Papers › MaxViT: Multi-Axis Vision Transformer

MaxViT: Multi-Axis Vision Transformer

4 Apr 2022arXiv:2204.01697archive 2025-07-28

Zhengzhong Tu, Hossein Talebi, Han Zhang, Feng Yang, Peyman Milanfar, Alan Bovik, Yinxiao Li

Transformers have recently gained significant attention in the computer vision community. However, the lack of scalability of self-attention mechanisms with respect to image size has limited their wide adoption in state-of-the-art vision backbones. In this paper we introduce an efficient and scalable attention model we call multi-axis attention, which consists of two aspects: blocked local and dilated global attention. These design choices allow global-local spatial interactions on arbitrary input resolutions with only linear complexity. We also present a new architectural element by effectively blending our proposed attention model with convolutions, and accordingly propose a simple hierarchical vision backbone, dubbed MaxViT, by simply repeating the basic building block over multiple stages. Notably, MaxViT is able to ''see'' globally throughout the entire network, even in earlier, high-resolution stages. We demonstrate the effectiveness of our model on a broad spectrum of vision tasks. On image classification, MaxViT achieves state-of-the-art performance under various settings: without extra data, MaxViT attains 86.5% ImageNet-1K top-1 accuracy; with ImageNet-21K pre-training, our model achieves 88.7% top-1 accuracy. For downstream tasks, MaxViT as a backbone delivers favorable performance on object detection as well as visual aesthetic assessment. We also show that our proposed model expresses strong generative modeling capability on ImageNet, demonstrating the superior potential of MaxViT blocks as a universal vision module. The source code and trained models will be available at https://github.com/google-research/maxvit.

PaperPDFCodeCode Syntology ranMethod as code

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 53 code samples harvested from 7 repositories linked to this paper: 37 ran (of which 15 constructed an object rather than computing a result; 23 with no instrument failure: 2 honoured, 0 violated, 21 with no contract checked; 14 where Syntology's instrument failed) · 16 unverified. Where Syntology's instrument failed: 3 where our contract draft was wrong; 11 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 11 samples from 1 repository, 4 ran (4 with no instrument failure; 0 where Syntology's instrument failed); community (archive-listed): 42 samples from 6 repositories, 33 ran (19 with no instrument failure; 14 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

No sample is linked to this paper from 3 of its repositories that Syntology harvested for other papers:

Levels written between 2026-09-22 and 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

15 repositories listed; official and paper-mentioned ones first.

google-research/maxvit officialmentioned in papermentioned on GitHubtfSyntology: 11 samples, 4 ran (0 instrument-failed) · 7 unverified report
ChristophReich1996/MaxViT mentioned on GitHubpytorchMITSyntology: 3 samples, 3 ran (0 instrument-failed) · 0 unverified report
RooKichenn/pytorch-MaxViT mentioned on GitHubpytorchSyntology: not harvested report
google-research/maxim mentioned on GitHubjaxSyntology: 7 samples, 5 ran (2 instrument-failed) · 2 unverified report
hankyul2/maxvit-pytorch mentioned on GitHubpytorchSyntology: 9 samples, 8 ran (5 instrument-failed) · 1 unverified report
lucidrains/imagen-pytorch mentioned on GitHubpytorchMITSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Tell2Design: A Dataset for Language-Guided Floor Plan Generation (arXiv:2311.15941; that paper's own run record: 24 ran (of which 0 constructed an object rather than computing a result; 18 with no instrument failure: 1 honoured, 5 violated, 12 with no contract checked; 6 where Syntology's instrument failed) · 4 unverified); nothing checks that they implement this paper's method. report
lucidrains/vit-pytorch mentioned on GitHubpytorchSyntology: 8 samples, 8 ran (1 instrument-failed) · 0 unverified report
qwopqwop200/MaxVIT-pytorch mentioned on GitHubpytorchMITSyntology: 3 samples, 2 ran (1 instrument-failed) · 1 unverified report
2024-MindSpore-1/Code3 mindsporeSyntology: not harvested report
Mind23-2/MindCode-101 mindsporeSyntology: not harvested report
Mind23-2/MindCode-3 mindsporeSyntology: not harvested report
huggingface/pytorch-image-models pytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Learning to Merge Tokens via Decoupled Embedding for Efficient Vision Transformers (arXiv:2412.10569; that paper's own run record: 28 ran (of which 1 constructed an object rather than computing a result; 12 with no instrument failure: 1 honoured, 0 violated, 11 with no contract checked; 16 where Syntology's instrument failed) · 6 unverified); nothing checks that they implement this paper's method. report
leondgarse/keras_cv_attention_models tfSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to RepViT: Revisiting Mobile CNN From ViT Perspective (arXiv:2307.09283; that paper's own run record: 6 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 3 where Syntology's instrument failed) · 6 unverified); nothing checks that they implement this paper's method. report
towhee-io/towhee pytorchSyntology: 12 samples, 7 ran (5 instrument-failed) · 5 unverified report
gitlab.com/birder/birder pytorchSyntology: not harvested report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

53 samples harvested: 37 ran (of which 15 constructed an object rather than computing a result; 23 with no instrument failure: 2 honoured, 0 violated, 21 with no contract checked; 14 where Syntology's instrument failed) · 16 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

2ran · honoured contract
3ran · our draft was wrong
11ran; the contract check could not run
21ran
16unverified

Licence: 9 of the 53 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 7 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. For 6 samples that ran, the graph records a signature this page does not show: for a class the fixture built without calling, the harness called the object with inputs it made up itself, so the recorded signature is the harness's probe, not the fixture's arguments. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: 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. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

google-research/maxvit official repository 11 samples, 4 ran (0 instrument-failed) · 7 unverified
Attention in 344 papers
census: 13 implementations from 12 papers, 10 distinct outputs across 4 buckets, one shared input per bucket
maxvit/models/maxvit.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · 2bc733f08e821139 · report fixture: returned Attention · level recorded 2026-09-22
BatchNormalization maxvit/models/maxvit.py ran Apache-2.0 (permissive) · e2718d640d43e769 · report fixture: signature inputs:EagerTensor/2/float, training:bool/0/bool · returned EagerTensor · level recorded 2026-09-24
TrailDense maxvit/models/maxvit.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · 839c67588aa822e2 · report fixture: returned TrailDense · level recorded 2026-09-22
float32_softmax maxvit/models/maxvit.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · 5ecfa57092a8cecc · report fixture: signature x:EagerTensor/2/float · returned EagerTensor · level recorded 2026-09-22
FFN in 37 papers maxvit/models/maxvit.py unverified Apache-2.0 (permissive) · f50211588bf5be8b · report
MBConvBlock in 10 papers maxvit/models/maxvit.py unverified Apache-2.0 (permissive) · 2d7a9595c2215efd · report
MaxViTBlock maxvit/models/maxvit.py unverified Apache-2.0 (permissive) · c99919623d98cd27 · report
SE in 13 papers maxvit/models/maxvit.py unverified Apache-2.0 (permissive) · ee80113a460390fe · report
TpuBatchNormalization maxvit/models/maxvit.py unverified Apache-2.0 (permissive) · d237bfe09ec44cfd · report
config_batch_norm maxvit/models/maxvit.py unverified Apache-2.0 (permissive) · 20213f171bc1c7e9 · report
create_config_from_dict maxvit/models/maxvit.py unverified Apache-2.0 (permissive) · 256b0e25fa8d8cee · report
ChristophReich1996/MaxViT community (archive-listed) 3 samples, 3 ran (0 instrument-failed) · 0 unverified
grid_partition maxvit/maxvit.py ran MIT (permissive) · bbc5501012bdee68 · report fixture: signature input:Tensor/4/float, grid_size:tuple/1/int · returned Tensor · level recorded 2026-09-24
window_partition in 377 papers
census: 13 implementations from 17 papers, 5 distinct outputs across 3 buckets, one shared input per bucket
maxvit/maxvit.py ran MIT (permissive) · 20f2f90659c203e1 · report fixture: signature input:Tensor/4/float, window_size:tuple/1/int · returned Tensor · level recorded 2026-09-24
window_reverse in 296 papers maxvit/maxvit.py ran MIT (permissive) · cd52f54038e59852 · report fixture: signature windows:Tensor/4/float, original_size:tuple/1/int, window_size:tuple/1/int · returned Tensor · level recorded 2026-09-24
google-research/maxim community (archive-listed) 7 samples, 5 ran (2 instrument-failed) · 2 unverified
BlockGatingUnit maxim/models/maxim.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · d8da107e27ecbe28 · report fixture: returned BlockGatingUnit · level recorded 2026-09-22
BlockGmlpLayer maxim/models/maxim.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · 325b31d5363dae7d · report fixture: returned BlockGmlpLayer · level recorded 2026-09-22
GridGatingUnit maxim/models/maxim.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · 86d6d05360b6903b · report fixture: returned GridGatingUnit · level recorded 2026-09-22
block_images_einops maxim/models/maxim.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · 8a67f23552a44d61 · report fixture: signature x:Tensor/4/float, patch_size:tuple/1/int · returned Tensor · level recorded 2026-09-22
unblock_images_einops maxim/models/maxim.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · 627ea4c76565d927 · report fixture: signature x:Tensor/4/float, grid_size:tuple/1/int, patch_size:tuple/1/int · returned Tensor · level recorded 2026-09-22
GridGmlpLayer maxim/models/maxim.py unverified Apache-2.0 (permissive) · bcddd7996b51db67 · report
ResidualSplitHeadMultiAxisGmlpLayer maxim/models/maxim.py unverified Apache-2.0 (permissive) · e7fa9568a4a28913 · report
hankyul2/maxvit-pytorch community (archive-listed) 9 samples, 8 ran (5 instrument-failed) · 1 unverified
Attention2d maxvit.py ran · metamorphic tier: deterministic no licence file found · pointer only · d227ba24d7d8d855 · report fixture: returned Attention2d · level recorded 2026-09-22
LayerScale2d maxvit.py ran · metamorphic tier: deterministic fingerprinted no licence file found · pointer only · 8fa30deee34804c0 · report fixture: returned LayerScale2d · level recorded 2026-09-22
MaxxVitTransformerCfg maxvit.py ran · metamorphic tier: deterministic no licence file found · pointer only · 08082022ac2c6e65 · report fixture: returned MaxxVitTransformerCfg · level recorded 2026-09-22
get_rel_pos_cls maxvit.py ran · our draft was wrong no licence file found · pointer only · 9d7347fa5b820e84 · report fixture: signature cfg:MaxxVitTransformerCfg/0/?, window_size:tuple/1/int · returned partial · level recorded 2026-09-22
grid_partition_nchw maxvit.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant no licence file found · pointer only · f41c6472d974b50b · report fixture: signature x:Tensor/4/float, grid_size:list/1/int · returned Tensor · level recorded 2026-09-22
grid_reverse_nchw maxvit.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant no licence file found · pointer only · 80cee5fcd8a9952b · report fixture: signature windows:Tensor/3/float, grid_size:list/1/int, img_size:list/1/int · returned Tensor · level recorded 2026-09-22
window_partition_nchw maxvit.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant no licence file found · pointer only · ae8a9a667b5d42f9 · report fixture: signature x:Tensor/4/float, window_size:list/1/int · returned Tensor · level recorded 2026-09-22
window_reverse_nchw maxvit.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant no licence file found · pointer only · c72f01470606b887 · report fixture: signature windows:Tensor/4/float, window_size:list/1/int, img_size:list/1/int · returned Tensor · level recorded 2026-09-22
PartitionAttention2d maxvit.py unverified no licence file found · pointer only · 641d6f2b8dbc0745 · report
lucidrains/vit-pytorch community (archive-listed) 8 samples, 8 ran (1 instrument-failed) · 0 unverified
Attention in 344 papers
census: 13 implementations from 12 papers, 10 distinct outputs across 4 buckets, one shared input per bucket
vit_pytorch/max_vit.py ran MIT (permissive) · 0ff4fe4361d49089 · report fixture: returned Attention · level recorded 2026-09-22
Dropsample vit_pytorch/max_vit.py ran fingerprinted MIT (permissive) · b3be1bfa6f51c051 · report fixture: returned Dropsample · level recorded 2026-09-22
FeedForward in 198 papers vit_pytorch/max_vit.py ran · metamorphic tier: invariant MIT (permissive) · 32bd4f3abdceaefe · report fixture: signature x:Tensor/2/float · returned Tensor · level recorded 2026-09-22
MBConv in 10 papers vit_pytorch/max_vit.py ran · our draft was wrong MIT (permissive) · 6b918699878401db · report fixture: signature dim_in:int/0/int, dim_out:int/0/int, downsample:bool/0/bool, expansion_rate:int/0/int, shrinkage_rate:float/0/float, dropout:float/0/float · returned Sequential · level recorded 2026-09-22
MBConvResidual vit_pytorch/max_vit.py ran fingerprinted MIT (permissive) · f4987984f431cfd0 · report fixture: returned MBConvResidual · level recorded 2026-09-22
MaxViT vit_pytorch/max_vit.py ran MIT (permissive) · 7e1c7833684c337a · report fixture: returned MaxViT · level recorded 2026-09-22
Residual in 24 papers vit_pytorch/max_vit.py ran fingerprinted MIT (permissive) · 636a208dabc2a711 · report fixture: returned Residual · level recorded 2026-09-22
SqueezeExcitation in 5 papers vit_pytorch/max_vit.py ran MIT (permissive) · b15b0232b5e08a9c · report fixture: returned SqueezeExcitation · level recorded 2026-09-22
qwopqwop200/MaxVIT-pytorch community (archive-listed) 3 samples, 2 ran (1 instrument-failed) · 1 unverified
block in 17 papers MaxVIT.py ran MIT (permissive) · 5d26f1a66e6a15e4 · report fixture: signature x:Tensor/4/float, window_size:tuple/1/int · returned Tensor · level recorded 2026-09-24
drop_path in 325 papers
census: 27 implementations from 61 papers, 5 distinct outputs across 2 buckets, one shared input per bucket this sample: bucket 1 of 2, arg 1: rank 4, kind float, dtype float32
MaxVIT.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant MIT (permissive) · 971ae8d2d8313e30 · report fixture: signature x:Tensor/4/float, drop_prob:float/0/float, training:bool/0/bool, scale_by_keep:bool/0/bool · returned Tensor · level recorded 2026-09-22
unblock MaxVIT.py unverified MIT (permissive) · 6eaf2972dbdcacc3 · report
towhee-io/towhee community (archive-listed) 12 samples, 7 ran (5 instrument-failed) · 5 unverified
RelativeSelfAttention towhee/models/max_vit/max_vit.py ran Apache-2.0 (permissive) · c72be25671079b1b · report fixture: returned RelativeSelfAttention · level recorded 2026-09-22
gelu_ignore_parameters towhee/models/max_vit/max_vit.py ran · our draft was wrong Apache-2.0 (permissive) · 48de7cab87cd24fa · report fixture: signature args:tuple/1/empty, kwargs:dict/0/? · returned GELU · level recorded 2026-09-22
get_relative_position_index in 11 papers towhee/models/max_vit/max_vit.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · 2e62acbb2e8ec0a9 · report fixture: signature win_h:int/0/int, win_w:int/0/int · returned Tensor · level recorded 2026-09-22
grid_partition towhee/models/max_vit/max_vit.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · 702e2dca4d951046 · report fixture: signature data:Tensor/4/float, grid_size:tuple/1/int · returned Tensor · level recorded 2026-09-22
grid_reverse towhee/models/max_vit/max_vit.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · 3b95c931d4a4355d · report fixture: signature grid:Tensor/4/float, original_size:tuple/1/int, grid_size:tuple/1/int · returned Tensor · level recorded 2026-09-22
window_partition in 377 papers
census: 13 implementations from 17 papers, 5 distinct outputs across 3 buckets, one shared input per bucket
towhee/models/max_vit/max_vit.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · ab7995b8493f8bd3 · report fixture: signature data:Tensor/4/float, window_size:tuple/1/int · returned Tensor · level recorded 2026-09-22
window_reverse in 296 papers towhee/models/max_vit/max_vit.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · 10fe1fb26b9e9e9d · report fixture: signature windows:Tensor/4/float, original_size:tuple/1/int, window_size:tuple/1/int · returned Tensor · level recorded 2026-09-22
MBConv in 10 papers towhee/models/max_vit/max_vit.py unverified Apache-2.0 (permissive) · 2df10618a97ad3c5 · report
MaxViT towhee/models/max_vit/max_vit.py unverified Apache-2.0 (permissive) · f6a21a6318030b5d · report
MaxViTBlock towhee/models/max_vit/max_vit.py unverified Apache-2.0 (permissive) · 04bab611e8e60c1e · report
MaxViTStage towhee/models/max_vit/max_vit.py unverified Apache-2.0 (permissive) · 1aad85bb6748a80a · report
MaxViTTransformerBlock towhee/models/max_vit/max_vit.py unverified Apache-2.0 (permissive) · da9db733c947bc49 · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. Siblings are assemblies lifted from this paper in the same batch for the same method: alternative readings kept side by side; none replaces or supersedes another.

5 assemblies on this paper: 5 harvested, 0 generated, listed in 2 groups (siblings lifted in one batch form one group).

harvested Assembly for multi-axis attention (graph method: multi-axis attention). 9 bricks; individually on a fixture: 8 ran (of which 3 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 5 where Syntology's instrument failed) · 1 unverified. The assembly has not been run end-to-end.

  1. PartitionAttention2d entry · level 0 · unverified · 641d6f2b8dbc0745
  2. Attention2d model · level 1 · ran · d227ba24d7d8d855
  3. LayerScale2d model · level 1 · ran · 8fa30deee34804c0
  4. MaxxVitTransformerCfg helper · level 1 · ran · 08082022ac2c6e65
  5. get_rel_pos_cls helper · level 1 · ran · our draft was wrong · 9d7347fa5b820e84
  6. grid_partition_nchw helper · level 1 · ran; the contract check could not run (RAISES) · f41c6472d974b50b
  7. grid_reverse_nchw helper · level 1 · ran; the contract check could not run (RAISES) · 80cee5fcd8a9952b
  8. window_partition_nchw helper · level 1 · ran; the contract check could not run (RAISES) · ae8a9a667b5d42f9
  9. window_reverse_nchw helper · level 1 · ran; the contract check could not run (RAISES) · c72f01470606b887

4 sibling assemblies for MaxViT (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.

harvested Assembly for MaxViT (graph method: MaxViT). 7 bricks; individually on a fixture: 5 ran (of which 3 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 2 where Syntology's instrument failed) · 2 unverified. The assembly has not been run end-to-end.

one of 4 siblings

  1. ResidualSplitHeadMultiAxisGmlpLayer entry · level 0 · unverified · e7fa9568a4a28913
  2. BlockGmlpLayer helper · level 1 · ran · 325b31d5363dae7d
  3. GridGmlpLayer helper · level 0 · unverified · bcddd7996b51db67
  4. BlockGatingUnit helper · level 1 · ran · d8da107e27ecbe28
  5. block_images_einops helper · level 1 · ran; the contract check could not run (RAISES) · 8a67f23552a44d61
  6. unblock_images_einops helper · level 1 · ran; the contract check could not run (RAISES) · 627ea4c76565d927
  7. GridGatingUnit helper · level 1 · ran · 86d6d05360b6903b

harvested Assembly for MaxViT (graph method: MaxViT). 11 bricks; individually on a fixture: 4 ran (of which 2 constructed an object rather than computing a result; 4 with no instrument failure: 1 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 7 unverified. The assembly has not been run end-to-end.

one of 4 siblings

  1. MaxViTBlock entry · level 0 · unverified · c99919623d98cd27
  2. Attention model · level 1 · ran · 2bc733f08e821139
  3. FFN model · level 0 · unverified · f50211588bf5be8b
  4. MBConvBlock model · level 0 · unverified · 2d7a9595c2215efd
  5. TrailDense helper · level 1 · ran · 839c67588aa822e2
  6. create_config_from_dict helper · level 0 · unverified · 256b0e25fa8d8cee
  7. float32_softmax helper · level 1 · ran · honoured contract · 5ecfa57092a8cecc
  8. SE model · level 0 · unverified · ee80113a460390fe
  9. config_batch_norm helper · level 0 · unverified · 20213f171bc1c7e9
  10. BatchNormalization helper · level 1 · ran · e2718d640d43e769
  11. TpuBatchNormalization helper · level 0 · unverified · d237bfe09ec44cfd

harvested Assembly for MaxViT (graph method: MaxViT). 10 bricks; individually on a fixture: 10 ran (of which 6 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 2 violated, 7 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. The assembly has not been run end-to-end.

one of 4 siblings

  1. MaxViT entry · level 1 · ran · 7e1c7833684c337a
  2. Attention model · level 1 · ran · 0ff4fe4361d49089
  3. FeedForward model · level 1 · ran · 32bd4f3abdceaefe
  4. MBConv model · level 1 · ran · our draft was wrong · 6b918699878401db
  5. Residual helper · level 1 · ran · 636a208dabc2a711
  6. default helper · level 1 · ran · violated contract · 60fff7c3c400d7ff
  7. MBConvResidual helper · level 1 · ran · f4987984f431cfd0
  8. SqueezeExcitation model · level 1 · ran · b15b0232b5e08a9c
  9. exists helper · level 1 · ran · violated contract · aa5486a3650902d8
  10. Dropsample helper · level 1 · ran · b3be1bfa6f51c051

harvested Assembly for MaxViT (graph method: MaxViT). 16 bricks; individually on a fixture: 11 ran (of which 2 constructed an object rather than computing a result; 3 with no instrument failure: 1 honoured, 0 violated, 2 with no contract checked; 8 where Syntology's instrument failed) · 5 unverified. The assembly has not been run end-to-end.

one of 4 siblings

  1. MaxViT entry · level 0 · unverified · f6a21a6318030b5d
  2. MaxViTStage helper · level 0 · unverified · 1aad85bb6748a80a
  3. drop_path helper · level 1 · ran; the contract check could not run · 39eace7e2822504f
  4. MaxViTBlock model · level 0 · unverified · 04bab611e8e60c1e
  5. MBConv helper · level 0 · unverified · 2df10618a97ad3c5
  6. MaxViTTransformerBlock model · level 0 · unverified · da9db733c947bc49
  7. grid_partition helper · level 1 · ran; the contract check could not run (RAISES) · 702e2dca4d951046
  8. grid_reverse helper · level 1 · ran; the contract check could not run (RAISES) · 3b95c931d4a4355d
  9. window_partition helper · level 1 · ran; the contract check could not run (RAISES) · ab7995b8493f8bd3
  10. window_reverse helper · level 1 · ran; the contract check could not run (RAISES) · 10fe1fb26b9e9e9d
  11. DropPath helper · level 1 · ran · 8f059bf3db11537b
  12. gelu_ignore_parameters helper · level 1 · ran · our draft was wrong · 48de7cab87cd24fa
  13. RelativeSelfAttention helper · level 1 · ran · c72be25671079b1b
  14. get_relative_position_index helper · level 1 · ran · honoured contract · 2e62acbb2e8ec0a9
  15. trunc_normal_ helper · level 1 · ran; the contract check could not run · f3e47c6e3250d13b
  16. _no_grad_trunc_normal_ helper · level 1 · ran; the contract check could not run · 607038475fdc388b

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2204.01697") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Image ClassificationObject Detectionimage-classificationobject-detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification ImageNet MaxViT-XL (512res, JFT) Top 1 Accuracy 89.53% #17 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-L (512res, JFT) Top 1 Accuracy 89.41% #20 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-XL (384res, JFT) Top 1 Accuracy 89.36% #21 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-L (384res, JFT) Top 1 Accuracy 89.12% #24 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-B (512res, JFT) Top 1 Accuracy 88.82% #29 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-XL (512res, 21K) Top 1 Accuracy 88.7% #32 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-B (384res, JFT) Top 1 Accuracy 88.69% #33 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-XL (384res, 21K) Top 1 Accuracy 88.51% #41 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-L (512res, 21K) Top 1 Accuracy 88.46% #46 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-B (512res, 21K) Top 1 Accuracy 88.38% #49 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-L (384res, 21K) Top 1 Accuracy 88.32% #53 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-B (512res) Top 1 Accuracy 86.7% #127 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-L (384res) Top 1 Accuracy 86.4% #144 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-B (384res) Top 1 Accuracy 86.34% #153 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-S (512res) Top 1 Accuracy 86.19% #171 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-T (384res) Top 1 Accuracy 85.72% #203 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-L (224res) GFLOPs 43.9 #252 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-L (224res) Number of params 212M #252 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-L (224res) Top 1 Accuracy 85.17% #252 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-B (224res) GFLOPs 23.4 #272 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-B (224res) Number of params 120M #272 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-B (224res) Top 1 Accuracy 84.94% #272 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-S (224res) GFLOPs 11.7 #315 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-S (224res) Number of params 69M #315 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-S (224res) Top 1 Accuracy 84.45% #315 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-T (224res) GFLOPs 5.6 #404 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-T (224res) Number of params 31M #404 of 1060 Archive leaderboard report
Image Classification ImageNet MaxViT-T (224res) Top 1 Accuracy 83.62% #404 of 1060 Archive leaderboard report
Object Detection COCO 2017 MaxViT-B AP 53.4 #1 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-B AP50 72.9 #1 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-B AP75 58.1 #1 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-B APM 45.7 #1 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-B APM50 70.3 #1 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-B APM75 50 #1 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-S AP 53.1 #2 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-S AP50 72.5 #2 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-S AP75 58.1 #2 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-S APM 45.4 #2 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-S APM50 69.8 #2 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-S APM75 49.5 #2 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-T AP 52.1 #3 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-T AP50 71.9 #3 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-T AP75 56.8 #3 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-T APM 44.6 #3 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-T APM50 69.1 #3 of 24 Archive leaderboard report
Object Detection COCO 2017 MaxViT-T APM75 48.4 #3 of 24 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections