DiffusionGemma：文本扩散模型实现4倍加速生成

AI 摘要

Google DeepMind 发布了实验性开放模型 DiffusionGemma，采用文本扩散技术，可在 GPU 上实现高达 4 倍的文本生成加速。该 26B MoE 模型激活参数仅 3.8B，支持双向注意力，适合本地低并发推理场景，但输出质量低于标准 Gemma 4。模型权重以 Apache 2.0 许可在 Hugging Face 上提供，并支持 vLLM、MLX 等工具。

正文节选

DiffusionGemma: 4x faster text generation Today, we’re introducing DiffusionGemma, an experimental open model that explores text diffusion, an exceptionally fast approach to text generation. Released under an Apache 2.0 license, this 26B Mixture of Experts (MoE) model moves beyond the sequential token-by-token processing of typical autoregressive Large Language Models (LLMs). Instead, it generates entire blocks of text simultaneously, delivering up to 4x faster text generation on GPUs. Built upo

发布时间：2026-06-11 00:24

抓取时间：2026-09-07 07:13

来源机构：Google DeepMind