ZGCM-1:面向数学与智能体搜索的全开源高效基础模型
原标题:ZGCM-1: A Fully Open and Extremely Efficient Foundation Model for Math and Agentic Search
AI 摘要
中关村学院与中关村人工智能研究院发布 ZGCM-1,一个完全开源的 7B 稠密基础模型,面向数学推理与智能体搜索,从零训练并开源权重、数据、代码与日志。该模型采用混合注意力、FP8 Muon 优化器和 MDP 中期训练等高效方案,在 16K 预训练上实现 4.2 倍时间到损失加速。评测显示其在 7B 规模推理基准上领先,并在部分数学与智能体搜索任务上接近参数量大得多的前沿模型。
正文节选
1]Zhongguancun Academy 2]Zhongguancun Institute of Artificial Intelligence \codehttps://github.com/zgcagi/ZGCM-1 \damodata[Model]https://huggingface.co/zgcagi/ZGCM-1-7B \damodata[Data]https://huggingface.co/datasets/zgcagi/ZGCM-1-Data ZGCM-1: A Fully Open and Extremely Efficient Foundation Model for Math and Agentic Search Abstract While foundation models continue to push the frontiers of mathematical reasoning and agentic problem solving, the broader academic community has been largely excluded