近日，Meta公司发布了一款名为Code Llama 70B的开源代码大模型，该模型的准确率优于广为人知的GPT-3.5。Meta公司CEO扎克伯格宣布，该公司最近对其代码生成AI模型Code Llama 70B进行了更新，宣称这是迄今为止“最大、性能最优的模型”。
Code Llama工具于去年8月发布，所有人都可以免费使用。新版Code Llama 70B现在能够处理比之前更多地查询。在HumanEval基准测试中，其准确率达到了53%，表现优于GPT-3.5的48.1%，但距离OpenAI论文中给出的GPT-4准确率（67%）还有一段差距。
这款模型的发布标志着Meta在人工智能领域取得了重要进展。Code Llama 70B的性能超越了GPT-3.5，使得Meta在人工智能领域的竞争力得到了提升。同时，这款模型的开源特性也意味着更多的研究人员和开发者可以参与到Code Llama 70B的改进和应用中，推动人工智能技术的发展。
尽管Code Llama 70B的准确率超过了GPT-3.5，但与GPT-4相比仍有差距。这表明Meta在人工智能领域还有提升空间。未来，Meta可能会继续优化Code Llama 70B的性能，以缩小与OpenAI的差距。
Meta的这一举措也引发了人们对人工智能技术未来发展的关注。随着人工智能技术的不断进步，Code Llama 70B等模型的性能将进一步提升，为各行各业带来更高效、更智能的解决方案。
英文翻译：
Title: Meta Unveils Open Source Large Model Code Llama 70B With Higher Accuracy Than GPT-3.5
Keywords: Meta, Code Llama, Artificial Intelligence
News content:
Recently, Meta公司 has released an open-source code large model called Code Llama 70B, which boasts a higher accuracy than the widely recognized GPT-3.5. Meta’s CEO Mark Zuckerberg announced that the company has recently updated its AI code generation model, Code Llama 70B, claiming it to be the “largest and most performant” model to date.
The Code Llama tool was released in August last year and is available to everyone for free. The updated Code Llama 70B can now handle more queries than before. In the HumanEval benchmark test, its accuracy reached 53%, outperforming GPT-3.5’s 48.1%, but there is still a gap to close compared to the accuracy reported for GPT-4 in OpenAI’s paper (67%).
The release of this model marks an important milestone for Meta in the field of artificial intelligence. The performance of Code Llama 70B surpasses that of GPT-3.5, enhancing Meta’s competitiveness in the field of artificial intelligence. Additionally, the open-source nature of this model means that more researchers and developers can contribute to the improvement and application of Code Llama 70B, driving the development of artificial intelligence technology.
Despite outperforming GPT-3.5 in accuracy, Code Llama 70B still lags behind GPT-4. This indicates that Meta has room for improvement in the field of artificial intelligence. In the future, Meta may continue to optimize the performance of Code Llama 70B to narrow the gap with OpenAI.
This move by Meta has also sparked attention about the future development of artificial intelligence technology. With the continuous advancement of artificial intelligence technology, the performance of models like Code Llama 70B will further improve, bringing more efficient and intelligent solutions to various industries.
【来源】https://www.ithome.com/0/747/917.htm
Views: 1