Prioritized Architecture Sampling with Monto-Carlo Tree Search
NAS-Bench-Macro
This repository includes the benchmark and code for NAS-Bench-Macro in paper “Prioritized Architecture Sampling with Monto-Carlo Tree Search”, CVPR2021.
NAS-Bench-Macro is a NAS benchmark on macro search space. The NAS-Bench-Macro consists of 6561 networks and their test accuracies, parameters, and FLOPs on CIFAR-10 dataset.
Each architecture in NAS-Bench-Macro is trained from scratch isolatedly.
Benchmark
All the evaluated architectures are stored in file nas-bench-macro_cifar10.json with the following format:
{
arch1: {
test_acc: [float, float, float], // the test accuracies of three independent training
mean_acc: float, // mean accuracy
std: float, // the standard deviation of test accuracies
params: int, // parameters
flops: int, // FLOPs
},
arch2: ......
}
Search Space
The search space of NAS-Bench-Macro is conducted with 8 searching layers; each layer contains 3