如今，Test-Time Scaling（测试时扩展）已成为提升模型推理能力的关键路径。而在这一浪潮中，块扩散语言模型（Block Diffusion Language Models, BDLMs） 凭借其独特的并行解码能力，被视为超越传统自回归（AR）模型推理效率的有力竞争者。然而，现有的 BDLMs 在面对长链推理时，陷入了一个两难的效率 - ...

这项由中国人民大学高瓴人工智能学院的聂晟、朱丰琪、游泽斌等研究者与蚂蚁集团联合完成的突破性研究发表于2025年2月，论文标题为《Large Language Diffusion Models》。有兴趣深入了解技术细节的读者可以通过arXiv:2502.09992访问完整论文，或访问项目主页https://ml-gsai ...

As generative artificial intelligence continues to influence how software is designed, built, and deployed, engineers and data professionals are increasingly expected to work directly with large ...

Rapid technological and scientific advances have fueled a huge wave of innovation over the past decades. The speed of global innovation is known to be dependent on the exchange of knowledge and skills ...

Unlock the full InfoQ experience by logging in! Stay updated with your favorite authors and topics, engage with content, and download exclusive resources. Cory Benfield discusses the evolution of ...

Want smarter insights in your inbox? Sign up for our weekly newsletters to get only what matters to enterprise AI, data, and security leaders. Subscribe Now Researchers from UCLA and Meta AI have ...