Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
taoht 1007629187 | 7 months ago | |
---|---|---|
.. | ||
docs | 7 months ago | |
README.md | 7 months ago |
PengCheng.Mind 又称鹏城·脑海是鹏城实验室开发、开源、开放的基于Transformer架构的自回归式语言模型。模型全流程基于中国算力网的全自主安全可控国产软硬件平台进行开发和训练,采用MindSpore框架实现在大规模集群上长期稳定的多维分布式并行训练。鹏城·脑海模型主要聚焦中文核心能力,兼顾英文和部分多语言能力。当前模型已完成训练1T Tokens数据量,仍在持续训练迭代中。
结构参数 | N(params) | N(layers) | D(model) | N(heads) | D(head) | seq_length | vocab_size |
---|---|---|---|---|---|---|---|
PengCheng.Mind 200B | 201.1 B | 104 | 12672 | 96 | 132 | 4096 | 49984 |
软硬件环境 | 数据并行 | 模型并行 | 流水线并行 | 优化器并行 | 位置编码 |
---|---|---|---|---|---|
Ascend 910A + MindSpore2.0beta | 48 | 4 | 18 | 16 | ROPE |
批大小 | 优化器 | beta1 | beta2 | 学习率 | dropout |
---|---|---|---|---|---|
3072 | adam | 0.9 | 0.96~0.98 | 5e-5~5e-6 | 0.1 |
大模型训练过程版本演化分析及工具开源问题:Open Issues in Analysis and Tools for the Evolution of LLM Version in Training Process
Python
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》