Open jia-zhuang opened 5 years ago
Hi, @jia-zhuang , I also met the same problem when training my 1-D signal. I tried many ways to adjust the parameters in my model-gc.json and reshape my input data, but the performance didn't seem to improve... Have you got some simple solution?
Hi, @RayShark . I even don't know how to config gc.json file for 1-D signal. Can you share it for reference?
For image features, e.g. 28x28, we can use FGWinLayer, what about sequence features, e.g. len=200, how to config gc.json file?