Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Meta Architecture Search

This repository contains the trained deep neural network architectures and weights, and training code for the BASE paper.

If you find this useful, or if you use it in your work, please cite:

    author = {Albert Shaw and Wei Wei and Weiyang Liu and Le Song and Bo Dai},
    title = {Meta Architecture Search},
    booktitle = {NeurIPS},
    year = {2019}


Python >= 3.6.0
PyTorch >= 1.0.1
torchvision >= 0.2.2
numpy >= 1.15.4


  1. Install the required packages.
  2. Clone this repository.
  3. Download and extract the Imagenet dataset to data/imagenet.


Use the script to evaluate the models. Logs are saved into the logs folder.

Training the networks on cifar10 requires one 1080 TI and 2 1080 TI to train Imagenet.

To evaluate the trained networks run:
python3 --model=get_cifar_tuned_model(True) --gpu 1 --eval 1
python3 --model=get_imagenet_tuned_model(True) --gpu 1 --eval 1

To train the found networks run:
python3 --model=get_cifar_tuned_model(False) --gpu 1
python3 --model=get_imagenet_tuned_model(False) --gpu 1

You can’t perform that action at this time.