返回全部动态

llama.cpp b10290 发布:新增 ggml_build_forward_order 修复音频图问题

原标题:b10290

llama.cpp Releases一手来源产品发布质量 72

AI 摘要

llama.cpp 发布 b10290 版本,新增 ggml_build_forward_order 函数,用于在计算图中插入节点而不设置计算标志,以解决 mtmd 音频图中 GEN_WAV 调用因陈旧输入触发断言的问题。该版本提供了多平台二进制文件,包括 macOS、Linux、Windows、Android 等。

以上摘要由 AI 生成,可能存在误差。事实请以原文为准。

正文节选

<details open> mtmd/ggml: add ggml_build_forward_order (#26649) * ggml: add ggml_build_forward_order ggml_build_forward_expand marks the tensor and all its ancestors for compute, so using it as a pure ordering hint (keeping q, k and v together) defeats ggml_build_forward_select: the unselected branch is forced to run with inputs that were never uploaded. In the mtmd audio graph this makes GEN_WAV calls execute the GEN_CODE branch with a stale inp_code0, hitting the get_rows bound assert on CP


发布时间:2026-08-06 09:33
抓取时间:2026-08-06 21:02
来源机构:ggml-org
阅读原文github.com