mlsanityData to 7 Oct 2026

Commit Locally, Exit Globally: Coordinating Adaptive Sampling and Early Exit in Diffusion Language Models

arXiv 2607.28166 ↗First posted 30 Jul 2026cs.CL cs.AIcited by 0

On the map: Diffusion models → Discrete and text diffusion

also in 1 more
Abstract (arXiv)

Diffusion language models expose a provisional prediction at every denoising step, and on many tasks the candidate answer inside it stabilizes before the step schedule is exhausted. This creates two acceleration opportunities, leaving a block early and stopping the sequence early, but the two require different criteria because block acceleration is local whereas sequence termination is global and freezes the graded answer. Existing methods usually optimize only one axis, and existing exit gates rely on fixed-region confidence or schedule-dependent rules rather than the candidate answer itself. We present C4, which coordinates the two axes by giving each decision its own gate. Confidence-Verified Early Exit (CVEE) decides when the sequence may stop, requiring confidence and sustained argmax stability over a candidate span re-extracted at every step. Commit-Core-Then-Confirm (CCTC) decides which token positions a step may commit by borrowing an autoregressive freezing order inside each block: it commits a boundary-anchored core and confirms deferred positions one step later, so the answer block can be accelerated without allowing local commits inside the answer span to determine sequence-level termination. On 12 zero-shot tasks with LLaDA and Dream, one frozen configuration removes 64--95% of decoding steps and delivers measured end-to-end speedups of 2.6 to 8.6 over full decoding. Code is available at https://github.com/ming053l/C4-dLLM.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Discrete and text diffusion, one of 14 directions in Diffusion models. Builds on 41 of its 46 references that are in the map. First posted on arXiv 30 Jul 2026.

Builds on

41 of 46 references.

Closest ancestors · 16

Same direction, newest first

  1. 2026TACG: Trajectory-Aware Commit Gating for Diffusion Language Model Decoding3 Jul
  2. 2026Reasoning on the Manifold: Bidirectional Consistency for Self-Verification in Diffusion Language Models17 Apr
  3. 2026Early Decisions Matter: Proximity Bias and Initial Trajectory Shaping in Non-Autoregressive Diffusion Language Models12 Apr
  4. 2026CORE: Context-Robust Remasking for Diffusion Language Models4 Feb
  5. 2025Fast-Decoding Diffusion Language Models via Progress-Aware Confidence Schedules2 Dec
  6. 2025KLASS: KL-Guided Fast Inference in Masked Diffusion Models7 Nov
  7. 2025Ultra-Fast Language Generation via Discrete Diffusion Divergence Instruct29 Sep
  8. 2025Dream 7B: Diffusion Large Language Models21 Aug
  9. 2025DPad: Efficient Diffusion Language Models with Suffix Dropout19 Aug
  10. 2025Large Language Diffusion Models14 Feb
  11. 2024Beyond Autoregression: Fast LLMs via Self-Distillation Through Time28 Oct
  12. 2024Masked Diffusion Models are Secretly Time-Agnostic Masked Models and Exploit Inaccurate Categorical Sampling4 Sep
4 more
  1. 2024Simple and Effective Masked Diffusion Language Models11 Jun
  2. 2024Simplified and Generalized Masked Diffusion for Discrete Data6 Jun
  3. 2024Your Absorbing Discrete Diffusion Secretly Models the Conditional Distributions of Clean Data6 Jun
  4. 2023Discrete Diffusion Modeling by Estimating the Ratios of the Data Distribution25 Oct

Other references · 25

  1. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  2. 2021Evaluating Large Language Models Trained on Codecited by 8,053
  3. 2020Measuring Massive Multitask Language Understandingcited by 7,768
  4. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  5. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  6. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,710
  7. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,018
  8. 2021Program Synthesis with Large Language Modelscited by 3,509
  9. 2019PIQA: Reasoning about Physical Commonsense in Natural Languagecited by 2,818
  10. 2021Structured Denoising Diffusion Models in Discrete State-Spacescited by 1,665
  11. 2021Are NLP Models really able to Solve Simple Math Word Problems?cited by 861
  12. 2022PAL: Program-aided Language Modelscited by 635
13 more
  1. 2021Argmax Flows and Multinomial Diffusion: Learning Categorical Distributionscited by 566
  2. 2022A Continuous Time Framework for Discrete Denoising Modelscited by 437
  3. 2025Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decodingcited by 404
  4. 2021A Diverse Corpus for Evaluating and Developing English Math Word Problem Solverscited by 314
  5. 2025Dynamic Early Exit in Reasoning Modelscited by 217
  6. 2025S-GRPO: Early Exit via Reinforcement Learning in Reasoning Modelscited by 87
  7. 2025Wide-In, Narrow-Out: Revokable Decoding for Efficient and Effective DLLMscited by 45
  8. 2025Accelerating Diffusion Large Language Models with SlowFast Sampling: The Three Golden Principlescited by 43
  9. 2026DART: Draft-Agreement Routing for Training-Free Adaptive Thinking Budgets in Hybrid Reasoning Modelscited by 7
  10. 2026R2-dLLM: Accelerating Diffusion Large Language Models via Spatio-Temporal Redundancy Reductioncited by 6
  11. 2026Statistical Early Stopping for Reasoning Modelscited by 4
  12. 2025Controlling the Risk of Corrupted Contexts for Language Models via Early-Exitingcited by 3
  13. 2026STDec: Spatio-Temporal Stability Guided Decoding for dLLMscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026LESS Is More: Mutual-Stability Sampling for Diffusion Language Models
  2. 2026Just on Time: Token-Level Early Stopping for Diffusion Language Models
  3. 2026Path-Dependent Denoising: A Non-Conservative Field Perspective on Order Collapse in Diffusion Language Models
  4. 2026Beyond Mode-Seeking RL: Trajectory-Balance Post-Training for Diffusion Language Models
  5. 2026Neither Parallel Nor Sequential: How DiffusionGemma Actually Commits Tokens
  6. 2026A Tale of Two Temperatures: Simple, Efficient, and Diverse Sampling from Diffusion Language Models
  7. 2026Don't Commit Alone: Joint Token Commitment in Diffusion Language Models
  8. 2026CAI-DLLM: Convergence Aware Inference for Diffusion Language Models