English(EN) b12126: qwen2 : add nvfp4 scales to build_ffn (#31657)

llama.cpp 为 Qwen2 模型添加 NVFP4 缩放支持

作者 PulseAugur 编辑部 · [1 个来源] · 2026-11-05 05:29

llama.cpp 项目发布了 b12126 版本，其中包含对 Qwen2 模型的更新。此更新解决了将模型转换为 NVFP4 格式时，由于 FFN（前馈网络）组件中的缩放未应用而导致模型崩溃的关键问题。此修复确保了这些缩放能够被正确传递和应用，从而防止了模型执行期间发生的断言失败。 AI

影响提高了 llama.cpp 框架内使用 NVFP4 格式的 Qwen2 模型的稳定性和兼容性。

排序理由这是对一个开源项目的软件更新，它提高了模型兼容性和稳定性，属于研究/基础设施类别。[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: 10 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 这是对一个开源项目的软件更新，它提高了模型兼容性和稳定性，属于研究/基础设施类别。[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: infra, model release; 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: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

llama.cpp — Releases TIER_1 English(EN) · danbev · 2026-11-05 05:29

b12126: qwen2 : 为 build_ffn 添加 nvfp4 缩放 (#31657)

This commit adds nvfp4 scales to qwen2's build_ffn call which are currently passed as NULL. The motivation for this is that a model converted to NVFP4 will currently crash because the scales are loaded but never applied in the FFN: <div class="high…