Papers › FBNetV3: Joint Architecture-Recipe Search using Predictor Pretraining

FBNetV3: Joint Architecture-Recipe Search using Predictor Pretraining

3 Jun 2020CVPR 2021 1arXiv:2006.02049archive 2025-07-28

Xiaoliang Dai, Alvin Wan, Peizhao Zhang, Bichen Wu, Zijian He, Zhen Wei, Kan Chen, Yuandong Tian, Matthew Yu, Peter Vajda, Joseph E. Gonzalez

Neural Architecture Search (NAS) yields state-of-the-art neural networks that outperform their best manually-designed counterparts. However, previous NAS methods search for architectures under one set of training hyper-parameters (i.e., a training recipe), overlooking superior architecture-recipe combinations. To address this, we present Neural Architecture-Recipe Search (NARS) to search both (a) architectures and (b) their corresponding training recipes, simultaneously. NARS utilizes an accuracy predictor that scores architecture and training recipes jointly, guiding both sample selection and ranking. Furthermore, to compensate for the enlarged search space, we leverage "free" architecture statistics (e.g., FLOP count) to pretrain the predictor, significantly improving its sample efficiency and prediction reliability. After training the predictor via constrained iterative optimization, we run fast evolutionary searches in just CPU minutes to generate architecture-recipe pairs for a variety of resource constraints, called FBNetV3. FBNetV3 makes up a family of state-of-the-art compact neural networks that outperform both automatically and manually-designed competitors. For example, FBNetV3 matches both EfficientNet and ResNeSt accuracy on ImageNet with up to 2.0x and 7.1x fewer FLOPs, respectively. Furthermore, FBNetV3 yields significant performance gains for downstream object detection tasks, improving mAP despite 18% fewer FLOPs and 34% fewer parameters than EfficientNet-based equivalents.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

rwightman/pytorch-image-models mentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Neural Architecture SearchObject Detectionobject-detection

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Neural Architecture Search ImageNet FBNetV3-G Accuracy 82.3 #5 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-G MACs 2.0G #5 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-G Top-1 Error Rate 17.7 #5 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-E Accuracy 80.4 #18 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-E MACs 752M #18 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-E Top-1 Error Rate 19.6 #18 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-C Accuracy 79.6 #29 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-C MACs 544M #29 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-C Top-1 Error Rate 20.4 #29 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-A Accuracy 78.0 #53 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-A MACs 343M #53 of 135 Archive leaderboard report
Neural Architecture Search ImageNet FBNetV3-A Top-1 Error Rate 22.0 #53 of 135 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections