SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2206.08477 · 2022

Backdoor Attacks on Vision Transformers

arXiv · PDF · Open in the Atlas

Code that ran

3 of 5 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, 0 violated it and 3 have no contract verdict; 0 failures of our instrument; 2 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/2206.08477.json · badge: /api/ran/2206.08477.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
ucdvision/backdoor_transformer canonical (classifier label)
classifier basis not recorded on the link
3 ran · 2 no recorded run
FunctionStatusWhere it lives
deit_base_patch16_224 Ran
ran on a synthesized input
ucdvision/backdoor_transformer/generate_poison_transformer.py
code served (permissive licence) · get_code("c81963c14587e622")
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 VisionTransformer. 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
from functools import partial
from timm.models.vision_transformer import _cfg

torch.manual_seed(0)

# The function deit_base_patch16_224 takes two parameters:
# 1. pretrained (bool, default True)
# 2. **kwargs (additional arguments passed to VisionTransformer)

# Looking at the function body:
# - If pretrained=True, it attempts to load from torch.hub (which will fail without network)
# - It constructs a VisionTransformer with specific parameters
# - The model needs to be a valid nn.Module that can accept load_state_dict

# To avoid network calls, we set pretrained=False
# For **kwargs, VisionTransformer accepts parameters like num_classes, img_size, etc.

ARGS = {
    'pretrained': False,
    'num_classes': 1000,
    'img_size': 224,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
timm, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type VisionTransformer — a VisionTransformer output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a VisionTransformer 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
deit_small_patch16_224 Ran
ran on a synthesized input
ucdvision/backdoor_transformer/generate_poison_transformer.py
code served (permissive licence) · get_code("2dbbcbdf7578ed09")
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 VisionTransformer. 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
from functools import partial

torch.manual_seed(0)

# The function deit_small_patch16_224 takes:
# - pretrained: bool (optional, defaults to True)
# - **kwargs: additional arguments to pass to VisionTransformer

# Since the function tries to download from torch.hub when pretrained=True,
# we'll set pretrained=False to avoid network calls
# The **kwargs would be passed to VisionTransformer constructor

ARGS = {
    'pretrained': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
timm, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type VisionTransformer — a VisionTransformer output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a VisionTransformer 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
deit_tiny_patch16_224 Ran
ran on a synthesized input
ucdvision/backdoor_transformer/generate_poison_transformer.py
code served (permissive licence) · get_code("22cdde7810d49c8e")
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 VisionTransformer. 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
from functools import partial
from timm.models.vision_transformer import VisionTransformer, _cfg

torch.manual_seed(0)

ARGS = {
    "pretrained": False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
timm, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type VisionTransformer — a VisionTransformer output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a VisionTransformer 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
grad_rollout Not yet run
unverified (harvested, no recorded run)
ucdvision/backdoor_transformer/vit_grad_rollout.py
code served (permissive licence) · get_code("cb8cb0f36e74db72")
How it ran

no invocation recorded.

grad_rollout_batch Not yet run
unverified (harvested, no recorded run)
ucdvision/backdoor_transformer/vit_grad_rollout.py
code served (permissive licence) · get_code("bfa739fab718ba36")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 4 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
grad_rollout 1 other paper
1 dated earlier
2010.11929 (2020-10)
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

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 Transformers (ViT) have recently demonstrated exemplary performance on a variety of vision tasks and are being used as an alternative to CNNs. Their design is based on a self-attention mechanism that processes images as a sequence of patches, which is quite different compared to CNNs. Hence it is interesting to study if ViTs are vulnerable to backdoor attacks. Backdoor attacks happen when an attacker poisons a small part of the training data for malicious purposes. The model performance is good on clean test images, but the attacker can manipulate the decision of the model by showing the trigger at test time. To the best of our knowledge, we are the first to show that ViTs are vulnerable to backdoor attacks. We also find an intriguing difference between ViTs and CNNs - interpretation algorithms effectively highlight the trigger on test images for ViTs but not for CNNs. Based on this observation, we propose a test-time image blocking defense for ViTs which reduces the attack success rate by a large margin. Code is available here: https://github.com/UCDvision/backdoor_transformer.git

For agents

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

get_harvested_code_for_paper("2206.08477")
get_code_for_paper("2206.08477")
have("2206.08477")

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

curl https://syntology.ai/api/ran/2206.08477.json

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

[![Syntology run record](https://syntology.ai/api/ran/2206.08477.svg)](https://syntology.ai/paper/2206.08477)

Connect an agent — have() is free.