SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2502.02513 · NeurIPS · 2025

Diffusion Generative Modeling on Lie Group Representations

Djork-Arné Clevert, Tuan Le, Marco Bertolini

arXiv · PDF · Open in the Atlas

Code that ran

20 of 25 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (4 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 4 have no contract verdict; 15 failures of our instrument; 5 no recorded run), measured 2026-11-22; 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/2502.02513.json · badge: /api/ran/2502.02513.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
jasonkyuyim/se3_diffusion community
Papers with Code archive
15 ran · 4 no recorded run
gcorso/DiffDock community
Papers with Code archive
5 ran · 1 no recorded run
FunctionStatusWhere it lives
BackboneUpdate Ran
ran on a synthesized input · constructed an object; no result was computed
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("89d66ab8c231b799")
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 an instance of its own class, BackboneUpdate: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "c_s": 4,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type BackboneUpdate — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a BackboneUpdate 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
EdgeTransition Ran
ran on a synthesized input · constructed an object; no result was computed
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("149672794a6ce709")
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 an instance of its own class, EdgeTransition: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "node_embed_size": 8,
    "edge_embed_in": 6,
    "edge_embed_out": 4,
    "num_layers": 2,
    "node_dilation": 2,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EdgeTransition — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a EdgeTransition 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
StructureModuleTransition Ran
ran on a synthesized input · constructed an object; no result was computed
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("b385876410970ada")
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 an instance of its own class, StructureModuleTransition: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'c': 4
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type StructureModuleTransition — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a StructureModuleTransition 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
TorsionAngles Ran
ran on a synthesized input · constructed an object; no result was computed
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("e673b21f3beecd05")
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 an instance of its own class, TorsionAngles: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# TorsionAngles.__init__ takes:
# - c: int (feature dimension)
# - num_torsions: int (number of torsion angles)
# - eps: float (default 1e-08)

# TorsionAngles.forward takes:
# - s: tensor of shape (batch, c)

ARGS = {
    "c": 4,
    "num_torsions": 3,
    "eps": 1e-08,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type TorsionAngles — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a TorsionAngles 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
axis_angle_to_matrix Ran
ran; our contract draft was wrong, not the code
gcorso/DiffDock/utils/diffusion_utils.py
code served (permissive licence) · get_code("914f2bbd56ed0c20")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# axis_angle_to_matrix expects axis_angle as a tensor of shape (..., 3)
# where the magnitude is the angle in radians around the vector's direction
# Let's create a small batch of axis-angle rotations

axis_angle = torch.randn(2, 3)

ARGS = {
    "axis_angle": axis_angle,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 3) — shape (2, 3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
axis_angle_to_quaternion Ran
ran; our contract draft was wrong, not the code
gcorso/DiffDock/utils/diffusion_utils.py
code served (permissive licence) · get_code("e79a535a0171afe6")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned 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)

# axis_angle_to_quaternion expects:
# - axis_angle: tensor of shape (..., 3) representing rotations as axis-angle vectors

# Create a small batch of axis-angle rotations
# Shape: (2, 3) - batch of 2 rotations, each with 3 components (axis * angle)
axis_angle = torch.randn(2, 3)

ARGS = {
    "axis_angle": axis_angle,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
identity_quats Ran
ran; our contract draft was wrong, not the code
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("0ed21f60eea50f93")
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 3, dtype family float, shape (2, 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
from typing import Tuple, Optional

torch.manual_seed(0)

ARGS = {
    'batch_dims': (2, 3),
    'dtype': torch.float32,
    'device': torch.device('cpu'),
    'requires_grad': True,
}
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 3, dtype family float, shape (2, 3, 4) — shape (2, 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
identity_rot_mats Ran
ran; our contract draft was wrong, not the code
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("1af8886de7bf2f78")
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, 3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

ARGS = {
    "batch_dims": (2, 3),
    "dtype": torch.float32,
    "device": torch.device("cpu"),
    "requires_grad": True,
}
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, 3, 3) — shape (2, 3, 3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
identity_trans Ran
ran; our contract draft was wrong, not the code
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("ca30dcd53a3e6051")
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 3, dtype family float, shape (2, 4, 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 torch
from typing import Optional, Tuple

torch.manual_seed(0)

# Construct arguments for identity_trans function
ARGS = {
    "batch_dims": (2, 4),
    "dtype": torch.float32,
    "device": torch.device("cpu"),
    "requires_grad": True,
}
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 3, dtype family float, shape (2, 4, 3) — shape (2, 4, 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
invert_quat Ran
ran; our contract draft was wrong, not the code
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("8a3167dc8230216b")
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)

# The function invert_quat takes a quaternion tensor and inverts it.
# It expects a tensor of shape (..., 4) where the last dimension contains
# quaternion components [w, x, y, z] or similar.

# Create a sample quaternion tensor with batch size 2 and dimension 4
quat = torch.randn(2, 4)

ARGS = {
    "quat": quat
}
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
invert_rot_mat Ran
ran, violated the contract we drafted
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("49fba1a002800760")
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 3, dtype family float, shape (2, 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 invert_rot_mat takes a rotation matrix and transposes its last two dimensions
# We need to provide a tensor that represents a rotation matrix
# Rotation matrices are typically square (n x n) but can also be batched
# Let's create a small batched rotation matrix tensor with shape (2, 4, 4)
# This gives us batch size 2, with 4x4 rotation matrices

rot_mat = torch.randn(2, 4, 4)

ARGS = {
    "rot_mat": rot_mat
}
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 3, dtype family float, shape (2, 4, 4) — shape (2, 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
modify_conformer_torsion_angles Ran
ran; our contract draft was wrong, not the code
gcorso/DiffDock/utils/diffusion_utils.py
code served (permissive licence) · get_code("35c7828fc29cfeec")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family float, shape (4, 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 torch
import numpy as np
from scipy.spatial.transform import Rotation as R

torch.manual_seed(0)

# Create small tensors and arrays
num_atoms = 4
num_edges = 3

# pos: positions of atoms (num_atoms x 3)
pos = torch.randn(num_atoms, 3, dtype=torch.float32)

# edge_index: (num_edges x 2) tensor with atom indices
edge_index = torch.tensor([
    [0, 1],
    [1, 2],
    [2, 3]
], dtype=torch.long)

# mask_rotate: (num_edges x num_atoms) boolean mask
# Indicates which atoms rotate for each edge
mask_rotate = torch.zeros((num_edges, num_atoms), dtype=torch.bool)
# For each edge, mark some atoms that should rotate
mask_rotate[0, 2:] = True  # atoms 2, 3 rotate for edge 0
mask_rotate[1, 3:] = True  # atom 3 rotates for edge 1
mask_rotate[2, 3:] = True  # atom 3 rotates for edge 2

# torsion_updates: (num_edges,) array of rotation angles
torsion_updates = np.array([0.5, 0.0, 0.3], dtype=np.float32)

ARGS = {
    "pos": pos,
    "edge_index": edge_index,
    "mask_rotate": mask_rotate,
    "torsion_updates": torsion_updates,
    "as_numpy": False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (4, 3) — shape (4, 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
quat_multiply Ran
ran; our contract draft was wrong, not the code
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("7504b8f0badd29b3")
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 numpy as np

torch.manual_seed(0)

# The function quat_multiply takes two quaternion tensors
# quat1 and quat2, where each quaternion has 4 components
# The function uses:
# - quat1.new_tensor() to create a tensor from numpy array
# - torch operations on the tensors
# - Broadcasting and tensor operations

# Create two quaternion tensors with batch dimension
# Quaternions have 4 components each (w, x, y, z)
# Shape: (batch_size, 4) for simple case, or more complex batch shapes

quat1 = torch.randn(2, 4)  # batch of 2 quaternions
quat2 = torch.randn(2, 4)  # batch of 2 quaternions

ARGS = {
    "quat1": quat1,
    "quat2": quat2,
}
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 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
quat_multiply_by_vec Ran
ran; our contract draft was wrong, not the code
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("7d8488e4260d21b8")
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 numpy as np

torch.manual_seed(0)

# Create a quaternion tensor: shape (batch=2, quat_dim=4)
quat = torch.randn(2, 4)

# Create a vector tensor: shape (batch=2, vec_dim=3)
# Note: The function uses vec[..., None, :, None] so it needs to be broadcastable
# The slicing [:, 1:, :] suggests the second dimension should be compatible with vectors
vec = torch.randn(2, 3)

ARGS = {
    "quat": quat,
    "vec": vec,
}
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 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
quat_to_rot Ran
ran; our contract draft was wrong, not the code
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("075f5885f662efa7")
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 3, dtype family float, shape (2, 3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# The function expects a quaternion tensor of shape [*, 4]
# Let's create a small batch of quaternions
quat_input = torch.randn(2, 4)  # batch size 2, 4 quaternion components

ARGS = {
    "quat": quat_input
}
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 3, dtype family float, shape (2, 3, 3) — shape (2, 3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
quaternion_to_matrix Ran
ran; our contract draft was wrong, not the code
gcorso/DiffDock/utils/diffusion_utils.py
code served (permissive licence) · get_code("757d8fbfcc15222b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create a batch of quaternions with shape (2, 4)
# Quaternions are represented as [w, x, y, z] where w is the real part
# We need to normalize them for valid rotation quaternions
quaternions = torch.randn(2, 4)
# Normalize to unit quaternions
quaternions = quaternions / torch.norm(quaternions, dim=-1, keepdim=True)

ARGS = {
    "quaternions": quaternions
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 3) — shape (2, 3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rigid_transform_Kabsch_3D_torch Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
gcorso/DiffDock/utils/diffusion_utils.py
code served (permissive licence) · get_code("87d39d0333f7619e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned 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 math

torch.manual_seed(0)

# Create two 3xN point clouds for rigid transformation
# A is the source point cloud (3 rows, N columns)
# B is the target point cloud (3 rows, N columns)
# Using N=5 points for a small example

A = torch.randn(3, 5, dtype=torch.float32)
B = torch.randn(3, 5, dtype=torch.float32)

ARGS = {
    "A": A,
    "B": B,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
scipy, 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
rot_matmul Ran
ran; our contract draft was wrong, not the code
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("2e879b992afd2e24")
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 3, dtype family float, shape (2, 3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create two rotation matrices with shape [batch, 3, 3]
# Using batch size 2 for testing
a = torch.randn(2, 3, 3)
b = torch.randn(2, 3, 3)

ARGS = {
    "a": a,
    "b": b,
}
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 3, dtype family float, shape (2, 3, 3) — shape (2, 3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rot_to_quat Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("2892b10c249d305e")
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)

# The function expects a tensor with shape (..., 3, 3) representing rotation matrices
# We'll create a batch of 2 rotation matrices, each of shape (3, 3)
# Total shape: (2, 3, 3)

# Create a valid rotation matrix using QR decomposition
batch_size = 2
random_matrix = torch.randn(batch_size, 3, 3)
# Convert to orthogonal matrices via QR decomposition
rot_matrices = torch.linalg.qr(random_matrix)[0]

ARGS = {
    "rot": rot_matrices
}
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
rot_vec_mul Ran
ran; our contract draft was wrong, not the code
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("8f2b10d716763add")
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, 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 torch

torch.manual_seed(0)

# Create rotation matrices: shape [2, 3, 3]
# Batch size 2, each element is a 3x3 rotation matrix
r = torch.eye(3).unsqueeze(0).repeat(2, 1, 1)
# Add some rotation by modifying slightly
r[0] = torch.tensor([[0.866, -0.5, 0.0],
                     [0.5, 0.866, 0.0],
                     [0.0, 0.0, 1.0]])
r[1] = torch.tensor([[1.0, 0.0, 0.0],
                     [0.0, 0.866, -0.5],
                     [0.0, 0.5, 0.866]])

# Create coordinate vectors: shape [2, 3]
# Batch size 2, each vector has 3 coordinates
t = torch.tensor([[1.0, 2.0, 3.0],
                  [4.0, 5.0, 6.0]])

ARGS = {
    "r": r,
    "t": t,
}
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, 3) — shape (2, 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
InvariantPointAttention Not yet run
unverified (harvested, no recorded run)
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("dfce785d0ad4df3b")
How it ran

no invocation recorded.

IpaScore Not yet run
unverified (harvested, no recorded run)
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("0ff59b29cf0ab633")
How it ran

no invocation recorded.

Rigid Not yet run
unverified (harvested, no recorded run)
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("ec6868fba749c722")
How it ran

no invocation recorded.

Rotation Not yet run
unverified (harvested, no recorded run)
jasonkyuyim/se3_diffusion/model/ipa_pytorch.py
code served (permissive licence) · get_code("de43748820aede00")
How it ran

no invocation recorded.

modify_conformer Not yet run
unverified (harvested, no recorded run)
gcorso/DiffDock/utils/diffusion_utils.py
code served (permissive licence) · get_code("eb444236e23ac6f3")
How it ran

no invocation recorded.

Also present in other papers' code

Of 25 samples lifted from this paper's linked repositories, 25 carry a function identity and 0 do not. Of the 25 distinct functions with an identity, 6 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 0 the same month, 1 later) and 19 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
quaternion_to_matrix
brick of this paper's assembly: helper
6 other papers
5 dated earlier, 1 later · 1 only as an assembly brick
2203.04810 (2022-03)
Benchmarking Graphormer on Large-Scale Molecular Modeling Datasets
2209.07589 (2022-09)
PIZZA: A Powerful Image-only Zero-Shot Zero-CAD Approach to 6 DoF Tracking
2210.01776 (2022-10)
DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking
and 3 more not named here
axis_angle_to_matrix
brick of this paper's assembly: helper
3 other papers
3 dated earlier · 1 only as an assembly brick
2203.04810 (2022-03)
Benchmarking Graphormer on Large-Scale Molecular Modeling Datasets
2207.05053 (2022-07)
Learning Continuous Grasping Function with a Dexterous Hand from Human Demonstrations
2210.01776 (2022-10)
DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking
axis_angle_to_quaternion
brick of this paper's assembly: helper
2 other papers
2 dated earlier · 1 only as an assembly brick
2203.04810 (2022-03)
Benchmarking Graphormer on Large-Scale Molecular Modeling Datasets
2210.01776 (2022-10)
DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking
identity_rot_mats
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2306.03117 (2023-06)
Str2Str: A Score-based Framework for Zero-shot Protein Conformation Sampling
rot_matmul
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2306.03117 (2023-06)
Str2Str: A Score-based Framework for Zero-shot Protein Conformation Sampling
rot_vec_mul
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2306.03117 (2023-06)
Str2Str: A Score-based Framework for Zero-shot Protein Conformation Sampling

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

We introduce a novel class of score-based diffusion processes that operate directly in the representation space of Lie groups. Leveraging the framework of Generalized Score Matching, we derive a class of Langevin dynamics that decomposes as a direct sum of Lie algebra representations, enabling the modeling of any target distribution on any (non-Abelian) Lie group. Standard score-matching emerges as a special case of our framework when the Lie group is the translation group. We prove that our generalized generative processes arise as solutions to a new class of paired stochastic differential equations (SDEs), introduced here for the first time. We validate our approach through experiments on diverse data types, demonstrating its effectiveness in real-world applications such as SO(3)-guided molecular conformer generation and modeling ligand-specific global SE(3) transformations for molecular docking, showing improvement in comparison to Riemannian diffusion on the group itself. We show that an appropriate choice of Lie group enhances learning efficiency by reducing the effective dimensionality of the trajectory space and enables the modeling of transitions between complex data distributions.

For agents

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

get_harvested_code_for_paper("2502.02513")
get_code_for_paper("2502.02513")
have("2502.02513")

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

curl https://syntology.ai/api/ran/2502.02513.json

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

[![Syntology run record](https://syntology.ai/api/ran/2502.02513.svg)](https://syntology.ai/paper/2502.02513)

Connect an agent — have() is free.