LayerNAS: Neural Architecture Search in Polynomial Complexity

Benchmark Model Rank Results
neural-architecture-search-on-imagenetLayerNAS-220M–Top-1 Error Rate: 24.4Params: 5.1MMACs: 229M
neural-architecture-search-on-imagenetLayerNAS-300M–Top-1 Error Rate: 22.9Params: 5.2MMACs: 322M
neural-architecture-search-on-imagenetLayerNAS-600M–Top-1 Error Rate: 21.4Params: 9.7MMACs: 627M
neural-architecture-search-on-imagenetLayerNAS-60M–Top-1 Error Rate: 31Params: 3.7MMACs: 61M
neural-architecture-search-on-nas-bench-101LayerNAS–Accuracy (%): 94.26%
neural-architecture-search-on-nats-bench-size-cifar-10LayerNAS–Test Accuracy: 93.2Validation Accuracy: 0.844
neural-architecture-search-on-nats-bench-size-cifar-100LayerNAS–Test Accuracy: 70.64Validation Accuracy: 60.67
neural-architecture-search-on-nats-bench-size-imagenet16-120LayerNAS–Test Accuracy: 45.37Validation Accuracy: 38.12
neural-architecture-search-on-nats-bench-topology-cifar-10LayerNAS–Test Accuracy: 94.34±0.12
neural-architecture-search-on-nats-bench-topology-cifar-100LayerNAS–Test Accuracy: 73.01±0.63
neural-architecture-search-on-nats-bench-topology-imagenet16-120LayerNAS–Test Accuracy: 46.58±0.59