SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.11846 · NeurIPS · 2024

UnlearnCanvas: Stylized Image Dataset for Enhanced Machine Unlearning Evaluation in Diffusion Models

Sijia Liu, Gaoyuan Zhang, Xiaoming Liu, Yimeng Zhang, Yuguang Yao, Jinghan Jia, Yihua Zhang, Gaowen Liu, Jiancheng Liu, Chongyu Fan, Ramana Kompella

arXiv · PDF · Open in the Atlas

Code that ran

55 of 74 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; 10 honoured the contract we drafted, 3 violated it and 2 have no contract verdict; 40 failures of our instrument; 19 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/2402.11846.json · badge: /api/ran/2402.11846.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
JingWu321/EraseDiff canonical (classifier label)
classifier basis: census
10 ran · 1 no recorded run
optml-group/unlearncanvas canonical (classifier label)
classifier basis: census
9 ran · 3 no recorded run
Con6924/SPM canonical (classifier label)
classifier basis: census
9 ran · 3 no recorded run
nupurkmr9/concept-ablation canonical (classifier label)
classifier basis: census
7 ran · 0 no recorded run
JingWu321/Scissorhands canonical (classifier label)
classifier basis: census
6 ran · 5 no recorded run
rohitgandikota/unified-concept-editing canonical (classifier label)
classifier basis: census
6 ran · 0 no recorded run
sen-mao/SuppressEOT canonical (classifier label)
classifier basis: census
5 ran · 4 no recorded run
SHI-Labs/Forget-Me-Not canonical (classifier label)
classifier basis: census
3 ran · 3 no recorded run
FunctionStatusWhere it lives
load_512 Ran
ran, honoured the contract we drafted
sen-mao/SuppressEOT/utils/null_text_inversion.py
pointer only (licence: NONE); the stored copy get_code serves, from sen-mao/suppresseot, is pointer only (licence: NONE) · get_code("067f451b2dbf739d")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sen-mao/SuppressEOT 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.

Convert Ran
ran; our contract draft was wrong, not the code
nupurkmr9/concept-ablation/compvis/evaluate.py
pointer only (licence: MIT); the stored copy get_code serves, from vlgiitr/unmasking-the-veil, is pointer only (licence: NONE) · get_code("254211bc4b378fe4")
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 vlgiitr/unmasking-the-veil, 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.

GetImageFolderLoader Ran
ran; our contract draft was wrong, not the code
JingWu321/EraseDiff/ddpm/classifier_evaluation.py
pointer only (licence: NONE) · get_code("29136f2953d790e8")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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.

aggregate_attention Ran
ran, honoured the contract we drafted
sen-mao/SuppressEOT/suppress_eot_w_nulltext.py
pointer only (licence: NONE); the stored copy get_code serves, from sen-mao/suppresseot, is pointer only (licence: NONE) · get_code("f3e6737ebc70cbbf")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sen-mao/SuppressEOT 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.

append_zero Ran
ran; our contract draft was wrong, not the code
optml-group/unlearncanvas/diffusion_model_finetuning/edm_sampler/sampling.py
pointer only (licence: NONE) · get_code("4b1e44a59b2094f3")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, optml-group/unlearncanvas 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.

calculate_matching_score Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Con6924/SPM/infer_spm.py
code served (permissive licence) · get_code("f53d21f5e97513c9")
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)

# Construct arguments for calculate_matching_score
# prompt_tokens: tensor of token IDs, shape [batch_size, seq_len]
prompt_tokens = torch.tensor([[101, 2054, 2003, 102], [101, 2054, 2054, 102]], dtype=torch.long)

# prompt_embeds: embeddings for prompts, shape [batch_size, seq_len, embed_dim]
# The function flattens dims 1,2 so we need enough dimensions
prompt_embeds = torch.randn(2, 4, 8, dtype=torch.float32)

# erased_prompt_tokens: tensor of token IDs for erased prompts, shape [batch_size, seq_len]
erased_prompt_tokens = torch.tensor([[101, 2054, 102, 102], [101, 2003, 102, 102]], dtype=torch.long)

# erased_prompt_embeds: embeddings for erased prompts, shape [batch_size, seq_len, embed_dim]
erased_prompt_embeds = torch.randn(2, 4, 8, dtype=torch.float32)

# matching_metric: one of 'clipcos', 'clipcos_tokenuni', 'tokenuni'
matching_metric = 'clipcos_tokenuni'

# special_token_ids: set of token IDs to exclude (like [CLS], [SEP], [PAD])
special_token_ids = {101, 102, 0}

# weight_dtype: torch dtype for weights
weight_dtype = torch.float32

ARGS = {
    'prompt_tokens': prompt_tokens,
    'prompt_embeds': prompt_embeds,
    'erased_prompt_tokens': erased_prompt_tokens,
    'erased_prompt_embeds': erased_prompt_embeds,
    'matching_metric': matching_metric,
    'special_token_ids': special_token_ids,
    'weight_dtype': weight_dtype,
}
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
chunk Ran
ran; our contract draft was wrong, not the code
nupurkmr9/concept-ablation/compvis/sample.py
pointer only (licence: MIT); the stored copy get_code serves, from vicgalle/stable-diffusion-aesthetic-gradients, is pointer only (licence: NOASSERTION) · get_code("8241c0562bc710fd")
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 vicgalle/stable-diffusion-aesthetic-gradients, 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.

collate_fn Ran
ran; our contract draft was wrong, not the code
SHI-Labs/Forget-Me-Not/src/train_attn.py
code served (permissive licence) · get_code("e5dbb0469de3dff9")
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 dict. 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

torch.manual_seed(0)

# Create example dicts that match the expected structure
example1 = {
    'instance_prompt_ids': torch.randint(0, 100, (1, 4)),
    'instance_prompt': 'a photo of a cat',
    'instance_images': torch.randn(3, 8, 8),
    'concept_positions': torch.tensor([[True, False, True, False]]),
    'class_prompt_ids': torch.randint(0, 100, (1, 4)),
    'class_images': torch.randn(3, 8, 8),
}

example2 = {
    'instance_prompt_ids': torch.randint(0, 100, (1, 4)),
    'instance_prompt': 'a photo of a dog',
    'instance_images': torch.randn(3, 8, 8),
    'concept_positions': torch.tensor([[False, True, False, True]]),
    'class_prompt_ids': torch.randint(0, 100, (1, 4)),
    'class_images': torch.randn(3, 8, 8),
}

examples = [example1, example2]

ARGS = {
    'examples': examples,
    'with_prior_preservation': False,
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
compute_rotation_matrix Ran
ran; our contract draft was wrong, not the code
Con6924/SPM/src/configs/prompt.py
code served (permissive licence) · get_code("cb0d0aa60feeb917")
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 (4, 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)

# The function expects a target vector as a FloatTensor
# It will reshape it to 1D, normalize it, then use Gram-Schmidt orthogonalization
# to build an orthonormal basis, and return the inverse

# Create a small target vector
target = torch.tensor([1.0, 2.0, 3.0, 4.0], dtype=torch.float32)

ARGS = {
    "target": target
}
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 (4, 4) — shape (4, 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
conv1x1 Ran
ran; our contract draft was wrong, not the code
JingWu321/Scissorhands/Classification/models/ResNet.py
code served (permissive licence) · get_code("d9def42110729a85")
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 Conv2d. 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 conv1x1 takes in_planes, out_planes, and stride as parameters
# It returns an nn.Conv2d layer configured for 1x1 convolution
# We need to provide valid integer arguments for these parameters

ARGS = {
    "in_planes": 3,
    "out_planes": 64,
    "stride": 1
}
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 Conv2d — a Conv2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Conv2d 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
conv3x3 Ran
ran; our contract draft was wrong, not the code
JingWu321/Scissorhands/Classification/models/ResNet.py
code served (permissive licence) · get_code("160bb14bd76201b4")
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 Conv2d. 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 = {
    "in_planes": 3,
    "out_planes": 64,
    "stride": 1,
    "groups": 1,
    "dilation": 1,
}
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 Conv2d — a Conv2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Conv2d 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
diffusion_step Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
sen-mao/SuppressEOT/utils/ptp_utils.py
pointer only (licence: NONE); the stored copy get_code serves, from nku-hlt/audioeditor, is pointer only (licence: NOASSERTION) · get_code("6d190d192ff5e777")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sen-mao/SuppressEOT 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.

dumps_data Ran
ran; our contract draft was wrong, not the code
JingWu321/Scissorhands/Classification/lmdb_dataset.py
code served (permissive licence) · get_code("cadb98df0b43c793")
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 bytes. 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 pickle
import torch
import numpy as np

# Set seed for reproducibility
torch.manual_seed(0)

# Create various serializable objects to test dumps_data
obj_dict = {
    "key1": "value1",
    "key2": 42,
    "key3": [1, 2, 3]
}

obj_list = [1, 2, 3, "test", {"nested": "dict"}]

obj_tensor = torch.randn(2, 4)

obj_numpy = np.array([[1, 2, 3], [4, 5, 6]])

class SimpleClass:
    def __init__(self):
        self.data = "test_data"
        self.value = 123

obj_custom = SimpleClass()

obj_tuple = (1, 2, "three", [4, 5])

obj_string = "This is a simple string"

obj_number = 42.5

# Test with dict - this is a common serialization target
ARGS = {
    "obj": obj_dict
}
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 bytes — a bytes output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a bytes 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
extract_all_images Ran
ran, honoured the contract we drafted
nupurkmr9/concept-ablation/compvis/evaluate.py
code served (permissive licence) · get_code("2cb4375648881b05")
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 float, shape (2, 5) as a later run recorded it. 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, 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
import numpy as np
from PIL import Image
import tempfile
import os

torch.manual_seed(0)

# Create a tiny stand-in model with encode_image method
class TinyClipModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.proj = nn.Linear(3, 5)
    
    def encode_image(self, x):
        # x shape: (batch, 3, 224, 224)
        # Return (batch, 5) features
        batch_size = x.shape[0]
        return self.proj(x.mean(dim=[2, 3]))

model = TinyClipModel()

# Create temporary image files for testing
temp_dir = tempfile.mkdtemp()
image_paths = []
for i in range(2):
    img = Image.new('RGB', (256, 256), color=(i*50, i*100, i*150))
    img_path = os.path.join(temp_dir, f'test_{i}.jpg')
    img.save(img_path)
    image_paths.append(img_path)

images = image_paths
device = 'cpu'
batch_size = 2
num_workers = 0

ARGS = {
    'images': images,
    'model': model,
    'device': device,
    'batch_size': batch_size,
    'num_workers': num_workers
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy, torch, torchvision, tqdm — 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 float, shape (2, 5) — shape (2, 5) 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
from_pil_image Ran
ran, honoured the contract we drafted
optml-group/unlearncanvas/diffusion_model_finetuning/edm_sampler/utils.py
pointer only (licence: NONE) · get_code("5c07965b9fb3ae74")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, optml-group/unlearncanvas 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_A Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
rohitgandikota/unified-concept-editing/evalscripts/debiasing_vl.py
code served (permissive licence) · get_code("9b4052692da3a8bc")
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 float, shape (4, 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 numpy as np

# Create small numpy arrays for z_i and z_j
# The function expects 1D arrays that will be reshaped to column vectors
z_i = np.array([1.0, 2.0, 3.0, 4.0])
z_j = np.array([5.0, 6.0, 7.0, 8.0])

ARGS = {
    "z_i": z_i,
    "z_j": z_j,
}
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 float, shape (4, 4) — shape (4, 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
get_M Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
rohitgandikota/unified-concept-editing/evalscripts/debiasing_vl.py
code served (permissive licence) · get_code("894009c45bf51dcf")
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 float, shape (4, 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 numpy as np

# Create small embeddings: shape (5, 4) means 5 embedding vectors of dimension 4
embeddings = np.random.randn(5, 4).astype(np.float32)

# Create a list of pairs S representing indices into embeddings
# Each element in S is a pair (i, j) of indices
S = [
    (0, 1),
    (1, 2),
    (2, 3),
]

ARGS = {
    "embeddings": embeddings,
    "S": S,
}
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 float, shape (4, 4) — shape (4, 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
get_param Ran
ran; our contract draft was wrong, not the code
JingWu321/EraseDiff/sd/train_scripts/erasediff.py
pointer only (licence: NONE) · get_code("2ccf562d07794d65")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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_sigmas_exponential Ran
ran, honoured the contract we drafted
optml-group/unlearncanvas/diffusion_model_finetuning/edm_sampler/sampling.py
pointer only (licence: NONE) · get_code("c7a8a88e04fd5351")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, optml-group/unlearncanvas 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_sigmas_karras Ran
ran, honoured the contract we drafted
optml-group/unlearncanvas/diffusion_model_finetuning/edm_sampler/sampling.py
pointer only (licence: NONE) · get_code("7064fad2ca846b74")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, optml-group/unlearncanvas 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_styled_prompt Ran
ran; our contract draft was wrong, not the code
nupurkmr9/concept-ablation/compvis/src/finetune_data.py
code served (permissive licence) · get_code("489a35aa27273ede")
How it ran

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

import numpy as np

np.random.seed(0)

ARGS = {
    'caption_target': 'Van Gogh',
    'caption': '  a starry night sky with hills  '
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_timestep_embedding Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
JingWu321/EraseDiff/ddpm/models/diffusion.py
pointer only (licence: NONE) · get_code("cb49209c125de1b4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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_x_y_from_data_dict Ran
ran; our contract draft was wrong, not the code
JingWu321/Scissorhands/Classification/imagenet.py
code served (permissive licence) · get_code("c7ae957080f6f740")
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. 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 small tensors with distinct dimensions
x_tensor = torch.randn(2, 4)  # batch size 2, feature dim 4
y_tensor = torch.randn(2, 3)  # batch size 2, label dim 3

# Case 1: data_dict with tensors (not lists)
data_dict_tensors = {
    'x': x_tensor,
    'y': y_tensor
}

device = 'cpu'

ARGS = {
    'data': data_dict_tensors,
    'device': device
}
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 tuple — a tuple 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
gram_matrix Ran
ran; our contract draft was wrong, not the code
rohitgandikota/unified-concept-editing/evalscripts/styleloss.py
pointer only (licence: MIT); the stored copy get_code serves, from Oldpan/Deep-Painterly-Harmonization-Pytorch, is pointer only (licence: NONE) · get_code("1090950b58e9358b")
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 Oldpan/Deep-Painterly-Harmonization-Pytorch, 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.

hf_datasets_augs_helper Ran
ran; our contract draft was wrong, not the code
optml-group/unlearncanvas/diffusion_model_finetuning/edm_sampler/utils.py
pointer only (licence: NONE) · get_code("b22174509b22ebfe")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, optml-group/unlearncanvas 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.

image_loader Ran
ran, honoured the contract we drafted
rohitgandikota/unified-concept-editing/evalscripts/styleloss.py
pointer only (licence: MIT); the stored copy get_code serves, from tuananhbui89/Erasing-Adversarial-Preservation, is pointer only (licence: NONE) · get_code("9ad49e53202d0110")
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 tuananhbui89/Erasing-Adversarial-Preservation, 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.

image_loader Ran
ran, honoured the contract we drafted
rohitgandikota/unified-concept-editing/evalscripts/lpips_eval.py
pointer only (licence: MIT); the stored copy get_code serves, from OPTML-Group/AdvUnlearn, is pointer only (licence: CC-BY-4.0) · get_code("56e74fc4361256b9")
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 OPTML-Group/AdvUnlearn, and CC-BY-4.0, as GitHub detects it on that repository, is not in the permissive set Syntology serves inline. 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.

isimage Ran
ran, violated the contract we drafted
nupurkmr9/concept-ablation/compvis/src/filter.py
pointer only (licence: MIT); the stored copy get_code serves, from wutao-cs/videomaker, is pointer only (licence: NOASSERTION) · get_code("ea4a1969281507a5")
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 wutao-cs/videomaker, 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.

isimage Ran
ran, violated the contract we drafted
optml-group/unlearncanvas/diffusion_model_finetuning/ldm/util.py
pointer only (licence: NONE) · get_code("b1368330cf0f5642")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, optml-group/unlearncanvas 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.

ismap Ran
ran, violated the contract we drafted
optml-group/unlearncanvas/diffusion_model_finetuning/ldm/util.py
pointer only (licence: NONE) · get_code("d72762b700feee6f")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, optml-group/unlearncanvas 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.

latent2image Ran
ran; our contract draft was wrong, not the code
sen-mao/SuppressEOT/utils/ptp_utils.py
pointer only (licence: NONE) · get_code("e3c3bb87bbe9b4a7")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sen-mao/SuppressEOT 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_yaml Ran
ran; our contract draft was wrong, not the code
Con6924/SPM/src/configs/generation_config.py
code served (permissive licence) · get_code("f0e2148d0a7aa2a2")
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 GenerationConfig. 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 yaml
import tempfile
import os
from pydantic import BaseModel

torch.manual_seed(0)

# Create a minimal YAML config file
yaml_content = """
prompts:
  - "a beautiful landscape"
  - "a portrait of a person"
negative_prompt: "bad quality,watermark"
unconditional_prompt: ""
width: 512
height: 512
num_inference_steps: 20
guidance_scale: 7.5
seed: 42
generate_num: 2
save_path: "/tmp/output"
"""

# Write to a temporary file
with tempfile.NamedTemporaryFile(mode='w', suffix='.yaml', delete=False) as f:
    f.write(yaml_content)
    temp_yaml_path = f.name

try:
    ARGS = {
        'cfg_path': temp_yaml_path
    }
finally:
    # Clean up will happen after execution
    pass
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pydantic, torch, yaml — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type GenerationConfig — a GenerationConfig output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a GenerationConfig 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
load_metadata_from_safetensors Ran
ran; our contract draft was wrong, not the code
Con6924/SPM/src/models/merge_spm.py
code served (permissive licence) · get_code("299ce035eef45373")
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 dict. 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 tempfile
import torch
from safetensors.torch import save_file

torch.manual_seed(0)

# Create a temporary safetensors file with metadata
temp_dir = tempfile.gettempdir()
safetensors_file = os.path.join(temp_dir, "test_model.safetensors")

# Create a simple tensor to save
tensors = {"weight": torch.randn(2, 4)}
metadata = {"test_key": "test_value"}

# Save the file with metadata
save_file(tensors, safetensors_file, metadata=metadata)

ARGS = {
    "safetensors_file": safetensors_file,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
safetensors, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
load_prompts_from_table Ran
ran; our contract draft was wrong, not the code
Con6924/SPM/src/configs/prompt.py
code served (permissive licence) · get_code("9e90320bd677ef48")
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 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 tempfile
import pandas as pd
from pathlib import Path

# Create a temporary CSV file with the required columns
temp_dir = tempfile.mkdtemp()
csv_path = Path(temp_dir) / "prompts.csv"

# Create a sample DataFrame with required columns
data = {
    'prompt': ['a cat', 'a dog', 'a bird'],
    'evaluation_seed': [42, 43, 44],
    'sd_seed': [100, 101, 102],
    'case_number': [0, 1, 2]
}
df = pd.DataFrame(data)
df.to_csv(csv_path, index=False)

ARGS = {
    'path': str(csv_path)
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas, pydantic — 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
load_prompts_from_yaml Ran
ran; our contract draft was wrong, not the code
Con6924/SPM/src/configs/prompt.py
code served (permissive licence) · get_code("1f3069e15698ab71")
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 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 tempfile
import yaml
from pathlib import Path

# Create a temporary YAML file with valid prompt configurations
temp_dir = tempfile.mkdtemp()
yaml_path = Path(temp_dir) / "prompts.yaml"

# Write a valid YAML file with prompt settings
prompts_data = [
    {
        'target': 'a dog',
        'positive': 'a happy dog',
        'unconditional': 'blurry',
        'neutral': 'neutral image',
        'action': 'erase',
        'guidance_scale': 7.5,
        'resolution': 512,
        'dynamic_resolution': False,
        'batch_size': 1,
        'dynamic_crops': False,
        'use_template': False,
        'la_strength': 1000.0,
        'sampling_batch_size': 4,
        'seed': 42,
        'case_number': 0
    },
    {
        'target': 'a cat',
        'action': 'erase_with_la',
        'guidance_scale': 5.0,
        'seed': 123
    }
]

with open(yaml_path, 'w') as f:
    yaml.dump(prompts_data, f)

ARGS = {
    'path': yaml_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pydantic, yaml — 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
loads_data Ran
ran; our contract draft was wrong, not the code
JingWu321/Scissorhands/Classification/lmdb_dataset.py
pointer only (licence: MIT); the stored copy get_code serves, from s-enmt/MMFT, is pointer only (licence: NOASSERTION) · get_code("2bc7277fbb33cc3f")
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 s-enmt/MMFT, 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.

merge_lora_models Ran
ran; our contract draft was wrong, not the code
Con6924/SPM/src/models/merge_spm.py
code served (permissive licence) · get_code("3c4f9becca932b96")
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 dict. 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 math
import tempfile
import os

torch.manual_seed(0)

# Create temporary LoRA model files to be "loaded"
# These will be .pt files (not .safetensors) to avoid file I/O complexity

temp_dir = tempfile.gettempdir()

# Create first LoRA model state dict
lora_model_1 = {
    'module_0.lora_down.weight': torch.randn(4, 8),
    'module_0.lora_up.weight': torch.randn(8, 4),
    'module_0.alpha': torch.tensor(8.0),
    'module_1.lora_down.weight': torch.randn(3, 6),
    'module_1.lora_up.weight': torch.randn(6, 3),
    'module_1.alpha': torch.tensor(6.0),
}

# Create second LoRA model state dict
lora_model_2 = {
    'module_0.lora_down.weight': torch.randn(4, 8),
    'module_0.lora_up.weight': torch.randn(8, 4),
    'module_0.alpha': torch.tensor(8.0),
    'module_1.lora_down.weight': torch.randn(3, 6),
    'module_1.lora_up.weight': torch.randn(6, 3),
    'module_1.alpha': torch.tensor(6.0),
}

# Save the models to temporary files
model_1_path = os.path.join(temp_dir, 'lora_model_1.pt')
model_2_path = os.path.join(temp_dir, 'lora_model_2.pt')

torch.save(lora_model_1, model_1_path)
torch.save(lora_model_2, model_2_path)

# Define ARGS dictionary with all required parameters for merge_lora_models
ARGS = {
    'models': [model_1_path, model_2_path],
    'ratios': [0.5, 0.5],
    'merge_dtype': torch.float32,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
safetensors, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
moving_average Ran
ran; our contract draft was wrong, not the code
JingWu321/EraseDiff/sd/train_scripts/erasediff.py
pointer only (licence: NONE); the stored copy get_code serves, from rcheng805/CORE-RL, is pointer only (licence: NONE) · get_code("f398bc270e4f4ffe")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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.

noise_estimation_loss Ran
ran; our contract draft was wrong, not the code
JingWu321/EraseDiff/ddpm/functions/losses.py
pointer only (licence: NONE) · get_code("36a45cceb1209d2f")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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.

noise_estimation_loss_conditional Ran
ran; our contract draft was wrong, not the code
JingWu321/EraseDiff/ddpm/functions/losses.py
pointer only (licence: NONE) · get_code("ca8a1bf21579d253")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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.

nonlinearity Ran
ran; our contract draft was wrong, not the code
JingWu321/EraseDiff/ddpm/models/diffusion.py
pointer only (licence: NONE) · get_code("3137073275f8c21a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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.

parse_extra_args Ran
ran; our contract draft was wrong, not the code
Con6924/SPM/evaluate_task.py
code served (permissive licence) · get_code("f4e5ea8e5d469b27")
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 dict. 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 numpy as np

torch.manual_seed(0)

# Test case 1: None input
test1 = None

# Test case 2: Empty list with empty string
test2 = ['']

# Test case 3: Integer argument
test3 = ['learning_rate=0.001']

# Test case 4: Float argument
test4 = ['weight_decay=0.0001']

# Test case 5: List of integers
test5 = ['hidden_dims=[64,128,256]']

# Test case 6: List of floats
test6 = ['dropout_rates=[0.1,0.2,0.3]']

# Test case 7: List of strings with spaces (using + as space replacement)
test7 = ['model_names=[bert+base,gpt+small]']

# Test case 8: Mixed arguments
test8 = ['epochs=10', 'lr=0.001', 'layers=[32,64,128]', 'optimizer=adam']

# Test case 9: Single string argument
test9 = ['activation=relu']

# Test case 10: Multiple arguments with various types
test10 = ['batch_size=32', 'threshold=0.5', 'seeds=[1,2,3,4,5]', 'model_type=transformer']

ARGS = {
    'extra_args': test8
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
parse_precision Ran
ran; our contract draft was wrong, not the code
Con6924/SPM/src/configs/config.py
code served (permissive licence) · get_code("ca142e6d863ef6a6")
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 dtype. 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

torch.manual_seed(0)

ARGS = {
    'precision': 'fp32'
}
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 dtype — a dtype output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dtype 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
parse_safeloras_embeds Ran
ran; our contract draft was wrong, not the code
SHI-Labs/Forget-Me-Not/src/patch_lora.py
code served (permissive licence) · get_code("74e4f1bd2e3487ff")
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 dict. 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
from safetensors.torch import save_file
import tempfile
import os

torch.manual_seed(0)

# Create a mock safetensors file with embeddings
class MockSafetensors:
    def __init__(self):
        self._tensors = {}
        self._metadata = {}
    
    def keys(self):
        return self._tensors.keys()
    
    def metadata(self):
        return self._metadata
    
    def get_tensor(self, key):
        return self._tensors[key]
    
    def add_embed(self, key, tensor):
        self._tensors[key] = tensor
        self._metadata[key] = '<embed>'
    
    def add_non_embed(self, key, tensor):
        self._tensors[key] = tensor
        self._metadata[key] = 'other_metadata'

# Construct the mock safeloras object
safeloras = MockSafetensors()

# Add some embeddings with the EMBED_FLAG metadata
embed_tensor_1 = torch.randn(4, 3)
embed_tensor_2 = torch.randn(2, 5)
safeloras.add_embed('token_embedding_1', embed_tensor_1)
safeloras.add_embed('token_embedding_2', embed_tensor_2)

# Add a non-embedding entry to test filtering
non_embed_tensor = torch.randn(3, 4)
safeloras.add_non_embed('non_embed_key', non_embed_tensor)

ARGS = {
    'safeloras': safeloras,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
safetensors, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
pil_to_jpg Ran
ran; our contract draft was wrong, not the code
nupurkmr9/concept-ablation/compvis/sample.py
pointer only (licence: MIT); the stored copy get_code serves, from vlgiitr/unmasking-the-veil, is pointer only (licence: NONE) · get_code("11ec361f7a77d02c")
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 vlgiitr/unmasking-the-veil, 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.

prob_mask_like Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
JingWu321/EraseDiff/ddpm/models/diffusion.py
pointer only (licence: NONE) · get_code("3c6433dd421724e0")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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.

raw_reader Ran
ran; our contract draft was wrong, not the code
JingWu321/Scissorhands/Classification/lmdb_dataset.py
code served (permissive licence) · get_code("51738010a21f76a1")
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 bytes. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import tempfile
import os

# Create a temporary binary file with some test data
temp_dir = tempfile.mkdtemp()
test_file = os.path.join(temp_dir, "test_data.bin")

# Write some binary data to the file
with open(test_file, 'wb') as f:
    f.write(b'\x00\x01\x02\x03\x04\x05')

ARGS = {
    'path': test_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bytes — a bytes output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a bytes 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
read_images_folder Ran
ran, honoured the contract we drafted
JingWu321/EraseDiff/ddpm/evaluator.py
pointer only (licence: NONE) · get_code("0bae0d076b1ae674")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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.

sample Ran
ran on a synthesized input
nupurkmr9/concept-ablation/compvis/src/utils.py
code served (permissive licence) · get_code("f7c1195d335ba6ea")
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. 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 os
import torch
import torch.nn as nn
from io import BytesIO
from pathlib import Path
import tempfile

torch.manual_seed(0)

# Create a minimal model that implements the required methods
class MinimalModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def get_learned_conditioning(self, prompts):
        # Return a tensor of shape (batch_size, 4)
        if isinstance(prompts, list):
            batch_size = len(prompts)
        else:
            batch_size = 1
        return torch.randn(batch_size, 4)
    
    def decode_first_stage(self, x):
        # Input: (batch, 4, 64, 64)
        # Output: (batch, 3, 64, 64) - RGB image
        batch_size = x.shape[0]
        return torch.randn(batch_size, 3, 64, 64)
    
    def ema_scope(self):
        # Context manager that does nothing
        class DummyContext:
            def __enter__(self):
                return self
            def __exit__(self, *args):
                pass
        return DummyContext()


# Create a minimal sampler
class MinimalSampler:
    def sample(self, S, conditioning, batch_size, shape, verbose, unconditional_guidance_scale, unconditional_conditioning, eta, x_T):
        # Return (samples, log) where samples has shape (batch_size, *shape)
        samples = torch.randn([batch_size] + shape)
        log = {}
        return samples, log


# Create temporary output directory
tmpdir = tempfile.mkdtemp()

# Create test data - list of batches where each batch is a tuple of prompts
data = [
    ("a red cat", "a blue dog"),
    ("a green tree",),
]

# Build ARGS
ARGS = {
    'data': data,
    'model': MinimalModel(),
    'sampler': MinimalSampler(),
    'outpath': tmpdir,
    'ddim_steps': 50,
    'n_samples': 2,
    'ddim_eta': 1.0,
    'n_iter': 1,
    'scale': 6.0,
    'batch_size': 2,
    'shape': (4, 64, 64),
    'fixed_code': False,
    'device': 'cpu',
    'skip_save': False,
    'skip_grid': True,
    'metadata': True,
    'base_count': 0,
    'n_rows': 1,
    'wandb_log': False,
    'ckptname': 'base',
    'rank': None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, einops, numpy, torch, torchvision, tqdm, wandb — 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
set_param Ran
ran; our contract draft was wrong, not the code
JingWu321/EraseDiff/sd/train_scripts/erasediff.py
pointer only (licence: NONE) · get_code("8168d4d8419b92c3")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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.

sorted_nicely Ran
ran; our contract draft was wrong, not the code
rohitgandikota/unified-concept-editing/evalscripts/CLIP_classify.py
code served (permissive licence) · get_code("8517a5a2a142c0ca")
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 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 re

def sorted_nicely(l):
    convert = lambda text: int(text) if text.isdigit() else text
    alphanum_key = lambda key: [convert(c) for c in re.split('([0-9]+)', key)]
    return sorted(l, key=alphanum_key)

# The function takes a list of strings and sorts them in a natural/human-readable way
# It handles strings with mixed alphanumeric characters
ARGS = {
    'l': ['file10.txt', 'file2.txt', 'file1.txt', 'item100', 'item20', 'item3', 'a1b2c3', 'a1b10c3']
}
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 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
text2img_dataloader Ran
ran on a synthesized input
SHI-Labs/Forget-Me-Not/src/train_ti.py
code served (permissive licence) · get_code("4ed7a5bbe748cb10")
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 DataLoader. 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 torch.utils.data import Dataset

torch.manual_seed(0)

# Minimal tokenizer stand-in
class MinimalTokenizer:
    def __init__(self):
        self.model_max_length = 77
    
    def pad(self, input_dict, padding=None, max_length=None, return_tensors=None):
        input_ids = input_dict['input_ids']
        # Pad each sequence to max_length
        padded = []
        for seq in input_ids:
            if len(seq) < max_length:
                seq = seq + [0] * (max_length - len(seq))
            else:
                seq = seq[:max_length]
            padded.append(seq)
        
        class PadResult:
            def __init__(self, ids):
                self.input_ids = torch.tensor(ids, dtype=torch.long)
        
        return PadResult(padded)

# Minimal VAE stand-in (not directly used in function body, but could be needed)
class MinimalVAE(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)

# Minimal text encoder stand-in (not directly used in function body, but could be needed)
class MinimalTextEncoder(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)

# Minimal dataset
class MinimalDataset(Dataset):
    def __init__(self, size=4):
        self.size = size
    
    def __len__(self):
        return self.size
    
    def __getitem__(self, idx):
        return {
            'instance_prompt_ids': [101, 2054, 2003, 102],  # 4 token IDs
            'uncond_prompt_ids': [101, 102, 0, 0],
            'instance_images': torch.randn(3, 8, 8),  # 3 channels, 8x8 image
            'class_prompt_ids': [101, 1045, 2572, 102],
            'class_images': torch.randn(3, 8, 8),
            'mask': torch.randn(1, 8, 8),
        }

# Construct arguments
ARGS = {
    'train_dataset': MinimalDataset(size=4),
    'train_batch_size': 2,
    'tokenizer': MinimalTokenizer(),
    'vae': MinimalVAE(),
    'text_encoder': MinimalTextEncoder(),
}
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 DataLoader — a DataLoader output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataLoader 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
text_under_image Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
sen-mao/SuppressEOT/utils/ptp_utils.py
pointer only (licence: NONE); the stored copy get_code serves, from TFNTF/PostEdit, is pointer only (licence: NONE) · get_code("85d5eb123674597d")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sen-mao/SuppressEOT 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.

to_cuda Ran
ran, honoured the contract we drafted
optml-group/unlearncanvas/diffusion_model_finetuning/evaluation/fid.py
pointer only (licence: NONE) · get_code("b4318fcc73fe317f")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, optml-group/unlearncanvas 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.

to_pil_image Ran
ran; our contract draft was wrong, not the code
optml-group/unlearncanvas/diffusion_model_finetuning/edm_sampler/utils.py
pointer only (licence: NONE) · get_code("ca508754772590c4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, optml-group/unlearncanvas 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.

all_but_one_class_dataset Not yet run
unverified (harvested, no recorded run)
JingWu321/EraseDiff/ddpm/save_base_dataset.py
pointer only (licence: NONE) · get_code("d278a27b738c2136")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, JingWu321/EraseDiff 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.

apply_learned_embed_in_clip Not yet run
unverified (harvested, no recorded run)
SHI-Labs/Forget-Me-Not/src/patch_lora.py
code served (permissive licence) · get_code("1f5610eafe95d0c8")
How it ran

no invocation recorded.

calculate_activation_statistics Not yet run
unverified (harvested, no recorded run)
optml-group/unlearncanvas/diffusion_model_finetuning/evaluation/fid.py
pointer only (licence: NONE) · get_code("f2517e59a9ce2ce4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, optml-group/unlearncanvas 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.

cifar100_dataloaders Not yet run
unverified (harvested, no recorded run)
JingWu321/Scissorhands/Classification/dataset.py
code served (permissive licence) · get_code("9ea29973ea0485a1")
How it ran

no invocation recorded.

cifar10_dataloaders_no_val Not yet run
unverified (harvested, no recorded run)
JingWu321/Scissorhands/Classification/dataset.py
code served (permissive licence) · get_code("361b798104c4eff0")
How it ran

no invocation recorded.

Showing the first 60. An agent gets every one from get_harvested_code_for_paper("2402.11846").

Also present in other papers' code

Of 74 samples lifted from this paper's linked repositories, 74 carry a function identity and 0 do not. Of the 74 distinct functions with an identity, 43 are also present in other papers whose code we hold (34 with at least one paper dated earlier, 13 the same month, 31 later) and 31 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
conv1x1 431 other papers
322 dated earlier, 7 the same month, 91 later, 10 same year, order not determinable, 1 undated · 38 only as an assembly brick
1409.1556 (2014-09)
Very Deep Convolutional Networks for Large-Scale Image Recognition
1506.01186 (2015-06)
Cyclical Learning Rates for Training Neural Networks
1507.05717 (2015-07)
An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognition
and 428 more not named here
conv3x3 312 other papers
221 dated earlier, 4 the same month, 76 later, 10 same year, order not determinable, 1 undated · 31 only as an assembly brick
1409.1556 (2014-09)
Very Deep Convolutional Networks for Large-Scale Image Recognition
1507.05717 (2015-07)
An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognition
1512.03385 (2015-12)
Deep Residual Learning for Image Recognition
and 309 more not named here
nonlinearity 156 other papers
64 dated earlier, 1 the same month, 87 later, 4 same year, order not determinable · 25 only as an assembly brick
2104.07652 (2021-04)
Geometry-Free View Synthesis: Transformers and no 3D Priors
2106.00132 (2021-06)
On Fast Sampling of Diffusion Probabilistic Models
2106.06819 (2021-06)
D2C: Diffusion-Denoising Models for Few-shot Conditional Generation
and 153 more not named here
resnet18 95 other papers
61 dated earlier, 2 the same month, 28 later, 3 same year, order not determinable, 1 undated
1805.00907 (2018-05)
Glow: Graph Lowering Compiler Techniques for Neural Networks
1902.08153 (2019-02)
Learned Step Size Quantization
1904.01355 (2019-04)
FCOS: Fully Convolutional One-Stage Object Detection
and 92 more not named here
get_timestep_embedding 91 other papers
49 dated earlier, 38 later, 4 same year, order not determinable · 7 only as an assembly brick
2104.07652 (2021-04)
Geometry-Free View Synthesis: Transformers and no 3D Priors
2106.00132 (2021-06)
On Fast Sampling of Diffusion Probabilistic Models
2106.06819 (2021-06)
D2C: Diffusion-Denoising Models for Few-shot Conditional Generation
and 88 more not named here
ismap 69 other papers
38 dated earlier, 1 the same month, 27 later, 3 same year, order not determinable
2204.06125 (2022-04)
Hierarchical Text-Conditional Image Generation with CLIP Latents
2206.02779 (2022-06)
Blended Latent Diffusion
2208.13753 (2022-08)
Frido: Feature Pyramid Diffusion for Complex Scene Image Synthesis
and 66 more not named here
isimage 61 other papers
35 dated earlier, 1 the same month, 23 later, 2 same year, order not determinable
2206.02779 (2022-06)
Blended Latent Diffusion
2208.13753 (2022-08)
Frido: Feature Pyramid Diffusion for Complex Scene Image Synthesis
2210.12965 (2022-10)
High-Resolution Image Editing via Multi-Stage Blended Diffusion
and 58 more not named here
log_txt_as_img 55 other papers
29 dated earlier, 23 later, 3 same year, order not determinable
2204.06125 (2022-04)
Hierarchical Text-Conditional Image Generation with CLIP Latents
2206.02779 (2022-06)
Blended Latent Diffusion
2210.12965 (2022-10)
High-Resolution Image Editing via Multi-Stage Blended Diffusion
and 52 more not named here
chunk 40 other papers
31 dated earlier, 1 the same month, 7 later, 1 same year, order not determinable
2106.14490 (2021-06)
Making Images Real Again: A Comprehensive Survey on Deep Image Composition
2112.10752 (2021-12)
High-Resolution Image Synthesis with Latent Diffusion Models
2204.11824 (2022-04)
Semi-Parametric Neural Image Synthesis
and 37 more not named here
import_model_class_from_model_name_or_path 24 other papers
9 dated earlier, 14 later, 1 same year, order not determinable
2212.04489 (2022-12)
SINE: SINgle Image Editing with Text-to-Image Diffusion Models
2306.06874 (2023-06)
VillanDiffusion: A Unified Backdoor Attack Framework for Diffusion Models
2306.07280 (2023-06)
Controlling Text-to-Image Diffusion by Orthogonal Finetuning
and 21 more not named here
noise_estimation_loss 14 other papers
9 dated earlier, 4 later, 1 same year, order not determinable
2106.06819 (2021-06)
D2C: Diffusion-Denoising Models for Few-shot Conditional Generation
2206.02262 (2022-06)
Diffusion-GAN: Training GANs with Diffusion
2211.14680 (2022-11)
A Physics-informed Diffusion Model for High-fidelity Flow Field Reconstruction
and 11 more not named here
text_under_image 14 other papers
6 dated earlier, 1 the same month, 6 later, 1 same year, order not determinable
2307.12493 (2023-07)
TF-ICON: Diffusion-Based Training-Free Cross-Domain Image Composition
2307.14352 (2023-07)
General Image-to-Image Translation with One-Shot Image Guidance
2309.14494 (2023-09)
Free-Bloom: Zero-Shot Text-to-Video Generator with LLM Director and LDM Animator
and 11 more not named here
latent2image 13 other papers
7 dated earlier, 1 the same month, 4 later, 1 same year, order not determinable
2305.15581 (2023-05)
Unsupervised Semantic Correspondence Using Stable Diffusion
2305.19164 (2023-05)
LANCE: Stress-testing Visual Models by Generating Language-guided Counterfactual Images
2308.06160 (2023-08)
DatasetDM: Synthesizing Data with Perception Annotations Using Diffusion Models
and 10 more not named here
gram_matrix 12 other papers
9 dated earlier, 3 later
1508.06576 (2015-08)
A Neural Algorithm of Artistic Style
1703.07511 (2017-03)
Deep Photo Style Transfer
1804.03189 (2018-04)
Deep Painterly Harmonization
and 9 more not named here
prob_mask_like 11 other papers
6 dated earlier, 1 the same month, 2 later, 2 same year, order not determinable · 2 only as an assembly brick
2202.04200 (2022-02)
MaskGIT: Masked Generative Image Transformer
2209.04439 (2022-09)
Improved Masked Image Generation with Token-Critic
2211.10658 (2022-11)
EDGE: Editable Dance Generation From Music
and 8 more not named here
raw_reader 8 other papers
5 dated earlier, 2 later, 1 same year, order not determinable
2202.07800 (2022-02)
Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations
2209.04996 (2022-09)
Switchable Online Knowledge Distillation
2210.05958 (2022-10)
Bridging the Gap Between Vision Transformers and Convolutional Neural Networks on Small Datasets
and 5 more not named here
loads_data 5 other papers
2 dated earlier, 2 later, 1 same year, order not determinable
2303.08500 (2023-03)
The Devil's Advocate: Shattering the Illusion of Unexploitable Data using Diffusion Models
aaai_28365 (2024 (year only))
AGS: Affordable and Generalizable Substitute Training for Transferable Adversarial Attack
2401.06187 (2024-01)
Scissorhands: Scrub Data Influence via Connection Sensitivity in Networks
and 2 more not named here
get_x_y_from_data_dict 4 other papers
3 dated earlier, 1 later · 1 only as an assembly brick
2304.04934 (2023-04)
Model Sparsity Can Simplify Machine Unlearning
2310.18574 (2023-10)
Breaking the Trilemma of Privacy, Utility, Efficiency via Controllable Machine Unlearning
2401.06187 (2024-01)
Scissorhands: Scrub Data Influence via Connection Sensitivity in Networks
and 1 more not named here
moving_average 4 other papers
3 dated earlier, 1 later
1905.05380 (2019-05)
Control Regularization for Reduced Variance Reinforcement Learning
2205.03146 (2022-05)
CLIP-CLOP: CLIP-Guided Collage and Photomontage
2401.05779 (2024-01)
Erasing Undesirable Influence in Diffusion Models
and 1 more not named here
sorted_nicely 4 other papers
2 dated earlier, 2 later
2106.14568 (2021-06)
Deep Ensembling with No Overhead for either Training or Testing: The All-Round Blessings of Dynamic Sparsity
2110.10809 (2021-10)
Hierarchical Skills for Efficient Exploration
2407.12383 (2024-07)
Reliable and Efficient Concept Erasure of Text-to-Image Diffusion Models
and 1 more not named here

Shown: the 20 functions present in the most other papers. 23 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

The technological advancements in diffusion models (DMs) have demonstrated unprecedented capabilities in text-to-image generation and are widely used in diverse applications. However, they have also raised significant societal concerns, such as the generation of harmful content and copyright disputes. Machine unlearning (MU) has emerged as a promising solution, capable of removing undesired generative capabilities from DMs. However, existing MU evaluation systems present several key challenges that can result in incomplete and inaccurate assessments. To address these issues, we propose UnlearnCanvas, a comprehensive high-resolution stylized image dataset that facilitates the evaluation of the unlearning of artistic styles and associated objects. This dataset enables the establishment of a standardized, automated evaluation framework with 7 quantitative metrics assessing various aspects of the unlearning performance for DMs. Through extensive experiments, we benchmark 9 state-of-the-art MU methods for DMs, revealing novel insights into their strengths, weaknesses, and underlying mechanisms. Additionally, we explore challenging unlearning scenarios for DMs to evaluate worst-case performance against adversarial prompts, the unlearning of finer-scale concepts, and sequential unlearning. We hope that this study can pave the way for developing more effective, accurate, and robust DM unlearning methods, ensuring safer and more ethical applications of DMs in the future. The dataset, benchmark, and codes are publicly available at https://unlearn-canvas.netlify.app/.

For agents

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

get_harvested_code_for_paper("2402.11846")
get_code_for_paper("2402.11846")
have("2402.11846")

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

curl https://syntology.ai/api/ran/2402.11846.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.11846.svg)](https://syntology.ai/paper/2402.11846)

Connect an agent — have() is free.