全球领先的代码托管平台GitHub近日宣布推出一项全新功能——GitHub Models，这一创新举措使得超过一亿开发者能够直接访问全球顶级的人工智能模型，极大地促进了人工智能工程师时代的到来。
GitHub Models功能允许开发者在一个交互式沙盒中试用包括微软的Phi-3、OpenAI的GPT-4o、Meta的Llama 3.1、Cohere的Command R+、Mistral AI的Mistral Large等在内的众多流行大模型。这一功能不仅为开发者提供了一个免费测试各家大模型的新途径，还简化了将合适模型导入到个人AI项目中的过程。GitHub创建了一条直接将模型放置在Codespaces和VS Code开发环境中的快速通道，降低了部署AI模型的门槛。
GitHub CEO Thomas Dohmke表示，GitHub Models功能的推出标志着GitHub的又一次转型，从通过开源协作创建AI到借助AI的力量创建软件，再到如今利用GitHub Models推动AI工程师的崛起。这一功能与GitHub对隐私和安全的持续承诺一致，任何提示或输出都不会共享给模型提供者，也不会用于训练或改进模型。
对于个人用户，GitHub Models每天限制访问次数为150次，每分钟不超过15次，每次请求最多可以处理8000个token，最多输出4000个token。这一功能预计在未来几个月内还将添加更多语言、视觉以及其他类型的模型。
GitHub Models的推出，无疑为全球开发者社区带来了新的活力和机遇，加速了人工智能技术的发展和应用，同时也为微软等科技巨头在人工智能领域的竞争提供了新的舞台。随着技术的不断进步和创新，我们可以期待在不久的将来，人工智能将在各行各业发挥更加重要的作用。
英语如下：
News Title: “GitHub Opens New Horizons for AI Engineers: Billions of Developers Can Test Top Models for Free”
Keywords: GitHub, AI Models, Developers
News Content: Global leader in code hosting platform GitHub recently announced the launch of a new feature called GitHub Models, which marks a significant innovation that allows over one billion developers to directly access the world’s top artificial intelligence models, greatly accelerating the advent of the era of AI engineers.
The GitHub Models feature enables developers to test a variety of popular large models, including Microsoft’s Phi-3, OpenAI’s GPT-4o, Meta’s Llama 3.1, Cohere’s Command R+, Mistral AI’s Mistral Large, and more, in an interactive sandbox. This feature not only provides developers with a new free avenue for testing various large models but also simplifies the process of importing suitable models into personal AI projects. GitHub has created a quick path to place models directly in Codespaces and VS Code development environments, lowering the barrier to deploying AI models.
GitHub CEO Thomas Dohmke stated that the launch of GitHub Models marks another transformation for GitHub, from creating AI through open-source collaboration to using AI to create software, and now pushing the rise of AI engineers through GitHub Models. This feature is consistent with GitHub’s ongoing commitment to privacy and security, with no prompts or outputs shared with model providers and not used for training or improving models.
For individual users, GitHub Models is limited to 150 accesses per day, with no more than 15 requests per minute, and each request can handle up to 8000 tokens, with a maximum output of 4000 tokens. This feature is expected to add more languages, visual models, and other types in the coming months.
The launch of GitHub Models undoubtedly brings new vitality and opportunities to the global developer community, accelerating the development and application of artificial intelligence technology and providing a new stage for the competition in the artificial intelligence field among tech giants like Microsoft. As technology continues to advance and innovate, we can look forward to the increasing role of artificial intelligence in various industries in the near future.
【来源】https://www.jiqizhixin.com/articles/2024-08-04-2
Views: 2