Code for paper:

Does Unsupervised Architecture Representation Learning Help Neural Architecture Search?
Shen Yan, Yu Zheng, Wei Ao, Xiao Zeng, Mi Zhang.
NeurIPS 2020.

Top: The supervision signal for representation learning comes from the accuracies of architectures selected by the search strategies. Bottom (ours): Disentangling architecture representation learning and architecture search through unsupervised pre-training.

The repository is built upon pytorch_geometric, pybnn, nas_benchmarks, bananas.

• NVIDIA GPU, Linux, Python3

pip install -r requirements.txt

Install nasbench and download nasbench_only108.tfrecord under ./data folder.

python preprocessing/gen_json.py

Data will be saved in ./data/data.json.

bash models/pretraining_nasbench101.sh

The pretrained model will be saved in ./pretrained/dim-16/.

bash run_scripts/extract_arch2vec.sh

The extracted arch2vec will be saved in ./pretrained/dim-16/.

Alternatively, you can download the pretrained arch2vec on NAS-Bench-101.

bash run_scripts/run_reinforce_supervised.sh 
bash run_scripts/run_reinforce_arch2vec.sh 

Search results will be saved in ./saved_logs/rl/dim16

Generate json file:

python plot_scripts/plot_reinforce_search_arch2vec.py 

bash run_scripts/run_dngo_supervised.sh 
bash run_scripts/run_dngo_arch2vec.sh 

Search results will be saved in ./saved_logs/bo/dim16.

Generate json file:

python plot_scripts/plot_dngo_search_arch2vec.py

python plot_scipts/plot_nasbench101_comparison.py

Download the search results from search_logs.

python plot_scripts/plot_cdf.py

Download the NAS-Bench-201-v1_0-e61699.pth under ./data folder.

python preprocessing/nasbench201_json.py

Data corresponding to the three datasets in NAS-Bench-201 will be saved in folder ./data/ as cifar10_valid_converged.json, cifar100.json, ImageNet16_120.json.

bash models/pretraining_nasbench201.sh

The pretrained model will be saved in ./pretrained/dim-16/.

Note that the pretrained model is shared across the 3 datasets in NAS-Bench-201.

bash run_scripts/extract_arch2vec_nasbench201.sh

The extracted arch2vec will be saved in ./pretrained/dim-16/ as cifar10_valid_converged-arch2vec.pt, cifar100-arch2vec.pt and ImageNet16_120-arch2vec.pt.

Alternatively, you can download the pretrained arch2vec on NAS-Bench-201.

CIFAR-10: ./run_scripts/run_reinforce_arch2vec_nasbench201_cifar10_valid.sh
CIFAR-100: ./run_scripts/run_reinforce_arch2vec_nasbench201_cifar100.sh
ImageNet-16-120: ./run_scripts/run_reinforce_arch2vec_nasbench201_ImageNet.sh

CIFAR-10: ./run_scripts/run_bo_arch2vec_nasbench201_cifar10_valid.sh
CIFAR-100: ./run_scripts/run_bo_arch2vec_nasbench201_cifar100.sh
ImageNet-16-120: ./run_scripts/run_bo_arch2vec_nasbench201_ImageNet.sh

python ./plot_scripts/summarize_nasbench201.py

The corresponding table will be printed to the console.

CIFAR-10 can be automatically downloaded by torchvision, ImageNet needs to be manually downloaded (preferably to a SSD) from http://image-net.org/download.

python preprocessing/gen_isomorphism_graphs.py

Data will be saved in ./data/data_darts_counter600000.json.

Alternatively, you can download the extracted data_darts_counter600000.json.

bash models/pretraining_darts.sh

The pretrained model is saved in ./pretrained/dim-16/.

bash run_scripts/extract_arch2vec_darts.sh

The extracted arch2vec will be saved in ./pretrained/dim-16/arch2vec-darts.pt.

Alternatively, you can download the pretrained arch2vec on DARTS search space.

bash run_scripts/run_reinforce_arch2vec_darts.sh

logs will be saved in ./darts-rl/.

Final search result will be saved in ./saved_logs/rl/dim16.

bash run_scripts/run_bo_arch2vec_darts.sh

logs will be saved in ./darts-bo/ .

Final search result will be saved in ./saved_logs/bo/dim16.

python darts/cnn/train.py --auxiliary --cutout --arch arch2vec_rl --seed 1
python darts/cnn/train.py --auxiliary --cutout --arch arch2vec_bo --seed 1

• Expected results (RL): 2.60% test error with 3.3M model params.
• Expected results (BO): 2.48% test error with 3.6M model params.

python darts/cnn/train_imagenet.py  --arch arch2vec_rl --seed 1 
python darts/cnn/train_imagenet.py  --arch arch2vec_bo --seed 1

• Expected results (RL): 25.8% test error with 4.8M model params and 533M mult-adds.
• Expected results (RL): 25.5% test error with 5.2M model params and 580M mult-adds.

python darts/cnn/visualize.py arch2vec_rl
python darts/cnn/visualize.py arch2vec_bo

Download pretrained supervised embeddings of nasbench101 and nasbench201.

bash plot_scripts/drawfig5-nas101.sh # visualization on nasbench-101
bash plot_scripts/drawfig5-nas201.sh # visualization on nasbench-201
bash plot_scripts/drawfig5-darts.sh  # visualization on darts

The plots will be saved in ./graphvisualization.

Install nas_benchmarks and download nasbench_full.tfrecord under the same directory.

python plot_scripts/distance_comparison_fig3.py

bash plot_scripts/drawfig4.sh

the plots will be saved in ./density.

Download predicted_accuracy under saved_logs/.

python plot_scripts/pearson_plot_fig2.py

If you find this useful for your work, please consider citing:

@InProceedings{yan2020arch,
  title = {Does Unsupervised Architecture Representation Learning Help Neural Architecture Search?},
  author = {Yan, Shen and Zheng, Yu and Ao, Wei and Zeng, Xiao and Zhang, Mi},
  booktitle = {NeurIPS},
  year = {2020}
}