【北京，中国讯】谷歌近日推出全新开源大模型Gemma，以其超越Meta Llama-2等竞品的卓越性能引发行业瞩目。Gemma模型系列包括2B（20亿参数）和7B（70亿）两种尺寸版本，其中2B版本甚至可在普通笔记本电脑上运行，体现了谷歌在AI领域的创新实力。
据钛媒体消息，Gemma模型是基于与Gemini相同的研究和技术，由Google DeepMind与其他谷歌团队共同开发，旨在推动负责任的AI发展。在18个关键的基准测试中，Gemma在语言理解、推理、数学等领域表现出色，有11个测试分数超过了Meta Llama-2等更大参数的开源模型。
这一突破性成果再次印证了谷歌在人工智能领域的领先地位。Gemma的开源特性，预计将极大促进全球AI研究和应用的发展，同时也为开源社区带来新的活力。未来，随着Gemma模型的进一步优化和应用，人工智能技术将更加贴近普通用户，为人们的工作和生活带来更多便利。
英文翻译：
Title: Google’s Gemma Model Leads the Open-source Arena
Keywords: Google, Gemma, Open-source Model
News content:
[Beijing, China] Google recently launched a new open-source large model, Gemma, which has attracted industry attention with its superior performance surpassing competitors like Meta Llama-2. The Gemma model series includes 2B (2 billion parameters) and 7B (70 billion) versions, with the 2B version even capable of running on a standard laptop, showcasing Google’s innovative strength in the AI field.
According to media reports, Gemma is based on the same research and technology as Gemini and was developed by Google DeepMind and other teams within Google, aiming to promote responsible AI development. In 18 critical benchmark tests, Gemma excelled in areas such as language understanding, reasoning, and mathematics, outperforming larger parameter open-source models like Meta Llama-2 in 11 tests.
This groundbreaking achievement再次印证了Google’s leading position in the field of artificial intelligence. The open-source nature of Gemma is expected to greatly promote the development of AI research and applications globally and bring new vitality to the open-source community. In the future, with further optimization and application of the Gemma model, artificial intelligence technology is set to become more accessible to the general public, bringing more convenience to people’s work and daily life.
【来源】https://www.tmtpost.com/6944205.html
Views: 1