{"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/big-self-supervised-models-are-strong-semi","title":"Big Self-Supervised Models are Strong Semi-Supervised Learners","arxiv_id":"2006.10029","date":"2020-06-17","proceeding":"NeurIPS 2020 12","authors":["Ting Chen","Simon Kornblith","Kevin Swersky","Mohammad Norouzi","Geoffrey Hinton"],"abstract":"One paradigm for learning from few labeled examples while making best use of a large amount of unlabeled data is unsupervised pretraining followed by supervised fine-tuning. Although this paradigm uses unlabeled data in a task-agnostic way, in contrast to common approaches to semi-supervised learning for computer vision, we show that it is surprisingly effective for semi-supervised learning on ImageNet. A key ingredient of our approach is the use of big (deep and wide) networks during pretraining and fine-tuning. We find that, the fewer the labels, the more this approach (task-agnostic use of unlabeled data) benefits from a bigger network. After fine-tuning, the big network can be further improved and distilled into a much smaller one with little loss in classification accuracy by using the unlabeled examples for a second time, but in a task-specific way. The proposed semi-supervised learning algorithm can be summarized in three steps: unsupervised pretraining of a big ResNet model using SimCLRv2, supervised fine-tuning on a few labeled examples, and distillation with unlabeled examples for refining and transferring the task-specific knowledge. This procedure achieves 73.9% ImageNet top-1 accuracy with just 1% of the labels ($\\le$13 labeled images per class) using ResNet-50, a $10\\times$ improvement in label efficiency over the previous state-of-the-art. With 10% of labels, ResNet-50 trained with our method achieves 77.5% top-1 accuracy, outperforming standard supervised training with all of the labels.","url_abs":"https://arxiv.org/abs/2006.10029v2","url_pdf":"https://arxiv.org/pdf/2006.10029v2.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":"big-self-supervised-models-are-strong-semi","repo_url":"https://github.com/google-research/simclr","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"big-self-supervised-models-are-strong-semi","repo_url":"https://github.com/mariaauslander/capstone_fall20_irrigation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"big-self-supervised-models-are-strong-semi","repo_url":"https://github.com/nikheelpandey/TAUP","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"big-self-supervised-models-are-strong-semi","repo_url":"https://github.com/nikheelpandey/TAUP-PyTorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"big-self-supervised-models-are-strong-semi","repo_url":"https://github.com/parkinkon1/simclr","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"big-self-supervised-models-are-strong-semi","repo_url":"https://github.com/sayakpaul/PAWS-TF","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"big-self-supervised-models-are-strong-semi","repo_url":"https://github.com/serre-lab/prj_selfsup","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"big-self-supervised-models-are-strong-semi","repo_url":"https://github.com/ta9ryuWalrus/simclr","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"big-self-supervised-models-are-strong-semi","repo_url":"https://github.com/lightly-ai/lightly","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"self-supervised-image-classification","task_name":"Self-Supervised Image Classification"},{"task_slug":"semi-supervised-image-classification","task_name":"Semi-Supervised Image Classification"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"average-pooling","method_name":"Average Pooling"},{"method_slug":"batch-normalization","method_name":"Batch Normalization"},{"method_slug":"bottleneck-residual-block","method_name":"Bottleneck Residual Block"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"global-average-pooling","method_name":"Global Average Pooling"},{"method_slug":"kaiming-initialization","method_name":"Kaiming Initialization"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"residual-block","method_name":"Residual Block"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"simclrv2","method_name":"SimCLRv2"}],"datasets_introduced":[],"methods_introduced":[{"slug":"simclrv2","name":"SimCLRv2","full_name":"SimCLRv2"}],"results":[{"leaderboard":"/sota/self-supervised-image-classification-on","task":"Self-Supervised Image Classification","dataset":"ImageNet","model":"SimCLRv2 (ResNet-152 x3, SK)","rank_in_archive_order":28,"of":144,"metrics":{"Number of Params":"795M","Top 1 Accuracy":"79.8%","Top 5 Accuracy":"94.9%"},"uses_additional_data":false},{"leaderboard":"/sota/self-supervised-image-classification-on","task":"Self-Supervised Image Classification","dataset":"ImageNet","model":"SimCLRv2 (ResNet-50 x2)","rank_in_archive_order":69,"of":144,"metrics":{"Number of Params":"94M","Top 1 Accuracy":"75.6%","Top 5 Accuracy":"92.7%"},"uses_additional_data":false},{"leaderboard":"/sota/self-supervised-image-classification-on","task":"Self-Supervised Image Classification","dataset":"ImageNet","model":"SimCLRv2 (ResNet-50)","rank_in_archive_order":97,"of":144,"metrics":{"Number of Params":"24M","Top 1 Accuracy":"71.7%","Top 5 Accuracy":"90.4%"},"uses_additional_data":false},{"leaderboard":"/sota/self-supervised-image-classification-on-1","task":"Self-Supervised Image Classification","dataset":"ImageNet (finetuned)","model":"SimCLRv2 (ResNet-152, 3×+SK)","rank_in_archive_order":46,"of":65,"metrics":{"Number of Params":"795M","Top 1 Accuracy":"83.1%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-1","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 1% labeled data","model":"SimCLRv2 self-distilled (ResNet-152 x3, SK)","rank_in_archive_order":10,"of":65,"metrics":{"Top 1 Accuracy":"76.6%","Top 5 Accuracy":"93.4%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-1","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 1% labeled data","model":"SimCLRv2 distilled (ResNet-50 x2, SK)","rank_in_archive_order":11,"of":65,"metrics":{"Top 1 Accuracy":"75.9%","Top 5 Accuracy":"93.0%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-1","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 1% labeled data","model":"SimCLRv2 (ResNet-152 x3, SK)","rank_in_archive_order":13,"of":65,"metrics":{"Top 1 Accuracy":"74.9%","Top 5 Accuracy":"92.3%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-1","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 1% labeled data","model":"SimCLRv2 distilled (ResNet-50)","rank_in_archive_order":14,"of":65,"metrics":{"Top 1 Accuracy":"73.9%","Top 5 Accuracy":"91.5%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-1","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 1% labeled data","model":"SimCLRv2 (ResNet-50 ×2)","rank_in_archive_order":30,"of":65,"metrics":{"Top 1 Accuracy":"66.3%","Top 5 Accuracy":"87.4%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-1","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 1% labeled data","model":"SimCLRv2 (ResNet-50)","rank_in_archive_order":42,"of":65,"metrics":{"Top 1 Accuracy":"57.9%","Top 5 Accuracy":"82.5%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-2","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 10% labeled data","model":"SimCLRv2 self-distilled (ResNet-152 x3, SK)","rank_in_archive_order":9,"of":75,"metrics":{"Top 1 Accuracy":"80.9%","Top 5 Accuracy":"95.5%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-2","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 10% labeled data","model":"SimCLRv2 distilled (ResNet-50 x2, SK)","rank_in_archive_order":11,"of":75,"metrics":{"Top 1 Accuracy":"80.2%","Top 5 Accuracy":"95.0%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-2","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 10% labeled data","model":"SimCLRv2 (ResNet-152 x3, SK)","rank_in_archive_order":12,"of":75,"metrics":{"Top 1 Accuracy":"80.1%","Top 5 Accuracy":"95.0%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-2","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 10% labeled data","model":"SimCLRv2 distilled (ResNet-50)","rank_in_archive_order":20,"of":75,"metrics":{"Top 1 Accuracy":"77.5%","Top 5 Accuracy":"93.4%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-2","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 10% labeled data","model":"SimCLRv2 (ResNet-50 x2)","rank_in_archive_order":33,"of":75,"metrics":{"Top 1 Accuracy":"73.9%","Top 5 Accuracy":"91.9%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-2","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 10% labeled data","model":"SimCLRv2 (ResNet-50)","rank_in_archive_order":43,"of":75,"metrics":{"Top 1 Accuracy":"68.4%","Top 5 Accuracy":"89.2%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2006.10029","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2006.10029"}},"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/google-research/simclr","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/lightly-ai/lightly","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/mariaauslander/capstone_fall20_irrigation","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/sayakpaul/PAWS-TF","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/serre-lab/prj_selfsup","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ta9ryuWalrus/simclr","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/parkinkon1/simclr","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/nikheelpandey/TAUP","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/nikheelpandey/TAUP-PyTorch","reach":{"status":"ok"}}],"summary":{"unverified":6},"by_repo_kind":{"listed":{"samples":6,"ran":0,"repositories":1}},"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":0,"samples":[{"code_sha256_prefix":"a7a5dcdaf8aafb54","entry":"attractive_loss","repo":"serre-lab/prj_selfsup","repo_kind":"listed","path":"model/objective.py","file_url":"https://github.com/serre-lab/prj_selfsup/blob/HEAD/model/objective.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"a7a5dcdaf8aafb54"}},{"code_sha256_prefix":"79e0efae44111eea","entry":"attractive_repulsive_loss","repo":"serre-lab/prj_selfsup","repo_kind":"listed","path":"model/objective.py","file_url":"https://github.com/serre-lab/prj_selfsup/blob/HEAD/model/objective.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"79e0efae44111eea"}},{"code_sha256_prefix":"14ec8ff095316493","entry":"get_train_steps","repo":"serre-lab/prj_selfsup","repo_kind":"listed","path":"model/model_util.py","file_url":"https://github.com/serre-lab/prj_selfsup/blob/HEAD/model/model_util.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"14ec8ff095316493"}},{"code_sha256_prefix":"ea360af54446b02b","entry":"learning_rate_schedule","repo":"serre-lab/prj_selfsup","repo_kind":"listed","path":"model/model_util.py","file_url":"https://github.com/serre-lab/prj_selfsup/blob/HEAD/model/model_util.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ea360af54446b02b"}},{"code_sha256_prefix":"ac78dbf80b2d832a","entry":"resnet_autoencoder_v1_generator","repo":"serre-lab/prj_selfsup","repo_kind":"listed","path":"model/resnet_ae.py","file_url":"https://github.com/serre-lab/prj_selfsup/blob/HEAD/model/resnet_ae.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ac78dbf80b2d832a"}},{"code_sha256_prefix":"e888c30f5b90e24a","entry":"supervised_loss","repo":"serre-lab/prj_selfsup","repo_kind":"listed","path":"model/objective.py","file_url":"https://github.com/serre-lab/prj_selfsup/blob/HEAD/model/objective.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"e888c30f5b90e24a"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}