alibaba/MNN

Issues

• mls app on Mac doesn't work when downloading models

  #3888 opened 2 months ago by limingchina
  2

• 本地加载的多模态模型，对话界面中不能显示“+”图片的按钮。是需要改动什么配置吗？

  #3885 opened 2 months ago by biker266
  1

• apps/Android/MnnLlmChat/app/src/main/cpp/llm_session.cpp这里可能有配置覆盖问题

  #3896 opened 2 months ago by af-74413592
  1

• master分支 iOS编译出来的库崩溃

  #3898 opened 2 months ago by wangsizhe626
  1

• 能否可以在应用内提供对生成内容直接进行更改

  #3897 opened 2 months ago by atmcy
  0

• 我在iOS上运行了MNNLLMiOS这个app，但是没法正常问答

  #3895 opened 2 months ago by sunweifeng
  4

• [Performance] 如何提升LLM在CPU上运行时的CPU利用率？

  #3894 opened 2 months ago by zhanweiw
  5

• 性能评测Qwen2.5 vl 3b 闪退

  #3893 opened 2 months ago by af-74413592
  1

• 在mac上运行 sherpa-mnn/kokoro-tts-zh-en-c-api，出现段错误崩溃

  #3845 opened 2 months ago by adi-hld
  9

• Build MNN for Windows on Arm failed: kleidiai can't be built successfully.

  #3851 opened 2 months ago by quic-zhanweiw
  7

• llm模块CMakeLists.txt语法错误

  #3892 opened 2 months ago by ArmstrongCN
  0

• Build MNN for Windows on Arm failed: error: undefined symbol: __declspec(dllimport) bool __cdecl MNN::AUDIO::save

  #3850 opened 2 months ago by quic-zhanweiw
  3

• Build MNN for Windows on Arm: failed to build it with MSVC cl.exe compile.

  #3853 opened 2 months ago by quic-zhanweiw
  3

• 请问MNN支持在手机部署MoviNet视频处理模型吗？5维输入

  #3867 opened 2 months ago by Young1403
  1

• 使用MNN::Express::Module推理kokoro语音合成，同样的input推理得到的结果不同

  #3868 opened 2 months ago by adi-hld
  1

• python3 build_deps.py llm,opencl,openmp,vulkan fail

  #3872 opened 2 months ago by andyt9527
  2

• Python API Float32 CPU前向推理数值稳定性问题

  #3880 opened 2 months ago by zpeng11
  1

• 请问是否调用高通SoC的NPU部分？

  #3884 opened 2 months ago by TriDefender
  2

• Ace-step and Wan 2.1 2.2 support

  #3886 opened 2 months ago by Amin456789
  0

• Delay after loading model and pressing back button

  #3887 opened 2 months ago by shhrayus
  0

• 在开启api网络服务后一定会闪退2次

  #3834 opened 2 months ago by Xiaokun19
  5

• 控制思考功能异常

  #3878 opened 2 months ago by bosco163
  5

• mnn 模型内存占用显著超过同规格 mlx 模型，请社区同学看下是否是配置问题

  #3883 opened 2 months ago by Remember2015
  3

• 为什么在x86电脑上推理fastvlm出现LLM init error错误

  #3876 opened 2 months ago by guoguo1314
  6

• 特定 win gpu 型号，opencl 报错 -5 然后崩溃

  #3870 opened 2 months ago by NikoLXQ
  2

• MNN opencl后端运行崩溃

  #3882 opened 2 months ago by Yoline777
  0

• 如何理解 memory 和 dynamic_option 参数？

  #3877 opened 2 months ago by YSF-A
  2

• 关于clip模型的一些疑问？

  #3881 opened 2 months ago by pange1802703882
  0

• _StridedSliceWrite在iOS MNN_FORWARD_CPU下的某些工作场景下，会出现复制不全

  #3856 opened 2 months ago by xiangyuanzhi
  3

• Slow first-token inference with Qwen3-0.6B on MNN

  #3859 opened 2 months ago by tianzunmengyan
  2

• (request) Add support for minicpm-v-4.5

  #3875 opened 2 months ago by shhrayus
  0

• llm_demo中文聊天崩溃

  #3855 opened 2 months ago by leapar
  0

• Tables created by the LLM misaligned in 0.7.2

  #3869 opened 2 months ago by shhrayus
  0

• LLM的表达式前向计算错误

  #3866 opened 2 months ago by huluhuluu
  2

• 前端页面

  #3864 opened 2 months ago by liuji308
  1

• video input not available on video models on 0.7.2

  #3862 opened 2 months ago by shhrayus
  1

• [Bug] Cannot use model(s) imported from outside

  #3854 opened 2 months ago by funnyMonkey18
  2

• MnnTaoAvatar 崩溃日志

  #3860 opened 2 months ago by BrickWorkerKing
  1

• Memory cannot be released on iOS

  #3858 opened 2 months ago by petyou
  0

• MNN是否支持vulkan的pipeline cache加速

  #3857 opened 2 months ago by junka
  0

• Please add internlm/InternVL3_5-1B-Instruct

  #3840 opened 2 months ago by abskvrm
  1

• gpt oss 20b crash also Qwen30b A3B

  #3849 opened 2 months ago by Synthacon
  2

• 在android上设置Precision_Low 模型推理失效

  #3848 opened 2 months ago by HankageCl
  1

• windows静态编译无法使用VULKAN

  #3837 opened 2 months ago by lvjinjie
  1

• python环境推理kokoro的mnn模型报错。

  #3844 opened 2 months ago by adi-hld
  2

• Gemma-3-270M: Model Forward Pass Fails with Compute Shape Error

  #3839 opened 2 months ago by ITM527
  0

• cross-compile pymnn for aarch64 devices

  #3838 opened 2 months ago by andyt9527
  0

• 请问对qwen3-0.6B,先与lora合并，再导出可以吗

  #3836 opened 2 months ago by noobccgg
  2

• ONNX::ConvInteger这个算子尚未支持吗

  #3835 opened 3 months ago by xanecdotex
  0

• .

  #3833 opened 3 months ago by Kivrar
  0