quark0 / darts

Differentiable architecture search for convolutional and recurrent networks
https://arxiv.org/abs/1806.09055
Apache License 2.0
3.92k stars 843 forks source link

Is this a normal case? #133

Closed mrk1992 closed 4 years ago

mrk1992 commented 4 years ago

I run this command

python train_search.py

and get a these results

image

Is it right ?

AlexeyGB commented 4 years ago

I have the same problem. It seems that something is wrong with torch Dataloader's multiprocessing. At least setting num_workers=0 for both train_queue and valid_queue makes the script work properly for me. But this crutch can not be considered as a satisfactory solution.

mrk1992 commented 4 years ago

@AlexeyGB Go to https://github.com/yuhuixu1993/PC-DARTS/issues/31#issuecomment-585484767