{"url":"/method/resnet-rs","slug":"resnet-rs","name":"ResNet-RS","full_name":"ResNet-RS","full_name_withheld":false,"description_markdown":"**ResNet-RS** is a family of [ResNet](https://paperswithcode.com/method/resnet) architectures that are 1.7x faster than [EfficientNets](https://paperswithcode.com/method/efficientnet) on TPUs, while achieving similar accuracies on ImageNet. The authors propose two new scaling strategies: (1) scale model depth in regimes where overfitting can occur (width scaling is preferable otherwise); (2) increase image resolution more slowly than previously recommended.\r\n\r\nAdditional improvements include the use of a [cosine learning rate schedule](https://paperswithcode.com/method/cosine-annealing), [label smoothing](https://paperswithcode.com/method/label-smoothing), [stochastic depth](https://paperswithcode.com/method/stochastic-depth), [RandAugment](https://paperswithcode.com/method/randaugment), decreased [weight decay](https://paperswithcode.com/method/weight-decay), [squeeze-and-excitation](https://paperswithcode.com/method/squeeze-and-excitation-block) and the use of the [ResNet-D](https://paperswithcode.com/method/resnet-d) architecture.","description_state":"present","introduced_year":null,"introduced_by":{"title":"Revisiting ResNets: Improved Training and Scaling Strategies","paper":"/paper/revisiting-resnets-improved-training-and","first_author":"Irwan Bello","n_authors":8,"url_abs":null,"archive_paper_url":"https://paperswithcode.com/paper/revisiting-resnets-improved-training-and"},"source":{"url":"https://arxiv.org/abs/2103.07579v1","title":"Revisiting ResNets: Improved Training and Scaling Strategies","url_on_a_paper_host":true},"code_snippet_url":null,"code_snippet_url_on_a_code_host":false,"categories":[{"area":"Computer Vision","area_id":"computer-vision","collection":"Convolutional Neural Networks","url":"/methods/category/convolutional-neural-networks","pwc_aliases":[]}],"n_papers_tagged":2,"archive_num_papers":2,"papers_newest_first":[{"paper":"/paper/revisiting-3d-resnets-for-video-recognition","title":"Revisiting 3D ResNets for Video Recognition","date":"2021-09-03","arxiv_id":"2109.01696","n_code_links":5,"syntology":null},{"paper":"/paper/revisiting-resnets-improved-training-and","title":"Revisiting ResNets: Improved Training and Scaling Strategies","date":"2021-03-13","arxiv_id":"2103.07579","n_code_links":3,"syntology":null}],"papers_shown":2,"tasks":[{"task":"/task/action-classification","name":"Action Classification","papers":2},{"task":"/task/contrastive-learning","name":"Contrastive Learning","papers":1},{"task":"/task/document-image-classification","name":"Document Image Classification","papers":1},{"task":"/task/image-classification","name":"Image Classification","papers":1},{"task":"/task/semantic-object-interaction-classification","name":"Semantic Object Interaction Classification","papers":1},{"task":"/task/video-classification","name":"Video Classification","papers":1},{"task":"/task/video-recognition","name":"Video Recognition","papers":1}],"tasks_shown":7,"n_tasks":7,"usage_by_year":[{"year":"2021","papers":2}],"row_source":"methods_table","archive":{"source":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","archive_url":"https://paperswithcode.com/method/resnet-rs"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}