{"url":"/sota/efficient-vits-on-imagenet-1k-with-lv-vit-s","task":{"name":"Efficient ViTs","url":"/task/efficient-vits","note":null},"dataset":{"name":"ImageNet-1K (With LV-ViT-S)","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":19,"rows_with_code":17,"rows_with_paper_page":19,"rows_dated":19,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"MCTF ($r=8$)","metrics":{"GFLOPs":"4.9","Top 1 Accuracy":"83.5"},"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":2,"model":"MCTF ($r=12$)","metrics":{"GFLOPs":"4.2","Top 1 Accuracy":"83.4"},"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":3,"model":"Base (LV-ViT-S)","metrics":{"GFLOPs":"6.6","Top 1 Accuracy":"83.3"},"uses_additional_data":false,"paper_date":"2021-04-22","paper":"/paper/token-labeling-training-a-85-5-top-1-accuracy","paper_url":"https://arxiv.org/abs/2104.10858v3","paper_title":"All Tokens Matter: Token Labeling for Training Better Vision Transformers","code":"https://github.com/zihangJiang/TokenLabeling","n_code_links":7,"syntology":{"n_ran":3,"n_unverified":2,"n_samples":5,"n_pointer_only_licence":3}},{"rank_in_archive_order":4,"model":"DynamicViT (90%)","metrics":{"GFLOPs":"5.8","Top 1 Accuracy":"83.3"},"uses_additional_data":false,"paper_date":"2021-06-03","paper":"/paper/dynamicvit-efficient-vision-transformers-with","paper_url":"https://arxiv.org/abs/2106.02034v2","paper_title":"DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification","code":"https://github.com/raoyongming/DynamicViT","n_code_links":2,"syntology":{"n_ran":7,"n_unverified":2,"n_samples":9,"n_pointer_only_licence":1}},{"rank_in_archive_order":5,"model":"DynamicViT (80%)","metrics":{"GFLOPs":"5.1","Top 1 Accuracy":"83.2"},"uses_additional_data":false,"paper_date":"2021-06-03","paper":"/paper/dynamicvit-efficient-vision-transformers-with","paper_url":"https://arxiv.org/abs/2106.02034v2","paper_title":"DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification","code":"https://github.com/raoyongming/DynamicViT","n_code_links":2,"syntology":{"n_ran":7,"n_unverified":2,"n_samples":9,"n_pointer_only_licence":1}},{"rank_in_archive_order":6,"model":"BAT","metrics":{"GFLOPs":"4.7","Top 1 Accuracy":"83.1"},"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":7,"model":"AS-LV-S (70%)","metrics":{"GFLOPs":"4.6","Top 1 Accuracy":"83.1"},"uses_additional_data":false,"paper_date":"2022-09-28","paper":"/paper/adaptive-sparse-vit-towards-learnable","paper_url":"https://arxiv.org/abs/2209.13802v2","paper_title":"Adaptive Sparse ViT: Towards Learnable Adaptive Token Pruning by Fully Exploiting Self-Attention","code":"https://github.com/cydia2018/as-vit","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":8,"model":"PPT","metrics":{"GFLOPs":"4.6","Top 1 Accuracy":"83.1"},"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":9,"model":"SPViT","metrics":{"GFLOPs":"4.3","Top 1 Accuracy":"83.1"},"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":10,"model":"EViT (70%)","metrics":{"GFLOPs":"4.7","Top 1 Accuracy":"83.0"},"uses_additional_data":false,"paper_date":"2022-02-16","paper":"/paper/not-all-patches-are-what-you-need-expediting","paper_url":"https://arxiv.org/abs/2202.07800v2","paper_title":"Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations","code":"https://github.com/youweiliang/evit","n_code_links":2,"syntology":{"n_ran":9,"n_unverified":6,"n_samples":15,"n_pointer_only_licence":12}},{"rank_in_archive_order":11,"model":"DynamicViT (70%)","metrics":{"GFLOPs":"4.6","Top 1 Accuracy":"83.0"},"uses_additional_data":false,"paper_date":"2021-06-03","paper":"/paper/dynamicvit-efficient-vision-transformers-with","paper_url":"https://arxiv.org/abs/2106.02034v2","paper_title":"DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification","code":"https://github.com/raoyongming/DynamicViT","n_code_links":2,"syntology":{"n_ran":7,"n_unverified":2,"n_samples":9,"n_pointer_only_licence":1}},{"rank_in_archive_order":12,"model":"DPS-LV-ViT-S","metrics":{"GFLOPs":"4.5","Top 1 Accuracy":"82.9"},"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":13,"model":"DiffRate","metrics":{"GFLOPs":"3.9","Top 1 Accuracy":"82.6"},"uses_additional_data":false,"paper_date":"2023-05-29","paper":"/paper/diffrate-differentiable-compression-rate-for","paper_url":"https://arxiv.org/abs/2305.17997v1","paper_title":"DiffRate : Differentiable Compression Rate for Efficient Vision Transformers","code":"https://github.com/opengvlab/diffrate","n_code_links":1,"syntology":null},{"rank_in_archive_order":14,"model":"AS-LV-S (60%)","metrics":{"GFLOPs":"3.9","Top 1 Accuracy":"82.6"},"uses_additional_data":false,"paper_date":"2022-09-28","paper":"/paper/adaptive-sparse-vit-towards-learnable","paper_url":"https://arxiv.org/abs/2209.13802v2","paper_title":"Adaptive Sparse ViT: Towards Learnable Adaptive Token Pruning by Fully Exploiting Self-Attention","code":"https://github.com/cydia2018/as-vit","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":15,"model":"dTPS","metrics":{"GFLOPs":"3.8","Top 1 Accuracy":"82.6"},"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":16,"model":"EViT (50%)","metrics":{"GFLOPs":"3.9","Top 1 Accuracy":"82.5"},"uses_additional_data":false,"paper_date":"2022-02-16","paper":"/paper/not-all-patches-are-what-you-need-expediting","paper_url":"https://arxiv.org/abs/2202.07800v2","paper_title":"Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations","code":"https://github.com/youweiliang/evit","n_code_links":2,"syntology":{"n_ran":9,"n_unverified":6,"n_samples":15,"n_pointer_only_licence":12}},{"rank_in_archive_order":17,"model":"eTPS","metrics":{"GFLOPs":"3.8","Top 1 Accuracy":"82.5"},"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":18,"model":"PS-LV-ViT-S","metrics":{"GFLOPs":"4.7","Top 1 Accuracy":"82.4"},"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":19,"model":"MCTF ($r=16$)","metrics":{"GFLOPs":"3.6","Top 1 Accuracy":"82.3"},"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}}],"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":15,"rows_with_any_sample_ran":15,"distinct_papers_with_graph_line":8,"distinct_papers_with_any_sample_ran":8,"samples_over_distinct_papers":{"n_ran":33,"n_unverified":20,"n_samples":53,"n_pointer_only_licence":20,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":65,"n_unverified":40,"n_samples":105,"n_pointer_only_licence":35,"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"}}}