SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2103.17239 · ICCV · 2021

Going deeper with Image Transformers

Matthieu Cord, Gabriel Synnaeve, Alexandre Sablayrolles, Hugo Touvron, Hervé Jégou

arXiv · PDF · Open in the Atlas

Code that ran

9 of 11 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; 0 honoured the contract we drafted, 2 violated it and 4 have no contract verdict; 3 failures of our instrument; 2 no recorded run), measured 2026-11-21; 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/2103.17239.json · badge: /api/ran/2103.17239.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
CPones/Paddle_CaiT community (classifier label)
classifier basis not recorded on the link
2 ran · 2 no recorded run
sayakpaul/cait-tf community (classifier label)
classifier basis not recorded on the link
2 ran · 0 no recorded run
copy not recorded community
not in the paper's declared repositories
2 ran · 0 no recorded run
DarshanDeshpande/jax-models reimplementation (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
conceptofmind/CaiT-Flax reimplementation (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
FunctionStatusWhere it lives
conv_transpose Ran
ran on a synthesized input
sayakpaul/cait-tf/utils/helpers.py
code served (permissive licence) · get_code("e218ba48c9d2c190")
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 ndarray, ndim 4, dtype family float, shape (4, 5, 3, 2) 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 numpy as np

# Create a small weight tensor for a PyTorch conv layer
# Conv layer weights in PyTorch have shape (out_channels, in_channels, kernel_h, kernel_w)
# Using distinct dimensions: 2, 3, 4, 5
w = np.random.randn(2, 3, 4, 5)

ARGS = {
    "w": w
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 4, dtype family float, shape (4, 5, 3, 2) — shape (4, 5, 3, 2) 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
dropout_layers Ran
ran; our contract draft was wrong, not the code
conceptofmind/CaiT-Flax/cait.py
code served (permissive licence) · get_code("7794eb689431c216")
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 list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 jax
import jax.numpy as jnp
from random import randrange

torch.manual_seed(0)

# Create a list of simple layer objects
# The function only uses len(layers) and iterates over layers with zip
# So we can use simple objects or even strings as placeholders
layers = ["layer_0", "layer_1", "layer_2", "layer_3"]

# dropout is a float between 0 and 1
dropout = 0.3

ARGS = {
    "layers": layers,
    "dropout": dropout,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
jax, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
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
exists Ran
ran, violated the contract we drafted
this paper's copy was not recorded; identical code first harvested from ThomasMrY/VCT
pointer only; the stored copy get_code serves, from ThomasMrY/VCT, is pointer only (licence: NONE) · get_code("aa5486a3650902d8")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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_args Ran
ran on a synthesized input
CPones/Paddle_CaiT/CaiT/infer.py
code served (permissive licence) · get_code("5fe12143b1680470")
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 Namespace. 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 argparse

def str2bool(v):
    return v.lower() in ('true', 't', '1')

ARGS = {
    'add_help': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Namespace — a Namespace output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Namespace 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
get_exclude_from_weight_decay_fn Ran
ran on a synthesized input
CPones/Paddle_CaiT/CaiT/utils.py
code served (permissive licence) · get_code("7cdb59fd6a14bd54")
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 function. 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 torch.nn as nn

torch.manual_seed(0)

# The function takes exclude_list as a parameter
# exclude_list should be a list of parameter name strings to exclude from weight decay
# Based on the function body, it checks if param names end with strings in exclude_list

# Test case 1: empty list (returns None)
ARGS = {
    "exclude_list": []
}

# Alternative test case with actual exclusion list
# This would test the exclude_fn logic
ARGS_WITH_EXCLUSIONS = {
    "exclude_list": ["bias", "norm.weight", "pos_embedding"]
}

# Use the version with exclusions to make the function more interesting
ARGS = {
    "exclude_list": ["bias", "norm.weight", "pos_embedding"]
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
mlp Ran
ran on a synthesized input
sayakpaul/cait-tf/cait/models.py
code served (permissive licence) · get_code("ae83452001c08f2a")
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 KerasTensor. 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 tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
import numpy as np

# Create a small input tensor (batch=2, features=4)
input_tensor = tf.keras.Input(shape=(4,))

# Define arguments for the mlp function
ARGS = {
    'x': input_tensor,
    'dropout_rate': 0.1,
    'hidden_units': [8, 16, 3]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type KerasTensor — a KerasTensor output has no array shape of its own; none is 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
pair Ran
ran, violated the contract we drafted
this paper's copy was not recorded; identical code first harvested from LINs-lab/DynMoE
pointer only · get_code("6ba8cee9f5daea41")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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.

window_partition Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
DarshanDeshpande/jax-models/jax_models/models/swin_transformer.py
code served (permissive licence) · get_code("07b1c6d8d9489187")
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 ArrayImpl, ndim 4, dtype family float, shape (12, 2, 2, 3) 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 jax.numpy as jnp

# Create a small input tensor with shape (batch, height, width, channels)
# batch=2, height=4, width=6, channels=3
# window_size must divide both height and width evenly
batch = 2
height = 4
width = 6
channels = 3
window_size = 2

# Create input array with JAX
x = jnp.ones((batch, height, width, channels))

ARGS = {
    "x": x,
    "window_size": window_size
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 4, dtype family float, shape (12, 2, 2, 3) — shape (12, 2, 2, 3) 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
window_reverse Ran
ran; our contract draft was wrong, not the code
DarshanDeshpande/jax-models/jax_models/models/swin_transformer.py
code served (permissive licence) · get_code("0859cb39e97935ba")
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 ArrayImpl, ndim 4, dtype family float, shape (2, 8, 8, 3) 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 jax.numpy as jnp
import numpy as np

# Analyze the function:
# - windows: a JAX array that gets reshaped
# - window_size: int, used in division and reshape
# - height: int, used in division and reshape
# - width: int, used in division and reshape
# The function reverses a windowing operation on image patches

# Set up parameters
window_size = 4
height = 8
width = 8
batch = 2

# Calculate the expected shape of windows
# From the reshape logic: windows.shape[0] = batch * (height * width / window_size / window_size)
# windows.shape[0] = 2 * (8 * 8 / 4 / 4) = 2 * 4 = 8
# The windows are reshaped to (batch, height//window_size, width//window_size, window_size, window_size, channels)
# So windows should have shape (8, 4, 4, num_channels) where num_channels is arbitrary
num_channels = 3
windows_shape = (batch * (height * width // (window_size * window_size)), window_size, window_size, num_channels)

# Create a small JAX array for windows
windows = jnp.ones(windows_shape, dtype=jnp.float32)

ARGS = {
    "windows": windows,
    "window_size": window_size,
    "height": height,
    "width": width,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
jax, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 4, dtype family float, shape (2, 8, 8, 3) — shape (2, 8, 8, 3) 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
get_config Not yet run
unverified (harvested, no recorded run)
CPones/Paddle_CaiT/CaiT/config.py
code served (permissive licence) · get_code("160fd4002bd2e51b")
How it ran

no invocation recorded.

update_config Not yet run
unverified (harvested, no recorded run)
CPones/Paddle_CaiT/CaiT/config.py
code served (permissive licence) · get_code("ee77e3fcb05ad040")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 5 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 4 the same month, 3 later) and 6 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
exists 486 other papers
35 dated earlier, 5 the same month, 444 later, 2 undated · 78 only as an assembly brick
1603.08155 (2016-03)
Perceptual Losses for Real-Time Style Transfer and Super-Resolution
1611.07004 (2016-11)
Image-to-Image Translation with Conditional Adversarial Networks
1804.04637 (2018-04)
EMBER: An Open Dataset for Training Static PE Malware Machine Learning Models
and 483 more not named here
pair 90 other papers
2 dated earlier, 1 the same month, 86 later, 1 undated · 8 only as an assembly brick
2004.13649 (2020-04)
Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixels
2010.11929 (2020-10)
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
2103.15691 (2021-03)
ViViT: A Video Vision Transformer
and 87 more not named here
get_config 2 other papers
2 later
2111.06377 (2021-11)
Masked Autoencoders Are Scalable Vision Learners
2111.09883 (2021-11)
Swin Transformer V2: Scaling Up Capacity and Resolution
window_partition 1 other paper
1 the same month
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
window_reverse 1 other paper
1 the same month
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

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 2026-11-21.

Repositories linked to this paper

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

Abstract

Transformers have been recently adapted for large scale image classification, achieving high scores shaking up the long supremacy of convolutional neural networks. However the optimization of vision transformers has been little studied so far. In this work, we build and optimize deeper transformer networks for image classification. In particular, we investigate the interplay of architecture and optimization of such dedicated transformers. We make two architecture changes that significantly improve the accuracy of deep transformers. This leads us to produce models whose performance does not saturate early with more depth, for instance we obtain 86.5% top-1 accuracy on Imagenet when training with no external data, we thus attain the current sate of the art with less floating-point operations and parameters. Our best model establishes the new state of the art on Imagenet with Reassessed labels and Imagenet-V2 / match frequency, in the setting with no additional training data. We share our code and models 1 .

For agents

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

get_harvested_code_for_paper("2103.17239")
get_code_for_paper("2103.17239")
have("2103.17239")

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

curl https://syntology.ai/api/ran/2103.17239.json

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

[![Syntology run record](https://syntology.ai/api/ran/2103.17239.svg)](https://syntology.ai/paper/2103.17239)

Connect an agent — have() is free.