近日，深度求索团队成功推出了首个国产开源MoE（多出口）大模型DeepSeek MoE，其性能表现与国际巨头相媲美。这款拥有160亿参数的专家模型在计算量仅为40%的情况下，表现完全不输给密集的Llama 2-7B模型。尤为值得一提的是，DeepSeek MoE在数学和代码能力上对Llama形成了碾压优势，堪称19边形战士。
这款国产开源大模型的问世，标志着我国在人工智能领域取得了重要突破。DeepSeek MoE的性能优异，同时主打节约计算量，这对于当前AI应用场景的广泛推广具有重要意义。
英文翻译：
News Title: Domestic open-source MoE large model debuts with stunning performance
Keywords: Domestic open-source, MoE large model, excellent performance, computing resource saving
News Content:
Recently, the DeepSeek team successfully launched the first domestic open-source MoE (multi-exit) large model. Its performance rivals international giants. This expert model with 16 billion parameters performs competitively against the dense Llama 2-7B model while only consuming 40% of the computing resources. Particularly noteworthy is the fact that DeepSeek MoE overwhelms Llama in terms of mathematical and coding capabilities, making it a 19-sided warrior.
The introduction of this domestic open-source large model signifies an important breakthrough in the field of artificial intelligence in our country. With its excellent performance and a focus on computing resource saving, DeepSeek MoE holds great promise for the widespread adoption of AI applications.
【来源】https://www.qbitai.com/2024/01/113381.html
Views: 1