苹果公司的研究团队近日发布了一款名为MM1的多模态大模型，这一模型的参数规模高达300亿，标志着苹果公司在该领域的重大突破。该模型由一系列密集模型和混合专家（MoE）变体组成，不仅在预训练指标中达到了行业领先水平，而且在已有多模态基准上进行监督微调后，仍能保持强大的竞争力。这一研究成果由多位作者共同撰写的一篇论文《MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training》中正式公布。MM1模型的推出不仅展示了苹果在人工智能技术方面的实力，也为未来的智能应用和交互方式打开了新的可能性。随着这一技术的不断发展和完善，我们有理由相信，MM1模型将成为推动人工智能领域向前发展的重要力量。
英文标题：Apple Unveils 30-Billion-Parameter Multimodal Large Language Model MM1
英文关键词：Apple, Multimodal, Large Language Model, MM1, MoE Architecture
英文新闻内容：
Apple’s research team has recently launched a multimodal large language model called MM1, which boasts a parameter scale of up to 30 billion, marking a significant breakthrough for the company in this field. The model comprises a series of dense models and mixed-expert (MoE) variants, achieving industry-leading performance in pre-training metrics and maintaining strong competitiveness on existing multimodal benchmarks after fine-tuning. This research was officially announced in a paper titled “MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training,” authored by multiple researchers. The release of the MM1 model not only demonstrates Apple’s prowess in artificial intelligence technology but also opens up new possibilities for future intelligent applications and interaction methods. As this technology continues to develop and mature, there is every reason to believe that the MM1 model will become a key driving force for the advancement of the artificial intelligence field.
【来源】https://mp.weixin.qq.com/s/i9bx6M32uk4Jq2KSRhv4ng
Views: 5