SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2003.10778 · 2020

PanNuke Dataset Extension, Insights and Baselines

arXiv · PDF · Open in the Atlas

Code that ran

35 of 40 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 28 have no contract verdict; 6 failures of our instrument; 5 no recorded run), measured 2026-11-21; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2003.10778.json · badge: /api/ran/2003.10778.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
kaiseem/pointnu-net community (classifier label)
classifier basis not recorded on the link
13 ran · 2 no recorded run
okunator/Dippa community (classifier label)
classifier basis not recorded on the link
11 ran · 2 no recorded run
alexjfoote/reetoolbox community (classifier label)
classifier basis not recorded on the link
10 ran · 1 no recorded run
copy not recorded community
not in the paper's declared repositories
1 ran · 0 no recorded run
FunctionStatusWhere it lives
AJI Ran
ran on a synthesized input
okunator/Dippa/src/metrics/metrics.py
code served (permissive licence) · get_code("70dfe5fac9f92989")
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 numpy as np

# Create simple labeled instance masks for testing AJI
# true: ground truth segmentation with instance IDs (0 = background)
# pred: predicted segmentation with instance IDs (0 = background)

# Create a 10x10 image with 2 instances in ground truth
true = np.zeros((10, 10), dtype=np.int32)
true[1:4, 1:4] = 1  # Instance 1
true[6:9, 6:9] = 2  # Instance 2

# Create a 10x10 predicted image with 2 instances (slightly different positions)
pred = np.zeros((10, 10), dtype=np.int32)
pred[1:5, 1:5] = 1  # Instance 1 (overlaps with true instance 1)
pred[6:8, 6:8] = 2  # Instance 2 (overlaps with true instance 2)

ARGS = {
    "true": true,
    "pred": pred,
}
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
AJI_plus Ran
ran on a synthesized input
okunator/Dippa/src/metrics/metrics.py
code served (permissive licence) · get_code("0ec0f8e9ea494c64")
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 numpy as np
from scipy.optimize import linear_sum_assignment

# Create sample ground truth and predicted masks
# Ground truth mask with 3 instances (labeled 1, 2, 3)
true_mask = np.zeros((10, 10), dtype=np.uint8)
true_mask[0:3, 0:3] = 1
true_mask[3:6, 3:6] = 2
true_mask[6:9, 6:9] = 3

# Predicted mask with 3 instances, with some overlap and some misalignment
pred_mask = np.zeros((10, 10), dtype=np.uint8)
pred_mask[0:4, 0:4] = 1  # Overlaps with true instance 1
pred_mask[3:7, 3:7] = 2  # Overlaps with true instances 2 and 3
pred_mask[7:10, 7:10] = 3  # Overlaps with true instance 3

ARGS = {
    'true': true_mask,
    'pred': pred_mask,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — 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
DICE2 Ran
ran on a synthesized input
okunator/Dippa/src/metrics/metrics.py
code served (permissive licence) · get_code("1942c4a97287927f")
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 numpy as np

# Create ground truth mask with labeled regions (H=4, W=3)
# Labels: 0 = background, 1 = first object, 2 = second object
true = np.array([
    [0, 1, 1],
    [0, 1, 2],
    [2, 2, 2],
    [0, 0, 0]
], dtype=np.uint8)

# Create predicted mask with similar but slightly different labeled regions
pred = np.array([
    [0, 1, 1],
    [0, 1, 1],
    [2, 2, 2],
    [0, 0, 0]
], dtype=np.uint8)

ARGS = {
    "true": true,
    "pred": pred,
}
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
accuracy Ran
ran on a synthesized input
alexjfoote/reetoolbox/reetoolbox/metrics.py
code served (permissive licence) · get_code("504d09c7575c571d")
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
from sklearn.metrics import accuracy_score

torch.manual_seed(0)

# The function expects a dict with 'labels' and 'outputs' keys
# - 'labels': ground truth labels (1D array-like)
# - 'outputs': model outputs (2D tensor, logits for each class)
# It uses torch.max(outputs, 1) to get predictions, so outputs should be 2D

# Create small tensors with batch size 2 and 4 classes
batch_size = 2
num_classes = 4

# Create random logits (outputs from model)
outputs = torch.randn(batch_size, num_classes)

# Create ground truth labels
labels = torch.tensor([1, 3])

# Construct the results dict
results = {
    'labels': labels,
    'outputs': outputs
}

ARGS = {
    'results': results
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
sklearn, 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
adversarial_accuracy Ran
ran on a synthesized input
alexjfoote/reetoolbox/reetoolbox/metrics.py
code served (permissive licence) · get_code("f2adf71dc1102ec8")
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 numpy as np
from sklearn.metrics import accuracy_score

torch.manual_seed(0)

# Construct a results dict with the required keys
batch_size = 2
num_classes = 4

# Create mock labels (integer class indices)
labels = np.array([0, 1])

# Create mock adversarial_outputs (logits before argmax)
# Shape: (batch_size, num_classes)
adversarial_outputs = torch.randn(batch_size, num_classes)

results = {
    'labels': labels,
    'adversarial_outputs': adversarial_outputs
}

ARGS = {
    'results': results
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, 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
apply_transforms Ran
ran on a synthesized input
alexjfoote/reetoolbox/reetoolbox/trainer.py
code served (permissive licence) · get_code("e45684da05f7df4a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import random

torch.manual_seed(0)

# Define a minimal model
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 2)
    
    def forward(self, x):
        return self.linear(x)

# Define a minimal adversarial optimizer
class SimpleAdvOptimizer:
    def __init__(self):
        self.model = None
    
    def optimise(self, inputs, targets, reset_weights=True):
        # Return dummy loss and modified inputs
        loss = torch.tensor(0.1)
        modified_inputs = inputs + 0.01 * torch.randn_like(inputs)
        return loss, modified_inputs

# Construct arguments
model = SimpleModel()
inputs = torch.randn(2, 4)  # batch_size=2, input_dim=4
labels = torch.randint(0, 2, (2,))  # batch_size=2, 2 classes
adv_optimisers = [SimpleAdvOptimizer() for _ in range(5)]
k = 3
reset_weights = True

ARGS = {
    "model": model,
    "inputs": inputs,
    "labels": labels,
    "adv_optimisers": adv_optimisers,
    "k": k,
    "reset_weights": reset_weights,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
binarize Ran
ran on a synthesized input
okunator/Dippa/src/utils/mask_utils.py
code served (permissive licence) · get_code("df537567c64fea4a")
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 (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 a small instance map with some labeled regions
# Instance maps typically contain integer labels where 0 is background
# and positive integers represent different instances/objects
inst_map = np.array([
    [0, 0, 1, 1],
    [0, 2, 2, 1],
    [3, 2, 0, 0],
    [3, 3, 0, 0]
], dtype=np.int32)

ARGS = {
    "inst_map": inst_map
}
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 (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
binary_focal_loss Ran
ran on a synthesized input
kaiseem/pointnu-net/losses/focal_loss.py
code served (permissive licence) · get_code("241b293cb3a3c43d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn.functional as F

torch.manual_seed(0)

# Construct arguments for binary_focal_loss
x = torch.sigmoid(torch.randn(2, 4))  # Batch size 2, 4 outputs, sigmoid to get probabilities in [0,1]
y = torch.randint(0, 2, (2, 4)).float()  # Binary targets, same shape as x

ARGS = {
    'x': x,
    'y': y,
    'alpha': 0.25,
    'gamma': 2.0,
    'reduction': 'none'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
check_sha1 Ran
ran on a synthesized input
kaiseem/pointnu-net/models/download.py
code served (permissive licence) · get_code("484859bf41155745")
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 bool, 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 hashlib
import tempfile
import os

# Create a temporary file with known content
temp_dir = tempfile.mkdtemp()
test_file = os.path.join(temp_dir, "test_file.txt")

# Write some test content to the file
test_content = b"Hello, this is test content for SHA1 hashing."
with open(test_file, 'wb') as f:
    f.write(test_content)

# Calculate the actual SHA1 hash of the content
sha1 = hashlib.sha1()
sha1.update(test_content)
expected_hash = sha1.hexdigest()

# Create arguments dictionary
ARGS = {
    "filename": test_file,
    "sha1_hash": expected_hash
}
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 bool, 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_edts_forPenalizedLoss Ran
ran on a synthesized input
kaiseem/pointnu-net/losses/dice_loss.py
code served (permissive licence) · get_code("b53ea61df431b694")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 4, dtype family float, shape (2, 4, 3, 5) 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
from scipy.ndimage import distance_transform_edt

# Create a small binary segmentation ground truth array
# Shape: (batch_size, x, y, z)
batch_size = 2
x_dim = 4
y_dim = 3
z_dim = 5

GT = np.random.randint(0, 2, size=(batch_size, x_dim, y_dim, z_dim)).astype(bool)

ARGS = {
    "GT": GT
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 4, dtype family float, shape (2, 4, 3, 5) — shape (2, 4, 3, 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
draw_stuff_contours Ran
ran on a synthesized input
okunator/Dippa/src/utils/viz_utils.py
code served (permissive licence) · get_code("42361797de298480")
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 (10, 10, 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 numpy as np
import cv2

# Create a small instance segmentation map (H=10, W=10)
# with instance IDs 1, 2, 3 and background 0
inst_map = np.zeros((10, 10), dtype=np.uint8)
inst_map[1:4, 1:4] = 1  # instance 1
inst_map[5:8, 5:8] = 2  # instance 2
inst_map[2:5, 6:9] = 3  # instance 3

# Create a semantic segmentation map (H=10, W=10)
# with class labels matching the instances
type_map = np.zeros((10, 10), dtype=np.uint8)
type_map[1:4, 1:4] = 1  # class 1 (epithelial)
type_map[5:8, 5:8] = 2  # class 2 (neoplastic)
type_map[2:5, 6:9] = 3  # class 3 (spindle)

# Create original image (H=10, W=10, 3 channels BGR)
image = np.ones((10, 10, 3), dtype=np.uint8) * 200

# Create classes dictionary mapping class names to integers
classes = {
    'background': 0,
    'epithelial': 1,
    'neoplastic': 2,
    'spindle': 3,
    'fibroblast': 4,
    'muscle': 5,
    'endothelial': 6,
    'connective': 7,
    'dead': 8
}

ARGS = {
    'inst_map': inst_map,
    'image': image,
    'type_map': type_map,
    'classes': classes,
    'fill_contours': False,
    'thickness': 2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, 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 (10, 10, 3) — shape (10, 10, 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
fix_duplicates Ran
ran on a synthesized input
okunator/Dippa/src/utils/mask_utils.py
code served (permissive licence) · get_code("b8109b79acc64068")
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 (6, 6) 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
from scipy import ndimage as ndi

# Create a simple instance map with some duplicated instances
# The map contains instance IDs, with 0 being background
# We'll create a scenario where after mirror padding, some instance IDs appear in disconnected regions
inst_map = np.array([
    [1, 1, 2, 2, 0, 0],
    [1, 1, 2, 2, 0, 0],
    [3, 3, 0, 0, 1, 1],
    [3, 3, 0, 0, 1, 1],
    [0, 0, 4, 4, 0, 0],
    [0, 0, 4, 4, 0, 0]
], dtype=np.uint8)

ARGS = {
    "inst_map": inst_map
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — 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 (6, 6) — shape (6, 6) 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_dataloader Ran
ran on a synthesized input
alexjfoote/reetoolbox/reetoolbox/utils.py
code served (permissive licence) · get_code("8fa7bbbf062a1aad")
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

torch.manual_seed(0)

# Create a minimal dataset-like object that supports indexing and len()
class MinimalDataset:
    def __init__(self, X, y):
        self.X = X
        self.y = y
    
    def __len__(self):
        return len(self.X)
    
    def __getitem__(self, idx):
        return (self.X[idx], self.y[idx])

# Create small tensors: batch size 2, feature dim 4, label dim 3
X = torch.randn(2, 4)
y = torch.randn(2, 3)

# Create dataset
dataset = MinimalDataset(X, y)

ARGS = {
    "dataset": dataset,
    "n": 2,
    "batch_size": 1,
    "num_workers": 0,
    "shuffle": 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 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
get_model_file Ran
ran on a synthesized input
kaiseem/pointnu-net/models/model_store.py
code served (permissive licence) · get_code("4bf53b623193a6f7")
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 os
import tempfile
import shutil

# Create a temporary directory to serve as the root
temp_dir = tempfile.mkdtemp()

# Create the models subdirectory
models_dir = os.path.join(temp_dir, '.torch', 'models')
os.makedirs(models_dir, exist_ok=True)

# Create a dummy model file
model_name = 'test_model'
model_file_path = os.path.join(models_dir, model_name + '.pth')
with open(model_file_path, 'w') as f:
    f.write('')

ARGS = {
    'name': 'test_model',
    'root': os.path.join(temp_dir, '.torch', 'models')
}
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 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_resnet_file Ran
ran on a synthesized input
kaiseem/pointnu-net/models/model_store.py
code served (permissive licence) · get_code("420ae9a3239bb203")
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 os
import zipfile
import tempfile
from unittest.mock import patch, MagicMock

# We need to mock the functions that would require network/file I/O
def mock_check_sha1(file_path, sha1_hash):
    """Mock function that always returns True"""
    return True

def mock_download(url, path, overwrite):
    """Mock function that creates a dummy zip file"""
    # Create a minimal valid .pth file
    os.makedirs(os.path.dirname(path), exist_ok=True)
    with zipfile.ZipFile(path, 'w') as zf:
        # Create a dummy model file inside the zip
        zf.writestr('resnet50-25c4b50.pth', b'dummy model content')

# Patch the functions before defining ARGS
import sys
current_module = sys.modules[__name__]
setattr(current_module, 'check_sha1', mock_check_sha1)
setattr(current_module, 'download', mock_download)

# Create a temporary directory for the model cache
temp_dir = tempfile.mkdtemp()

ARGS = {
    'name': 'resnet50',
    'root': temp_dir
}
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 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
iou Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
kaiseem/pointnu-net/losses/lovasz_losses.py
code served (permissive licence) · get_code("d94639b28b2134ff")
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 1, dtype family float, shape (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
import torch

torch.manual_seed(0)

# Construct valid arguments for the iou function
# preds: prediction array/tensor with class indices
# labels: ground truth array/tensor with class indices
# C: number of classes
# EMPTY: value to return for empty unions (default 1.0)
# ignore: class index to ignore (default None)
# per_image: whether to compute per-image IoU (default False)

# Create small prediction and label arrays
# Using shape (2, 4) - batch of 2, spatial dimension of 4
preds = np.array([[0, 1, 2, 3],
                   [1, 2, 0, 1]], dtype=np.int64)

labels = np.array([[0, 1, 1, 3],
                    [1, 2, 2, 1]], dtype=np.int64)

# Number of classes
C = 4

# EMPTY value for classes with no union
EMPTY = 1.0

# Class to ignore during computation
ignore = None

# Whether to compute per-image IoU
per_image = False

ARGS = {
    "preds": preds,
    "labels": labels,
    "C": C,
    "EMPTY": EMPTY,
    "ignore": ignore,
    "per_image": per_image
}
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 ndarray, ndim 1, dtype family float, shape (4,) — shape (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
iou_binary Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
kaiseem/pointnu-net/losses/lovasz_losses.py
code served (permissive licence) · get_code("1211530cd527bd0b")
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 numpy as np

def isnan(x):
    return x != x

def mean(l, ignore_nan=False, empty=0):
    """
    nanmean compatible with generators.
    """
    l = iter(l)
    if ignore_nan:
        from itertools import filterfalse
        l = filterfalse(isnan, l)
    try:
        n = 1
        acc = next(l)
    except StopIteration:
        if empty == 'raise':
            raise ValueError('Empty mean')
        return empty
    for n, v in enumerate(l, 2):
        acc += v
    if n == 1:
        return acc
    return acc / n

def iou_binary(preds, labels, EMPTY=1.0, ignore=None, per_image=True):
    """
    IoU for foreground class
    binary: 1 foreground, 0 background
    """
    if not per_image:
        preds, labels = ((preds,), (labels,))
    ious = []
    for pred, label in zip(preds, labels):
        intersection = ((label == 1) & (pred == 1)).sum()
        union = ((label == 1) | (pred == 1) & (label != ignore)).sum()
        if not union:
            iou = EMPTY
        else:
            iou = float(intersection) / float(union)
        ious.append(iou)
    iou = mean(ious)
    return 100 * iou

# Setup code
np.random.seed(0)

# Create small prediction and label arrays for binary segmentation
# Shape: (height=4, width=3) for two images
preds_img1 = np.array([[0, 1, 1],
                       [1, 1, 0],
                       [0, 1, 1],
                       [1, 0, 0]], dtype=np.int32)

labels_img1 = np.array([[0, 1, 0],
                        [1, 1, 1],
                        [0, 1, 0],
                        [1, 0, 1]], dtype=np.int32)

preds_img2 = np.array([[1, 1, 0],
                       [0, 1, 1],
                       [1, 1, 0],
                       [0, 0, 1]], dtype=np.int32)

labels_img2 = np.array([[1, 0, 0],
                        [0, 1, 1],
                        [1, 0, 0],
                        [0, 1, 1]], dtype=np.int32)

# Arguments for iou_binary function
ARGS = {
    'preds': np.array([preds_img1, preds_img2]),
    'labels': np.array([labels_img1, labels_img2]),
    'EMPTY': 1.0,
    'ignore': 0,
    'per_image': True,
}
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 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
label_patch Ran
ran on a synthesized input
alexjfoote/reetoolbox/reetoolbox/utils.py
code served (permissive licence) · get_code("9b00b93154e766c2")
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 numpy as np

torch.manual_seed(0)

# Analyzing the function:
# - patch: used as-is, returned as first element of tuple. Could be any object.
# - count: accessed with count[0], compared with integers (>=5, ==0). Must be indexable.

# Creating valid arguments:
patch = torch.randn(2, 4)  # Small tensor as patch

count = [7]  # List with count[0] >= 5, will trigger first condition

ARGS = {
    "patch": patch,
    "count": count
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — 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
load_resnet Ran
ran on a synthesized input
alexjfoote/reetoolbox/reetoolbox/utils.py
code served (permissive licence) · get_code("e7a3997691bd97d3")
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 ResNet. 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 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 torchvision import models
import tempfile
import os

torch.manual_seed(0)

# Create a temporary model file to load from
temp_dir = tempfile.mkdtemp()
model_path = os.path.join(temp_dir, "test_model.pt")

# Create a dummy ResNet18 model with the expected structure
dummy_model = models.resnet18(pretrained=False)
dummy_model.fc = nn.Sequential(
    nn.Linear(512, 2),
    nn.LogSoftmax(dim=1)
)

# Save the model state dict
torch.save(dummy_model.state_dict(), model_path)

ARGS = {
    'model_path': model_path,
    'n_classes': 2,
    'hidden_size': 512,
    'pretrained': False,
    'device': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ResNet — a ResNet output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ResNet 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
lovasz_grad Ran
ran, violated the contract we drafted
kaiseem/pointnu-net/losses/lovasz_losses.py
code served (permissive licence) · get_code("e85b5d21dc5b8382")
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 (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)

# gt_sorted should be a 1D tensor of binary/boolean values
# representing ground truth sorted by some error metric
# The function expects a tensor with sum(), float(), cumsum() methods
gt_sorted = torch.tensor([0, 0, 1, 1], dtype=torch.long)

ARGS = {
    "gt_sorted": gt_sorted
}
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 (4,) — shape (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_one_hot Ran
ran on a synthesized input
kaiseem/pointnu-net/losses/dice_loss.py
code served (permissive licence) · get_code("f93cd0d2bb2e4795")
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 4, dtype family float, shape (1, 4, 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 torch
import numpy as np

torch.manual_seed(0)

# Create a small input tensor of shape [N, 1, *]
# N=2 (batch size), 1 (channel dimension), and spatial dims [3, 4]
input_tensor = torch.tensor([[[[0, 1, 2, 1],
                               [3, 0, 1, 2],
                               [1, 2, 0, 3]]]], dtype=torch.long)

# num_classes: the maximum class index is 3, so we need at least 4 classes
num_classes = 4

ARGS = {
    "input": input_tensor,
    "num_classes": num_classes
}
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 Tensor, ndim 4, dtype family float, shape (1, 4, 3, 4) — shape (1, 4, 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
normalize Ran
ran on a synthesized input
okunator/Dippa/src/utils/img_utils.py
code served (permissive licence) · get_code("247e39ae4944e031")
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 float, shape (4, 3, 2) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a small test image with HWC format
np.random.seed(0)
img_hwc = np.random.randn(4, 3, 2).astype(np.float32) * 10 + 50  # Height=4, Width=3, Channels=2

# Create a small test image with CHW format
img_chw = np.random.randn(2, 4, 3).astype(np.float32) * 10 + 50  # Channels=2, Height=4, Width=3

ARGS = {
    "img": img_hwc,
    "channels": "HWC",
    "standardize": True
}
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 3, dtype family float, shape (4, 3, 2) — shape (4, 3, 2) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
numericalSort Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from rr-learning/disentanglement_dataset
pointer only; the stored copy get_code serves, from rr-learning/disentanglement_dataset, is pointer only (licence: CC-BY-4.0) · get_code("d39d1cd46009e948")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

percentile_normalize Ran
ran on a synthesized input
okunator/Dippa/src/utils/img_utils.py
code served (permissive licence) · get_code("dc269540663f3fb7")
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 float, shape (4, 3, 2) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

np.random.seed(0)

# Create a small test image with HWC format
# Height=4, Width=3, Channels=2
img_hwc = np.random.rand(4, 3, 2).astype(np.float32) * 255

ARGS = {
    'img': img_hwc,
    'channels': 'HWC'
}
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 3, dtype family float, shape (4, 3, 2) — shape (4, 3, 2) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
percentile_normalize_and_clamp Ran
ran on a synthesized input
okunator/Dippa/src/utils/img_utils.py
code served (permissive licence) · get_code("12218d17d813eb74")
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 float, shape (4, 3, 2) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a small test image with shape (H, W, C)
# Using distinct dimensions: H=4, W=3, C=2
np.random.seed(0)
im = np.random.randn(4, 3, 2) * 100  # Scale up to have meaningful percentiles

ARGS = {
    "im": im,
    "a_min": -1,
    "a_max": 1,
}
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 3, dtype family float, shape (4, 3, 2) — shape (4, 3, 2) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
poly2mask Ran
ran on a synthesized input
okunator/Dippa/src/utils/save_utils.py
code served (permissive licence) · get_code("ac05aad67c119188")
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 (20, 20) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a simple polygon contour (exterior coordinates)
# This represents a simple square polygon with 5 points (closed polygon)
contour = np.array([
    [0, 0],
    [10, 0],
    [10, 10],
    [0, 10],
    [0, 0]
], dtype=np.float64)

# Define the shape of the output mask (height, width)
shape = (20, 20)

# Define optional offsets
x_off = 0
y_off = 0

ARGS = {
    "contour": contour,
    "shape": shape,
    "x_off": x_off,
    "y_off": y_off,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, skimage — 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 (20, 20) — shape (20, 20) 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
reduce_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
kaiseem/pointnu-net/losses/generalise_focal_loss.py
code served (permissive licence) · get_code("93358ba7aa223343")
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 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.functional as F

torch.manual_seed(0)

# Create a small elementwise loss tensor
loss = torch.randn(2, 4)

# Test with different reduction modes
ARGS = {
    "loss": loss,
    "reduction": "mean"
}
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 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
remove_small_objects Ran
ran on a synthesized input
okunator/Dippa/src/utils/mask_utils.py
code served (permissive licence) · get_code("15b380d239dc63a4")
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 (3, 5) 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
from scipy import ndimage as ndi

# Create a simple labeled array with some small and large objects
ar = np.array([[0, 0, 0, 1, 0],
               [1, 1, 1, 0, 0],
               [1, 1, 1, 0, 1]], dtype=np.int32)

# Alternative: boolean array
ar_bool = np.array([[0, 0, 0, 1, 0],
                    [1, 1, 1, 0, 0],
                    [1, 1, 1, 0, 1]], dtype=bool)

# Create output array for out parameter
out_array = np.zeros_like(ar, dtype=np.int32)

ARGS = {
    'ar': ar_bool,
    'min_size': 64,
    'connectivity': 1,
    'in_place': False,
    'out': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — 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 (3, 5) — shape (3, 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
rmse Ran
ran on a synthesized input
alexjfoote/reetoolbox/reetoolbox/metrics.py
code served (permissive licence) · get_code("30f91c811cbc554f")
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 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

torch.manual_seed(0)

# The rmse function takes a single parameter 'a' which is used with:
# - torch.pow(a, 2) - element-wise squaring
# - torch.mean() - computing mean
# - torch.sqrt() - square root
# So 'a' should be a tensor

a = torch.randn(2, 4)

ARGS = {
    "a": a
}
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 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
run_val Ran
ran on a synthesized input
alexjfoote/reetoolbox/reetoolbox/trainer.py
code served (permissive licence) · get_code("db1bae643c60dd31")
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 torch
import torch.nn as nn

torch.manual_seed(0)

# Build a minimal model that works with the function
class SimpleClassifier(nn.Module):
    def __init__(self, num_classes=3):
        super().__init__()
        self.linear = nn.Linear(4, num_classes)
    
    def forward(self, x):
        return self.linear(x)

# Build a minimal data loader (list of batches)
inputs_1 = torch.randn(2, 4)
labels_1 = torch.tensor([0, 1])
inputs_2 = torch.randn(2, 4)
labels_2 = torch.tensor([2, 0])
loader = [(inputs_1, labels_1), (inputs_2, labels_2)]

# Build model and criterion
model = SimpleClassifier(num_classes=3)
criterion = nn.CrossEntropyLoss()

ARGS = {
    'model': model,
    'loader': loader,
    'criterion': criterion,
    'device': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
short_hash Ran
ran; our contract draft was wrong, not the code
kaiseem/pointnu-net/models/model_store.py
code served (permissive licence) · get_code("77bc65d2018e9d9e")
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 torch

# The function short_hash takes a 'name' parameter and looks it up in _model_sha1
# It checks if the name exists in the dictionary, and if so, returns the first 8 characters of the checksum

ARGS = {
    'name': 'resnet50'
}
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 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
stack_prediction Ran
ran on a synthesized input
kaiseem/pointnu-net/infer_pannuke.py
code served (permissive licence) · get_code("274a4c2795ca8dd4")
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 float, shape (256, 256, 6) 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

# Construct seg_masks: list of 2D numpy arrays (256x256), each representing a segmentation mask
# The function zips these with cate_labels, so we need matching lengths
seg_masks = [
    np.random.rand(256, 256).astype(np.float32),
    np.random.rand(256, 256).astype(np.float32),
    np.random.rand(256, 256).astype(np.float32),
]

# Construct cate_labels: list of category labels (integers 0-4, NOT 5 as per assertion)
cate_labels = [0, 1, 2]

ARGS = {
    "seg_masks": seg_masks,
    "cate_labels": cate_labels,
}
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 3, dtype family float, shape (256, 256, 6) — shape (256, 256, 6) 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
training_bout Ran
ran on a synthesized input
alexjfoote/reetoolbox/reetoolbox/trainer.py
code served (permissive licence) · get_code("d6bd6ff6fe6aad52")
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. 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 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 copy

torch.manual_seed(0)

# Define a minimal model
class SimpleModel(nn.Module):
    def __init__(self):
        super(SimpleModel, self).__init__()
        self.fc = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.fc(x)

# Define a minimal attack class
class SimpleAttack:
    def __init__(self):
        self.model = None
    
    def optimise(self, inputs, targets, reset_weights=True):
        # Return a tuple of (something, adversarial_inputs)
        return (None, inputs.clone())

# Create data loaders as lists of (inputs, labels) tuples
def create_loader(num_batches=2, batch_size=2, input_dim=4, num_classes=3):
    loader = []
    for _ in range(num_batches):
        inputs = torch.randn(batch_size, input_dim)
        labels = torch.randint(0, num_classes, (batch_size,))
        loader.append((inputs, labels))
    return loader

# Instantiate components
model = SimpleModel()
train_loader = create_loader(num_batches=2, batch_size=2, input_dim=4, num_classes=3)
val_loader = create_loader(num_batches=2, batch_size=2, input_dim=4, num_classes=3)
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)
attack = SimpleAttack()

# Define arguments dictionary
ARGS = {
    'model': model,
    'train_loader': train_loader,
    'val_loader': val_loader,
    'optimizer': optimizer,
    'criterion': criterion,
    'adversarial': False,
    'attacks': [],
    'epochs': 2,
    'm': 1,
    'last_n': 1,
    'k': 2,
    'device': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — 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
untargeted_loss Ran
ran on a synthesized input
alexjfoote/reetoolbox/reetoolbox/optimisers.py
code served (permissive licence) · get_code("432e211219845a2b")
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)

# outputs: a tensor that will be indexed with .gather(1, ...)
# The gather operation gathers values from dimension 1 using indices from labels
# outputs should have shape (batch_size, num_classes)
outputs = torch.randn(2, 4)

# labels: indices to gather from outputs along dimension 1
# labels should have shape (batch_size,) with values in range [0, num_classes)
labels = torch.tensor([0, 2])

ARGS = {
    "outputs": outputs,
    "labels": labels,
}
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
weight_reduce_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
kaiseem/pointnu-net/losses/generalise_focal_loss.py
code served (permissive licence) · get_code("8553862254d49840")
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 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.functional as F

torch.manual_seed(0)

# Create tiny tensors for the function parameters
loss_tensor = torch.randn(2, 4)  # Elementwise loss tensor
weight_tensor = torch.ones(2, 4)  # Element-wise weights

ARGS = {
    'loss': loss_tensor,
    'weight': weight_tensor,
    'reduction': 'mean',
    'avg_factor': None
}
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 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
download Not yet run
unverified (harvested, no recorded run)
kaiseem/pointnu-net/models/download.py
code served (permissive licence) · get_code("8faa1b60508d7b44")
How it ran

no invocation recorded.

draw_thing_contours Not yet run
unverified (harvested, no recorded run)
okunator/Dippa/src/utils/viz_utils.py
code served (permissive licence) · get_code("5c30f53f7cde0bda")
How it ran

no invocation recorded.

quality_focal_loss Not yet run
unverified (harvested, no recorded run)
kaiseem/pointnu-net/losses/generalise_focal_loss.py
code served (permissive licence) · get_code("e5457e2f64864dd0")
How it ran

no invocation recorded.

targeted_loss Not yet run
unverified (harvested, no recorded run)
alexjfoote/reetoolbox/reetoolbox/optimisers.py
code served (permissive licence) · get_code("21780435c6ca1913")
How it ran

no invocation recorded.

viz_patches Not yet run
unverified (harvested, no recorded run)
okunator/Dippa/src/patching/viz_patches.py
code served (permissive licence) · get_code("86befdff7aeeaec4")
How it ran

no invocation recorded.

Also present in other papers' code

Of 40 samples lifted from this paper's linked repositories, 40 carry a function identity and 0 do not. Of the 40 distinct functions with an identity, 7 are also present in other papers whose code we hold (5 with at least one paper dated earlier, 6 the same month, 6 later) and 33 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
lovasz_grad 40 other papers
4 dated earlier, 2 the same month, 34 later · 1 only as an assembly brick
1505.04597 (2015-05)
U-Net: Convolutional Networks for Biomedical Image Segmentation
1807.10165 (2018-07)
UNet++: A Nested U-Net Architecture for Medical Image Segmentation
1811.11721 (2018-11)
CCNet: Criss-Cross Attention for Semantic Segmentation
and 37 more not named here
iou 24 other papers
2 dated earlier, 2 the same month, 20 later
1807.10165 (2018-07)
UNet++: A Nested U-Net Architecture for Medical Image Segmentation
1811.11721 (2018-11)
CCNet: Criss-Cross Attention for Semantic Segmentation
2003.13948 (2020-03)
Segmenting Transparent Objects in the Wild
and 21 more not named here
iou_binary 24 other papers
2 dated earlier, 2 the same month, 20 later
1807.10165 (2018-07)
UNet++: A Nested U-Net Architecture for Medical Image Segmentation
1811.11721 (2018-11)
CCNet: Criss-Cross Attention for Semantic Segmentation
2003.13948 (2020-03)
Segmenting Transparent Objects in the Wild
and 21 more not named here
reduce_loss 18 other papers
1 the same month, 17 later
2003.10152 (2020-03)
SOLOv2: Dynamic and Fast Instance Segmentation
2107.08430 (2021-07)
YOLOX: Exceeding YOLO Series in 2021
2108.06747 (2021-08)
SOTR: Segmenting Objects with Transformers
and 15 more not named here
short_hash 11 other papers
5 dated earlier, 2 the same month, 4 later
1803.08904 (2018-03)
Context Encoding for Semantic Segmentation
1809.02983 (2018-09)
Dual Attention Network for Scene Segmentation
1905.00641 (2019-05)
RetinaFace: Single-stage Dense Face Localisation in the Wild
and 8 more not named here
weight_reduce_loss 7 other papers
1 the same month, 6 later
2003.10152 (2020-03)
SOLOv2: Dynamic and Fast Instance Segmentation
2107.08430 (2021-07)
YOLOX: Exceeding YOLO Series in 2021
2023.findings-emnlp.517 (2023 (year only))
Text2Tree: Aligning Text Representation to the Label Tree Hierarchy for Imbalanced Medical Classification
and 4 more not named here
numericalSort 3 other papers
3 dated earlier
1603.06059 (2016-03)
Generating Natural Questions About an Image
1703.04105 (2017-03)
Combining Residual Networks with LSTMs for Lipreading
1906.03292 (2019-06)
On the Transfer of Inductive Bias from Simulation to the Real World: a New Disentanglement Dataset

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-21.

Repositories linked to this paper

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

Abstract

The emerging area of computational pathology (CPath) is ripe ground for the application of deep learning (DL) methods to healthcare due to the sheer volume of raw pixel data in whole-slide images (WSIs) of cancerous tissue slides. However, it is imperative for the DL algorithms relying on nuclei-level details to be able to cope with data from `the clinical wild', which tends to be quite challenging. We study, and extend recently released PanNuke dataset consisting of ~200,000 nuclei categorized into 5 clinically important classes for the challenging tasks of segmenting and classifying nuclei in WSIs. Previous pan-cancer datasets consisted of only up to 9 different tissues and up to 21,000 unlabeled nuclei and just over 24,000 labeled nuclei with segmentation masks. PanNuke consists of 19 different tissue types that have been semi-automatically annotated and quality controlled by clinical pathologists, leading to a dataset with statistics similar to the clinical wild and with minimal selection bias. We study the performance of segmentation and classification models when applied to the proposed dataset and demonstrate the application of models trained on PanNuke to whole-slide images. We provide comprehensive statistics about the dataset and outline recommendations and research directions to address the limitations of existing DL tools when applied to real-world CPath applications.

For agents

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

get_harvested_code_for_paper("2003.10778")
get_code_for_paper("2003.10778")
have("2003.10778")

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

curl https://syntology.ai/api/ran/2003.10778.json

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

[![Syntology run record](https://syntology.ai/api/ran/2003.10778.svg)](https://syntology.ai/paper/2003.10778)

Connect an agent — have() is free.