SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2106.12673 · 2021

Conditional Deformable Image Registration with Convolutional Neural Network

arXiv · PDF · Open in the Atlas

Code that ran

14 of 17 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 13 have no contract verdict; 1 failure of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2106.12673.json · badge: /api/ran/2106.12673.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
cwmok/Conditional_LapIRN canonical (classifier label)
classifier basis: census
8 ran · 0 no recorded run
cwmok/dirac community (classifier label)
classifier basis not recorded on the link
6 ran · 3 no recorded run
FunctionStatusWhere it lives
JacboianDet Ran
ran on a synthesized input
cwmok/Conditional_LapIRN/Code/miccai2021_model.py
code served (permissive licence) · get_code("faf9dce71d5373ba")
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 (2, 3, 4, 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 torch
import numpy as np

torch.manual_seed(0)

# The function takes y_pred and sample_grid, both tensors
# Looking at the operations:
# - J = y_pred + sample_grid (element-wise addition)
# - J is indexed as J[:, 1:, :-1, :-1, :] etc., so it's 5D tensor
# - The last dimension is indexed [0], [1], [2], so it has at least 3 elements
# - Need batch dimension, 3 spatial dimensions, and feature dimension

# Create y_pred: shape (batch, dim1, dim2, dim3, 3)
# batch=2, spatial dims: 4x5x6, feature=3
y_pred = torch.randn(2, 4, 5, 6, 3)

# Create sample_grid with same shape
sample_grid = torch.randn(2, 4, 5, 6, 3)

ARGS = {
    "y_pred": y_pred,
    "sample_grid": sample_grid,
}
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 (2, 3, 4, 5) — shape (2, 3, 4, 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
compute_tre Ran
ran on a synthesized input
cwmok/dirac/Code/BRATS_test_DIRAC.py
code served (permissive licence) · get_code("d2c770b494ce1f32")
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 (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

# The function compute_tre takes:
# - x: array-like of shape (N, D) where N is batch size and D is dimension
# - y: array-like of shape (N, D)
# - spacing: tuple of length 3 (default=(1, 1, 1))

# Looking at the function body:
# - x and y are subtracted: (x - y)
# - result is multiplied element-wise by spacing
# - np.linalg.norm is computed along axis=1
# This means x and y should be 2D arrays where axis=1 represents the spatial dimensions

# Create small test arrays
batch_size = 2
num_dims = 3

x = np.array([
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0]
])

y = np.array([
    [1.5, 2.5, 3.5],
    [4.5, 5.5, 6.5]
])

spacing = (1.0, 1.0, 1.0)

ARGS = {
    "x": x,
    "y": y,
    "spacing": spacing
}
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 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
continuous_contrastive_loss Ran
ran on a synthesized input
cwmok/dirac/Code/bratsreg_model_stage.py
code served (permissive licence) · get_code("81e311bc7a3294ba")
How it ran

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

import torch

torch.manual_seed(0)

# Examine the function:
# - smo_1, smo_2: appear to be tensors that support multiplication and subtraction
# - w_1, w_2: appear to be scalars or tensors that support multiplication with smo_1 and smo_2

ARGS = {
    "smo_1": torch.randn(2, 4),
    "smo_2": torch.randn(2, 4),
    "w_1": torch.tensor(0.5),
    "w_2": torch.tensor(0.3),
}
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
create_image_pyramid Ran
ran on a synthesized input
cwmok/dirac/Code/Functions.py
code served (permissive licence) · get_code("d0dc3465c6af23ad")
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. 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)

# Create a small 3D image tensor (1, 1, H, W, D)
# Using distinct sizes: 1, 1, 4, 3, 2 for batch, channel, height, width, depth
image = torch.randn(1, 1, 4, 3, 2)

# Down sample factors
down_sample_factor = [4, 2]

# Device
device = 'cpu'

# Channel
channel = 1

ARGS = {
    'image': image,
    'down_sample_factor': down_sample_factor,
    'device': device,
    'channel': channel
}
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 list — a list 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
dice Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
cwmok/Conditional_LapIRN/Code/Train_cLapIRN_2D.py
code served (permissive licence) · get_code("7c8b1b17367fd701")
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, 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

# Create two small numpy arrays representing segmentation masks
# im1: predicted segmentation
# atlas: ground truth/reference segmentation

np.random.seed(0)

# Create small segmentation masks with shape (4, 3)
# Values represent class labels (0 for background, 1-2 for foreground classes)
im1 = np.array([
    [0, 1, 2],
    [1, 1, 2],
    [0, 1, 1],
    [2, 2, 0]
], dtype=np.int32)

atlas = np.array([
    [0, 1, 1],
    [1, 1, 2],
    [0, 1, 2],
    [2, 2, 0]
], dtype=np.int32)

ARGS = {
    "im1": im1,
    "atlas": atlas
}
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, 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
dice Ran
ran on a synthesized input
cwmok/dirac/Code/BRATS_test_DIRAC.py
code served (permissive licence) · get_code("5e2ed8f9e07d1692")
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

# The dice function takes two parameters: im1 and atlas
# Both are used with numpy array indexing and comparison operations
# They should be numpy arrays of the same shape
# The function compares class predictions (im1) against an atlas/ground truth
# and computes Dice coefficient for each class (excluding class 0)

# Create small test arrays with shape (4, 3) to keep them tiny
# im1: predicted segmentation with classes 0, 1, 2
im1 = np.array([
    [0, 1, 2],
    [1, 2, 1],
    [0, 2, 2],
    [1, 1, 0]
], dtype=np.int32)

# atlas: ground truth segmentation with same classes
atlas = np.array([
    [0, 1, 1],
    [1, 2, 2],
    [0, 2, 2],
    [1, 0, 0]
], dtype=np.int32)

ARGS = {
    "im1": im1,
    "atlas": atlas
}
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
generate_grid Ran
ran on a synthesized input
cwmok/Conditional_LapIRN/Code/Functions.py
code served (permissive licence) · get_code("526f7aabe47bb29e")
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 int, shape (4, 5, 6, 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

# The function takes an imgshape parameter which should be a tuple/list of 3 integers
# representing the shape of a 3D image (depth, height, width)
imgshape = (4, 5, 6)

ARGS = {
    "imgshape": imgshape
}
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 4, dtype family int, shape (4, 5, 6, 3) — shape (4, 5, 6, 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
generate_grid_unit Ran
ran on a synthesized input
cwmok/Conditional_LapIRN/Code/Functions.py
code served (permissive licence) · get_code("e5fb50af358b26e5")
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 (4, 5, 6, 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

def generate_grid_unit(imgshape):
    x = (np.arange(imgshape[0]) - (imgshape[0] - 1) / 2) / (imgshape[0] - 1) * 2
    y = (np.arange(imgshape[1]) - (imgshape[1] - 1) / 2) / (imgshape[1] - 1) * 2
    z = (np.arange(imgshape[2]) - (imgshape[2] - 1) / 2) / (imgshape[2] - 1) * 2
    grid = np.rollaxis(np.array(np.meshgrid(z, y, x)), 0, 4)
    grid = np.swapaxes(grid, 0, 2)
    grid = np.swapaxes(grid, 1, 2)
    return grid

# The function takes a single parameter: imgshape
# imgshape is used with np.arange() three times, so it needs to be indexable
# with indices 0, 1, 2 and support arithmetic operations
# It should be a tuple or list of 3 integers representing image dimensions

ARGS = {
    "imgshape": (4, 5, 6)
}
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 4, dtype family float, shape (4, 5, 6, 3) — shape (4, 5, 6, 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
get_gaussian_kernel Ran
ran on a synthesized input
cwmok/dirac/Code/Functions.py
code served (permissive licence) · get_code("55263643ea3ef4b6")
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 5, dtype family float, shape (1, 1, 5, 5, 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 torch
import math

torch.manual_seed(0)

# Based on the function signature and body:
# - kernel_size: tuple of integers representing spatial dimensions
# - sigma: tuple of floats representing standard deviations for each dimension
# 
# The function creates a Gaussian kernel by:
# 1. Creating meshgrids for each dimension in kernel_size
# 2. Computing a Gaussian distribution for each dimension with given sigma
# 3. Multiplying them together and normalizing
# 4. Reshaping to (1, 1, *kernel.size())

ARGS = {
    "kernel_size": (5, 5, 5),
    "sigma": (1.0, 1.0, 1.0),
}
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 5, dtype family float, shape (1, 1, 5, 5, 5) — shape (1, 1, 5, 5, 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
neg_Jdet_loss Ran
ran on a synthesized input
cwmok/Conditional_LapIRN/Code/miccai2021_model.py
code served (permissive licence) · get_code("8ebf7b299843ca5f")
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)

# Construct y_pred: a 5D tensor (batch, depth, height, width, channels)
# The function slices along dimensions 1-3 and accesses dimension 4 for x,y,z components
# Shape: (batch=2, depth=4, height=3, width=5, channels=3)
y_pred = torch.randn(2, 4, 3, 5, 3)

# Construct sample_grid: same shape as y_pred
# The function adds y_pred + sample_grid, then does differential operations
sample_grid = torch.randn(2, 4, 3, 5, 3)

ARGS = {
    "y_pred": y_pred,
    "sample_grid": sample_grid,
}
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
smoothloss Ran
ran on a synthesized input
cwmok/Conditional_LapIRN/Code/miccai2021_model.py
code served (permissive licence) · get_code("b683a710c6b4f28a")
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 function expects y_pred to be a 5D tensor
# It computes differences along dimensions 2, 3, and 4
# Shape: (batch, channels, depth, height, width)
y_pred = torch.randn(2, 3, 4, 5, 6)

ARGS = {
    "y_pred": y_pred
}
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
smoothloss Ran
ran on a synthesized input
cwmok/Conditional_LapIRN/Code/miccai2021_model_2D.py
code served (permissive licence) · get_code("90d1a77d4263a27e")
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)

# y_pred is indexed as [:, :, 1:, :] and [:, :, :-1, :] for height dimension
# and [:, :, :, 1:] and [:, :, :, :-1] for width dimension
# This suggests a 4D tensor: (batch, channels, height, width)
# We need batch=2, channels=4, height=3, width=5 to have distinct dimensions

y_pred = torch.randn(2, 4, 3, 5)

ARGS = {
    "y_pred": y_pred
}
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
transform_unit_flow_to_flow Ran
ran on a synthesized input
cwmok/Conditional_LapIRN/Code/Functions.py
code served (permissive licence) · get_code("05bfed21c3b36e8a")
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 (2, 3, 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 tensor with shape (x, y, z, 4)
# where it unpacks the last dimension and scales the first 3 channels
# by factors derived from x, y, z dimensions

# Create a small flow tensor with distinct dimensions
# shape: (x=2, y=3, z=4, channels=4)
flow = torch.randn(2, 3, 4, 4, dtype=torch.float32)

ARGS = {
    "flow": flow
}
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 4, dtype family float, shape (2, 3, 4, 4) — shape (2, 3, 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
weighted_smoothloss Ran
ran on a synthesized input
cwmok/dirac/Code/bratsreg_model_stage.py
code served (permissive licence) · get_code("f80c634a4ec99824")
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)

# Create y_pred tensor with shape (batch, channels, depth, height, width)
# Using distinct sizes: batch=2, channels=3, depth=4, height=5, width=6
y_pred = torch.randn(2, 3, 4, 5, 6)

# Create weight tensor with same shape as y_pred
weight = torch.ones(2, 3, 4, 5, 6)

ARGS = {
    "y_pred": y_pred,
    "weight": weight,
}
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
affine_aug Not yet run
unverified (harvested, no recorded run)
cwmok/dirac/Code/BRATS_train_DIRAC.py
code served (permissive licence) · get_code("72a85c1e6b65e2a9")
How it ran

no invocation recorded.

affine_aug Not yet run
unverified (harvested, no recorded run)
cwmok/dirac/Code/BRATS_train_DIRAC_D.py
code served (permissive licence) · get_code("308ec5997d538495")
How it ran

no invocation recorded.

jacobian_determinant Not yet run
unverified (harvested, no recorded run)
cwmok/dirac/Code/Functions.py
code served (permissive licence) · get_code("ce29fe9090724b39")
How it ran

no invocation recorded.

Also present in other papers' code

Of 17 samples lifted from this paper's linked repositories, 17 carry a function identity and 0 do not. Of the 17 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 16 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
dice 1 other paper
1 dated earlier
2003.09514 (2020-03)
Fast Symmetric Diffeomorphic Image Registration with Convolutional Neural Networks

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

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

Abstract

Recent deep learning-based methods have shown promising results and runtime advantages in deformable image registration. However, analyzing the effects of hyperparameters and searching for optimal regularization parameters prove to be too prohibitive in deep learning-based methods. This is because it involves training a substantial number of separate models with distinct hyperparameter values. In this paper, we propose a conditional image registration method and a new self-supervised learning paradigm for deep deformable image registration. By learning the conditional features that are correlated with the regularization hyperparameter, we demonstrate that optimal solutions with arbitrary hyperparameters can be captured by a single deep convolutional neural network. In addition, the smoothness of the resulting deformation field can be manipulated with arbitrary strength of smoothness regularization during inference. Extensive experiments on a large-scale brain MRI dataset show that our proposed method enables the precise control of the smoothness of the deformation field without sacrificing the runtime advantage or registration accuracy.

For agents

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

get_harvested_code_for_paper("2106.12673")
get_code_for_paper("2106.12673")
have("2106.12673")

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

curl https://syntology.ai/api/ran/2106.12673.json

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

[![Syntology run record](https://syntology.ai/api/ran/2106.12673.svg)](https://syntology.ai/paper/2106.12673)

Connect an agent — have() is free.