ENAS CIFAR-10 Implementation in PyTorch

"Efficient Neural Architecture Search via Parameter Sharing" implementation in PyTorch

Includes code for CIFAR-10 image classification task.

Paper: https://arxiv.org/abs/1802.03268

Authors: Hieu Pham*, Melody Y. Guan*, Barret Zoph, Quoc V. Le, Jeff Dean

Prerequisites

• PyTorch 0.4.0+

CIFAR-10

To run the ENAS experiments on the micro search space, please use:

python train_search.py

TODO

• [x] Search in micro search space in CIFAR-10
• [ ] Train micro architecture from scratch in CIFAR-10
• [ ] Search in macro search space in CIFAR-10
• [ ] Train macro architecture from scratch in CIFAR-10

Acknowledgements

This implementation is based on

• melodyguan/enas
• carpedm20/ENAS-pytorch
• quark0/darts