communities.

Chat is not available.

Schedule

Time zone: Conference local time zone

8:15 AM

Opening Remarks

8:25 AM

Stefano Ermon

Stefano Ermon

9:00 AM

Subham Sahoo

Subham Sahoo

9:35 AM

Poster Session I + Coffee

10:35 AM

Arash Vahdat

Arash Vahdat

11:10 AM

Oral Talks 1

11:40 AM

Irina Belousova

Irina Belousova

12:05 PM

Lunch Break

1:00 PM

Volodymyr Kuleshov

Volodymyr Kuleshov

1:35 PM

Aditya Grover

Aditya Grover

2:10 PM

Poster Session II + Coffee

3:10 PM

Itai Gat

Itai Gat

3:45 PM

Oral Talks 2

4:05 PM

Panel Discussion

Subham Sahoo ⋅ Arash Vahdat

4:50 PM

Closing Remarks + Awards

DACA-GRPO: Denoising-Aware Credit Assignment for Reinforcement Learning in Diffusion Language Models

Dominic Culver ⋅ Yizhe Zhang

Measuring Parallel Decoding in Diffusion Language Models: Decoder Settings, Generated Text, and Commit Order

Parsa Rahimi ⋅ Azim Dehghani Amirabad

Error Feedback Enables Activation Compression for Pipeline-Parallel Diffusion Language Model Inference

Jinseok Chung ⋅ Namhoon Lee

Flow Reasoning Models: Turning Flows Into Efficient Recurrent Reasoners

Alec Helbling ⋅ Andrey Bryutkin ⋅ Mauro Martino ⋅ Duen Horng Chau ⋅ Nima Dehmamy ⋅ Hendrik Strobelt

SpanBench: When Does Bidirectional Generation Help? Controlled Multi-Span Infilling and Local Editing

Ashraf Shaik

Label Collapse in Diffusion Language Model In-Context Learning

Ziye Guo ⋅ Ye Yuan

Temporal Expert Caching for Accelerating Inference in MoE Diffusion Language Models

Vikhyath Kothamasu ⋅ Sandeep Kumar ⋅ Deepika Palagani

GoldiMask: Fine-Tuning Diffusion Language Models in the Measured Zone of Learnability

Loay Mualem ⋅ Lluis Pastor Perez ⋅ Vinh Tong ⋅ Mathias Niepert

From Interface to Inference: Eliciting Any-Order Inference from Any-Order Models

Seunggeun Kim ⋅ Jaeyeon Kim ⋅ Taekyun Lee ⋅ Yuyuan Chen ⋅ Yilun Du ⋅ Sham Kakade ⋅ Sitan Chen

Temporal Self-Distillation: Faster Inference in Discrete Diffusion Language Models

Andrea Miele ⋅ Shijian Xu ⋅ Metod Jazbec ⋅ Volker Roth ⋅ Eric Nalisnick ⋅ Ilija Bogunovic

AnchorRank: Training-Free Anchor based Plan-and-Execute Decoding in DLMs

Neharika Jali ⋅ Jane Lee ⋅ Anupam Nayak ⋅ Chu-Cheng Lin ⋅ Jinoo Baek

Trace–Answer Compatibility Emerges at Depth and Mediates Prediction in Diffusion Language Models

Ashutosh Ojha ⋅ Pranath Reddy ⋅ Sergei Gleyzer

From Truncation to Commitment: Persistent Context in Uniform Discrete Diffusion

Satoshi Hayakawa

Two Lenses, Two Inference Modes: Comparing Autoregressive and Diffusion Language Models

Jacob Maimon

Hidden Messages Between Diffusion Language Model Agents: Causal Evidence for Latent Communication

Alan-Barsag Gazzaev ⋅ Alexey Gavrilov ⋅ Sergey Muravyov ⋅ Ilya Makarov

One-Step Generation via Riemannian Wasserstein Gradient Flows

David Li ⋅ Chanhyuk Lee ⋅ Jaehoon Yoo ⋅ Nikita Gushchin ⋅ Floor Eijkelboom ⋅ Eric Moulines ⋅ Maxim Panov ⋅ Aleksandr Korotin ⋅ Seunghoon Hong ⋅ Jinwoo Kim

E-MoE: Enhanced Mixture-of-Experts for Non-Factorized Discrete Diffusion Language Models

Arseny Ivanov ⋅ Alexander Kolesov ⋅ Aleksandr Korotin ⋅ Ivan Oseledets ⋅ Mikhail Goncharov

Coverage, Not Ranking: Re-examining KV-Staleness Signals for Block-Diffusion Language Models

Shis Maheta ⋅ Dev Patel ⋅ Krrish Bhagat

Low-Confidence Remasking Traps Flexibility: Realizing Arbitrary-Order Potential for Diverse Rollouts in Diffusion LLMs

Moongyu Jeon ⋅ Dongjae Jeon ⋅ Bumjun Kim ⋅ Mingyu Kim ⋅ Albert No

Phase-MoE: A Co-Design to Bound the Expert Explosion in Block Diffusion Language Models

Van-Hoan Trinh ⋅ Vincent Pauline ⋅ Stefan Bauer

On the Trainability of Masked Diffusion Language Models via Blockwise Locality

Yuxiang Wang ⋅ Xiang Yu ⋅ Baojian Zhou ⋅ Zhao ⋅ Keyue Jiang ⋅ Yanghua Xiao ⋅ Xiaoxiao Xu

DenMark: Robust Semantic Watermarking for Diffusion Language Models

Tianhao Ma ⋅ Weihao Xuan ⋅ Dong-Dong Wu ⋅ Farshid Nooshi ⋅ Takashi Ishida ⋅ Naoto YOKOYA ⋅ Masashi Sugiyama

DUEL: Exact Likelihood for Masked Diffusion via Deterministic Unmasking

Gilad Turok ⋅ Yair Schiff ⋅ Christopher De Sa ⋅ Volodymyr Kuleshov

Revision Breaks Monotone Decoding: Termination and Progress in Diffusion Language Models

Srinivas Harish

The Entropy–Drift Correlation Is Not One Quantity

Dev Patel ⋅ Krrish Bhagat ⋅ Shis Maheta

Architecture Outweighs Budget: a Controlled Comparison of Low-Budget AR-to-Diffusion Conversion in MoE LLMs

Wentao Lu ⋅ Tianyu Zhu ⋅ Jesse Clark

Optimizers for Discrete Diffusion Models: A Controlled Benchmark

Arman Bolatov ⋅ Egor Shulgin ⋅ David Li ⋅ Abduragim Shtanchaev ⋅ Sebastian Stich ⋅ Peter Richtarik ⋅ Martin Takac

How Much Jointness Must a Denoiser Predict? Harmonic Elicitation Limits for Parallel Diffusion Language Models

Manoj Saravanan

Discrete Diffusion Models with Interpolation-Based Controllable Resampling

Sirine Ayadi ⋅ Marcel Kollovieh ⋅ Stephan Günnemann

Latent Space Co-training for Few-Step Text Generation

Maksim Semenov ⋅ Nikita Balagansky ⋅ Ilya Drobyshevskiy ⋅ Dmitry Baranchuk ⋅ Nikita Gushchin ⋅ Daniil Gavrilov ⋅ Aibek Alanov ⋅ Viacheslav Meshchaninov

Fork-dLLM: Avoiding the Flexibility Trap in Diffusion Language Models

Stipe Frković ⋅ Metod Jazbec ⋅ Christian Andersson Naesseth

Trajectory as the Teacher: Few-Step Discrete Flow Matching via Energy-Navigated Distillation

Dominic Culver

When Does Decoding Order Cost Coverage in Diffusion Large Language Models?

Sean Wu ⋅ Jaewon Lee ⋅ Hongjian Zhou ⋅ Fenglin Liu ⋅ Zachary Shinnick ⋅ David Clifton ⋅ Junchi Yu

How Much Dependency Can a Marginal Rule See? Total Correlation in Diffusion LM Parallel Decoding

Dovud Asadov ⋅ Rustambek Urokov ⋅ Qobiljon Toshnazarov ⋅ KAYUMOV ABDUAZIZ ⋅ Mardon Hazratov ⋅ Husanboy Mansuraliyev

When Do Denoising Errors Propagate? A Theory of Downstream Sensitivity in Diffusion Language Models

Preeti ⋅ Venkata S Dhara ⋅ KIRAN RAVISH ⋅ Ankita Kushwaha ⋅ Pawan Kumar

Steering Embedded Language Flows with Formal Constraints

TUXIAO ⋅ Yuntong Gu ⋅ Xiangming Meng

The Hidden Cost of MoE Diffusion Decoding: Profiling and Fixing the Sampler

Harshit Samrat ⋅ Ramakrishna P Kompella

Diff Diffusion: Predicting Edit Paths with Diffusion Models for Efficient Code Editing

xuxiaolong ⋅ Hao Jiang ⋅ Bo Jiang ⋅ Zhiwen Deng ⋅ Ningyuan Sun ⋅ Chen Bingzhou ⋅ Jin chen ⋅ Chenxing Wei

TwinQuant: Decoding-Aware Dual-Centred Quantization for Diffusion Language Models

Seokho Han ⋅ Chaehyeon Min ⋅ Junwon Lee ⋅ Jong Hwan Ko

What Does a Discrete Diffusion Model Learn? One Reverse Process in Three Coordinates

Rodrigo Casado Noguerales ⋅ Bernhard Schölkopf ⋅ Thomas Hofmann ⋅ Aran Raoufi

Plan and Patch : Diffusion Language Models for Agentic Planning

Syamantak Kumar ⋅ Jiang Guo ⋅ Hassan Hamad ⋅ Hideo Kobayashi ⋅ Yi Xiang ⋅ Yezhou Yang ⋅ Yanjun Qi ⋅ Daniele Bonadiman ⋅ Jiarong Jiang

BitSieve: One Bit Budget for KV Selection and Quantization in Block Diffusion LLMs

Gleb Molodtsov ⋅ Ekaterina Alimaskina ⋅ Artur Zagitov ⋅ Evgeny Uskov ⋅ Aleksandr Beznosikov

The Wrong Data at the Right Time: Off-Domain Mixing for Discrete Diffusion

Julian Kleutgens ⋅ Mauricio Tec ⋅ Claudio Battiloro ⋅ Francesca Dominici ⋅ Giannis Daras

ConGO: Consistency-guided Optimization for Parallel Decoding in Discrete Diffusion Models

Jelena Trisovic ⋅ Michael Hersche ⋅ Nicolas Menet ⋅ Abbas Rahimi

Decentralized Diffusion Language Models

Yuehao Wang ⋅ Jinghan Ke ⋅ Yuqing Jian ⋅ Peihao Wang ⋅ Jiajun Zhu ⋅ Yifan Yu ⋅ Qinsi Wang ⋅ Qixing Huang ⋅ Chenfeng Xu ⋅ Zhangyang "Atlas" Wang

Diffusion Language Models Are Natively Length-Aware

Vittorio Rossi ⋅ Giacomo Cirò ⋅ Davide Beltrame ⋅ Luca Gandolfi ⋅ Paul Röttger ⋅ Dirk Hovy

Understanding Parallel Samplers in Masked Diffusion via Random Walks on Graphs

Vansh Bansal ⋅ Cholyeon Cho ⋅ Syamantak Kumar ⋅ Sujay Sanghavi ⋅ Purnamrita Sarkar

GLUQuant: Leveraging Expert Activation Geometry for 2-Bit Diffusion MoEs

Honjar Xing

QWaF: Ultra-Low-Bit KV-Cache Quantization for Block-Diffusion Language Models

Junwon Lee ⋅ Jae-sun Seo ⋅ Jong Hwan Ko

Refinement Buys Intelligibility, Search Buys Identity: What Test-Time Compute Buys in Masked-Diffusion TTS

Nityanand Mathur ⋅ Hamees Sayed ⋅ Ayush P Singh

Where Arbitrary Order Helps: Future-First Story Generation with Diffusion Language Models

Yiran Li

d3LLM-v2: Ultra-Fast Agentic Diffusion LLM via Reward Gating and Online Distillation

Yu-Yang Qian ⋅ Yan-Yu Long ⋅ Lanxiang Hu ⋅ Shen Nie ⋅ Chongxuan LI ⋅ Peng Zhao ⋅ Hao Zhang

Faster Density Estimation and Inference-Time Steering for Flow Language Models

Hanlin Yu ⋅ Sibylle Marcotte ⋅ Arto Klami ⋅ Christian Andersson Naesseth ⋅ Omar Chehab

Before Token Revelation: Linguistic Structure Across Denoising in Diffusion Language Models

Sai Raghuram Jandhyam ⋅ Ryan Dhalbisoi ⋅ Jennifer Seok ⋅ Daniel Huang ⋅ Justice Sefas

From Autoregression to Block Diffusion: Adapting Language Models for Efficient Parallel Decoding

Rouzbeh Amin Tafreshi ⋅ Jack Fan ⋅ Edoardo Mosca ⋅ Mathias Lechner ⋅ Alexander Amini

Block Parallelism for Efficient Distributed Long-Context Diffusion Language Model Training

Tarun Suresh ⋅ Pranshu Chaturvedi ⋅ Hangoo Kang ⋅ Parth Shroff ⋅ Ishan Khare ⋅ Hermann Kumbong ⋅ Azalia Mirhoseini

Efficient Diffusion Language Model Serving beyond Autoregressive Assumptions

Guan-Ming Chiu ⋅ Jeng-Yue Liu

DMLLM-Cache: Efficient Caching for Multimodal Generation with Diffusion Language Models

Zhibo Ren ⋅ Guanxi Lu ⋅ Zhican Wang ⋅ Hao Chen ⋅ Wenxi Gao ⋅ Hongxiang Fan

End-to-End Guarantees for Masked Diffusion Models under Conditional Mixing

Fareed Sheriff

Representation-Space MMD for Few-step Continuous Diffusion Language Models

Ilya Drobyshevskiy ⋅ Maksim Semenov ⋅ Ilia Sudakov ⋅ Nikita Gushchin ⋅ Nikita Balagansky ⋅ Viacheslav Meshchaninov ⋅ Dmitry Baranchuk

Outcome-Routed Distillation and Reinforcement Learning for Diffusion Language Models

Wentao Lu ⋅ Tianyu Zhu ⋅ Jesse Clark

SPIRA: Sparse Pre-RoPE Informed Routing Attention for Diffusion Language Models

Chaehyeon Min ⋅ Seokho Han ⋅ Junwon Lee ⋅ Jong Hwan Ko