SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2110.09468 · NeurIPS · 2021

Improving Robustness using Generated Data

Sylvestre-Alvise Rebuffi, London, Timothy Deepmind, Sven Gowal, Dan Calian, Florian Stimberg, Olivia Wiles

arXiv · PDF · Open in the Atlas

Code that ran

6 of 7 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 0 have no contract verdict; 6 failures of our instrument; 1 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/2110.09468.json · badge: /api/ran/2110.09468.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
imrahulr/adversarial_robustness_pytorch community (classifier label)
classifier basis not recorded on the link
6 ran · 1 no recorded run
FunctionStatusWhere it lives
accuracy Ran
ran; our contract draft was wrong, not the code
imrahulr/adversarial_robustness_pytorch/core/metrics.py
pointer only (licence: MIT); the stored copy get_code serves, from p2333/score, is pointer only (licence: NONE) · get_code("9aa516330b945a07")
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 p2333/score, 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.

calc_l2distsq Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
imrahulr/adversarial_robustness_pytorch/core/attacks/utils.py
code served (permissive licence) · get_code("6d57c3097e379953")
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 Tensor, ndim 1, dtype family float, shape (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 torch

torch.manual_seed(0)

# Create two small tensors
# x: batch size 2, feature dimension 4
x = torch.randn(2, 4)

# y: batch size 2, feature dimension 4
y = torch.randn(2, 4)

ARGS = {
    "x": x,
    "y": y,
}
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 Tensor, ndim 1, dtype family float, shape (2,) — shape (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
perturb_deepfool Ran
ran; our contract draft was wrong, not the code
imrahulr/adversarial_robustness_pytorch/core/attacks/deepfool.py
code served (permissive licence) · get_code("8a6e2e06ca96fb3c")
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 tuple, ndim 5, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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 numpy as np
import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a simple neural network for prediction
class SimplePredictor(nn.Module):
    def __init__(self, input_dim=4, num_classes=3):
        super().__init__()
        self.fc1 = nn.Linear(input_dim, 8)
        self.fc2 = nn.Linear(8, num_classes)
    
    def forward(self, x):
        # x shape: (batch_size, 1, 2, 2) or similar
        # Flatten spatial dimensions
        batch_size = x.shape[0]
        x_flat = x.view(batch_size, -1)
        x = torch.relu(self.fc1(x_flat))
        logits = self.fc2(x)
        return logits

# Create batch of input images: batch_size=2, channels=1, height=2, width=2
batch_size = 2
xvar = torch.randn(batch_size, 1, 2, 2)

# Create ground truth labels
yvar = torch.tensor([0, 2], dtype=torch.long)

# Create predictor model
predictor = SimplePredictor(input_dim=4, num_classes=3)
predictor.eval()

# Device (CPU only)
device = torch.device('cpu')

# Build arguments dictionary
ARGS = {
    'xvar': xvar,
    'yvar': yvar,
    'predict': predictor,
    'nb_iter': 5,
    'overshoot': 0.02,
    'ord': np.inf,
    'clip_min': 0.0,
    'clip_max': 1.0,
    'search_iter': 2,
    'device': device
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 5, dtype family float — a tuple 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
preact_resnet Ran
ran; our contract draft was wrong, not the code
imrahulr/adversarial_robustness_pytorch/core/models/preact_resnet.py
code served (permissive licence) · get_code("a99efdab28da25d8")
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 PreActResNet. 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)

ARGS = {
    'name': 'preact-resnet18',
    'num_classes': 10,
    'pretrained': False,
    'device': 'cpu',
}
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 PreActResNet — a PreActResNet output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a PreActResNet 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
replicate_input Ran
ran; our contract draft was wrong, not the code
imrahulr/adversarial_robustness_pytorch/core/attacks/utils.py
code served (permissive licence) · get_code("096e065144e00e11")
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 Tensor, ndim 2, dtype family float, shape (2, 4) 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

torch.manual_seed(0)

# Create a small tensor that will be used as input
x = torch.randn(2, 4)

# Map parameters to values
ARGS = {
    "x": x
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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
replicate_input_withgrad Ran
ran; our contract draft was wrong, not the code
imrahulr/adversarial_robustness_pytorch/core/attacks/utils.py
code served (permissive licence) · get_code("706e45721616b050")
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 Tensor, ndim 2, dtype family float, shape (2, 4) 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

torch.manual_seed(0)

# Create a small input tensor with requires_grad initially False
x = torch.randn(2, 4, requires_grad=False)

ARGS = {
    "x": x
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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
perturb_iterative Not yet run
unverified (harvested, no recorded run)
imrahulr/adversarial_robustness_pytorch/core/attacks/pgd.py
code served (permissive licence) · get_code("ec5326fd666c5823")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 7 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 7 later) and 0 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
accuracy 4 other papers
4 later
2202.10103 (2022-02)
Robustness and Accuracy Could Be Reconcilable by (Proper) Definition
2405.01817 (2024-05)
Uniformly Stable Algorithms for Adversarial Training and Beyond
2606.00738 (2026-06)
SORA: Free Second-Order Attacks in Fast Adversarial Training
and 1 more not named here
replicate_input 4 other papers
4 later
2302.03015 (2023-02)
Exploring and Exploiting Decision Boundary Dynamics for Adversarial Robustness
2405.16260 (2024-05)
Enhancing Consistency-Based Image Generation via Adversarialy-Trained Classification and Energy-Based Discrimination
2406.13283 (2024-06)
Large-Scale Dataset Pruning in Adversarial Training through Data Importance Extrapolation
and 1 more not named here
calc_l2distsq 3 other papers
3 later
2405.16260 (2024-05)
Enhancing Consistency-Based Image Generation via Adversarialy-Trained Classification and Energy-Based Discrimination
2406.13283 (2024-06)
Large-Scale Dataset Pruning in Adversarial Training through Data Importance Extrapolation
2606.02267 (2026-06)
A combination of noise and bilateral filters achieve supralinear and scalable adversarial robustness in CNNs
perturb_deepfool 3 other papers
3 later
2406.13283 (2024-06)
Large-Scale Dataset Pruning in Adversarial Training through Data Importance Extrapolation
2606.00738 (2026-06)
SORA: Free Second-Order Attacks in Fast Adversarial Training
2606.02267 (2026-06)
A combination of noise and bilateral filters achieve supralinear and scalable adversarial robustness in CNNs
preact_resnet 3 other papers
3 later
2310.07931 (2023-10)
D2 Pruning: Message Passing for Balancing Diversity and Difficulty in Data Pruning
2406.13283 (2024-06)
Large-Scale Dataset Pruning in Adversarial Training through Data Importance Extrapolation
2606.02267 (2026-06)
A combination of noise and bilateral filters achieve supralinear and scalable adversarial robustness in CNNs
replicate_input_withgrad 3 other papers
3 later
2405.16260 (2024-05)
Enhancing Consistency-Based Image Generation via Adversarialy-Trained Classification and Energy-Based Discrimination
2406.13283 (2024-06)
Large-Scale Dataset Pruning in Adversarial Training through Data Importance Extrapolation
2606.02267 (2026-06)
A combination of noise and bilateral filters achieve supralinear and scalable adversarial robustness in CNNs
perturb_iterative 2 other papers
2 later
2406.13283 (2024-06)
Large-Scale Dataset Pruning in Adversarial Training through Data Importance Extrapolation
2606.02267 (2026-06)
A combination of noise and bilateral filters achieve supralinear and scalable adversarial robustness in CNNs

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

Recent work argues that robust training requires substantially larger datasets than those required for standard classification. On CIFAR-10 and CIFAR-100, this translates into a sizable robust-accuracy gap between models trained solely on data from the original training set and those trained with additional data extracted from the "80 Million Tiny Images" dataset (80M-TI). In this paper, we explore how generative models trained solely on the original training set can be leveraged to artificially increase the size of the original training set and improve adversarial robustness to p norm-bounded perturbations. We identify the sufficient conditions under which incorporating additional generated data can improve robustness, and demonstrate that it is possible to significantly reduce the robust-accuracy gap to models trained with additional real data. Surprisingly, we show that even the addition of non-realistic random data (generated by Gaussian sampling) can improve robustness. We evaluate our approach on CIFAR-10, CIFAR-100, SVHN and TINYIMAGENET against ∞ and 2 norm-bounded perturbations of size = 8/255 and = 128/255, respectively. We show large absolute improvements in robust accuracy compared to previous state-of-the-art methods. Against ∞ normbounded perturbations of size = 8/255, our models achieve 66.10% and 33.49% robust accuracy on CIFAR-10 and CIFAR-100, respectively (improving upon the state-of-the-art by +8.96% and +3.29%). Against 2 norm-bounded perturbations of size = 128/255, our model achieves 78.31% on CIFAR-10 (+3.81%). These results beat most prior works that use external data.

For agents

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

get_harvested_code_for_paper("2110.09468")
get_code_for_paper("2110.09468")
have("2110.09468")

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

curl https://syntology.ai/api/ran/2110.09468.json

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

[![Syntology run record](https://syntology.ai/api/ran/2110.09468.svg)](https://syntology.ai/paper/2110.09468)

Connect an agent — have() is free.