Customize Consent Preferences

We use cookies to help you navigate efficiently and perform certain functions. You will find detailed information about all cookies under each consent category below.

The cookies that are categorized as "Necessary" are stored on your browser as they are essential for enabling the basic functionalities of the site. ... 

Always Active

Necessary cookies are required to enable the basic features of this site, such as providing secure log-in or adjusting your consent preferences. These cookies do not store any personally identifiable data.

No cookies to display.

Functional cookies help perform certain functionalities like sharing the content of the website on social media platforms, collecting feedback, and other third-party features.

No cookies to display.

Analytical cookies are used to understand how visitors interact with the website. These cookies help provide information on metrics such as the number of visitors, bounce rate, traffic source, etc.

No cookies to display.

Performance cookies are used to understand and analyze the key performance indexes of the website which helps in delivering a better user experience for the visitors.

No cookies to display.

Advertisement cookies are used to provide visitors with customized advertisements based on the pages you visited previously and to analyze the effectiveness of the ad campaigns.

No cookies to display.

shanghaishanghai
0

今日凌晨,全球科技巨头马斯克旗下的大模型公司 xAI 发布了一项重大开源举措,宣布其3140亿参数的混合专家(MoE)模型“Grok-1”正式开源,包括模型的权重和网络架构。这一行动使Grok-1一举成为目前公开参数量最大的语言模型。

据机器之心报道,Grok-1的基础构建基于海量文本数据训练而成,未针对特定任务进行微调,保持了广泛的通用性。该模型的特色在于,其在每个token上的激活权重仅为25%,这意味着在处理信息时,Grok-1能够以高效的方式筛选和理解文本内容。

xAI公司透露,Grok-1是在2023年10月利用JAX库和Rust语言构建的自定义训练堆栈从零开始训练的,显示了公司在高性能计算和编程语言选择上的创新实践。此外,该公司遵循开源社区的友好原则,选择了Apache 2.0许可证来授权Grok-1的权重和架构,鼓励开发者和研究者进行二次开发和合作。

这一开源决定不仅彰显了马斯克和xAI对于人工智能技术的开放态度,也预示着全球AI研究和应用将迎来新的发展机遇。Grok-1的开源,无疑将为科研人员和工程师提供一个强大的工具,进一步推动自然语言处理技术的边界。

英语如下:

News Title: “Musk’s xAI Stuns with Open-Source Release: Grok-1, the World’s Largest Language Model with 3140 Billion Parameters!”

Keywords: Musk, Grok-1, Open-source Large Model

News Content: Early this morning, tech giant Elon Musk’s big model company xAI made a groundbreaking open-source announcement, unveiling that its 3140 billion parameter Mixed Expert (MoE) model, “Grok-1,” is now available to the public, including both model weights and network architecture. This move instantly establishes Grok-1 as the largest publicly disclosed parameterized language model to date.

According to Machine之心, Grok-1 is built on a foundation of extensive text data training,未经特定任务的微调,保留了广泛的通用性。The model’s standout feature is that it activates only 25% of its weights per token, enabling Grok-1 to efficiently filter and comprehend text content.

xAI disclosed that Grok-1 was trained from scratch in October 2023 using a custom training stack built with the JAX library and Rust programming language, showcasing the company’s innovative practices in high-performance computing and language selection. Furthermore, in line with the principles of the open-source community, the company has chosen the Apache 2.0 license to govern Grok-1’s weights and architecture, fostering secondary development and collaboration among developers and researchers.

This open-source decision not only underscores Musk and xAI’s commitment to openness in AI technology but also foreshadows new opportunities for global AI research and application. The open-source availability of Grok-1 undoubtedly provides researchers and engineers with a powerful tool, pushing the boundaries of natural language processing technology further.

【来源】https://mp.weixin.qq.com/s/hvt5zwoazDx26KOaKuTs_w

Views: 1

0

发表回复

您的邮箱地址不会被公开。 必填项已用 * 标注