slai-t-rex-full-parameter-post-training-of-the-dee

name	slai-t-rex-full-parameter-post-training-of-the-dee
description	Skill generated from arXiv paper 2607.20145: SLAI T-Rex: Full-Parameter Post-training of the DeepSeek-V4 Family on Ascend SuperPOD
metadata	{"arxiv":{"id":"2607.20145","title":"SLAI T-Rex: Full-Parameter Post-training of the DeepSeek-V4 Family on Ascend SuperPOD","authors":["Dongfang Li","Xiaodong Luo","Ruoyu Sun","Xuhui Chen","Linyuan Qiu","Jian Meng","Zhengxuan Lu","Yiting Wang","Yucheng Xie","Tao Guo","Tianxiang Fang","Jing Li","Sihang Chen","Shihao Hong","Chang Liu","Weihua Dai","Zirong Zeng","Ziwei Zhu","Zhuohan Wang","Zhengjun Yue","Igor Vasilyev","Min Liu","Weijian Sun","Xin Chen","Yingmeng Gao","Jinhua Zhou","Taolue Chen","Chenwei Wu","Dong Zhang","Wenlong Jin","Jinmin Xiang","Barkova Maria","[Truncated]"],"published":"2026-07-22","categories":["cs.CL","cs.AI"],"url":"https://arxiv.org/abs/2607.20145","utility":1}}

SLAI T-Rex: Full-Parameter Post-training of the DeepSeek-V4 Family on Ascend SuperPOD

arXiv: 2607.20145
Published: 2026-07-22
Authors: Dongfang Li, Xiaodong Luo, Ruoyu Sun, Xuhui Chen, Linyuan Qiu, Jian Meng, Zhengxuan Lu, Yiting Wang, Yucheng Xie, Tao Guo, Tianxiang Fang, Jing Li, Sihang Chen, Shihao Hong, Chang Liu, Weihua Dai, Zirong Zeng, Ziwei Zhu, Zhuohan Wang, Zhengjun Yue, Igor Vasilyev, Min Liu, Weijian Sun, Xin Chen, Yingmeng Gao, Jinhua Zhou, Taolue Chen, Chenwei Wu, Dong Zhang, Wenlong Jin, Jinmin Xiang, Barkova Maria, Ushakov Anton, Xianfei Jin, Tian Ding, Zhihang Lin, Qian Chen, Linxin Yang, Mingzhe Yang, Bingwei Zhang, Hongzhang Yang, Fangxue Zhang, Shijun Qin, Jie Yu, Cuihua Hu, Tolstykh Vasiliy, Nosov Ivan, Abdullin Amir, Zhichen Zhou, Xin Zhang, Zhixiong Ning, Xutong Zhao, Junjie Huang, Jiajun Liu, Weiyan Kong, Zheng Zhang, Wenhan Luo, Lin Hu, Yangbo Guo, Li Zeng, Shihao Zeng, Baotian Hu, Min Zhang, Haizhou Li, Zhiquan Luo
Categories: cs.CL, cs.AI
Utility: 1.00

Key Innovation

Full-parameter post-training of trillion-parameter-scale MoE models introduces substantial system-level challenges for large-scale distributed training, including severe memory pressure, non-overlapped communication overhead, and inefficient kernel execution. While most large-scale LLM training systems are built around GPU-based clusters, this report presents an end-to-end optimization practice on the Ascend NPU SuperPOD. Using the DeepSeek-V4 model family as the target workload, we develop a hi...

Potential Application

This paper presents advancements that could be applied to enhance agent capabilities in the areas of cs.CL, cs.AI.

References

• arXiv: https://arxiv.org/abs/2607.20145