{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/fbnetv2-differentiable-neural-architecture","title":"FBNetV2: Differentiable Neural Architecture Search for Spatial and Channel Dimensions","arxiv_id":"2004.05565","date":"2020-04-12","proceeding":"CVPR 2020 6","authors":["Alvin Wan","Xiaoliang Dai","Peizhao Zhang","Zijian He","Yuandong Tian","Saining Xie","Bichen Wu","Matthew Yu","Tao Xu","Kan Chen","Peter Vajda","Joseph E. Gonzalez"],"abstract":"Differentiable Neural Architecture Search (DNAS) has demonstrated great success in designing state-of-the-art, efficient neural networks. However, DARTS-based DNAS's search space is small when compared to other search methods', since all candidate network layers must be explicitly instantiated in memory. To address this bottleneck, we propose a memory and computationally efficient DNAS variant: DMaskingNAS. This algorithm expands the search space by up to $10^{14}\\times$ over conventional DNAS, supporting searches over spatial and channel dimensions that are otherwise prohibitively expensive: input resolution and number of filters. We propose a masking mechanism for feature map reuse, so that memory and computational costs stay nearly constant as the search space expands. Furthermore, we employ effective shape propagation to maximize per-FLOP or per-parameter accuracy. The searched FBNetV2s yield state-of-the-art performance when compared with all previous architectures. With up to 421$\\times$ less search cost, DMaskingNAS finds models with 0.9% higher accuracy, 15% fewer FLOPs than MobileNetV3-Small; and with similar accuracy but 20% fewer FLOPs than Efficient-B0. Furthermore, our FBNetV2 outperforms MobileNetV3 by 2.6% in accuracy, with equivalent model size. FBNetV2 models are open-sourced at https://github.com/facebookresearch/mobile-vision.","url_abs":"https://arxiv.org/abs/2004.05565v1","url_pdf":"https://arxiv.org/pdf/2004.05565v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"fbnetv2-differentiable-neural-architecture","repo_url":"https://github.com/facebookresearch/mobile-vision","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"caffe2","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"architecture-search","task_name":"Neural Architecture Search"}],"methods":[{"method_slug":"dnas","method_name":"DNAS"},{"method_slug":"gumbel-softmax","method_name":"Gumbel Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/neural-architecture-search-on-imagenet","task":"Neural Architecture Search","dataset":"ImageNet","model":"FBNetV2-L1","rank_in_archive_order":70,"of":135,"metrics":{"Accuracy":"77.2","MACs":"325M","Top-1 Error Rate":"22.8"},"uses_additional_data":false},{"leaderboard":"/sota/neural-architecture-search-on-imagenet","task":"Neural Architecture Search","dataset":"ImageNet","model":"FBNetV2-F4","rank_in_archive_order":96,"of":135,"metrics":{"Accuracy":"76.0","MACs":"238M","Top-1 Error Rate":"24.0"},"uses_additional_data":false},{"leaderboard":"/sota/neural-architecture-search-on-imagenet","task":"Neural Architecture Search","dataset":"ImageNet","model":"FBNetV2-F3","rank_in_archive_order":125,"of":135,"metrics":{"Accuracy":"73.2","MACs":"126M","Top-1 Error Rate":"26.8"},"uses_additional_data":false},{"leaderboard":"/sota/neural-architecture-search-on-imagenet","task":"Neural Architecture Search","dataset":"ImageNet","model":"FBNetV2-F1","rank_in_archive_order":130,"of":135,"metrics":{"Accuracy":"68.3","MACs":"56M","Top-1 Error Rate":"31.7"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2004.05565","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}