SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2106.04263 · ICLR · 2022

On the Connection between Local Attention and Dynamic Depth-wise Convolution

Jingdong Wang, Qi Han, Zejia Fan, Qi Dai, Lei Sun, Ming-Ming Cheng, Jiaying Liu, † Tklndst, Nankai Univerisy

arXiv · PDF · Open in the Atlas

Code that ran

11 of 16 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 2 violated it and 0 have no contract verdict; 7 failures of our instrument; 5 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2106.04263.json · badge: /api/ran/2106.04263.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
Atten4Vis/DemystifyLocalViT canonical (classifier label)
classifier basis: census
11 ran · 5 no recorded run
FunctionStatusWhere it lives
GET_BLOCKS Ran
ran, violated the contract we drafted
Atten4Vis/DemystifyLocalViT/models/idynamic.py
code served (permissive licence) · get_code("c5acf386fe00c470")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# The function GET_BLOCKS takes a single parameter N (an integer representing number of elements)
# and calculates how many blocks are needed given CUDA_NUM_THREADS = 1024

N = 2048  # Example: number of elements to process

ARGS = {
    'N': N
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
auto_resume_helper Ran
ran; our contract draft was wrong, not the code
Atten4Vis/DemystifyLocalViT/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from leaplabthu/inline, is pointer only (licence: NONE) · get_code("0c32459bef92ff8c")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from leaplabthu/inline, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

build_optimizer Ran
ran; our contract draft was wrong, not the code
Atten4Vis/DemystifyLocalViT/optimizer.py
pointer only (licence: MIT); the stored copy get_code serves, from CBalance/SSD, is pointer only (licence: NONE) · get_code("e35a4b2e4114b1a5")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from CBalance/SSD, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

build_scheduler Ran
ran; our contract draft was wrong, not the code
Atten4Vis/DemystifyLocalViT/lr_scheduler.py
pointer only (licence: MIT); the stored copy get_code serves, from ghrua/seqpe, is pointer only (licence: NONE) · get_code("c02cc4e0b5720211")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ghrua/seqpe, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

check_keywords_in_name Ran
ran, violated the contract we drafted
Atten4Vis/DemystifyLocalViT/optimizer.py
pointer only (licence: MIT); the stored copy get_code serves, from leaplabthu/inline, is pointer only (licence: NONE) · get_code("b0a5beb34716d5a6")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from leaplabthu/inline, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

create_logger Ran
ran; our contract draft was wrong, not the code
Atten4Vis/DemystifyLocalViT/logger.py
pointer only (licence: MIT); the stored copy get_code serves, from ghrua/seqpe, is pointer only (licence: NONE) · get_code("80450600f7f09b0a")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ghrua/seqpe, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_final_iter Ran
ran, honoured the contract we drafted
Atten4Vis/DemystifyLocalViT/downstreams/segmentation/.dev/gather_models.py
pointer only (licence: MIT); the stored copy get_code serves, from zmhhmz/resfu, is pointer only (licence: NONE) · get_code("88d68650d669bb11")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from zmhhmz/resfu, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_final_results Ran
ran; our contract draft was wrong, not the code
Atten4Vis/DemystifyLocalViT/downstreams/detection/.dev_scripts/gather_models.py
pointer only (licence: MIT); the stored copy get_code serves, from zhaoyangwei123/sapnet, is pointer only (licence: NONE) · get_code("2268355e37eebb82")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from zhaoyangwei123/sapnet, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_grad_norm Ran
ran, honoured the contract we drafted
Atten4Vis/DemystifyLocalViT/utils.py
code served (permissive licence) · get_code("eec1e7cba51d5e8e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a simple model with parameters that have gradients
model = nn.Linear(4, 3)

# Forward pass to create some gradients
x = torch.randn(2, 4)
y = model(x)
loss = y.sum()
loss.backward()

# Now model.parameters() have gradients attached
parameters = list(model.parameters())

ARGS = {
    "parameters": parameters,
    "norm_type": 2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
process_checkpoint Ran
ran; our contract draft was wrong, not the code
Atten4Vis/DemystifyLocalViT/downstreams/detection/.dev_scripts/gather_models.py
code served (permissive licence) · get_code("e76a491553cf9fcf")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import tempfile
import os

torch.manual_seed(0)

# Create a temporary directory for checkpoint files
temp_dir = tempfile.mkdtemp()

# Create a minimal checkpoint dict with a model state and optimizer
checkpoint = {
    'model': {
        'layer1.weight': torch.randn(4, 3),
        'layer1.bias': torch.randn(4),
    },
    'optimizer': {
        'state': {},
        'param_groups': [],
    },
    'epoch': 1,
}

# Save the checkpoint to a temporary file
in_file = os.path.join(temp_dir, 'checkpoint.pth')
torch.save(checkpoint, in_file)

# Define output file path
out_file = os.path.join(temp_dir, 'checkpoint_processed.pth')

ARGS = {
    'in_file': in_file,
    'out_file': out_file,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
set_weight_decay Ran
ran; our contract draft was wrong, not the code
Atten4Vis/DemystifyLocalViT/optimizer.py
pointer only (licence: MIT); the stored copy get_code serves, from ztx-100/efficient_vit_with_dw, is pointer only (licence: NONE) · get_code("b33222a09fc93bec")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ztx-100/efficient_vit_with_dw, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Dtype Not yet run
unverified (harvested, no recorded run)
Atten4Vis/DemystifyLocalViT/models/idynamic.py
pointer only (licence: MIT); the stored copy get_code serves, from thomasverelst/dynconv, is pointer only (licence: NONE) · get_code("1e624f979ea53ae4")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from thomasverelst/dynconv, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

build_model Not yet run
unverified (harvested, no recorded run)
Atten4Vis/DemystifyLocalViT/models/build.py
code served (permissive licence) · get_code("dd74736c6244d64b")
How it ran

no invocation recorded.

get_config Not yet run
unverified (harvested, no recorded run)
Atten4Vis/DemystifyLocalViT/config.py
code served (permissive licence) · get_code("c32302734842a753")
How it ran

no invocation recorded.

get_final_results Not yet run
unverified (harvested, no recorded run)
Atten4Vis/DemystifyLocalViT/downstreams/segmentation/.dev/gather_models.py
code served (permissive licence) · get_code("6c10c287d59a984c")
How it ran

no invocation recorded.

load_checkpoint Not yet run
unverified (harvested, no recorded run)
Atten4Vis/DemystifyLocalViT/utils.py
code served (permissive licence) · get_code("d690bbafcc136a37")
How it ran

no invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 14 are also present in other papers whose code we hold (12 with at least one paper dated earlier, 8 the same month, 14 later) and 2 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
get_config 51 other papers
1 dated earlier, 3 the same month, 47 later
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
2106.03650 (2021-06)
Shuffle Transformer: Rethinking Spatial Shuffle for Vision Transformer
2106.03746 (2021-06)
Efficient Training of Visual Transformers with Small Datasets
and 48 more not named here
process_checkpoint 45 other papers
13 dated earlier, 31 later, 1 undated
1904.11490 (2019-04)
RepPoints: Point Set Representation for Object Detection
1906.07155 (2019-06)
MMDetection: Open MMLab Detection Toolbox and Benchmark
1908.05612 (2019-08)
R3Det: Refined Single-Stage Detector with Feature Refinement for Rotating Object
and 42 more not named here
create_logger 35 other papers
1 dated earlier, 2 the same month, 32 later
2104.13840 (2021-04)
Twins: Revisiting the Design of Spatial Attention in Vision Transformers
2106.03650 (2021-06)
Shuffle Transformer: Rethinking Spatial Shuffle for Vision Transformer
2106.03746 (2021-06)
Efficient Training of Visual Transformers with Small Datasets
and 32 more not named here
check_keywords_in_name 31 other papers
1 dated earlier, 2 the same month, 28 later
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
2106.03650 (2021-06)
Shuffle Transformer: Rethinking Spatial Shuffle for Vision Transformer
2106.03746 (2021-06)
Efficient Training of Visual Transformers with Small Datasets
and 28 more not named here
get_grad_norm 23 other papers
2 dated earlier, 1 the same month, 20 later
1910.13616 (2019-10)
Multimodal Model-Agnostic Meta-Learning via Task-Aware Modulation
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
2106.03650 (2021-06)
Shuffle Transformer: Rethinking Spatial Shuffle for Vision Transformer
and 20 more not named here
auto_resume_helper 20 other papers
1 dated earlier, 1 the same month, 18 later
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
2106.03650 (2021-06)
Shuffle Transformer: Rethinking Spatial Shuffle for Vision Transformer
2107.08391 (2021-07)
AS-MLP: An Axial Shifted MLP Architecture for Vision
and 17 more not named here
set_weight_decay 14 other papers
1 dated earlier, 2 the same month, 11 later
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
2106.03650 (2021-06)
Shuffle Transformer: Rethinking Spatial Shuffle for Vision Transformer
2106.03746 (2021-06)
Efficient Training of Visual Transformers with Small Datasets
and 11 more not named here
build_scheduler 13 other papers
2 the same month, 11 later
2106.03650 (2021-06)
Shuffle Transformer: Rethinking Spatial Shuffle for Vision Transformer
2106.03746 (2021-06)
Efficient Training of Visual Transformers with Small Datasets
2107.08391 (2021-07)
AS-MLP: An Axial Shifted MLP Architecture for Vision
and 10 more not named here
get_final_iter 13 other papers
3 dated earlier, 10 later
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
2105.05409 (2021-05)
A Large-Scale Benchmark for Food Image Segmentation
2105.15203 (2021-05)
SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers
and 10 more not named here
get_final_results 10 other papers
3 dated earlier, 7 later
1906.07155 (2019-06)
MMDetection: Open MMLab Detection Toolbox and Benchmark
1911.10683 (2019-11)
Image-based table recognition: data, model, and evaluation
2008.13367 (2020-08)
VarifocalNet: An IoU-aware Dense Object Detector
and 7 more not named here
build_optimizer 8 other papers
2 the same month, 6 later
2106.03650 (2021-06)
Shuffle Transformer: Rethinking Spatial Shuffle for Vision Transformer
2106.03746 (2021-06)
Efficient Training of Visual Transformers with Small Datasets
2107.08391 (2021-07)
AS-MLP: An Axial Shifted MLP Architecture for Vision
and 5 more not named here
GET_BLOCKS 5 other papers
1 dated earlier, 4 later
1710.10903 (2017-10)
Graph Attention Networks
2107.08391 (2021-07)
AS-MLP: An Axial Shifted MLP Architecture for Vision
2204.07143 (2022-04)
Neighborhood Attention Transformer
and 2 more not named here
Dtype 4 other papers
1 dated earlier, 3 later
1912.03203 (2019-12)
Dynamic Convolutions: Exploiting Spatial Sparsity for Faster Inference
2107.08391 (2021-07)
AS-MLP: An Axial Shifted MLP Architecture for Vision
and 1 more not named here
get_final_results 4 other papers
2 dated earlier, 2 later
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
2105.05409 (2021-05)
A Large-Scale Benchmark for Food Image Segmentation
Guo_ISDNet_Integrating_Shallow_and_Deep_Networks_for_Efficient_Ultra-High_Resolution_CVPR_2022_paper (2022 (year only))
ISDNet: Integrating Shallow and Deep Networks for Efficient Ultra-high Resolution Segmentation
and 1 more not named here

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Vision Transformer (ViT) attains state-of-the-art performance in visual recognition, and the variant, Local Vision Transformer, makes further improvements. The major component in Local Vision Transformer, local attention, performs the attention separately over small local windows. We rephrase local attention as a channel-wise locally-connected layer and analyze it from two network regularization manners, sparse connectivity and weight sharing, as well as dynamic weight computation. We point out that local attention resembles depth-wise convolution and its dynamic variants in sparse connectivity: there is no connection across channels, and each position is connected to the positions within a small local window. The main differences lie in (i) weight sharing -depth-wise convolution shares connection weights (kernel weights) across spatial positions and attention shares the connection weights across channels, and (ii) dynamic weight computation manners -local attention is based on dot-products between pairwise positions in the local window, and dynamic convolution is based on linear projections conducted on the center representation or the globally pooled representation. The connection between local attention and dynamic depth-wise convolution is empirically verified by the ablation study about weight sharing and dynamic weight computation in Local Vision Transformer and (dynamic) depth-wise convolution based network, namely (dynamic) DWNet. We empirically observe that the depth-wise convolution based DWNet and its dynamic variants with lower computation complexity perform on-par with or slightly better than Swin Transformer, an instance of Local Vision Transformer, for ImageNet classification, COCO object detection and ADE semantic segmentation. Code is available at https://github.com/Atten4Vis/DemystifyLocalViT.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2106.04263")
get_code_for_paper("2106.04263")
have("2106.04263")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2106.04263.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2106.04263.svg)](https://syntology.ai/paper/2106.04263)

Connect an agent — have() is free.