carpedm20/ENAS-pytorch
A PyTorch implementation of ENAS that reduces NAS computational cost by 1000x through parameter sharing to discover optimal RNN and CNN architectures.
Not currently ranked — collecting fresh signals.
star history
This repository implements Efficient Neural Architecture Search (ENAS), an AutoML technique that uses a reinforcement learning controller to search for optimal neural network architectures. The controller learns to sample architectures from a shared computation graph, dramatically reducing GPU-hours compared to traditional NAS. It demonstrates state-of-the-art performance on Penn Treebank language modeling by discovering optimal recurrent cell structures.
Frequently asked
- What is carpedm20/ENAS-pytorch?
- A PyTorch implementation of ENAS that reduces NAS computational cost by 1000x through parameter sharing to discover optimal RNN and CNN architectures.
- Is ENAS-pytorch open source?
- Yes — carpedm20/ENAS-pytorch is open source, released under the Apache-2.0 license.
- What language is ENAS-pytorch written in?
- carpedm20/ENAS-pytorch is primarily written in Python.
- How popular is ENAS-pytorch?
- carpedm20/ENAS-pytorch has 2.7k stars on GitHub.
- Where can I find ENAS-pytorch?
- carpedm20/ENAS-pytorch is on GitHub at https://github.com/carpedm20/ENAS-pytorch.