{"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/large-batch-training-of-convolutional","title":"Large Batch Training of Convolutional Networks","arxiv_id":"1708.03888","date":"2017-08-13","proceeding":null,"authors":["Yang You","Igor Gitman","Boris Ginsburg"],"abstract":"A common way to speed up training of large convolutional networks is to add\ncomputational units. Training is then performed using data-parallel synchronous\nStochastic Gradient Descent (SGD) with mini-batch divided between computational\nunits. With an increase in the number of nodes, the batch size grows. But\ntraining with large batch size often results in the lower model accuracy. We\nargue that the current recipe for large batch training (linear learning rate\nscaling with warm-up) is not general enough and training may diverge. To\novercome this optimization difficulties we propose a new training algorithm\nbased on Layer-wise Adaptive Rate Scaling (LARS). Using LARS, we scaled Alexnet\nup to a batch size of 8K, and Resnet-50 to a batch size of 32K without loss in\naccuracy.","url_abs":"http://arxiv.org/abs/1708.03888v3","url_pdf":"http://arxiv.org/pdf/1708.03888v3.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":"large-batch-training-of-convolutional","repo_url":"https://github.com/NUS-HPC-AI-Lab/LARS-ImageNet-PyTorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/facebookresearch/ClassyVision","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/facebookresearch/vissl","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/frgfm/Holocron","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/jxbz/fromage","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/kakaobrain/torchlars","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/krafton-ai/mini-batch-cl","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/lulor/ood-class-separation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/noahgolmant/pytorch-lars","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/optimization-ai/sogclr","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/sylviesss/pytorch-simclr","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"large-batch-training-of-convolutional","repo_url":"https://github.com/PaddlePaddle/FleetX/blob/develop/examples/resnet/train_fleet_lars.py","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"paddle","reach":null}],"tasks":[{"task_slug":null,"task_name":"8k"}],"methods":[{"method_slug":"lars","method_name":"LARS"},{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[{"slug":"lars","name":"LARS","full_name":"LARS"}],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1708.03888","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1708.03888"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/facebookresearch/vissl","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/sylviesss/pytorch-simclr","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/frgfm/Holocron","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/jxbz/fromage","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/PaddlePaddle/FleetX/blob/develop/examples/resnet/train_fleet_lars.py","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/lulor/ood-class-separation","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/facebookresearch/ClassyVision","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/NUS-HPC-AI-Lab/LARS-ImageNet-PyTorch","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/noahgolmant/pytorch-lars","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/optimization-ai/sogclr","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/krafton-ai/mini-batch-cl","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kakaobrain/torchlars","reach":{"status":"unanswered"}},{"provenance":"deterministic:regex_extraction","url":"https://github.com/BVLC/caffe","reach":{"status":"unanswered"}},{"provenance":"deterministic:regex_extraction","url":"https://github.com/borisgin/nvcaffe-0.16","reach":{"status":"unanswered"}}],"summary":{"ran":4,"unverified":1},"by_repo_kind":{"listed":{"samples":5,"ran":4,"repositories":5}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":1,"samples":[{"code_sha256_prefix":"4c2823d62e6b5ca1","entry":"LARS","repo":"frgfm/Holocron","repo_kind":"listed","path":"holocron/optim/lars.py","file_url":"https://github.com/frgfm/Holocron/blob/HEAD/holocron/optim/lars.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"4c2823d62e6b5ca1"}},{"code_sha256_prefix":"350b8435032a7c95","entry":"LARS","repo":"krafton-ai/mini-batch-cl","repo_kind":"listed","path":"sogclr/optimizer.py","file_url":"https://github.com/krafton-ai/mini-batch-cl/blob/HEAD/sogclr/optimizer.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"350b8435032a7c95"}},{"code_sha256_prefix":"2b456b26b38fabeb","entry":"LARS","repo":"noahgolmant/pytorch-lars","repo_kind":"listed","path":"lars.py","file_url":"https://github.com/noahgolmant/pytorch-lars/blob/HEAD/lars.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"2b456b26b38fabeb"}},{"code_sha256_prefix":"f189d069869671ac","entry":"Lars","repo":"jxbz/fromage","repo_kind":"listed","path":"generate-cifar/lars.py","file_url":"https://github.com/jxbz/fromage/blob/HEAD/generate-cifar/lars.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"f189d069869671ac"}},{"code_sha256_prefix":"af8abbe0acb62ad6","entry":"Lars","repo":"NUS-HPC-AI-Lab/LARS-ImageNet-PyTorch","repo_kind":"listed","path":"lars.py","file_url":"https://github.com/NUS-HPC-AI-Lab/LARS-ImageNet-PyTorch/blob/HEAD/lars.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"af8abbe0acb62ad6"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}