Bin Li

How did you install the Android SDK? You can find how we install it in MACE docker around [here](https://github.com/XiaoMi/mace/blob/2a56318518b8aacb15d2d1e522d3f630cbee772f/docker/mace-dev-lite/Dockerfile#L98).

Hi @apivovarov. This is due to the wrong limitations on the unsigned PD of hexagon firmware on SM8150. `hexagon_nn_execute` calls some APIs that unsigned PD is (wrongly) prohibited to access....

Well, hexagon_nn may use some APIs that unsigned PD happens to reject to access... [Redmi K30 5G](https://www.gsmarena.com/xiaomi_redmi_k30_5g-9979.php) is expected to be released next month...

Can you share your model and quantize_range_file so that we can test it?

Dupped with https://github.com/XiaoMi/mace/issues/286#issuecomment-449937586 What's the version of Tensorflow when running `python tools/converter.py convert --config_file=/path/to/your/model_deployment_file`?

那个回答的意思是1.8.0以上就支持`remove_control_dependencies`了，建议你都用同一个版本1.12吧

这是在什么情况下遇到的问题？请列出详细步骤和日志。

Thanks for the contribution! However, as it will change the default behavior for qualcomm socs, there is something more that needs to done to make the change work. 1. cmake...

我从https://github.com/tensorflow/tensorflow/blob/r1.12/tensorflow/contrib/lite/g3doc/models.md 下载了Mobilenet_v2_1.0_224_quant，并用mace master转换成功。 你用的哪个模型，mace是哪个版本？

可以提供一个能复现该问题的模型吗