MagicSource

Yeah sure, just wonder tensorflow is onnx the best to deploy model.

@guyjacob Link dead again man, why not provide a google drive link?????

``` apt update -y The operation couldn’t be completed. Unable to locate a Java Runtime. Please visit http://www.java.com for information on installing Java. make: *** [apt.done] Error 1 ``` mac...

@yt605155624 请教一下，为什么有了g2pw 就不需要多因子预测了，比如下面的句子能预测对马： ``` 孩子，别吃了，这里的肉脏，走，跟我去太平间 ```

@MoonWolf9067 single line lidar can not generate anything, 16 lines lidar should work but with less accuracy

@edgarriba I don't know which one is right, but these are old code from VPoser, a human prior project. I don't know why they give me different result.

@edgarriba for vposer, they using old torchgeometry get a right result. In my reasoning, so, if new kornia gives a different result, then the "right" result I want is different....

@amonszpart thanks for your investigation. Can u test right with torchgeometry old code from directly and than compare with kornia new code? above code shows result are not equal. odl...

@robin1001 Hi, macoS m1 can not built.

我也感觉训练有点问题，貌似loss难以降低． 可以试一下这个: https://github.com/Okery/YOLOv5-PyTorch