模型

TeleChat 系列 Xing 4 发布:285B 参数基于昇腾 910C 训练

TeleChat to Xing4: Advancing Frontier Agentic Models on Domestic Chips and Frameworks

精选理由

TeleChat 的 Xing 4 在 8192 张昇腾 910C 上训了 285B 参数模型,还配套开源了 TeleGym 智能体训练平台。想了解不靠 NVIDIA 怎么做大规模预训练和 agent 后训练的,这篇技术报告细节很扎实。

中国电信 TeleChat 系列发布 Xing 4 大模型技术报告,最大模型参数量达 285B,预训练使用 8,192 张昇腾 910C 卡完成。这是少数完全绕开 NVIDIA GPU 生态、在国产芯片与 MindSpore 框架上跑到该规模的公开案例。

正文

技术细节上,Xing 4 在架构层面权衡了稀疏激活、注意力设计、模型深度与宽度以及投机解码,目标是在模型容量与计算、访存和执行开销之间取得平衡。训练基础设施基于 MindSpore HyperParallel,采用拓扑感知混合并行分片、分布式优化器,并通过 CPU 亲和性与核心隔离降低宿主机调度干扰。

后训练阶段推出 TeleGym 智能体执行平台,把执行、评测、轨迹构建与策略优化打通,采用异步执行、有界过采样和部分 rollout 来处理异构任务时长。对关注国产算力路线的团队来说,报告提供了架构与系统工程的具体参考。

原文 · arXiv cs.AI

TeleChat to Xing4: Advancing Frontier Agentic Models on Domestic Chips and Frameworks

We present Xing 4, a series of large language models that extends the TeleChat lineage through coordinated advances in model architecture and training infrastructure on Ascend. This report describes the design principles and systems engineering behind the series, spanning architecture selection, large-scale pretraining, and reasoning and agentic post-training. We study how sparse activation, attention design, model depth and width, and speculative decoding balance model capacity with computation, memory access, and execution overhead. Our pretraining infrastructure builds on MindSpore HyperParallel, combining topology-aware hybrid sharding, efficient distributed optimizer execution, and coordinated management of long-sequence workloads and activation memory. CPU affinity and core isolation further reduce host scheduling interference. Together, these mechanisms support the pretraining of a 285B-parameter model on 8,192 Ascend 910C cards, providing a concrete setting for the architectural and systems studies. For post-training, we introduce TeleGym, a unified agent execution platform that connects execution, evaluation, and trajectory construction to policy optimization. Disaggregated asynchronous execution, bounded oversampling, and partial rollout accommodate heterogeneous task durations while preserving trajectory continuity across policy updates. Together, these advances establish an Ascend-oriented training foundation for the continued development of Xing 4 across model scales and increasingly demanding reasoning and agent workloads.

产业坐标与信源核验

本条情报归属 AITOP「模型」核心栏目,重点聚焦中国 AI 产业与国产大模型生态。在跨区域产业观察坐标中,该动态直接关联「中美大模型技术代差与商业化对照」「AI 开发者工具链、Coding Agent 与工程化实践」研究主线,反映出当前技术演进与商业落地的最新风向。涉及核心实体与技术议题包括:TeleChat、昇腾910C、MindSpore、推理模型,读者可通过下方标签进入对应专题页查阅历史脉络与关联对比。

信源核验说明:本文原始线索采编自「arXiv cs.AI」(@Xingchen AGI Lab, :, Baotian Hu, Bingkai Yang, Boyang Wang, Changzai Pan, Chao Lyu, Chao Wang, Chen Li, Chenshuo Pan, Chenyu Lin, Chenyuan Zhang, Chongming Liu, Dawei Fan, Di Wu, Dong Guo, Dong Li, Dongfang Li, Enlin Liu, Fanye Xiao, Fengkai Li, Fubei Yao, Gang Lin, Han Gao, Hanming Wu, Haowei He, Huapeng Liu, Hua Wang, Huinan Xu, Jiahong Xiong, Jiaxin Peng, Jianxiang Ma, Jianxing Zhuge, Jie Zhang, Ji Xiong, Jinglin Chen, Jingqi Li, Jingxuan Li, Jiwang Ye, Junchen Li, Junbiao Li, Junyang Wu, Kaidong Yu, Kaipeng Jia, Kehai Chen, Kuai Li, Laobin Liu, Lei Li, Libo Qin, Lingling Shi, Luwen Pu, Menghao Xu, Mengjiao Bao, Mengqi Chen, Min Zhang, Mingrui Wang, Naying Wang, Peicong Han, Peiming Guo, Pengfei Li, Qiang Feng, Qi Wang, Qingjun Zhu, Qingli Li, Qitai Ding, Qiyi Xie, Quanchao Zhao, Rongbao Han, Rui Xue, Ruiyu Fang, Ruiting Song, Shijie Li, Shiquan Wang, Shuangyong Song, Sishi Xiong, Teng Su, Wankang Bao, Weijie Zhang, Wei Mao, Weiwei Fu, Weiyang Huang, Wenmin Deng, Wenqing Hu, Wujie Zhong, Xi Yu, Xiaohui Hu, Xiaomin Fang, Xiaomeng Huang, Xiaoqiong Shi, Xiaotong Shang, Xiaoyu Li, Xinglei Xu, Xin Jiang, Xin Wang, Xin Zhang, Xinzhang Liu, Xue Liu, Xuefeng Bai, Xuewei He, Yahan Li, Yanhan Zhang, Yanlei Xue, Yanzhang Wu, Yang Yu, Yi Liu, Yin Zhang, Yingyan Li, Yitong Yao, Yixuan Li, Yongqiang Xia, Yongxiang Li, Yinzhou Zhu, Yu Zhao, Yuhao Jiang, Yunpeng Zhang, Yuting Lin, Yuxuan Wu, Yuyao Huang, Yuyi Wang, Zebo Peng, Zhenwu Pei, Zhenxiong Li, Zhao Wang, Zhaoxi Zhang, Zhihao Yang, Zhiqiang Dong, Zhiwei Yan, Zhilong Lu, Zhonghao Guo, Zhongheng Wu, Zhongjiang He, Zhongqiu Li, Zhu Yuan, Zhuoru Jiang, Zhuoyi Wei, Zijia Guo, Zilu Huang, Zhipeng Guo, Zirui Li, Zihan Wang),经 AITOP 多维评分管线结构化提炼。