NAS-Bench-360

A NAS Benchmark for Diverse Tasks


Neural architecture search (NAS) benchmarks and methods prioritize performance on well-studied tasks, e.g., image classification on CIFAR and ImageNet. To mitigate this bias, NAS-Bench-360 is a benchmark suite for evaluating state-of-the-art NAS methods on a diverse set of tasks. The selection spans different application domains, dataset sizes, problem dimensionalities, and learning objectives. Please contact Renbo Tu (renbo@cmu.edu) if you have questions or wish to contribute to the benchmark.

GitHub

Datasets




Citation

Please cite the following if you use NAS-Bench-360:

NAS-Bench-360: Benchmarking Diverse Tasks for Neural Architecture Search
Renbo Tu, Mikhail Khodak, Nicholas Roberts, Ameet Talwalkar

			@misc{tu2021nasbench360 
			      title={NAS-Bench-360: Benchmarking Diverse Tasks for Neural Architecture Search}, 
			      author={Renbo Tu and Mikhail Khodak and Nicholas Roberts and Ameet Talwalkar},
			      year={2021},    
			      eprint={2110.05668},
			      archivePrefix={arXiv},
			      primaryClass={cs.CV}
			      
			}


License

NAS-Bench-360 is released under the MIT license.