SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1906.06818 · 2019

Stacked Capsule Autoencoders

arXiv · PDF · Open in the Atlas

Code that ran

25 of 29 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 23 have no contract verdict; 1 failure of our instrument; 4 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/1906.06818.json · badge: /api/ran/1906.06818.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
akosiorek/stacked_capsule_autoencoders canonical (classifier label)
classifier basis: census
3 ran · 4 no recorded run
KohavTal/SCAE_Project community (classifier label)
classifier basis not recorded on the link
16 ran · 0 no recorded run
bdsaglam/torch-scae community (classifier label)
classifier basis not recorded on the link
4 ran · 0 no recorded run
YuZiHanorz/stacked_capsule_autoencoders community (classifier label)
classifier basis not recorded on the link
2 ran · 0 no recorded run
FunctionStatusWhere it lives
BatchMLP Ran
ran on a synthesized input
bdsaglam/torch-scae/torch_scae/nn_ext.py
code served (permissive licence) · get_code("d6debf9640bd00f2")
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 Sequential. 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 = {
    'n_mlps': 3,
    'sizes': [4, 8, 6, 2],
    'activation': nn.ReLU,
    'activate_final': False,
    'bias_final': 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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
BestStack Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/nn_ext.py
code served (permissive licence) · get_code("3a64054158d703c5")
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 Sequential. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "in_channels": 3,
    "out_channels": [16, 32, 64],
    "kernel_sizes": [3, 3, 3],
    "strides": [1, 2, 1],
    "activation": nn.ReLU,
    "activate_final": True,
    "dropout": 0.1
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
Conv2dStack Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/nn_ext.py
code served (permissive licence) · get_code("1f7981add3e51350")
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 Sequential. 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)

# Define arguments for Conv2dStack
ARGS = {
    'in_channels': 3,
    'out_channels': [16, 32, 64],
    'kernel_sizes': [3, 3, 3],
    'strides': [1, 2, 1],
    'activation': nn.ReLU,
    'activate_final': 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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
MLP Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/nn_ext.py
code served (permissive licence) · get_code("018f862a3ae6ed96")
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 Sequential. 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 = {
    'sizes': [4, 8, 16, 3],
    'activation': nn.ReLU,
    'activate_final': True,
    'bias': 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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
affine_grid Ran
ran on a synthesized input
bdsaglam/torch-scae/torch_scae/part_decoder.py
code served (permissive licence) · get_code("7972654ad56da3c4")
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, 4, 5, 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)

# The function takes:
# - matrix: a tensor of shape (batch, 6) representing affine transformation parameters
# - size: a tuple/list where size[-2:] gives (height, width)

# Create a small batch of affine transformation matrices
# Shape: (batch_size=2, 6 parameters)
matrix = torch.randn(2, 6)

# Create a size tuple - can be any shape as long as last 2 elements are height and width
# Using (1, 3, 4, 5) to be distinct: 1 channel, 3 batch(?), 4 height, 5 width
size = (1, 3, 4, 5)

ARGS = {
    "matrix": matrix,
    "size": size,
}
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, 4, 5, 2) — shape (2, 4, 5, 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
capsule_l2_loss Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/object_decoder.py
code served (permissive licence) · get_code("dbc3e25948c02028")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create a small tensor for caps_presence
# Shape: (batch_size=2, num_caps=4)
# Values should be in a reasonable range for capsule activations (e.g., 0-1)
caps_presence = torch.rand(2, 4)

# Number of classes
n_classes = 2

# Optional within_example_constant (set to None to use default calculation)
within_example_constant = None

ARGS = {
    "caps_presence": caps_presence,
    "n_classes": n_classes,
    "within_example_constant": within_example_constant,
}
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
combined_shape Ran
ran; our contract draft was wrong, not the code
KohavTal/SCAE_Project/torch_scae/general_utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from yinglunxu/reward_poisoning_attack_drl, is pointer only (licence: NONE) · get_code("57542f87e248db45")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from yinglunxu/reward_poisoning_attack_drl, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

conv_output_shape Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/nn_utils.py
code served (permissive licence) · get_code("46a280d5631eb673")
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 int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from typing import Tuple

torch.manual_seed(0)

# The function conv_output_shape takes:
# - input_shape: Tuple[int, int, int] - a 3-tuple representing (channels, height, width)
# - out_channels: int - number of output channels
# - kernel_size: int - size of the convolutional kernel
# - stride: int - stride of the convolution (default=1)
# - padding: int - padding applied to input (default=0)

# Construct valid arguments
input_shape = (3, 32, 32)  # 3 input channels, 32x32 spatial dimensions
out_channels = 16
kernel_size = 3
stride = 1
padding = 0

ARGS = {
    "input_shape": input_shape,
    "out_channels": out_channels,
    "kernel_size": kernel_size,
    "stride": stride,
    "padding": padding,
}
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, ndim 1, dtype family int — 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
conv_output_size Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/nn_utils.py
code served (permissive licence) · get_code("6c8cdd002826e750")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

ARGS = {
    "in_size": 32,
    "kernel_size": 5,
    "stride": 1,
    "padding": 0
}
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 int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
cross_entropy_safe Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/math_ops.py
code served (permissive licence) · get_code("f65130ec27f57098")
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 small tensors for the function parameters
# true_probs: probability distribution (should sum to 1 along dim)
# probs: predicted probabilities (should sum to 1 along dim)
batch_size = 2
num_classes = 4

# Create true_probs as a valid probability distribution
true_probs = torch.tensor([
    [0.1, 0.2, 0.3, 0.4],
    [0.25, 0.25, 0.25, 0.25]
], dtype=torch.float32)

# Create probs as predicted probabilities
probs = torch.tensor([
    [0.15, 0.25, 0.35, 0.25],
    [0.2, 0.3, 0.25, 0.25]
], dtype=torch.float32)

ARGS = {
    "true_probs": true_probs,
    "probs": probs,
    "dim": -1
}
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
geometric_transform Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/cv_ops.py
code served (permissive licence) · get_code("eaacf4c75671e1c5")
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, 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 torch
import math

torch.manual_seed(0)

# Create a small pose tensor with shape [..., 6]
# Using batch size 2, so shape is (2, 6)
pose_tensor = torch.randn(2, 6)

ARGS = {
    "pose_tensor": pose_tensor,
    "similarity": False,
    "nonlinear": True,
    "as_matrix": 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 Tensor, ndim 2, dtype family float, shape (2, 6) — shape (2, 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
geometric_transform Ran
ran on a synthesized input
bdsaglam/torch-scae/torch_scae/cv_ops.py
code served (permissive licence) · get_code("aee48248137bef44")
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, 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 torch
import math

torch.manual_seed(0)

# Create a pose tensor with shape [..., 6]
# Using batch size 2, so shape is [2, 6]
pose_tensor = torch.randn(2, 6)

ARGS = {
    "pose_tensor": pose_tensor,
    "similarity": False,
    "nonlinear": True,
    "as_matrix": 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 Tensor, ndim 2, dtype family float, shape (2, 6) — shape (2, 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_latest_file_iteration Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/general_utils.py
code served (permissive licence) · get_code("7b5f49577ae87818")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pathlib
import tempfile
import os

# Create a temporary directory with test files
temp_dir = tempfile.mkdtemp()

# Create some test files with numbers in their names
test_files = [
    "model_iteration_5.pt",
    "model_iteration_12.pt",
    "model_iteration_8.pt",
    "checkpoint_3.pth",
]

for filename in test_files:
    filepath = os.path.join(temp_dir, filename)
    with open(filepath, 'w') as f:
        f.write("dummy content")

ARGS = {
    'folder': temp_dir,
    'pattern': '*'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
hex_to_rgb Ran
ran on a synthesized input
akosiorek/stacked_capsule_autoencoders/capsules/plot.py
code served (permissive licence) · get_code("4c1d49543510213d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

def hex_to_rgb(value):
    value = value.lstrip('#')
    lv = len(value)
    return tuple((int(value[i:i + lv // 3], 16) / 255.0 for i in range(0, lv, lv // 3)))

# The function takes a hex color string and converts it to RGB
# It expects a string starting with '#' followed by hex digits
# Common formats: '#RGB' (3 chars), '#RRGGBB' (6 chars), '#RRGGBBAA' (8 chars)

ARGS = {
    'value': '#FF5733'
}
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, 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
invert_affine Ran
ran on a synthesized input
bdsaglam/torch-scae/torch_scae/part_decoder.py
code served (permissive licence) · get_code("00f1c79b7fde1b74")
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, 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)

# Construct a batch of (..., 2, 3) affine matrices
# Shape: (batch_size, 2, 3) where:
#   - First 2 rows, 2 cols: linear transformation (2x2)
#   - First 2 rows, last col: translation (2x1)
matrix = torch.tensor([
    [[1.0, 0.5, 2.0],
     [0.3, 1.2, 3.0]],
    [[0.8, 0.2, 1.5],
     [0.1, 0.9, 2.5]]
], dtype=torch.float32)

ARGS = {
    "matrix": matrix
}
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, 2, 3) — shape (2, 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
load_data_from_mat Ran
ran on a synthesized input
YuZiHanorz/stacked_capsule_autoencoders/load_affnist.py
code served (permissive licence) · get_code("dba87d282c88c543")
How it ran

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

import scipy.io as sio
import tempfile
import os
import numpy as np

# Create a temporary .mat file with test data
temp_dir = tempfile.mkdtemp()
mat_path = os.path.join(temp_dir, "test_data.mat")

# Create a simple structure to save
test_struct = {}
test_struct['simple_array'] = np.array([1, 2, 3, 4])
test_struct['nested_value'] = 42

# Save as .mat file
sio.savemat(mat_path, test_struct)

ARGS = {
    'path': mat_path
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
log_safe Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/math_ops.py
code served (permissive licence) · get_code("dcc02359a095982c")
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)

# Create a small tensor with positive values to test log_safe
# Using a 2x4 tensor with values > eps to test the main path
tensor = torch.tensor([[0.5, 1.0, 2.0, 0.1],
                       [0.01, 100.0, 0.0001, 0.5]], dtype=torch.float32)

eps = 1e-16

ARGS = {
    "tensor": tensor,
    "eps": eps
}
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
measure_shape Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/nn_utils.py
code served (permissive licence) · get_code("835457d36160df6b")
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 Size. 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)

class SimpleNetwork(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 8)
        self.linear2 = nn.Linear(8, 3)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

network = SimpleNetwork()
input_shape = (4,)
input_dtype = torch.float32

ARGS = {
    "network": network,
    "input_shape": input_shape,
    "input_dtype": input_dtype,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Size — a Size output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Size output the recorded ndim 1 and dtype family int 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
memoize_arg_transform Ran
ran on a synthesized input
akosiorek/stacked_capsule_autoencoders/capsules/models/model.py
code served (permissive licence) · get_code("8c55f32be627e0ca")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# The function memoize_arg_transform takes args and kwargs as parameters
# It transforms unhashable types to their string representation
# args should be a tuple/list of arguments
# kwargs should be a dictionary

# Create test arguments with mixed hashable and unhashable types
test_args = (
    42,  # hashable int
    "hello",  # hashable string
    torch.tensor([1, 2, 3, 4]),  # unhashable tensor
    [1, 2, 3],  # unhashable list
    {"key": "value"},  # unhashable dict
    np.array([1, 2, 3]),  # unhashable numpy array
)

test_kwargs = {
    "hashable_int": 100,
    "hashable_str": "world",
    "unhashable_tensor": torch.tensor([[1, 2], [3, 4]]),
    "unhashable_list": [10, 20, 30],
    "unhashable_dict": {"nested": "dict"},
}

ARGS = {
    "args": test_args,
    "kwargs": test_kwargs,
}
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, 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
normalize Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/math_ops.py
code served (permissive licence) · get_code("557120b6207db13a")
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)

# Create a small tensor with shape (2, 4)
# The function normalizes along a specified dimension
tensor = torch.randn(2, 4)

# The dim parameter specifies which dimension to sum over
dim = 1

ARGS = {
    "tensor": tensor,
    "dim": dim,
}
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
prepare_model_params Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/factory.py
code served (permissive licence) · get_code("114ca279f4f910b1")
How it ran

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

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'image_shape': (3, 32, 32),
    'n_classes': 10,
    'n_part_caps': 16,
    'n_obj_caps': 8,
    'pcae_cnn_encoder_params': None,
    'pcae_encoder_params': None,
    'pcae_template_generator_params': None,
    'pcae_decoder_params': None,
    'ocae_encoder_set_transformer_params': None,
    'ocae_decoder_capsule_params': None,
    'scae_params': 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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
prod Ran
ran, violated the contract we drafted
KohavTal/SCAE_Project/torch_scae/general_utils.py
code served (permissive licence) · get_code("6d5acfb0087d1ee7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import operator
from functools import reduce

def prod(iterable):
    return reduce(operator.mul, iterable, 1)

# Test cases: prod works with iterables of numbers
# The function multiplies all elements together using reduce

ARGS = {
    'iterable': [2, 3, 4, 5]
}
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 int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
qkv_attention Ran
ran on a synthesized input
KohavTal/SCAE_Project/torch_scae/set_transformer.py
code served (permissive licence) · get_code("11fd17f376687a4c")
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, 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 torch
import numpy as np

torch.manual_seed(0)

# Define dimensions
batch_size = 2
num_queries = 3
num_keys = 4
d_k = 5
d_v = 6

# Create queries: [B, N, d_k]
queries = torch.randn(batch_size, num_queries, d_k)

# Create keys: [B, M, d_k]
keys = torch.randn(batch_size, num_keys, d_k)

# Create values: [B, M, d_v]
values = torch.randn(batch_size, num_keys, d_v)

# Create presence: [B, M] (optional, can be None)
presence = torch.rand(batch_size, num_keys)

ARGS = {
    "queries": queries,
    "keys": keys,
    "values": values,
    "presence": presence,
}
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, 6) — shape (2, 3, 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
relu1 Ran
ran on a synthesized input
akosiorek/stacked_capsule_autoencoders/capsules/math_ops.py
code served (permissive licence) · get_code("7ac5d15a4e829dd7")
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 EagerTensor, 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 tensorflow as tf
import numpy as np

def relu1(x):
    return tf.nn.relu6(x * 6.0) / 6.0

# Create a small TensorFlow tensor
x_value = tf.constant([[1.5, -2.0, 0.5, 3.0],
                       [0.2, -1.5, 2.5, -0.8]], dtype=tf.float32)

ARGS = {
    'x': x_value
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
toNum Ran
ran on a synthesized input
YuZiHanorz/stacked_capsule_autoencoders/handle_csv.py
code served (permissive licence) · get_code("a0c52efb1af7f6d5")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

# The function toNum takes a single parameter: str
# It processes a string to extract a numeric part and a unit part
# The function expects a string input

ARGS = {
    'str': '42.5 kg'
}
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, 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
gaussian_blobs Not yet run
unverified (harvested, no recorded run)
akosiorek/stacked_capsule_autoencoders/capsules/plot.py
code served (permissive licence) · get_code("c94d19b7ab25522a")
How it ran

no invocation recorded.

gaussian_blobs_const_scale Not yet run
unverified (harvested, no recorded run)
akosiorek/stacked_capsule_autoencoders/capsules/plot.py
code served (permissive licence) · get_code("8e494526a6edfb93")
How it ran

no invocation recorded.

safe_ce Not yet run
unverified (harvested, no recorded run)
akosiorek/stacked_capsule_autoencoders/capsules/math_ops.py
code served (permissive licence) · get_code("094969771f521bf6")
How it ran

no invocation recorded.

safe_log Not yet run
unverified (harvested, no recorded run)
akosiorek/stacked_capsule_autoencoders/capsules/math_ops.py
code served (permissive licence) · get_code("d06435ecd510267e")
How it ran

no invocation recorded.

Also present in other papers' code

Of 29 samples lifted from this paper's linked repositories, 29 carry a function identity and 0 do not. Of the 29 distinct functions with an identity, 2 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 2 later) and 27 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
combined_shape 14 other papers
14 later
1912.13465 (2019-12)
Reward-Conditioned Policies
2006.07442 (2020-06)
Self-Imitation Learning via Generalized Lower Bound Q-learning
2010.09635 (2020-10)
Deep Reinforcement Learning with Population-Coded Spiking Neural Network for Continuous Control
and 11 more not named here
prod 7 other papers
1 dated earlier, 6 later
1904.08489 (2019-04)
Semantic Adversarial Attacks: Parametric Transformations That Fool Deep Classifiers
2106.02081 (2021-06)
Solving Schrödinger Bridges via Maximum Likelihood
2311.03153 (2023-11)
Architectural Sweet Spots for Modeling Human Label Variation by the Example of Argument Quality: It's Best to Relate Perspectives!
and 4 more not named here

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

Repositories linked to this paper

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

Abstract

Objects are composed of a set of geometrically organized parts. We introduce an unsupervised capsule autoencoder (SCAE), which explicitly uses geometric relationships between parts to reason about objects. Since these relationships do not depend on the viewpoint, our model is robust to viewpoint changes. SCAE consists of two stages. In the first stage, the model predicts presences and poses of part templates directly from the image and tries to reconstruct the image by appropriately arranging the templates. In the second stage, SCAE predicts parameters of a few object capsules, which are then used to reconstruct part poses. Inference in this model is amortized and performed by off-the-shelf neural encoders, unlike in previous capsule networks. We find that object capsule presences are highly informative of the object class, which leads to state-of-the-art results for unsupervised classification on SVHN (55%) and MNIST (98.7%). The code is available at https://github.com/google-research/google-research/tree/master/stacked_capsule_autoencoders

For agents

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

get_harvested_code_for_paper("1906.06818")
get_code_for_paper("1906.06818")
have("1906.06818")

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

curl https://syntology.ai/api/ran/1906.06818.json

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

[![Syntology run record](https://syntology.ai/api/ran/1906.06818.svg)](https://syntology.ai/paper/1906.06818)

Connect an agent — have() is free.