Jingbo Lin

Hello @brooklyn1900 I have tried LovaszSoftmaxLoss on different datasets. In one of them, I have the same problem with you, loss was always swing around at 0.30 and got nan...

> Is search config used in obtain_search_args? > And I found that AutoML is a little consistent with [official_deeplab ](https://github.com/tensorflow/models/tree/master/research/deeplab) as follows > > ```python > if prev_layer is None:...

@HankKung Hi, Thanks for your hard work! >This is the one I've added pre_pre_input for those edge tensors. Does the way you add pre_pre_input for those edge tensors like the...

Oh, that's very helpful! I made a mistake in the above list. In the case of the two-layer stem: ``` 4-1 output of stem0, output of stem1 4-2 output of...

Just start to retrain after the ~40 epochs searching. searching for 40 epochs -> retrain from scratch

Thanks for your reply @Annbless, waiting for your good news. But if you are too busy to update this issue, you could give me some further hints to solve this...

depthwise_conv2d implemented using nn.Conv2d(in channels, out channels, ksize, groups=inchannels) also performed badly in my experiment, is there something need to config specifically? I am using pytorch 1.0.1