谷歌DiffusionGemma：无需从头训练即可构建文本扩散模型
原标题：Google's DiffusionGemma proves you don't need to train from scratch to build a text diffusion model
AI 摘要
谷歌DeepMind发布了DiffusionGemma的技术报告，该模型基于Gemma-4-26B-A4B改造而成，无需从头训练，采用文本扩散方法并行生成256个token，在H100上达到每秒1500 token的速度。通过两阶段训练（重建和SD·RL）平衡质量与速度，在推理基准上平均提升10分，但绝对性能仍低于自回归模型，且存在重复循环和多用户吞吐量下降等问题。模型以Apache 2.0许可证发布在Hugging Face上，旨在加速文本扩散研究。
正文节选
Google's DiffusionGemma proves you don't need to train from scratch to build a text diffusion model Instead of training a new model from scratch, Google DeepMind retrofitted Gemma 4 into a diffusion model. The newly published report explains how it works and where the tradeoffs are. Google DeepMind released DiffusionGemma as a model in mid-June and has now followed up with the technical report. Unlike standard language models that generate text one token at a time, DiffusionGemma refines blocks