# Improving the Throughput of Diffusion-based Large Language Models via a Training-Free Confidence-Aware Calibration

arXiv: https://arxiv.org/abs/2512.07173 · HTML: https://mlsanity.com/paper/2512.07173/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 7 of its 20 references that are in the map.
- First posted on arXiv 8 Dec 2025.

## Builds on (most cited first, 7 of 7)

- [Structured Denoising Diffusion Models in Discrete State-Spaces](https://mlsanity.com/paper/2107.03006/) (2021)
- [Hierarchical Neural Story Generation](https://mlsanity.com/paper/1805.04833/) (2018)
- [A Continuous Time Framework for Discrete Denoising Models](https://mlsanity.com/paper/2205.14987/) (2022)
- [Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding](https://mlsanity.com/paper/2505.22618/) (2025)
- [Learning to Write with Cooperative Discriminators](https://mlsanity.com/paper/1805.06087/) (2018)
- [Diffusion Language Models Know the Answer Before Decoding](https://mlsanity.com/paper/2508.19982/) (2025)
- [AdaBlock-dLLM: Semantic-Aware Diffusion LLM Inference via Adaptive Block Size](https://mlsanity.com/paper/2509.26432/) (2025)

## Built on by (most cited first, 7 of 7)

- [Learning Unmasking Policies for Diffusion Language Models](https://mlsanity.com/paper/2512.09106/) (2025)
- [Beyond Hard Masks: Progressive Token Evolution for Diffusion Language Models](https://mlsanity.com/paper/2601.07351/) (2026)
- [R2-dLLM: Accelerating Diffusion Large Language Models via Spatio-Temporal Redundancy Reduction](https://mlsanity.com/paper/2604.18995/) (2026)
- [TAD: Temporal-Aware Trajectory Self-Distillation for Fast and Accurate Diffusion LLM](https://mlsanity.com/paper/2605.09536/) (2026)
- [GeoBlock: Inferring Block Granularity from Dependency Geometry in Diffusion Language Models](https://mlsanity.com/paper/2603.26675/) (2026)
- [Polestar: Drift-Aware Cache Calibration and Token Commitment for Efficient Inference of Diffusion LLMs](https://mlsanity.com/paper/2607.14107/) (2026)
- [OLED-MoE: Accelerating MoE-Based dLLM Inference via Inter-Iteration Locality-Aware Expert Offloading](https://mlsanity.com/paper/2609.33385/) (2026)

## Abstract (arXiv)

We present CadLLM, a training-free method to accelerate the inference throughput of diffusion-based LLMs (dLLMs). We first investigate the dynamic nature of token unmasking confidence across blocks and steps. Based on this observation, we present a lightweight adaptive approach that controls the generation block size, step size, and threshold based on the average confidence of unmasked tokens. We further reduce softmax overhead by dynamically leveraging a subset of the vocabulary to regulate sampling breadth. CadLLM is a plug-and-play, model-agnostic method compatible with KV-cache-based dLLMs. Extensive experiments on four popular tasks demonstrate that CadLLM yields up to 1.1-2.28x throughput improvement over the state-of-the-art baseline with competitive accuracy.