| Sharpness-Aware Minimization for Efficiently Improving Generalization | ✓ Link | 97.10 | 2.90% | | | EffNet-L2 (SAM) | 2020-10-03 |
| Big Transfer (BiT): General Visual Representation Learning | ✓ Link | 96.62 | 3.38% | | | BiT-L (ResNet) | 2019-12-24 |
| A Continual Development Methodology for Large-scale Multitask Dynamic ML Systems | ✓ Link | 95.5 | | | | µ2Net+ (ViT-L/16) | 2022-09-15 |
| An Evolutionary Approach to Dynamic Introduction of Tasks in Large-scale Multitask Learning Systems | ✓ Link | 95.3 | | | | µ2Net (ViT-L/16) | 2022-05-25 |
| Big Transfer (BiT): General Visual Representation Learning | ✓ Link | 94.47 | 5.53% | | | BiT-M (ResNet) | 2019-12-24 |
| Compounding the Performance Improvements of Assembled Techniques in a Convolutional Neural Network | ✓ Link | 94.3% | 5.7 | | | Assemble-ResNet-FGVC-50 | 2020-01-17 |
| Neural Architecture Transfer | ✓ Link | 94.3 | 5.7% | 744M | 8.5M | NAT-M4 | 2020-05-12 |
| Neural Architecture Transfer | ✓ Link | 94.1 | 5.9% | 471M | 5.7M | NAT-M3 | 2020-05-12 |
| Neural Architecture Transfer | ✓ Link | 93.5 | 6.5% | 306M | 5.5M | NAT-M2 | 2020-05-12 |
| When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations | ✓ Link | 93.3 | | | | ResNet-152-SAM | 2021-06-03 |
| When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations | ✓ Link | 93.1 | | | | ViT-B/16- SAM | 2021-06-03 |
| When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations | ✓ Link | 92.9 | | | | ViT-S/16- SAM | 2021-06-03 |
| When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations | ✓ Link | 92.5 | | | | Mixer-B/16- SAM | 2021-06-03 |
| When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations | ✓ Link | 91.6 | | | | ResNet-50-SAM | 2021-06-03 |
| When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations | ✓ Link | 88.7 | | | | Mixer-S/16- SAM | 2021-06-03 |
| Stochastic Subsampling With Average Pooling | | 86.011 | | | | SE-ResNet-101 (SAP) | 2024-09-25 |
| How to Use Dropout Correctly on Residual Networks with Batch Normalization | ✓ Link | 85.5897 | | | | PreResNet-101 | 2023-02-13 |
| On the Ideal Number of Groups for Isometric Gradient Propagation | | 77.076 | | | | ResNet-101 (ideal number of groups) | 2023-02-07 |
| An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale | ✓ Link | | 6.2% | | | ViT-B/16 | 2020-10-22 |