curtis.abcd

It will be almost same. you should check the alpha variable's value, if they seem to be converged, you are safe to use 49th epoch's genotype.

@inmoonlight Can you tell me how long it will take to solve this problem? I haven't signed up for a few weeks.

@marble25 Thanks, It seems to be resolved now. cc @inmoonlight

@ZheC Can you post the loss curve?

@ZheC at least you can tell the last loss value so that we can compare.

@ZheC Thanks for sharing!!

@trathpai I accidentally removed the file. I will update the new model weight If I have a chance to train it. But for now, I have no plan to do...

According to the author's email reply, I changed some parameters. ### With epoch 180 (paper(RandAugment)'s epoch) M\N | 1 | 2 | 3 -- | -- | -- | --...

https://github.com/tensorflow/tpu/blob/8462d083dd89489a79e3200bcc8d4063bf362186/models/official/efficientnet/autoaugment.py#L663 @BarretZoph Is this the code you mentioned? Some parts are mismatched with this code, I will change them and conduct experiments soon.

@BarretZoph Is this correct? https://github.com/tensorflow/tpu/blob/8462d083dd89489a79e3200bcc8d4063bf362186/models/official/efficientnet/autoaugment.py#L181 Since SolarizeAdd have a addition parameter with "0"(zero), this doesn't affect the image.