Repository Issue Activity (beta)

#4795 [bug] tflite2MNNNet crashes on TFLite models whose tensor, buffer and input indices are out of range

#4833 [Android] Fatal crash on SanaSession.initNative - MNN Chat 0.8.3 (P0)

#4822 qwen3-tts-12hz-0.6b 在cpu上运行时间过长，并且不支持在opencl(gpu)上面运行

#4824 怎么保留Qwen等模型中的MatMul，不被转换成Convolution。

#4828 MNN在联发科天玑9500设备上开启SME2存在输出异常

#4798 [bug] MNNConvert crashes in Program::create and the ONNX/TFLite extra-op transforms when conversion leaves null inputs

#4335 Benchmark: MiniMind2-65M & Samsung A15 - "Infinity" Result.

#4796 [bug] MNNConvert dereferences null builtin_options and empty shape vectors on a malformed tflite model

#4804 AWQ quantizer crashes on multimodal models: get_position_ids() called without input_ids

#4788 iOS Metal: live-runtime counter leaves intermittent 600+ MiB spikes in staged SD pipeline

#4818 输出表格的时候总是会闪

#4802 Qwen2.5-Omni-3B/7B 在最新代码上opencl上推理崩溃

#4810 模型胡说八道

#4815 Android 内存泄露

#4816 Qwen2.5-Omni-7B 在最新的master代码上面推理崩溃

#4813 模型在流式输出时文本会上下跳动

#4814 sd生成有问题

#4812 模型不行，找更好的。

#4799 不支持NPU

#4811 Android API inference crash: SIGABRT in MNN::DeferBufferAllocator::apply() / MNN::Session::resize()

#4797 [bug] MNNConvert divides by zero when a model sets a convolution group, stride or dilation to 0

#4224 [BUG] Model Inference Crash / Application Force Close on MediaTek Helio G80 (8GB RAM)

#4793 希望 MNN 支持 MTP（Multi-Token Prediction）模型的导出和推理适配

#4800 手机卡屏问题。

#4801 鸿蒙系统适配计划