谷歌发布了一款名为Gemma的开源大型语言模型，该模型在18个关键基准测试中有11个超过了包括Meta Llama-2在内的其他更大参数的开源模型。Gemma模型系列分为2B（20亿参数）和7B（70亿）两种尺寸版本，其中2B版本可以在笔记本电脑上直接运行。该模型由Google DeepMind和谷歌其他团队开发，专为负责任的AI开发而设计。这一发布标志着谷歌在人工智能领域的又一重要进展，同时也为研究人员和开发者提供了一个高效、轻量级且性能强大的工具。
英文标题：Google Releases Open Source Large Language Model Gemma, Outperforming Meta Llama-2 and Other Competitors
英文关键词：Google, Gemma, Open Source
英文新闻内容：
Google has announced the launch of Gemma, a new series of open source large language models that deliver superior performance in 18 key benchmarks, surpassing larger parameter models like Meta’s Llama-2 in 11 of these tests. The Gemma model series comes in two sizes: a 2B version (200 million parameters) that can run directly on a laptop, and a 7B version (700 million parameters). Developed by Google DeepMind and other teams within Google, Gemma is designed for responsible AI development. This release represents another significant advancement by Google in the field of artificial intelligence and provides researchers and developers with a powerful, lightweight, and efficient tool.
【来源】https://www.tmtpost.com/6944205.html
Views: 1