English(EN) Huawei Open-Sources openPangu-2.0 Pretrain, SFT and RL Training Code on Ascend

华为开源openPangu-2.0训练代码，支持Ascend栈

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-28 07:41

华为发布了其openPangu-2.0模型的训练代码，该模型专为华为的Ascend原生混合专家（MoE）栈设计。此次发布包括预训练、监督微调（SFT）和人类反馈强化学习（RL）的代码。开源代码是对先前发布的模型权重（特别是Pro (505B)和Flash (92B)版本）的补充。 AI

影响此次发布为研究人员和开发人员提供了在华为Ascend硬件上训练和微调大规模MoE模型的工具，有望促进人工智能开发的创新。

排序理由大型语言模型训练代码的开源发布。[lever_c_demoted from research: ic=1 ai=1.0]

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 大型语言模型训练代码的开源发布。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: model release, infra; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: 2 days old; Coverage has settled into its steady-state source set.

完整方法见我们的编辑标准。

Pandaily TIER_1 English(EN) · [email protected] (Pandaily) · 2026-09-28 07:41

华为开源 openPangu-2.0 预训练、SFT 和 RL 训练代码（基于 Ascend）

On September 28, 2026 Huawei open-sourced openPangu-2.0 pretrain, SFT and post-training RL code for its Ascend-native MoE stack, complementing earlier Pro (505B) and Flash (92B) weight releases.