paper-with-me

Papers

Diffusion LLM with Native Variable Generation Lengths: Let [EOS] Lead the Way

2025-10-28 · Yicun Yang, Cong Wang, Shaobo Wang, Zichen Wen, Biqing Qi, Hanlin Xu, Linfeng Zhang arxiv

Diffusion-based large language models (dLLMs) have exhibited substantial potential for parallel text generation, which may enable more efficient generation compared to autoregressive models. However, current dLLMs suffer from fixed generation lengths, which indicates the generation lengths of dLLMs have to be determined before decoding as a hyper-parameter, leading to issues in efficiency and flexibility. To solve these problems, in this work, we propose to train a diffusion LLM with native variable generation lengths, abbreviated as dLLM-Var. Concretely, we aim to train a model to accurately predict the [EOS] token in the generated text, which makes a dLLM be able to natively infer in a block diffusion manner, while still maintaining the ability of global bi-directional (full) attention and high parallelism. Experiments on standard benchmarks demonstrate that our method achieves a 30.1x speedup over traditional dLLM inference paradigms and a 2.4x speedup relative to autoregressive models such as Qwen and Llama. Our method achieves higher accuracy and faster inference, elevating dLLMs beyond mere academic novelty and supporting their practical use in real-world applications. Codes and models have been released.

📄 PDF Abstract BibTeX arXiv:2510.24605

Code (0)

등록된 구현이 없습니다.

Tasks

Text Generation

Similar Papers 제목 키워드 기반

Improving Variable-Length Generation in Diffusion Language Models via Length Regularization

2026-02-07 · Zicong Cheng, Ruixuan Jia, Jia Li, Guo-Wei Yang 외 arxiv

Diffusion Large Language Models (DLLMs) are inherently ill-suited for variable-length generation, as their inference is defined on a fixed-length canvas and implicitly assumes a known target length. When the length is un…

Variable-Length Tokenization via Learnable Global Merging for Diffusion Transformers

2026-06-18 · Dong Hoon Lee, Seunghoon Hong arxiv

Latent Diffusion Models (LDMs) have become dominant in visual synthesis, but their quality-compute trade-off is largely constrained by the tokenizer's fixed compression ratio. Variable-length tokenizers (VLTs) promise ad…

Beyond Fixed: Training-Free Variable-Length Denoising for Diffusion Large Language Models

2025-08-01 · Jinsong Li, Xiaoyi Dong, Yuhang Zang, Yuhang Cao 외 arxiv

Diffusion Large Language Models (DLLMs) are emerging as a powerful alternative to the dominant Autoregressive Large Language Models, offering efficient parallel generation and capable global context modeling. However, th…

Computational Efficiency

Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusion

2024-07-01 · Boyuan Chen, Diego Marti Monso, Yilun Du, Max Simchowitz 외

This paper presents Diffusion Forcing, a new training paradigm where a diffusion model is trained to denoise a set of tokens with independent per-token noise levels. We apply Diffusion Forcing to sequence generative mode…

Decision MakingPrediction

ElasticAST: An Audio Spectrogram Transformer for All Length and Resolutions

2024-07-11 · Jiu Feng, Mehmet Hamza Erol, Joon Son Chung, Arda Senocak

Transformers have rapidly overtaken CNN-based architectures as the new standard in audio classification. Transformer-based models, such as the Audio Spectrogram Transformers (AST), also inherit the fixed-size input parad…

AllAudio Classification