Skip to content
PyTorch implementation of PNASNet-5 on ImageNet
Branch: master
Clone or download
Latest commit c3c830d Dec 8, 2018


PyTorch implementation of PNASNet-5. Specifically, PyTorch code from this repository is adapted to completely match both my implemetation and the official implementation of PNASNet-5, both written in TensorFlow. This complete match allows the pretrained TF model to be exactly converted to PyTorch: see

If you use the code, please cite:

  author    = {Chenxi Liu and
               Barret Zoph and
               Maxim Neumann and
               Jonathon Shlens and
               Wei Hua and
               Li{-}Jia Li and
               Li Fei{-}Fei and
               Alan L. Yuille and
               Jonathan Huang and
               Kevin Murphy},
  title     = {Progressive Neural Architecture Search},
  booktitle = {European Conference on Computer Vision},
  year      = {2018}


  • TensorFlow 1.8.0 (for image preprocessing)
  • PyTorch 0.4.0
  • torchvision 0.2.1

Data and Model Preparation

  • Download the ImageNet validation set and move images to labeled subfolders. To do the latter, you can use this script. Make sure the folder val is under data/.
  • Download PNASNet.TF and follow its README to download the PNASNet-5_Large_331 pretrained model.
  • Convert TensorFlow model to PyTorch model:

Notes on Model Conversion

  • In both TensorFlow implementations, net[0] means prev and net[1] means prev_prev. However, in the PyTorch implementation, states[0] means prev_prev and states[1] means prev. I followed the PyTorch implemetation in this repository. This is why the 0 and 1 in PNASCell specification are reversed.
  • The default value of eps in BatchNorm layers is 1e-3 in TensorFlow and 1e-5 in PyTorch. I changed all BatchNorm eps values to 1e-3 (see to exactly match the TensorFlow pretrained model.
  • The TensorFlow pretrained model uses tf.image.resize_bilinear to resize the image (see I cannot find a python function that exactly matches this function's behavior (also see this thread and this post on this topic), so currently in I call TensorFlow to do the image preprocessing, in order to guarantee both models have the identical input.
  • When converting the model from TensorFlow to PyTorch (i.e., I use input image size of 323 instead of 331. This is because the 'SAME' padding in TensorFlow may differ from padding in PyTorch in some layers (see this link; basically TF may only pad 1 right and bottom, whereas PyTorch always pads 1 for all four margins). However, they behave exactly the same when image size is 323: conv0 does not have padding, so feature size becomes 161, then 81, 41, etc.
  • The exact conversion when image size is 323 is also corroborated by the following table:
Image Size Official TensorFlow Model Converted PyTorch Model
(331, 331) (0.829, 0.962) (0.828, 0.961)
(323, 323) (0.827, 0.961) (0.827, 0.961)



The last printed line should read:

Test: [50000/50000]	Prec@1 0.828	Prec@5 0.961
You can’t perform that action at this time.