paper-with-me

Network Pruning 벤치마크

Network Pruning on ImageNet

17개 결과 · ⬇ CSV · JSON

Accuracy

57.5 62.82 68.15 73.47 78.79 2016-02 2026-09 SqueezeNet (6-bit Deep Compression) — 57.5 (2016-02-24) ResNet50-2.3 GFLOPs — 78.79 (2016-08-31) ResNet50-1.5 GFLOPs — 78.07 (2016-08-31) ResNet50-1G FLOPs — 76.376 (2016-08-31) TAS-pruned ResNet-50 — 76.2 (2019-05-23) ResNet50 2.5 GFLOPS — 78.0 (2020-02-19) ResNet50 2.0 GFLOPS — 77.7 (2020-02-19) ResNet50-3G FLOPs — 77.1 (2020-07-06) ResNet50-2G FLOPs — 76.4 (2020-07-06) ResNet50-1G FLOPs — 74.2 (2020-07-06) ResNet50-1G FLOPs — 74.2 (2020-07-06) MobileNetV1-50% FLOPs — 70.7 (2020-07-06) ResNet50 — 75.59 (2021-05-07) ResNet50 — 73.14 (2021-06-23) RegX-1.6G — 77.97 (2021-08-02) MobileNetV2 — 73.42 (2021-08-02) AgenticPruner — 77.04 (2026-01-18) SqueezeNet (6-bit Deep Compression) — 57.5 (2016-02-24) ResNet50-2.3 GFLOPs — 78.79 (2016-08-31)
RankModel AccuracyGFLOPsMParams PaperCodeYear
1 ResNet50-2.3 GFLOPs 78.792.33514.811 Pruning Filters for Efficient ConvNets PaddlePaddle/PaddleOCR · VainF/Torch-Pruning · he-y/filter-pruning-geometric-median · +18 2016
2 ResNet50-1.5 GFLOPs 78.071.63510.511 Pruning Filters for Efficient ConvNets PaddlePaddle/PaddleOCR · VainF/Torch-Pruning · he-y/filter-pruning-geometric-median · +18 2016
3 ResNet50 2.5 GFLOPS 78.02.5– Knapsack Pruning with Inner Distillation yoniaflalo/knapsack_pruning 2020
4 RegX-1.6G 77.971.5889.3 Group Fisher Pruning for Practical Network Compression jshilong/FisherPruning · Ben-Louis/FisherPruning-Pytorch 2021
5 ResNet50 2.0 GFLOPS 77.702– Knapsack Pruning with Inner Distillation yoniaflalo/knapsack_pruning 2020
6 ResNet50-3G FLOPs 77.1–– EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning anonymous47823493/EagleEye 2020
7 AgenticPruner 자동 추출 77.04–– AgenticPruner: MAC-Constrained Neural Network Compression via LLM-Driven Strategy Search 2026
8 ResNet50-2G FLOPs 76.4–– EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning anonymous47823493/EagleEye 2020
9 ResNet50-1G FLOPs 76.3761.0756.954 Pruning Filters for Efficient ConvNets PaddlePaddle/PaddleOCR · VainF/Torch-Pruning · he-y/filter-pruning-geometric-median · +18 2016
10 TAS-pruned ResNet-50 76.202.3– Network Pruning via Transformable Architecture Search D-X-Y/GDAS · D-X-Y/NAS-Projects · D-X-Y/AutoDL-Projects · +1 2019
11 ResNet50 75.59–– Network Pruning That Matters: A Case Study on Retraining Variants lehduong/NPTM 2021
12 ResNet50-1G FLOPs 74.2–– EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning anonymous47823493/EagleEye 2020
12 ResNet50-1G FLOPs 74.2–– EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning anonymous47823493/EagleEye 2020
14 MobileNetV2 73.420.293.31 Group Fisher Pruning for Practical Network Compression jshilong/FisherPruning · Ben-Louis/FisherPruning-Pytorch 2021
15 ResNet50 73.14–– AC/DC: Alternating Compressed/DeCompressed Training of Deep Neural Networks IST-DASLab/ACDC · IST-DASLab/sparseprop 2021
16 MobileNetV1-50% FLOPs 70.7–– EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning anonymous47823493/EagleEye 2020
17 SqueezeNet (6-bit Deep Compression) 57.5%–1.24 SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size pytorch/vision · PaddlePaddle/PaddleClas · osmr/imgclsmob · +56 2016
1–17 / 17 페이지당 10 20 50 100