SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2206.10550 · 2022

(Certified!!) Adversarial Robustness for Free!

arXiv · PDF · Open in the Atlas

Code that ran

28 of 34 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; 4 honoured the contract we drafted, 0 violated it and 10 have no contract verdict; 14 failures of our instrument; 6 no recorded run); 1 of them was also run on the host, not in the jail, by the peer-agreement lane that set its level, 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/2206.10550.json · badge: /api/ran/2206.10550.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
ethz-privsec/diffusion_denoised_smoothing canonical (classifier label)
classifier basis not recorded on the link
12 ran · 6 no recorded run
blaisedelattre/bridging_the_gap_rs community (classifier label)
classifier basis not recorded on the link
16 ran · 0 no recorded run
FunctionStatusWhere it lives
center_crop_arr Ran
ran; our contract draft was wrong, not the code
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/image_datasets.py
code served (permissive licence) · get_code("8ab9a97df8161f31")
How it ran

Ran in Syntology's jail on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned ndarray, ndim 3, dtype family int, shape (64, 64, 3) as a later run recorded it. The image it ran in was not recorded. The comparison behind this sample's level is a different run: the peer-agreement lane ran it again on the host, not in the jail (report dated 2026-09-05T00:51:18Z). 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.

from PIL import Image
import numpy as np

# Create a small PIL Image for testing
# Make it larger than 2 * image_size so we can test the resizing logic
pil_image = Image.new('RGB', (256, 256), color='red')

# Define image_size parameter
image_size = 64

# Construct ARGS dictionary
ARGS = {
    'pil_image': pil_image,
    'image_size': image_size
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
PIL, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (64, 64, 3) — shape (64, 64, 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
approx_standard_normal_cdf Ran
ran; our contract draft was wrong, not the code
blaisedelattre/bridging_the_gap_rs/code/improved_diffusion/losses.py
pointer only (licence: MIT); the stored copy get_code serves, from quandao10/slct, is pointer only (licence: NONE) · get_code("cfd76fd0d89574a4")
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 quandao10/slct, 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.

betas_for_alpha_bar Ran
ran, honoured the contract we drafted
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/gaussian_diffusion.py
pointer only (licence: MIT); the stored copy get_code serves, from NVlabs/DiffPure, is pointer only (licence: NOASSERTION) · get_code("2ab2316ac6fdd869")
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 NVlabs/DiffPure, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

compute_batchnorm_lipschitz Ran
ran on a synthesized input
blaisedelattre/bridging_the_gap_rs/code/PUB/compute_lip_PUB_resnet110.py
code served (permissive licence) · get_code("e24d913c8ab83bf6")
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 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 BatchNorm2d layer with affine=True
# The function accesses: layer.affine, layer.weight.data, layer.running_var.data, layer.eps
batchnorm_layer = nn.BatchNorm2d(num_features=4, affine=True, eps=1e-5)

# Initialize the running_var to ensure it has valid values
batchnorm_layer.running_var.data = torch.ones(4) * 0.5

ARGS = {
    "layer": batchnorm_layer
}
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 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
compute_lipschitz_constant_smoothed_classifier Ran
ran on a synthesized input
blaisedelattre/bridging_the_gap_rs/code/compute_rs_lipschitz_radius.py
code served (permissive licence) · get_code("bc700205671c1850")
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 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 math

ARGS = {
    "L": 2.5,
    "r": 1.0,
    "sigma": 0.3
}
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 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
compute_module_lipschitz Ran
ran on a synthesized input
blaisedelattre/bridging_the_gap_rs/code/PUB/compute_lip_PUB_resnet110.py
code served (permissive licence) · get_code("a5ec2c7b30f09261")
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 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)

# Build a small Conv2d layer
conv_layer = nn.Conv2d(in_channels=3, out_channels=16, kernel_size=3, padding=1)

# Build a small Linear layer
linear_layer = nn.Linear(in_features=64, out_features=32)

# Build a BatchNorm2d layer
bn_layer = nn.BatchNorm2d(num_features=16)

# Build a simple module for testing
simple_module = nn.ReLU()

# Build a Sequential module
sequential_module = nn.Sequential(
    nn.Conv2d(3, 8, kernel_size=3, padding=1),
    nn.BatchNorm2d(8),
    nn.ReLU()
)

# Input shape for Conv2d: (batch_size, channels, height, width)
conv_input_shape = (2, 3, 8, 8)

# Input shape for Linear: (batch_size, features)
linear_input_shape = (2, 64)

# Input shape for BatchNorm2d: (batch_size, channels, height, width)
bn_input_shape = (2, 16, 8, 8)

# Input shape for ReLU: (batch_size, channels, height, width)
relu_input_shape = (2, 16, 8, 8)

# Input shape for Sequential: (batch_size, channels, height, width)
sequential_input_shape = (2, 3, 8, 8)

ARGS = {
    "module": conv_layer,
    "input_shape": conv_input_shape
}
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 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
compute_radius_tsuzuku_for_smoothed_classifier Ran
ran on a synthesized input
blaisedelattre/bridging_the_gap_rs/code/compute_rs_lipschitz_radius.py
code served (permissive licence) · get_code("079533fab6549a40")
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 float64, ndim 0, 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 math
import numpy as np

# Set random seed for reproducibility
np.random.seed(0)

# Define arguments for compute_radius_tsuzuku_for_smoothed_classifier
ARGS = {
    "margin": 0.5,          # Classification margin between top and second class logits
    "r": 1.0,               # Upper bound on the output range of classifier logits
    "L": 2.5,               # Lipschitz constant of the original network
    "sigma": 0.3,           # Standard deviation of Gaussian noise
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
compute_spectral_norm Ran
ran on a synthesized input
blaisedelattre/bridging_the_gap_rs/code/PUB/compute_lip_PUB_resnet110.py
code served (permissive licence) · get_code("5a6f228bb471d3ba")
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 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)

# Test with Conv2d layer
conv_layer = nn.Conv2d(in_channels=3, out_channels=4, kernel_size=3, padding=1)

# Input shape for Conv2d: (batch, channels, height, width)
input_shape_conv = (2, 3, 8, 8)

# Test with Linear layer
linear_layer = nn.Linear(in_features=5, out_features=6)

# Input shape for Linear (not directly used in the function but needed for consistency)
input_shape_linear = (2, 5)

# Define ARGS with Conv2d layer
ARGS = {
    "layer": conv_layer,
    "input_shape": input_shape_conv,
}
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 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
conv3x3 Ran
ran; our contract draft was wrong, not the code
blaisedelattre/bridging_the_gap_rs/code/archs/cifar_resnet.py
pointer only (licence: MIT); the stored copy get_code serves, from lixiaotong97/DSU, is pointer only (licence: NONE) · get_code("fac5364e2f53c6db")
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 lixiaotong97/DSU, 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_named_schedule_sampler Ran
ran; our contract draft was wrong, not the code
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/resample.py
pointer only (licence: MIT); the stored copy get_code serves, from ucsb-nlp-chang/diffusion_resampling, is pointer only (licence: NONE) · get_code("35572566acf61c78")
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 ucsb-nlp-chang/diffusion_resampling, 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.

discretized_gaussian_log_likelihood Ran
ran; our contract draft was wrong, not the code
blaisedelattre/bridging_the_gap_rs/code/improved_diffusion/losses.py
pointer only (licence: MIT); the stored copy get_code serves, from quandao10/slct, is pointer only (licence: NONE) · get_code("cd33283d615fb3d7")
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 quandao10/slct, 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_dataset Ran
ran; our contract draft was wrong, not the code
blaisedelattre/bridging_the_gap_rs/code/datasets.py
code served (permissive licence) · get_code("04c56ca301f7731e")
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 MockCIFAR10. 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 os
import torch
from torchvision import transforms, datasets
from torch.utils.data import Dataset

torch.manual_seed(0)

# Mock CIFAR10 dataset to avoid downloads
class MockCIFAR10(Dataset):
    def __init__(self, root, train=True, download=True, transform=None):
        self.train = train
        self.transform = transform
        self.data = torch.randn(10, 3, 32, 32)
        self.targets = torch.randint(0, 10, (10,))
    
    def __len__(self):
        return len(self.data)
    
    def __getitem__(self, idx):
        img = self.data[idx]
        target = self.targets[idx].item()
        if self.transform:
            img = self.transform(img)
        return img, target

# Mock ImageFolder dataset to avoid file I/O
class MockImageFolder(Dataset):
    def __init__(self, root, transform=None):
        self.transform = transform
        self.data = torch.randn(10, 3, 256, 256)
        self.targets = torch.randint(0, 1000, (10,))
    
    def __len__(self):
        return len(self.data)
    
    def __getitem__(self, idx):
        img = self.data[idx]
        target = self.targets[idx].item()
        if self.transform:
            img = self.transform(img)
        return img, target

# Patch the datasets module to use mock classes
original_cifar10 = datasets.CIFAR10
original_imagefolder = datasets.ImageFolder

datasets.CIFAR10 = MockCIFAR10
datasets.ImageFolder = MockImageFolder

# Set ImageNet environment variable
os.environ['IMAGENET_DIR'] = '/tmp/imagenet_mock'

# Define arguments for get_dataset
ARGS = {
    'dataset': 'cifar10',
    'split': 'train'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MockCIFAR10 — a MockCIFAR10 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
get_named_beta_schedule Ran
ran, honoured the contract we drafted
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/gaussian_diffusion.py
pointer only (licence: MIT); the stored copy get_code serves, from NVlabs/DiffPure, is pointer only (licence: NOASSERTION) · get_code("a086d6286a40b889")
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 NVlabs/DiffPure, and GitHub could not classify the licence of that repository (NOASSERTION). 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_num_classes Ran
ran, honoured the contract we drafted
blaisedelattre/bridging_the_gap_rs/code/datasets.py
code served (permissive licence) · get_code("8c99213918181c27")
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 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

torch.manual_seed(0)

ARGS = {
    'dataset': 'cifar10'
}
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 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
get_param_groups_and_shapes Ran
ran; our contract draft was wrong, not the code
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/fp16_util.py
pointer only (licence: MIT); the stored copy get_code serves, from ZhendongWang6/DIRE, is pointer only (licence: NONE) · get_code("e41367ad14ff58fd")
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 ZhendongWang6/DIRE, 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.

load_config_from_json Ran
ran on a synthesized input
blaisedelattre/bridging_the_gap_rs/code/certify_cpm_from_log.py
code served (permissive licence) · get_code("75fcfc8a765c3ee8")
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 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 json
import tempfile
import os

# Create a temporary JSON config file
config_data = {
    'dataset': 'cifar10',
    'base_classifier': 'resnet50',
    'sigma': 0.25,
    'skip': 2,
    'max': 100,
    'N': 5000,
    'N0': 50
}

# Write to a temporary file
temp_dir = tempfile.gettempdir()
config_file = os.path.join(temp_dir, 'test_config.json')
with open(config_file, 'w') as f:
    json.dump(config_data, f)

ARGS = {
    'config_file': config_file
}
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 tuple, ndim 1, 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
load_counts Ran
ran on a synthesized input
blaisedelattre/bridging_the_gap_rs/code/certify_cpm_from_log.py
code served (permissive licence) · get_code("84851ec4bb865b9e")
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 2, dtype family int, shape (3, 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 tempfile
import os
import numpy as np

# Create a temporary file with test data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file_path = temp_file.name

# Write header and sample data
temp_file.write("header\n")
temp_file.write("row1\tcol1\t[1, 2, 3]\n")
temp_file.write("row2\tcol2\t[4, 5, 6]\n")
temp_file.write("row3\tcol3\t[7, 8, 9]\n")
temp_file.close()

ARGS = {
    'file_path': temp_file_path
}
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 2, dtype family int, shape (3, 3) — shape (3, 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
load_labels Ran
ran on a synthesized input
blaisedelattre/bridging_the_gap_rs/code/certify_cpm_from_log.py
code served (permissive licence) · get_code("b3199d231cdd7b96")
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 2, dtype family int, shape (3, 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 tempfile
import os
import numpy as np

# Create a temporary file with test data in the expected format
with tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt') as f:
    temp_file = f.name
    # Write header
    f.write("header\n")
    # Write data lines with tab-separated values
    # Format: identifier\tcounts_as_string
    f.write("sample1\t[1, 2, 3, 4]\n")
    f.write("sample2\t[5, 6, 7, 8]\n")
    f.write("sample3\t[9, 10, 11, 12]\n")

ARGS = {
    "file_path": temp_file
}
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 2, dtype family int, shape (3, 4) — shape (3, 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
make_master_params Ran
ran; our contract draft was wrong, not the code
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/fp16_util.py
pointer only (licence: MIT); the stored copy get_code serves, from ZhendongWang6/DIRE, is pointer only (licence: NONE) · get_code("e20dd5102da3b050")
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 ZhendongWang6/DIRE, 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.

make_output_format Ran
ran; our contract draft was wrong, not the code
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/logger.py
code served (permissive licence) · get_code("bcd8b4acab199405")
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 HumanOutputFormat. 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 os
import sys
import tempfile

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

ARGS = {
    'format': 'log',
    'ev_dir': temp_dir,
    'log_suffix': '_test'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type HumanOutputFormat — a HumanOutputFormat output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a HumanOutputFormat 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
mpi_weighted_mean Ran
ran; our contract draft was wrong, not the code
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/logger.py
pointer only (licence: MIT); the stored copy get_code serves, from HeadLiuYun/NeuroDiff, is pointer only (licence: NONE) · get_code("e515a67f7f32e76d")
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 HeadLiuYun/NeuroDiff, 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.

multinomial_event_probability Ran
ran on a synthesized input
blaisedelattre/bridging_the_gap_rs/counter_example.py
code served (permissive licence) · get_code("cfe3d3b5d12792a5")
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 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.

from math import comb
from typing import Tuple, Callable

# Define the event callable - example: probability that x0 >= 2
def example_event(counts: Tuple[int, int, int]) -> bool:
    """Event: first category count is at least 2"""
    x0, x1, x2 = counts
    return x0 >= 2

# Set up arguments
p = (0.5, 0.3, 0.2)  # Probabilities for 3 categories (sum to 1)
n = 5  # Total number of trials
event = example_event  # Callable that checks if event occurred

ARGS = {
    "p": p,
    "n": n,
    "event": event
}
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 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
normal_kl Ran
ran, honoured the contract we drafted
blaisedelattre/bridging_the_gap_rs/code/improved_diffusion/losses.py
pointer only (licence: MIT); the stored copy get_code serves, from facebookresearch/agrol, is pointer only (licence: NOASSERTION) · get_code("cf2798b666b231ca")
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 facebookresearch/agrol, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

phi_sigma Ran
ran on a synthesized input
blaisedelattre/bridging_the_gap_rs/code/compute_rs_lipschitz_radius.py
code served (permissive licence) · get_code("865a29f21fdb0344")
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 float64, ndim 0, 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 numpy as np
from scipy.stats import norm

torch.manual_seed(0)

# phi_sigma takes two scalar parameters: s and sigma
# s: a float value at which to evaluate the CDF
# sigma: a float standard deviation of the Gaussian

ARGS = {
    "s": 1.5,
    "sigma": 2.0
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
random_crop_arr Ran
ran; our contract draft was wrong, not the code
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/image_datasets.py
code served (permissive licence) · get_code("53d5be4fdbcfac23")
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 3, dtype family int, shape (64, 64, 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 math
import random
from PIL import Image
import numpy as np

# Create a simple PIL Image for testing
# Make it large enough that cropping operations make sense
width, height = 256, 256
pil_image = Image.new('RGB', (width, height), color='red')

# Fill with some variation so it's not completely uniform
pixels = pil_image.load()
for i in range(width):
    for j in range(height):
        pixels[i, j] = (i % 256, j % 256, (i + j) % 256)

# Define parameters
image_size = 64
min_crop_frac = 0.8
max_crop_frac = 1.0

ARGS = {
    'pil_image': pil_image,
    'image_size': image_size,
    'min_crop_frac': min_crop_frac,
    'max_crop_frac': max_crop_frac,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (64, 64, 3) — shape (64, 64, 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
space_timesteps Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/respace.py
pointer only (licence: MIT); the stored copy get_code serves, from LehongWu/MacDiff, is pointer only (licence: NONE) · get_code("9b6e4e070473e76f")
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 LehongWu/MacDiff, 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.

unflatten_master_params Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/fp16_util.py
pointer only (licence: MIT); the stored copy get_code serves, from ZhendongWang6/DIRE, is pointer only (licence: NONE) · get_code("64fff1e30802b815")
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 ZhendongWang6/DIRE, 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.

zero_module Ran
ran; our contract draft was wrong, not the code
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/nn.py
code served (permissive licence) · get_code("129b804760b3115f")
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 SimpleModule. 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)

# Create a simple module with some parameters
class SimpleModule(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
        self.conv = nn.Conv1d(2, 2, 3)
    
    def forward(self, x):
        return x

module = SimpleModule()

# Verify the module has non-zero parameters before zeroing
for p in module.parameters():
    assert p.abs().sum() > 0, "Parameters should be non-zero initially"

ARGS = {
    "module": module
}
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 SimpleModule — a SimpleModule output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a SimpleModule 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
avg_pool_nd Not yet run
unverified (harvested, no recorded run)
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/nn.py
code served (permissive licence) · get_code("ecd0fc28815b65ae")
How it ran

no invocation recorded.

conv_nd Not yet run
unverified (harvested, no recorded run)
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/nn.py
code served (permissive licence) · get_code("fe4eb545bbb728e0")
How it ran

no invocation recorded.

create_classifier_and_diffusion Not yet run
unverified (harvested, no recorded run)
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/script_util.py
code served (permissive licence) · get_code("5e5d99a81705a07f")
How it ran

no invocation recorded.

create_model Not yet run
unverified (harvested, no recorded run)
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/script_util.py
code served (permissive licence) · get_code("35ea054532112e53")
How it ran

no invocation recorded.

create_model_and_diffusion Not yet run
unverified (harvested, no recorded run)
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/script_util.py
code served (permissive licence) · get_code("6a3cfc58dafbe392")
How it ran

no invocation recorded.

profile Not yet run
unverified (harvested, no recorded run)
ethz-privsec/diffusion_denoised_smoothing/imagenet/guided_diffusion/logger.py
pointer only (licence: MIT); the stored copy get_code serves, from HeadLiuYun/NeuroDiff, is pointer only (licence: NONE) · get_code("0c6607473a4c4c55")
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 HeadLiuYun/NeuroDiff, 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.

Also present in other papers' code

Of 34 samples lifted from this paper's linked repositories, 34 carry a function identity and 0 do not. Of the 34 distinct functions with an identity, 24 are also present in other papers whose code we hold (23 with at least one paper dated earlier, 15 the same month, 24 later) and 10 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
conv3x3 686 other papers
394 dated earlier, 3 the same month, 278 later, 11 same year, order not determinable · 31 only as an assembly brick
1406.2661 (2014-06)
Generative Adversarial Networks
1409.0575 (2014-09)
ImageNet Large Scale Visual Recognition Challenge
1412.6572 (2014-12)
Explaining and Harnessing Adversarial Examples
and 683 more not named here
zero_module 196 other papers
12 dated earlier, 2 the same month, 182 later · 49 only as an assembly brick
2006.11239 (2020-06)
Denoising Diffusion Probabilistic Models
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2105.05233 (2021-05)
Diffusion Models Beat GANs on Image Synthesis
and 193 more not named here
normal_kl 160 other papers
13 dated earlier, 2 the same month, 145 later · 19 only as an assembly brick
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2105.05233 (2021-05)
Diffusion Models Beat GANs on Image Synthesis
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
and 157 more not named here
approx_standard_normal_cdf 154 other papers
13 dated earlier, 2 the same month, 139 later · 17 only as an assembly brick
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2105.05233 (2021-05)
Diffusion Models Beat GANs on Image Synthesis
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
and 151 more not named here
discretized_gaussian_log_likelihood 128 other papers
13 dated earlier, 2 the same month, 113 later · 16 only as an assembly brick
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2105.05233 (2021-05)
Diffusion Models Beat GANs on Image Synthesis
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
and 125 more not named here
conv_nd 121 other papers
10 dated earlier, 2 the same month, 109 later · 32 only as an assembly brick
2006.11239 (2020-06)
Denoising Diffusion Probabilistic Models
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
and 118 more not named here
avg_pool_nd 113 other papers
11 dated earlier, 2 the same month, 100 later · 31 only as an assembly brick
2006.11239 (2020-06)
Denoising Diffusion Probabilistic Models
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2110.11291 (2021-10)
Likelihood Training of Schrödinger Bridge using Forward-Backward SDEs Theory
and 110 more not named here
betas_for_alpha_bar 92 other papers
11 dated earlier, 2 the same month, 79 later · 3 only as an assembly brick
2006.11239 (2020-06)
Denoising Diffusion Probabilistic Models
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2102.09672 (2021-02)
Improved Denoising Diffusion Probabilistic Models
and 89 more not named here
make_output_format 73 other papers
26 dated earlier, 1 the same month, 46 later
1611.05763 (2016-11)
Learning to reinforcement learn
1804.06459 (2018-04)
On Learning Intrinsic Rewards for Policy Gradient Methods
1806.02473 (2018-06)
Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generation
and 70 more not named here
unflatten_master_params 72 other papers
6 dated earlier, 66 later
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
2112.03145 (2021-12)
Diffusion Models for Implicit Image Segmentation Ensembles
and 69 more not named here
make_master_params 70 other papers
6 dated earlier, 64 later
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
2112.03145 (2021-12)
Diffusion Models for Implicit Image Segmentation Ensembles
and 67 more not named here
get_param_groups_and_shapes 68 other papers
6 dated earlier, 62 later
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
2112.03145 (2021-12)
Diffusion Models for Implicit Image Segmentation Ensembles
and 65 more not named here
get_named_beta_schedule 63 other papers
10 dated earlier, 2 the same month, 51 later
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2108.02938 (2021-08)
ILVR: Conditioning Method for Denoising Diffusion Probabilistic Models
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
and 60 more not named here
profile 61 other papers
13 dated earlier, 1 the same month, 47 later
1611.05763 (2016-11)
Learning to reinforcement learn
1905.06750 (2019-05)
Random Expert Distillation: Imitation Learning via Expert Policy Support Estimation
1906.05841 (2019-06)
Deep Reinforcement Learning for Industrial Insertion Tasks with Visual Inputs and Natural Rewards
and 58 more not named here
space_timesteps 57 other papers
5 dated earlier, 1 the same month, 51 later · 2 only as an assembly brick
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
and 54 more not named here
mpi_weighted_mean 52 other papers
7 dated earlier, 1 the same month, 44 later
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
and 49 more not named here
create_named_schedule_sampler 45 other papers
4 dated earlier, 41 later
2010.02502 (2020-10)
Denoising Diffusion Implicit Models
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
and 42 more not named here
center_crop_arr 33 other papers
4 dated earlier, 29 later
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
2203.15636 (2022-03)
Diffusion Models for Counterfactual Explanations
and 30 more not named here
random_crop_arr 32 other papers
5 dated earlier, 27 later
2111.11672 (2021-11)
Few-shot Image Generation with Mixup-based Distance Learning
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
2203.04306 (2022-03)
Diffusion Models for Medical Anomaly Detection
and 29 more not named here
create_classifier_and_diffusion 9 other papers
1 dated earlier, 1 the same month, 7 later
2111.14818 (2021-11)
Blended Diffusion for Text-driven Editing of Natural Images
2206.00941 (2022-06)
Improving Diffusion Models for Inverse Problems using Manifold Constraints
2212.00490 (2022-12)
Zero-Shot Image Restoration Using Denoising Diffusion Null-Space Model
and 6 more not named here

Shown: the 20 functions present in the most other papers. 4 more shared functions are not listed 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

In this paper we show how to achieve state-of-the-art certified adversarial robustness to 2-norm bounded perturbations by relying exclusively on off-the-shelf pretrained models. To do so, we instantiate the denoised smoothing approach of Salman et al. 2020 by combining a pretrained denoising diffusion probabilistic model and a standard high-accuracy classifier. This allows us to certify 71% accuracy on ImageNet under adversarial perturbations constrained to be within an 2-norm of 0.5, an improvement of 14 percentage points over the prior certified SoTA using any approach, or an improvement of 30 percentage points over denoised smoothing. We obtain these results using only pretrained diffusion models and image classifiers, without requiring any fine tuning or retraining of model parameters.

For agents

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

get_harvested_code_for_paper("2206.10550")
get_code_for_paper("2206.10550")
have("2206.10550")

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

curl https://syntology.ai/api/ran/2206.10550.json

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

[![Syntology run record](https://syntology.ai/api/ran/2206.10550.svg)](https://syntology.ai/paper/2206.10550)

Connect an agent — have() is free.