Search results

ONNX转MNN后，无法直接在sherpa-mnn使用？

Status: Open.

#4084 In alibaba/MNN;

· 1006245347 opened on Dec 26, 2025

MNN支持OpenCL后端的int8量化推理么？

Status: Open.

#4080 In alibaba/MNN;

· letsudo opened on Dec 23, 2025

推理fastvlm，cpu推理正常，gpu推理全是感叹号

Status: Open.

#4077 In alibaba/MNN;

· guoguo1314 opened on Dec 22, 2025

functiongemma支持存在问题

Status: Open.

#4074 In alibaba/MNN;

· dqh147258 opened on Dec 21, 2025

使用vulkan推理PP-OCRv5数据异常

Status: Open.

#4073 In alibaba/MNN;

· wuyanzhiyan opened on Dec 20, 2025

Unable to Convert onnx-community/LFM2-350M-Extract-ONNX to MNN

Status: Open.

#4071 In alibaba/MNN;

· sheldonrobinson opened on Dec 19, 2025

OpenCL Backend: Zero-copy input using external cl_mem (DMA buffer) via clMemoryImportARM on Mali

Status: Open.

#4065 In alibaba/MNN;

· katolikov opened on Dec 18, 2025

MNN-TaoAvatar 换人脸重建

Status: Open.

#4064 In alibaba/MNN;

· wujianwei0420 opened on Dec 18, 2025

在使用官方预编译的 MNN 3.3.0 Windows 版本部署时，在特定硬件环境下遇到了两个严重问题。

Status: Open.

#4063 In alibaba/MNN;

· liyh1999 opened on Dec 18, 2025

有如何跑通lookahead、EAGLE-3两种投机解码的readme？

Status: Open.

#4062 In alibaba/MNN;

· guoguo1314 opened on Dec 17, 2025

请问目前app支持最大的本地化模型是多少B的？

Status: Open.

#4060 In alibaba/MNN;

· Destiny-Hao opened on Dec 16, 2025

转模型完成了MNN权重量化&激活值量化，但是推理时按照文档设置low mem调用NPU后端（QNN、NNAPI）进行计算无加速效果，但精度下降

Status: Open.

#4059 In alibaba/MNN;

· 13737546369 opened on Dec 16, 2025