mlsanityData to 7 Oct 2026

Discrete and text diffusion

681 papers11 this weekHigh confidence in this grouping: the papers form one clear topicHigh confidence in this grouping: the papers form one clear topic · About

Keywords: language, diffusion, diffusion language, decoding, masked

Key facts. Discrete and text diffusion: 681 papers on arXiv since 2021 (data through 7 Oct 2026). 353 of its 681 mapped papers (52%) cite “Large Language Diffusion Models” (2025, arXiv 2502.09992). 351 of its 681 mapped papers (52%) cite “Simple and Effective Masked Diffusion Language Models” (2024, arXiv 2406.07524). 78 new papers 17 Sep–7 Oct.

What’s new

17 Sep – 7 Oct 2026

78 new papers, 17 Sep – 7 Oct; 11 this week.

  1. This week · 11
  2. Unrolled Flow Models for Reasoning
  3. Beyond Token Revision: Investigating Mask-and-Replace Diffusion for Zero-Shot Text-to-Speech also fits: Audio, speech and music
  4. Efficient Reasoning with Flow Language Models
  5. Denoising Blocks, Not Tokens: Efficient Compressed Continuous Diffusion with Branching Token Realization
  6. Noise Your Prompt: Noising Conditioning Tokens in Continuous Diffusion Language Models
  7. Denoising Hierarchical Representations: Joint Continuous Diffusion for Language Modeling
  8. Disentangling Paradigm, Identifier, and Decoding in Generative Retrieval
  9. Enhancing Diffusion Language Models with Autoregressive Post-Training Weights
  10. SoloQ: Calibration-Free Quantization for Diffusion Language Models
  11. Representation-Space MMD for Diffusion Language Models
All 78
  1. Noise Out, Bias In: Targeted Bias Injection in Diffusion Language Models via Closed-Loop Activation Steering
  2. Earlier · 67
  3. Mask-Guided KV Cache Eviction in Block Diffusion Language Models
  4. Towards Unbiased On-Policy Distillation for Block Diffusion Language Models
  5. Bayesian Entropy-based Reordering for Calibrated Diffusion Language Models
  6. SpecFold: Folding Multi-Branch Redundancy for Faster Speculative Decoding in Diffusion Language Models
  7. Stabilizing language models under continual learning via condition-anchored distillation
  8. ALoDLM: Adaptively Looped Diffusion Language Models
  9. Pivot-SD: Efficient Self-Distillation for Masked Diffusion Language Models
  10. Preserving Mathematical Reasoning in Compressed Diffusion Language Models via Trajectory-Aware Low-Rank Approximation
  11. Refinement Buys Intelligibility, Search Buys Identity: What Test-Time Compute Buys in Masked-Diffusion TTS also fits: Audio, speech and music
  12. Large Language Continuous Diffusion Models
  13. Context-Tower Conversion Preserves Generation While Freezing Retains Knowledge: Low-Budget AR-to-Diffusion Conversion of MoE LLMs
  14. Hierarchical Continuous Diffusion Language Models
  15. SALD: Self-Referenced Advantage Learning for Diffusion Models also fits: Diffusion foundations and sampling
  16. ITC-MoE: Importance-guided Token-aware Compression for MoE Diffusion Language Models
  17. Know When to Hold 'em: Correct-Token Retention in Uniform-State Diffusion Language Models
  18. Temporally-Resolved Token Attribution Reveals the Generation Dynamics of Diffusion Language Models
  19. Scaling and Distilling Text Embeddings for Better Diffusibility
  20. Two Clocks in Diffusion MLLMs: When Answers Stabilize Before Rationales Unfold
  21. Clock Diffusion: Efficient Semi-Autoregressive Continuous Diffusion Language Models
  22. Blackboard Intelligence Can Surpass Autoregressive on Globally Constrained Problems
  23. Exploring More, Reasoning Better: Stepwise Risk-Sensitive GRPO for Diffusion Language Models
  24. Distribution Matching Distillation for Continuous Diffusion Language Models
  25. Fork-dLLM: Avoiding the Flexibility Trap in Diffusion Language Models
  26. GFD-OPD: Guidance-Folded On-Policy Distillation of Diffusion Models Across Scales
  27. Self-Repulsive Sampling for Diffusion Language Models
  28. Does This Action Still Explain the Task? Reverse Scoring for Diffusion Language Model Agents
  29. Fine-Tuning Diffusion Language Models with Context Selection and Target Weighting
  30. Acceleration of Diffusion Language Model through Discrete Average Generator
  31. How Should Diffusion Language Models Edit Code?
  32. Alpha Diffusion Language Models: Factorization Alone Is Not the Problem
  33. On Trajectory-Aware Training for Masked Diffusion Language Models
  34. Time-Anchored Diffusion Language Models: Latent-Space Caching for Fast Generation
  35. E-MoE: Enhanced Mixture-of-Experts for Non-Factorized Diffusion Language Models
  36. Rethinking Soft Tokens for Parallel Decoding in Diffusion Language Models
  37. Reliable Parallel Decoding in Masked Diffusion Language Models
  38. Reasoning with Continuous Latent Diffusion
  39. CoDR: Training-Free Confidence-Drift Remasking for Diffusion Language Models
  40. FastGuide: Accelerating Reward Guidance for Diffusion Large Language Models
  41. Grab a Coffee: Future-Aware Guidance for Discrete Diffusion with Compiled Objectives also fits: Diffusion foundations and sampling
  42. Twist, Don't Tilt: Trajectory-Exact Constrained Decoding for Masked Diffusion Models
  43. Why Deterministic PRM Guidance Underperforms in Discrete Diffusion Reasoning
  44. d-OPD: Future-Aware On-Policy Distillation for Block Diffusion Language Models
  45. Before the Token Commits: Trajectory-Level Benchmarking of Visual Hallucinations in Diffusion VLMs
  46. Low-Confidence Remasking Traps Flexibility: Realizing Arbitrary-Order Potential for Diverse Rollouts in Diffusion LLMs
  47. DreamingGoose: Staged Distillation from Autoregressive Transformers to Bidirectional Recurrent Diffusion Language Models
  48. Simple Diffusion Language Models Are More Effective Few-Step Generators Than Reported
  49. Faster Block-Diffusion Serving with Distribution-Free Risk Guarantees
  50. One Latent, Many Tokens: Jointly Learning Compressed Embeddings for Efficient Language Diffusion
  51. Safety Reconstructed: Generative Modeling via Masked Diffusion Builds Strong Safety Guardrails also fits: Robustness, privacy and adversarial aspects
  52. From Position Risks to Block Survival: Faster Generation for Diffusion Language Models
  53. Unmask the State: When Does State Adaptation Matter for Masked Diffusion Language Models
  54. Hesitation-Aware On-Policy Distillation for Diffusion Language Models
  55. Constraints Are Graphs, Not Chains: Exact Decoding for Diffusion Language Models
  56. ForkLeft: Entropy-First Rollouts for Prefix-Aligned Autoregressive-to-Diffusion Distillation
  57. Does Uniform Discrete Diffusion Need Time?
  58. Why Jailbreaks Succeed in Diffusion Language Models: An Energy Landscape Analysis
  59. ELF-REG: Scaling Continuous Diffusion Language Models to Reasoning Tasks
  60. Towards Efficient Reasoning: Learning Causal Shortcuts for Diffusion Language Models
  61. LOCKR: A Hidden-State Trajectory-Guided Planner for Detecting and Repairing Stable-but-Wrong Lock-In in Diffusion Language Models
  62. Less Uniform Discrete Diffusion is More Powerful and Scalable
  63. LADDER: Graph-Guided Diffusion Language Models for Efficient Multi-Hop Reasoning
  64. Optimizers for Diffusion Models: A Controlled Benchmark also fits: Diffusion foundations and sampling
  65. CaLR: Causal Latent Revision for Robust Diffusion Reasoning
  66. dQwen3.5: Hybrid-Attention Diffusion Language Models
  67. Parallelism, critical windows, and separations among diffusion language models
  68. Uni-LaDiR: Latent Diffusion Unifies Multimodal Reasoning also fits: Image generation, text-to-image and editing
  69. Zarya: A Hybrid Autoregressive--Masked Diffusion Language Model with Flexible Training and Dual-Mode Inference

Foundations

Most cited by papers here

  1. Large Language Diffusion Models2025cited by 353 here
  2. Simple and Effective Masked Diffusion Language Models2024cited by 351 here
  3. Discrete Diffusion Modeling by Estimating the Ratios of the Data Distribution2023cited by 316 here
  4. Simplified and Generalized Masked Diffusion for Discrete Data2024cited by 286 here
  5. Dream 7B: Diffusion Large Language Models2025cited by 276 here

Lineage

The papers this direction builds on, oldest first.

Each paper on the line cites the one before.

  1. 2020Denoising Diffusion Probabilistic Modelscited by 297 here
  2. 2020Denoising Diffusion Implicit Modelscited by 111 here
  3. 2021Argmax Flows and Multinomial Diffusion: Learning Categorical Distributionscited by 152 here
  4. 2022DiffuSeq: Sequence to Sequence Text Generation with Diffusion Modelscited by 141 here
  5. 2022DiffusionBERT: Improving Generative Masked Language Models with Diffusion Modelscited by 110 here
  6. 2023Discrete Diffusion Modeling by Estimating the Ratios of the Data Distributioncited by 316 here
  7. 2024Your Absorbing Discrete Diffusion Secretly Models the Conditional Distributions of Clean Datacited by 198 here
  8. 2024Simple and Effective Masked Diffusion Language Modelscited by 351 here
  9. 2024Scaling Diffusion Language Models via Adaptation from Autoregressive Modelscited by 153 here
  10. 2024Scaling up Masked Diffusion Models on Textcited by 110 here
  11. 2025Large Language Diffusion Modelscited by 353 here
  12. 2025LLaDA 1.5: Variance-Reduced Preference Optimization for Large Language Diffusion Modelscited by 122 here
  13. 2025DiffuCoder: Understanding and Improving Masked Diffusion Models for Code Generationcited by 110 here
11 more it builds on, not in this chain
  1. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 164 here
  2. 2021Structured Denoising Diffusion Models in Discrete State-Spacescited by 465 here
  3. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 105 here
  4. 2022A Continuous Time Framework for Discrete Denoising Modelscited by 157 here
  5. 2022SSD-LM: Semi-autoregressive Simplex-based Diffusion Language Model for Text Generation and Modular Controlcited by 98 here
  6. 2024Simplified and Generalized Masked Diffusion for Discrete Datacited by 286 here
  7. 2024Masked Diffusion Models are Secretly Time-Agnostic Masked Models and Exploit Inaccurate Categorical Samplingcited by 114 here
  8. 2025Train for the Worst, Plan for the Best: Understanding Token Ordering in Masked Diffusionscited by 125 here
  9. 2025Remasking Discrete Diffusion Models with Inference-Time Scalingcited by 111 here
  10. 2025Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Modelscited by 208 here
  11. 2025Dream 7B: Diffusion Large Language Modelscited by 276 here

The 24 most cited of 1,370.

Browse

30 of 681, ranked 6–35

  1. 6Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models2025208
  2. 7Your Absorbing Discrete Diffusion Secretly Models the Conditional Distributions of Clean Data2024198
  3. 8Scaling Diffusion Language Models via Adaptation from Autoregressive Models2024153
  4. 9DiffuSeq: Sequence to Sequence Text Generation with Diffusion Models2022141
  5. 10Train for the Worst, Plan for the Best: Understanding Token Ordering in Masked Diffusions2025125
  6. 11Masked Diffusion Models are Secretly Time-Agnostic Masked Models and Exploit Inaccurate Categorical Sampling2024114
  7. 12Remasking Discrete Diffusion Models with Inference-Time Scaling2025111
  8. 13DiffuCoder: Understanding and Improving Masked Diffusion Models for Code Generation2025110
  9. 14Scaling up Masked Diffusion Models on Text2024110
  10. 15DiffusionBERT: Improving Generative Masked Language Models with Diffusion Models2022110
All 30
  1. 16SSD-LM: Semi-autoregressive Simplex-based Diffusion Language Model for Text Generation and Modular Control202298
  2. 17Seed Diffusion: A Large-Scale Diffusion Language Model with High-Speed Inference202595
  3. 18Likelihood-Based Diffusion Language Models202388
  4. 19Beyond Autoregression: Discrete Diffusion for Complex Reasoning and Planning202482
  5. 20A Reparameterized Discrete Diffusion Model for Text Generation202382
  6. 21dKV-Cache: The Cache for Diffusion Language Models202581
  7. 22Continuous diffusion for categorical data202281
  8. 23Accelerated Sampling from Masked Diffusion Models via Entropy Bounded Unmasking202576
  9. 24LLaDA2.0: Scaling Up Diffusion Language Models to 100B202572
  10. 25The Diffusion Duality202565
  11. 26Diffusion LLMs Can Do Faster-Than-AR Inference via Discrete Diffusion Forcing202561
  12. 27SDAR: A Synergistic Diffusion-AutoRegression Paradigm for Scalable Sequence Generation202561
  13. 28Self-conditioned Embedding Diffusion for Text Generation202259
  14. 29Autoregressive Diffusion Models202156
  15. 30Path Planning for Masked Diffusion Model Sampling202553
  16. 31Dream-Coder 7B: An Open Diffusion Language Model for Code202549
  17. 32Generalized Interpolating Discrete Diffusion202545
  18. 33Latent Diffusion for Language Generation202243
  19. 34FlashDLM: Accelerating Diffusion Language Model Inference via Efficient KV Caching and Guided Diffusion202542
  20. 35wd1: Weighted Policy Optimization for Reasoning in Diffusion Language Models202542