{"url":"/sota/knowledge-distillation-on-imagenet","task":{"name":"Knowledge Distillation","url":"/task/knowledge-distillation","note":null},"dataset":{"name":"ImageNet","url":"/dataset/imagenet"},"category":"Computer Vision","categories":["Computer Vision","Natural Language Processing"],"category_note":null,"description":"Knowledge distillation is the process of transferring knowledge from a large model to a smaller one. While large models (such as very deep neural networks or ensembles of many models) have higher knowledge capacity than small models, this capacity might not be fully utilized.","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["Top-1 accuracy %","model size","CRD training setting",""],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Top-1 accuracy %":"higher","model size":null,"CRD training setting":null,"":null}},"counts":{"rows":52,"rows_with_code":52,"rows_with_paper_page":52,"rows_dated":52,"rows_using_additional_data":1},"rows":[{"rank_in_archive_order":1,"model":"ScaleKD (T:BEiT-L S:ViT-B/14)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"86.43","model size":"87M"},"uses_additional_data":false,"paper_date":"2024-11-11","paper":"/paper/scalekd-strong-vision-transformers-could-be","paper_url":"https://arxiv.org/abs/2411.06786v1","paper_title":"ScaleKD: Strong Vision Transformers Could Be Excellent Teachers","code":"https://github.com/deep-optimization/scalekd","n_code_links":1,"syntology":null},{"rank_in_archive_order":2,"model":"ScaleKD (T:Swin-L S:ViT-B/16)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"85.53","model size":"87M"},"uses_additional_data":false,"paper_date":"2024-11-11","paper":"/paper/scalekd-strong-vision-transformers-could-be","paper_url":"https://arxiv.org/abs/2411.06786v1","paper_title":"ScaleKD: Strong Vision Transformers Could Be Excellent Teachers","code":"https://github.com/deep-optimization/scalekd","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"ScaleKD (T:Swin-L S:ViT-S/16)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"83.93","model size":"22M"},"uses_additional_data":false,"paper_date":"2024-11-11","paper":"/paper/scalekd-strong-vision-transformers-could-be","paper_url":"https://arxiv.org/abs/2411.06786v1","paper_title":"ScaleKD: Strong Vision Transformers Could Be Excellent Teachers","code":"https://github.com/deep-optimization/scalekd","n_code_links":1,"syntology":null},{"rank_in_archive_order":4,"model":"ScaleKD (T:Swin-L S:Swin-T)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"83.8","model size":"27M"},"uses_additional_data":false,"paper_date":"2024-11-11","paper":"/paper/scalekd-strong-vision-transformers-could-be","paper_url":"https://arxiv.org/abs/2411.06786v1","paper_title":"ScaleKD: Strong Vision Transformers Could Be Excellent Teachers","code":"https://github.com/deep-optimization/scalekd","n_code_links":1,"syntology":null},{"rank_in_archive_order":5,"model":"KD++(T: regnety-16GF S:ViT-B)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"83.60","model size":"87M"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":6,"model":"VkD (T:RegNety 160 S:DeiT-S)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"82.9","model size":"22M"},"uses_additional_data":false,"paper_date":"2024-03-10","paper":"/paper/v-kd-improving-knowledge-distillation-using","paper_url":"https://arxiv.org/abs/2403.06213v1","paper_title":"$V_kD:$ Improving Knowledge Distillation using Orthogonal Projections","code":"https://github.com/roymiles/vkd","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":1,"n_samples":5,"n_pointer_only_licence":5}},{"rank_in_archive_order":7,"model":"SpectralKD (T:Swin-S S:Swin-T)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"82.7","model size":"22M"},"uses_additional_data":false,"paper_date":"2024-12-26","paper":"/paper/spectralkd-understanding-and-optimizing","paper_url":"https://arxiv.org/abs/2412.19055v3","paper_title":"SpectralKD: A Unified Framework for Interpreting and Distilling Vision Transformers via Spectral Analysis","code":"https://github.com/thy960112/SpectralKD","n_code_links":1,"syntology":null},{"rank_in_archive_order":8,"model":"ScaleKD (T:Swin-L S:ResNet-50)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"82.55","model size":"22M"},"uses_additional_data":false,"paper_date":"2024-11-11","paper":"/paper/scalekd-strong-vision-transformers-could-be","paper_url":"https://arxiv.org/abs/2411.06786v1","paper_title":"ScaleKD: Strong Vision Transformers Could Be Excellent Teachers","code":"https://github.com/deep-optimization/scalekd","n_code_links":1,"syntology":null},{"rank_in_archive_order":9,"model":"DiffKD (T:Swin-L S: Swin-T)","metrics":{"Top-1 accuracy %":"82.5"},"uses_additional_data":false,"paper_date":"2023-05-25","paper":"/paper/knowledge-diffusion-for-distillation-1","paper_url":"https://arxiv.org/abs/2305.15712v2","paper_title":"Knowledge Diffusion for Distillation","code":"https://github.com/hunto/diffkd","n_code_links":1,"syntology":null},{"rank_in_archive_order":10,"model":"DIST (T: Swin-L S: Swin-T)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"82.3","model size":"29M"},"uses_additional_data":true,"paper_date":"2022-05-21","paper":"/paper/knowledge-distillation-from-a-stronger","paper_url":"https://arxiv.org/abs/2205.10536v3","paper_title":"Knowledge Distillation from A Stronger Teacher","code":"https://github.com/yoshitomo-matsubara/torchdistill","n_code_links":3,"syntology":{"n_ran":10,"n_unverified":1,"n_samples":11,"n_pointer_only_licence":0}},{"rank_in_archive_order":11,"model":"SpectralKD (T:Cait-S24 S:DeiT-S)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"82.2","model size":"22M"},"uses_additional_data":false,"paper_date":"2024-12-26","paper":"/paper/spectralkd-understanding-and-optimizing","paper_url":"https://arxiv.org/abs/2412.19055v3","paper_title":"SpectralKD: A Unified Framework for Interpreting and Distilling Vision Transformers via Spectral Analysis","code":"https://github.com/thy960112/SpectralKD","n_code_links":1,"syntology":null},{"rank_in_archive_order":12,"model":"SRD (T:RegNety 160 S:DeiT-S)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"82.1","model size":"22M"},"uses_additional_data":false,"paper_date":"2023-03-20","paper":"/paper/a-closer-look-at-the-training-dynamics-of","paper_url":"https://arxiv.org/abs/2303.11098v5","paper_title":"Understanding the Role of the Projector in Knowledge Distillation","code":"https://github.com/yoshitomo-matsubara/torchdistill","n_code_links":4,"syntology":null},{"rank_in_archive_order":13,"model":"OFA (T: ViT-B S: ResNet-50)","metrics":{"Top-1 accuracy %":"81.33"},"uses_additional_data":false,"paper_date":"2023-10-30","paper":"/paper/one-for-all-bridge-the-gap-between-1","paper_url":"https://arxiv.org/abs/2310.19444v1","paper_title":"One-for-All: Bridge the Gap Between Heterogeneous Architectures in Knowledge Distillation","code":"https://github.com/hao840/ofakd","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":3,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":14,"model":"DiffKD (T:Swin-L S: ResNet-50)","metrics":{"Top-1 accuracy %":"80.5"},"uses_additional_data":false,"paper_date":"2023-05-25","paper":"/paper/knowledge-diffusion-for-distillation-1","paper_url":"https://arxiv.org/abs/2305.15712v2","paper_title":"Knowledge Diffusion for Distillation","code":"https://github.com/hunto/diffkd","n_code_links":1,"syntology":null},{"rank_in_archive_order":15,"model":"VkD (T:RegNety 160 S:DeiT-Ti)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"79.2","model size":"6M"},"uses_additional_data":false,"paper_date":"2024-03-10","paper":"/paper/v-kd-improving-knowledge-distillation-using","paper_url":"https://arxiv.org/abs/2403.06213v1","paper_title":"$V_kD:$ Improving Knowledge Distillation using Orthogonal Projections","code":"https://github.com/roymiles/vkd","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":1,"n_samples":5,"n_pointer_only_licence":5}},{"rank_in_archive_order":16,"model":"KD++(T:resnet-152 S:resnet-101)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"79.15","model size":"44.5M"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":17,"model":"ADLIK-MO-P25(T:SeNet154, ResNet152b  S:ResNet-50-prune25%)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"78.79","model size":"56.9M"},"uses_additional_data":false,"paper_date":"2019-09-17","paper":"/paper/ensemble-knowledge-distillation-for-learning","paper_url":"https://arxiv.org/abs/1909.08097v3","paper_title":"Ensemble Knowledge Distillation for Learning Improved and Efficient Networks","code":"https://github.com/Adlik/model_optimizer","n_code_links":2,"syntology":{"n_ran":5,"n_unverified":1,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":18,"model":"ADLIK-MO-P375(T:SeNet154, ResNet152b  S:ResNet-50-prune37.5)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"78.07","model size":"40.5M"},"uses_additional_data":false,"paper_date":"2019-09-17","paper":"/paper/ensemble-knowledge-distillation-for-learning","paper_url":"https://arxiv.org/abs/1909.08097v3","paper_title":"Ensemble Knowledge Distillation for Learning Improved and Efficient Networks","code":"https://github.com/Adlik/model_optimizer","n_code_links":2,"syntology":{"n_ran":5,"n_unverified":1,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":19,"model":"KD++(T:resnet-152 S:resnet-50)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"77.48"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":20,"model":"SpectralKD (T:Cait-S24 S:DeiT-T)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"77.4","model size":"6M"},"uses_additional_data":false,"paper_date":"2024-12-26","paper":"/paper/spectralkd-understanding-and-optimizing","paper_url":"https://arxiv.org/abs/2412.19055v3","paper_title":"SpectralKD: A Unified Framework for Interpreting and Distilling Vision Transformers via Spectral Analysis","code":"https://github.com/thy960112/SpectralKD","n_code_links":1,"syntology":null},{"rank_in_archive_order":21,"model":"SRD (T:RegNety 160 S:DeIT-Ti)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"77.2","model size":"6M"},"uses_additional_data":false,"paper_date":"2023-03-20","paper":"/paper/a-closer-look-at-the-training-dynamics-of","paper_url":"https://arxiv.org/abs/2303.11098v5","paper_title":"Understanding the Role of the Projector in Knowledge Distillation","code":"https://github.com/yoshitomo-matsubara/torchdistill","n_code_links":4,"syntology":null},{"rank_in_archive_order":22,"model":"ADLIK-MO(T: ResNet101 S: ResNet50)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"77.14","model size":"99M"},"uses_additional_data":false,"paper_date":"2015-03-09","paper":"/paper/distilling-the-knowledge-in-a-neural-network","paper_url":"http://arxiv.org/abs/1503.02531v1","paper_title":"Distilling the Knowledge in a Neural Network","code":"https://github.com/labmlai/annotated_deep_learning_paper_implementations","n_code_links":64,"syntology":{"n_ran":16,"n_unverified":21,"n_samples":37,"n_pointer_only_licence":10}},{"rank_in_archive_order":23,"model":"WTTM (T: DeiT III-Small S:DeiT-Tiny)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"77.03"},"uses_additional_data":false,"paper_date":"2024-02-17","paper":"/paper/knowledge-distillation-based-on-transformed","paper_url":"https://arxiv.org/abs/2402.11148v2","paper_title":"Knowledge Distillation Based on Transformed Teacher Matching","code":"https://github.com/zkxufo/TTM","n_code_links":1,"syntology":{"n_ran":5,"n_unverified":2,"n_samples":7,"n_pointer_only_licence":0}},{"rank_in_archive_order":24,"model":"ADLIK-MO-P50(T:SeNet154, ResNet152b  S:ResNet-50-half)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"76.376","model size":"27M"},"uses_additional_data":false,"paper_date":"2019-09-17","paper":"/paper/ensemble-knowledge-distillation-for-learning","paper_url":"https://arxiv.org/abs/1909.08097v3","paper_title":"Ensemble Knowledge Distillation for Learning Improved and Efficient Networks","code":"https://github.com/Adlik/model_optimizer","n_code_links":2,"syntology":{"n_ran":5,"n_unverified":1,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":25,"model":"KD++(T:resnet152 S:resnet34)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"75.53"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":26,"model":"WTTM (T:resnet50, S:mobilenet-v1)","metrics":{"Top-1 accuracy %":"73.09"},"uses_additional_data":false,"paper_date":"2024-02-17","paper":"/paper/knowledge-distillation-based-on-transformed","paper_url":"https://arxiv.org/abs/2402.11148v2","paper_title":"Knowledge Distillation Based on Transformed Teacher Matching","code":"https://github.com/zkxufo/TTM","n_code_links":1,"syntology":{"n_ran":5,"n_unverified":2,"n_samples":7,"n_pointer_only_licence":0}},{"rank_in_archive_order":27,"model":"ReviewKD++(T:resnet50, S:mobilenet-v1)","metrics":{"Top-1 accuracy %":"72.96"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":28,"model":"KD++(T:resnet-152 S:resnet18)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"72.54"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":29,"model":"KD++(T:renset101 S:resnet18)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"72.54"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":30,"model":"KD++(T:resnet50 S:resnet18)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"72.53"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":31,"model":"HSAKD  (T: ResNet-34  S:ResNet-18)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"72.39"},"uses_additional_data":false,"paper_date":"2021-07-29","paper":"/paper/hierarchical-self-supervised-augmented","paper_url":"https://arxiv.org/abs/2107.13715v2","paper_title":"Hierarchical Self-supervised Augmented Knowledge Distillation","code":"https://github.com/winycg/HSAKD","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":1,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":32,"model":"ICKD (T: ResNet-34  S:ResNet-18)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"72.19"},"uses_additional_data":false,"paper_date":"2021-01-01","paper":"/paper/exploring-inter-channel-correlation-for","paper_url":"http://openaccess.thecvf.com//content/ICCV2021/html/Liu_Exploring_Inter-Channel_Correlation_for_Diversity-Preserved_Knowledge_Distillation_ICCV_2021_paper.html","paper_title":"Exploring Inter-Channel Correlation for Diversity-Preserved Knowledge Distillation","code":"https://github.com/yoshitomo-matsubara/torchdistill","n_code_links":2,"syntology":null},{"rank_in_archive_order":33,"model":"WTTM (T: ResNet-34 S:ResNet-18)","metrics":{"CRD training setting":"✓","Top-1 accuracy %":"72.19"},"uses_additional_data":false,"paper_date":"2024-02-17","paper":"/paper/knowledge-distillation-based-on-transformed","paper_url":"https://arxiv.org/abs/2402.11148v2","paper_title":"Knowledge Distillation Based on Transformed Teacher Matching","code":"https://github.com/zkxufo/TTM","n_code_links":1,"syntology":{"n_ran":5,"n_unverified":2,"n_samples":7,"n_pointer_only_licence":0}},{"rank_in_archive_order":34,"model":"DIST (T: ResNet-34 S:ResNet-18)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"72.07"},"uses_additional_data":false,"paper_date":"2022-05-21","paper":"/paper/knowledge-distillation-from-a-stronger","paper_url":"https://arxiv.org/abs/2205.10536v3","paper_title":"Knowledge Distillation from A Stronger Teacher","code":"https://github.com/yoshitomo-matsubara/torchdistill","n_code_links":3,"syntology":{"n_ran":10,"n_unverified":1,"n_samples":11,"n_pointer_only_licence":0}},{"rank_in_archive_order":35,"model":"KD++(T: ResNet-34 S:ResNet-18)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"72.07"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":36,"model":"WSL (T: ResNet-34  S:ResNet-18)","metrics":{"Top-1 accuracy %":"72.04"},"uses_additional_data":false,"paper_date":"2021-02-01","paper":"/paper/rethinking-soft-labels-for-knowledge-1","paper_url":"https://arxiv.org/abs/2102.00650v1","paper_title":"Rethinking Soft Labels for Knowledge Distillation: A Bias-Variance Tradeoff Perspective","code":"https://github.com/open-mmlab/mmrazor","n_code_links":4,"syntology":{"n_ran":0,"n_unverified":1,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":37,"model":"CRCD (T: ResNet-34  S:ResNet-18)","metrics":{"CRD training setting":"✓","Top-1 accuracy %":"71.96"},"uses_additional_data":false,"paper_date":"2021-03-29","paper":"/paper/complementary-relation-contrastive","paper_url":"https://arxiv.org/abs/2103.16367v1","paper_title":"Complementary Relation Contrastive Distillation","code":"https://github.com/Lechatelia/CRCD","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":4}},{"rank_in_archive_order":38,"model":"SRD (T: ResNet-34 S:ResNet-18)","metrics":{"CRD training setting":"✓","Top-1 accuracy %":"71.87"},"uses_additional_data":false,"paper_date":"2023-03-20","paper":"/paper/a-closer-look-at-the-training-dynamics-of","paper_url":"https://arxiv.org/abs/2303.11098v5","paper_title":"Understanding the Role of the Projector in Knowledge Distillation","code":"https://github.com/yoshitomo-matsubara/torchdistill","n_code_links":4,"syntology":null},{"rank_in_archive_order":39,"model":"KD++(T:ViT-B, S:resnet18)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"71.84"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":40,"model":"LSHFM (T: ResNet-34 S:ResNet-18)","metrics":{"Top-1 accuracy %":"71.72"},"uses_additional_data":false,"paper_date":"2020-11-03","paper":"/paper/in-defense-of-feature-mimicking-for-knowledge","paper_url":"https://arxiv.org/abs/2011.01424v2","paper_title":"Distilling Knowledge by Mimicking Features","code":"https://github.com/DoctorKey/LSHFM.singleclassification","n_code_links":3,"syntology":{"n_ran":5,"n_unverified":1,"n_samples":6,"n_pointer_only_licence":6}},{"rank_in_archive_order":41,"model":"ITRD (T: ResNet-34 S:ResNet-18)","metrics":{"CRD training setting":"✓","Top-1 accuracy %":"71.68","model size":"11.69M"},"uses_additional_data":false,"paper_date":"2021-12-01","paper":"/paper/information-theoretic-representation","paper_url":"https://arxiv.org/abs/2112.00459v3","paper_title":"Information Theoretic Representation Distillation","code":"https://github.com/roymiles/ITRD","n_code_links":1,"syntology":null},{"rank_in_archive_order":42,"model":"GLD (T: ResNet-34  S:ResNet-18)","metrics":{"Top-1 accuracy %":"71.63"},"uses_additional_data":false,"paper_date":"2021-01-01","paper":"/paper/distilling-global-and-local-logits-with","paper_url":"http://openaccess.thecvf.com//content/ICCV2021/html/Kim_Distilling_Global_and_Local_Logits_With_Densely_Connected_Relations_ICCV_2021_paper.html","paper_title":"Distilling Global and Local Logits With Densely Connected Relations","code":"https://github.com/Youmin-Kim/GLD","n_code_links":1,"syntology":null},{"rank_in_archive_order":43,"model":"SSKD (T: ResNet-34  S:ResNet-18)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"71.62"},"uses_additional_data":false,"paper_date":"2020-06-12","paper":"/paper/knowledge-distillation-meets-self-supervision","paper_url":"https://arxiv.org/abs/2006.07114v2","paper_title":"Knowledge Distillation Meets Self-Supervision","code":"https://github.com/yoshitomo-matsubara/torchdistill","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":44,"model":"Knowledge Review (T: ResNet-34  S:ResNet-18)","metrics":{"CRD training setting":"✓","Top-1 accuracy %":"71.61"},"uses_additional_data":false,"paper_date":"2021-04-19","paper":"/paper/distilling-knowledge-via-knowledge-review","paper_url":"https://arxiv.org/abs/2104.09044v1","paper_title":"Distilling Knowledge via Knowledge Review","code":"https://github.com/yoshitomo-matsubara/torchdistill","n_code_links":7,"syntology":{"n_ran":6,"n_unverified":2,"n_samples":8,"n_pointer_only_licence":5}},{"rank_in_archive_order":45,"model":"Adaptive (T:ResNet-50 S:ResNet-18)","metrics":{"Top-1 accuracy %":"71.61"},"uses_additional_data":false,"paper_date":"2021-10-19","paper":"/paper/adaptive-distillation-aggregating-knowledge","paper_url":"https://arxiv.org/abs/2110.09674v2","paper_title":"Adaptive Distillation: Aggregating Knowledge from Multiple Paths for Efficient Distillation","code":"https://github.com/wyze-AI/AdaptiveDistillation","n_code_links":1,"syntology":null},{"rank_in_archive_order":46,"model":"KD++(T: ViT-S, S:resnet18)","metrics":{"CRD training setting":"✘","Top-1 accuracy %":"71.46"},"uses_additional_data":false,"paper_date":"2023-05-26","paper":"/paper/improving-knowledge-distillation-via-1","paper_url":"https://arxiv.org/abs/2305.17007v1","paper_title":"Improving Knowledge Distillation via Regularizing Feature Norm and Direction","code":"https://github.com/wangyz1608/knowledge-distillation-via-nd","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":47,"model":"AFD (T: ResNet-34  S:ResNet-18)","metrics":{"Top-1 accuracy %":"71.38"},"uses_additional_data":false,"paper_date":"2021-02-05","paper":"/paper/show-attend-and-distill-knowledge","paper_url":"https://arxiv.org/abs/2102.02973v1","paper_title":"Show, Attend and Distill:Knowledge Distillation via Attention-based Feature Matching","code":"https://github.com/clovaai/attention-feature-distillation","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":2,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":48,"model":"CRD  (T: ResNet-34  S:ResNet-18)","metrics":{"Top-1 accuracy %":"71.38"},"uses_additional_data":false,"paper_date":"2019-10-23","paper":"/paper/contrastive-representation-distillation-1","paper_url":"https://arxiv.org/abs/1910.10699v3","paper_title":"Contrastive Representation Distillation","code":"https://github.com/HobbitLong/RepDistiller","n_code_links":4,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":49,"model":"Overhual (T: ResNet-34 S:ResNet-18)","metrics":{"Top-1 accuracy %":"70.81"},"uses_additional_data":false,"paper_date":"2019-04-03","paper":"/paper/a-comprehensive-overhaul-of-feature","paper_url":"https://arxiv.org/abs/1904.01866v2","paper_title":"A Comprehensive Overhaul of Feature Distillation","code":"https://github.com/clovaai/overhaul-distillation","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":5,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":50,"model":"KD (T: ResNet-34 S:ResNet-18)","metrics":{"CRD training setting":"✓","Top-1 accuracy %":"70.66"},"uses_additional_data":false,"paper_date":"2015-03-09","paper":"/paper/distilling-the-knowledge-in-a-neural-network","paper_url":"http://arxiv.org/abs/1503.02531v1","paper_title":"Distilling the Knowledge in a Neural Network","code":"https://github.com/labmlai/annotated_deep_learning_paper_implementations","n_code_links":64,"syntology":{"n_ran":16,"n_unverified":21,"n_samples":37,"n_pointer_only_licence":10}},{"rank_in_archive_order":51,"model":"AT (T: ResNet-34 S:ResNet-18)","metrics":{"":"70.70"},"uses_additional_data":false,"paper_date":"2016-12-12","paper":"/paper/paying-more-attention-to-attention-improving","paper_url":"http://arxiv.org/abs/1612.03928v3","paper_title":"Paying More Attention to Attention: Improving the Performance of Convolutional Neural Networks via Attention Transfer","code":"https://github.com/yoshitomo-matsubara/torchdistill","n_code_links":7,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":52,"model":"AT  (T: ResNet-34  S:ResNet-18)","metrics":{"CRD training setting":"✓"},"uses_additional_data":false,"paper_date":"2016-12-12","paper":"/paper/paying-more-attention-to-attention-improving","paper_url":"http://arxiv.org/abs/1612.03928v3","paper_title":"Paying More Attention to Attention: Improving the Performance of Convolutional Neural Networks via Attention Transfer","code":"https://github.com/yoshitomo-matsubara/torchdistill","n_code_links":7,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,885 of the 9,623 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9623,"papers_checked":6885,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2737},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-25T09:33:49+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":35,"rows_with_any_sample_ran":32,"distinct_papers_with_graph_line":17,"distinct_papers_with_any_sample_ran":14,"samples_over_distinct_papers":{"n_ran":65,"n_unverified":42,"n_samples":107,"n_pointer_only_licence":40,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":146,"n_unverified":71,"n_samples":217,"n_pointer_only_licence":86,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}