{"url":"/sota/efficient-vits-on-imagenet-1k-with-deit-t","task":{"name":"Efficient ViTs","url":"/task/efficient-vits","note":null},"dataset":{"name":"ImageNet-1K (with DeiT-T)","url":null},"category":"Computer Vision","categories":["Adversarial","Computer Vision"],"category_note":null,"description":"Increasing the efficiency of ViTs without the modification of the architecture. (i.e., Key & Query Sparsification, Token pruning & merging)","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","GFLOPs"],"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","GFLOPs":null}},"counts":{"rows":22,"rows_with_code":20,"rows_with_paper_page":22,"rows_dated":22,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"dTPS","metrics":{"GFLOPs":"0.8","Top 1 Accuracy":"72.9"},"uses_additional_data":false,"paper_date":"2023-04-21","paper":"/paper/joint-token-pruning-and-squeezing-towards","paper_url":"https://arxiv.org/abs/2304.10716v1","paper_title":"Joint Token Pruning and Squeezing Towards More Aggressive Compression of Vision Transformers","code":"https://github.com/megvii-research/tps-cvpr2023","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":6,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":2,"model":"MCTF ($r=8$)","metrics":{"GFLOPs":"1.0","Top 1 Accuracy":"72.9"},"uses_additional_data":false,"paper_date":"2024-03-15","paper":"/paper/multi-criteria-token-fusion-with-one-step","paper_url":"https://arxiv.org/abs/2403.10030v3","paper_title":"Multi-criteria Token Fusion with One-step-ahead Attention for Efficient Vision Transformers","code":"https://github.com/mlvlab/mctf","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":2,"n_samples":5,"n_pointer_only_licence":0}},{"rank_in_archive_order":3,"model":"MCTF ($r=16$)","metrics":{"GFLOPs":"0.7","Top 1 Accuracy":"72.7"},"uses_additional_data":false,"paper_date":"2024-03-15","paper":"/paper/multi-criteria-token-fusion-with-one-step","paper_url":"https://arxiv.org/abs/2403.10030v3","paper_title":"Multi-criteria Token Fusion with One-step-ahead Attention for Efficient Vision Transformers","code":"https://github.com/mlvlab/mctf","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":2,"n_samples":5,"n_pointer_only_licence":0}},{"rank_in_archive_order":4,"model":"BAT","metrics":{"GFLOPs":"0.8","Top 1 Accuracy":"72.3"},"uses_additional_data":false,"paper_date":"2022-11-21","paper":"/paper/beyond-attentive-tokens-incorporating-token","paper_url":"https://arxiv.org/abs/2211.11315v1","paper_title":"Beyond Attentive Tokens: Incorporating Token Importance and Diversity for Efficient Vision Transformers","code":"https://github.com/BWLONG/BeyondAttentiveTokens","n_code_links":1,"syntology":null},{"rank_in_archive_order":5,"model":"eTPS","metrics":{"GFLOPs":"0.8","Top 1 Accuracy":"72.3"},"uses_additional_data":false,"paper_date":"2023-04-21","paper":"/paper/joint-token-pruning-and-squeezing-towards","paper_url":"https://arxiv.org/abs/2304.10716v1","paper_title":"Joint Token Pruning and Squeezing Towards More Aggressive Compression of Vision Transformers","code":"https://github.com/megvii-research/tps-cvpr2023","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":6,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":6,"model":"SPViT (1.0G)","metrics":{"GFLOPs":"1.0","Top 1 Accuracy":"72.2"},"uses_additional_data":false,"paper_date":"2021-12-27","paper":"/paper/spvit-enabling-faster-vision-transformers-via","paper_url":"https://arxiv.org/abs/2112.13890v2","paper_title":"SPViT: Enabling Faster Vision Transformers via Soft Token Pruning","code":"https://github.com/peiyanflying/spvit","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":7,"model":"Base (DeiT-T)","metrics":{"GFLOPs":"1.2","Top 1 Accuracy":"72.2"},"uses_additional_data":false,"paper_date":"2020-12-23","paper":"/paper/training-data-efficient-image-transformers","paper_url":"https://arxiv.org/abs/2012.12877v2","paper_title":"Training data-efficient image transformers & distillation through attention","code":"https://github.com/huggingface/transformers","n_code_links":40,"syntology":{"n_ran":12,"n_unverified":7,"n_samples":19,"n_pointer_only_licence":3}},{"rank_in_archive_order":8,"model":"DPS-ViT","metrics":{"GFLOPs":"0.6","Top 1 Accuracy":"72.1"},"uses_additional_data":false,"paper_date":"2021-06-05","paper":"/paper/patch-slimming-for-efficient-vision","paper_url":"https://arxiv.org/abs/2106.02852v2","paper_title":"Patch Slimming for Efficient Vision Transformers","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":9,"model":"PPT","metrics":{"GFLOPs":"0.8","Top 1 Accuracy":"72.1"},"uses_additional_data":false,"paper_date":"2023-10-03","paper":"/paper/ppt-token-pruning-and-pooling-for-efficient","paper_url":"https://arxiv.org/abs/2310.01812v3","paper_title":"PPT: Token Pruning and Pooling for Efficient Vision Transformers","code":"https://github.com/mindspore-lab/models","n_code_links":2,"syntology":{"n_ran":5,"n_unverified":2,"n_samples":7,"n_pointer_only_licence":0}},{"rank_in_archive_order":10,"model":"SPViT (0.9G)","metrics":{"GFLOPs":"0.9","Top 1 Accuracy":"72.1"},"uses_additional_data":false,"paper_date":"2021-12-27","paper":"/paper/spvit-enabling-faster-vision-transformers-via","paper_url":"https://arxiv.org/abs/2112.13890v2","paper_title":"SPViT: Enabling Faster Vision Transformers via Soft Token Pruning","code":"https://github.com/peiyanflying/spvit","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":11,"model":"PS-ViT","metrics":{"GFLOPs":"0.7","Top 1 Accuracy":"72.0"},"uses_additional_data":false,"paper_date":"2021-06-05","paper":"/paper/patch-slimming-for-efficient-vision","paper_url":"https://arxiv.org/abs/2106.02852v2","paper_title":"Patch Slimming for Efficient Vision Transformers","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":12,"model":"EvoViT","metrics":{"GFLOPs":"0.8","Top 1 Accuracy":"72.0"},"uses_additional_data":false,"paper_date":"2021-08-03","paper":"/paper/evo-vit-slow-fast-token-evolution-for-dynamic","paper_url":"https://arxiv.org/abs/2108.01390v5","paper_title":"Evo-ViT: Slow-Fast Token Evolution for Dynamic Vision Transformer","code":"https://github.com/YifanXu74/Evo-ViT","n_code_links":1,"syntology":{"n_ran":5,"n_unverified":4,"n_samples":9,"n_pointer_only_licence":0}},{"rank_in_archive_order":13,"model":"LTMP (80%)","metrics":{"GFLOPs":"1.0","Top 1 Accuracy":"72.0"},"uses_additional_data":false,"paper_date":"2023-07-20","paper":"/paper/learned-thresholds-token-merging-and-pruning","paper_url":"https://arxiv.org/abs/2307.10780v2","paper_title":"Learned Thresholds Token Merging and Pruning for Vision Transformers","code":"https://github.com/mxbonn/ltmp","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":1,"n_samples":5,"n_pointer_only_licence":5}},{"rank_in_archive_order":14,"model":"ToMe ($r=8$)","metrics":{"GFLOPs":"0.9","Top 1 Accuracy":"71.7"},"uses_additional_data":false,"paper_date":"2022-10-17","paper":"/paper/token-merging-your-vit-but-faster","paper_url":"https://arxiv.org/abs/2210.09461v3","paper_title":"Token Merging: Your ViT But Faster","code":"https://github.com/dbolya/tomesd","n_code_links":5,"syntology":{"n_ran":5,"n_unverified":0,"n_samples":5,"n_pointer_only_licence":4}},{"rank_in_archive_order":15,"model":"LTMP (60%)","metrics":{"GFLOPs":"0.8","Top 1 Accuracy":"71.5"},"uses_additional_data":false,"paper_date":"2023-07-20","paper":"/paper/learned-thresholds-token-merging-and-pruning","paper_url":"https://arxiv.org/abs/2307.10780v2","paper_title":"Learned Thresholds Token Merging and Pruning for Vision Transformers","code":"https://github.com/mxbonn/ltmp","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":"MCTF ($r=20$)","metrics":{"GFLOPs":"0.6","Top 1 Accuracy":"71.4"},"uses_additional_data":false,"paper_date":"2024-03-15","paper":"/paper/multi-criteria-token-fusion-with-one-step","paper_url":"https://arxiv.org/abs/2403.10030v3","paper_title":"Multi-criteria Token Fusion with One-step-ahead Attention for Efficient Vision Transformers","code":"https://github.com/mlvlab/mctf","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":2,"n_samples":5,"n_pointer_only_licence":0}},{"rank_in_archive_order":17,"model":"ToMe ($r=12$)","metrics":{"GFLOPs":"0.8","Top 1 Accuracy":"71.4"},"uses_additional_data":false,"paper_date":"2022-10-17","paper":"/paper/token-merging-your-vit-but-faster","paper_url":"https://arxiv.org/abs/2210.09461v3","paper_title":"Token Merging: Your ViT But Faster","code":"https://github.com/dbolya/tomesd","n_code_links":5,"syntology":{"n_ran":5,"n_unverified":0,"n_samples":5,"n_pointer_only_licence":4}},{"rank_in_archive_order":18,"model":"ToMe ($r=16$)","metrics":{"GFLOPs":"0.6","Top 1 Accuracy":"70.7"},"uses_additional_data":false,"paper_date":"2022-10-17","paper":"/paper/token-merging-your-vit-but-faster","paper_url":"https://arxiv.org/abs/2210.09461v3","paper_title":"Token Merging: Your ViT But Faster","code":"https://github.com/dbolya/tomesd","n_code_links":5,"syntology":{"n_ran":5,"n_unverified":0,"n_samples":5,"n_pointer_only_licence":4}},{"rank_in_archive_order":19,"model":"SPViT","metrics":{"GFLOPs":"1.0","Top 1 Accuracy":"70.7"},"uses_additional_data":false,"paper_date":"2021-11-23","paper":"/paper/pruning-self-attentions-into-convolutional","paper_url":"https://arxiv.org/abs/2111.11802v4","paper_title":"Pruning Self-attentions into Convolutional Layers in Single Path","code":"https://github.com/zhuang-group/spvit","n_code_links":3,"syntology":{"n_ran":4,"n_unverified":3,"n_samples":7,"n_pointer_only_licence":0}},{"rank_in_archive_order":20,"model":"S$^2$ViTE","metrics":{"GFLOPs":"0.9","Top 1 Accuracy":"70.1"},"uses_additional_data":false,"paper_date":"2021-06-08","paper":"/paper/chasing-sparsity-in-vision-transformers-an","paper_url":"https://arxiv.org/abs/2106.04533v3","paper_title":"Chasing Sparsity in Vision Transformers: An End-to-End Exploration","code":"https://github.com/VITA-Group/SViTE","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":5,"n_samples":7,"n_pointer_only_licence":0}},{"rank_in_archive_order":21,"model":"LTMP (45%)","metrics":{"GFLOPs":"0.7","Top 1 Accuracy":"69.8"},"uses_additional_data":false,"paper_date":"2023-07-20","paper":"/paper/learned-thresholds-token-merging-and-pruning","paper_url":"https://arxiv.org/abs/2307.10780v2","paper_title":"Learned Thresholds Token Merging and Pruning for Vision Transformers","code":"https://github.com/mxbonn/ltmp","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":1,"n_samples":5,"n_pointer_only_licence":5}},{"rank_in_archive_order":22,"model":"HVT-Ti-1","metrics":{"GFLOPs":"0.6","Top 1 Accuracy":"69.6"},"uses_additional_data":false,"paper_date":"2021-03-19","paper":"/paper/scalable-visual-transformers-with","paper_url":"https://arxiv.org/abs/2103.10619v2","paper_title":"Scalable Vision Transformers with Hierarchical Pooling","code":"https://github.com/BR-IDL/PaddleViT","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":1,"n_samples":2,"n_pointer_only_licence":0}}],"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":19,"rows_with_any_sample_ran":19,"distinct_papers_with_graph_line":11,"distinct_papers_with_any_sample_ran":11,"samples_over_distinct_papers":{"n_ran":46,"n_unverified":31,"n_samples":77,"n_pointer_only_licence":15,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":75,"n_unverified":43,"n_samples":118,"n_pointer_only_licence":36,"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"}}}