近日，Meta 发布了一款名为 Code Llama 70B 的人工智能大模型，该模型在代码生成领域的性能表现优异。据该公司首席执行官扎克伯格透露，Code Llama 70B 是迄今为止“最大、性能最优的模型”。相较于前版本，Code Llama 70B 能够处理更多查询，在 HumanEval 基准测试中的准确率达到了 53%，超过了 GPT-3.5 的 48.1%。然而，相较于 OpenAI 论文中给出的 GPT-4 准确率（67%），Code Llama 70B 仍有一段差距。
Meta 开源的 Code Llama 70B 模型为所有人免费使用，旨在推动人工智能技术的发展和应用。尽管 Code Llama 70B 在代码生成领域的表现已经相当出色，但 Meta 并未止步于此，未来还将持续优化和完善该模型，以期在更多领域取得突破。
英文翻译：
News Title: Meta Releases Open Source Giant Model Code Llama 70B, Outperforming GPT-3.5
Keywords: Meta, Code Llama, GPT-3.5, Artificial Intelligence
News Content:
Recently, Meta released an artificial intelligence giant model called Code Llama 70B, which boasts excellent performance in the field of code generation. According to CEO Zuckerberg, Code Llama 70B is the “largest and best-performing model” to date. Compared to the previous version, Code Llama 70B can handle more queries and achieve an accuracy of 53% in HumanEval benchmark tests, outperforming GPT-3.5’s 48.1%. However, there is still a gap between Code Llama 70B and GPT-4’s accuracy rate of 67% mentioned in the OpenAI paper.
Meta’s open-source Code Llama 70B model is available for free to everyone and aims to promote the development and application of artificial intelligence technology. Although Code Llama 70B has already demonstrated outstanding performance in the field of code generation, Meta plans to continue optimizing and improving the model for future breakthroughs in more areas.
【来源】https://www.ithome.com/0/747/917.htm
Views: 1