SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2001.08957 · 2020

An Explicit Local and Global Representation Disentanglement Framework with Applications in Deep Clustering and Unsupervised Object Detection

arXiv · PDF · Open in the Atlas

Code that ran

11 of 18 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 11 have no contract verdict; 0 failures of our instrument; 7 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/2001.08957.json · badge: /api/ran/2001.08957.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
51616/split-vae canonical (classifier label)
classifier basis: census
11 ran · 7 no recorded run
FunctionStatusWhere it lives
calculateIntersection Ran
ran on a synthesized input
51616/split-vae/spair/data.py
code served (permissive licence) · get_code("b1f2137631d4df35")
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

torch.manual_seed(0)

# Based on the function signature and implementation,
# calculateIntersection takes 4 numeric parameters: a0, a1, b0, b1
# These represent interval boundaries and are used in numeric comparisons and arithmetic

ARGS = {
    "a0": 2,
    "a1": 5,
    "b0": 1,
    "b1": 8
}
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 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
calculate_overlap Ran
ran on a synthesized input
51616/split-vae/spair/data.py
code served (permissive licence) · get_code("5af6f50a761a3cd2")
How it ran

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

import torch
import numpy as np

torch.manual_seed(0)

# Define arguments for calculate_overlap function
rand_x = 10  # x-coordinate of random position
rand_y = 20  # y-coordinate of random position

# List of drawn boxes, each box is [x, y, ...]
# Creating boxes with some overlap and some without
drawn_boxes = [
    [15, 25],  # Box that will overlap with rand_x, rand_y
    [50, 50],  # Box that won't overlap
    [8, 18],   # Box with potential partial overlap
]

ARGS = {
    "rand_x": rand_x,
    "rand_y": rand_y,
    "drawn_boxes": drawn_boxes,
}
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 bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
concrete_binary_pre_sigmoid_sample Ran
ran on a synthesized input
51616/split-vae/spair/utils.py
code served (permissive licence) · get_code("15a44e208eb95089")
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

# Set seed for reproducibility
tf.random.set_seed(0)

# Create small test tensors
log_odds = tf.constant([[0.5, -0.3, 1.2, -0.8],
                        [0.1, 0.9, -0.5, 0.4]], dtype=tf.float32)

temperature = tf.constant(0.5, dtype=tf.float32)

eps = 1e-08

ARGS = {
    "log_odds": log_odds,
    "temperature": temperature,
    "eps": eps
}
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
concrete_binary_sample_kl Ran
ran on a synthesized input
51616/split-vae/spair/trainer.py
code served (permissive licence) · get_code("0f6ffc1f82fd1092")
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

# Set random seeds for reproducibility
tf.random.set_seed(0)
np.random.seed(0)

# Create small tensor arguments
batch_size = 2
feature_dim = 4

# Generate sample tensors
pre_sigmoid_sample = tf.constant(
    np.random.randn(batch_size, feature_dim).astype(np.float32)
)

prior_log_odds = tf.constant(
    np.random.randn(batch_size, feature_dim).astype(np.float32)
)

prior_temperature = tf.constant(
    np.abs(np.random.randn(batch_size, feature_dim)).astype(np.float32) + 0.1
)

posterior_log_odds = tf.constant(
    np.random.randn(batch_size, feature_dim).astype(np.float32)
)

posterior_temperature = tf.constant(
    np.abs(np.random.randn(batch_size, feature_dim)).astype(np.float32) + 0.1
)

eps = 1e-08

ARGS = {
    "pre_sigmoid_sample": pre_sigmoid_sample,
    "prior_log_odds": prior_log_odds,
    "prior_temperature": prior_temperature,
    "posterior_log_odds": posterior_log_odds,
    "posterior_temperature": posterior_temperature,
    "eps": eps,
}
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
discretised_logistic_loss Ran
ran on a synthesized input
51616/split-vae/vae/trainer.py
code served (permissive licence) · get_code("2158436f4619f9af")
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

# Create small test tensors
batch_size = 2
feature_dim = 4

# x: the input data, should be in range [-1, 1] or close to it for discretised logistic
x = tf.constant(np.random.uniform(-0.99, 0.99, size=(batch_size, feature_dim)), dtype=tf.float32)

# m: mean/location parameter, same shape as x
m = tf.constant(np.random.randn(batch_size, feature_dim) * 0.1, dtype=tf.float32)

# log_scales: log of scale parameter, same shape as x
log_scales = tf.constant(np.random.randn(batch_size, feature_dim) * 0.5, dtype=tf.float32)

ARGS = {
    "x": x,
    "m": m,
    "log_scales": log_scales
}
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
kl_divergence Ran
ran on a synthesized input
51616/split-vae/vae/trainer.py
code served (permissive licence) · get_code("978b2197d45a49c5")
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 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 tensorflow as tf
import numpy as np

# Create small test tensors for z_mean and z_sig
# Batch size: 2, latent dimension: 4
z_mean = tf.constant(np.random.randn(2, 4).astype(np.float32))
z_sig = tf.constant(np.abs(np.random.randn(2, 4)).astype(np.float32) + 0.1)  # Ensure positive values

ARGS = {
    "z_mean": z_mean,
    "z_sig": z_sig,
}
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 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
kl_divergence Ran
ran on a synthesized input
51616/split-vae/spair/trainer.py
code served (permissive licence) · get_code("06d6e95f7b14b7dd")
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 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 tensorflow as tf
import numpy as np

# Set seed for reproducibility
tf.random.set_seed(0)

# Test case 1: 2D tensors (batch_size=2, latent_dim=4)
z_mean_2d = tf.constant(np.random.randn(2, 4).astype(np.float32))
z_sig_2d = tf.constant(np.abs(np.random.randn(2, 4)).astype(np.float32) + 0.1)

# Test case 2: 4D tensors (batch_size=2, height=3, width=3, channels=4)
z_mean_4d = tf.constant(np.random.randn(2, 3, 3, 4).astype(np.float32))
z_sig_4d = tf.constant(np.abs(np.random.randn(2, 3, 3, 4)).astype(np.float32) + 0.1)

# Use 2D case as default
ARGS = {
    'z_mean': z_mean_2d,
    'z_sig': z_sig_2d
}
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 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
kl_divergence_two_gauss Ran
ran on a synthesized input
51616/split-vae/vae/trainer.py
code served (permissive licence) · get_code("dfdbb3b5c5bd51e4")
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 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 tensorflow as tf
import numpy as np

# Create small tensors for KL divergence computation
# The function computes KL divergence between two Gaussians
# It expects means, sigmas with shape (batch_size, dim)

batch_size = 2
dim = 4

# First Gaussian parameters
mean1 = tf.constant(np.random.randn(batch_size, dim).astype(np.float32))
sig1 = tf.constant(np.abs(np.random.randn(batch_size, dim)).astype(np.float32) + 0.1)

# Second Gaussian parameters
mean2 = tf.constant(np.random.randn(batch_size, dim).astype(np.float32))
sig2 = tf.constant(np.abs(np.random.randn(batch_size, dim)).astype(np.float32) + 0.1)

ARGS = {
    "mean1": mean1,
    "sig1": sig1,
    "mean2": mean2,
    "sig2": sig2
}
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 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
kl_divergence_two_gauss Ran
ran on a synthesized input
51616/split-vae/spair/trainer.py
code served (permissive licence) · get_code("e1b9bac44422dfff")
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 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 tensorflow as tf
import numpy as np

# Set seed for reproducibility
tf.random.set_seed(0)
np.random.seed(0)

# Create small tensor arguments with distinct dimensions
# Batch size: 2, spatial dims: 4x3x5 (distinct sizes for probing)
batch_size = 2
dim1 = 4
dim2 = 3
dim3 = 5

# mean1: shape [batch_size, dim1, dim2, dim3]
mean1 = tf.constant(np.random.randn(batch_size, dim1, dim2, dim3), dtype=tf.float32)

# sig1: shape [batch_size, dim1, dim2, dim3] - standard deviation, keep positive
sig1 = tf.constant(np.abs(np.random.randn(batch_size, dim1, dim2, dim3)) + 0.1, dtype=tf.float32)

# mean2: shape [batch_size, dim1, dim2, dim3]
mean2 = tf.constant(np.random.randn(batch_size, dim1, dim2, dim3), dtype=tf.float32)

# sig2: shape [batch_size, dim1, dim2, dim3] - standard deviation, keep positive
sig2 = tf.constant(np.abs(np.random.randn(batch_size, dim1, dim2, dim3)) + 0.1, dtype=tf.float32)

ARGS = {
    "mean1": mean1,
    "sig1": sig1,
    "mean2": mean2,
    "sig2": sig2,
}
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 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
random_crop Ran
ran on a synthesized input
51616/split-vae/spair/data.py
code served (permissive licence) · get_code("f83c05a84860edf5")
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 3, dtype family float, shape (4, 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 tensorflow as tf
import numpy as np

# Create a sample input image tensor
# TensorFlow expects [height, width, channels] format
# Using shape [8, 8, 3] for a small 8x8 RGB image
input_image = tf.constant(
    np.random.rand(8, 8, 3).astype(np.float32)
)

# Define crop size
crop_size = 4

# Construct the arguments dictionary
ARGS = {
    "input_image": input_image,
    "crop_size": crop_size,
}
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 3, dtype family float, shape (4, 4, 3) — shape (4, 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
reconstruction_bbox Ran
ran on a synthesized input
51616/split-vae/spair/visualizer.py
code served (permissive licence) · get_code("42bf7a6e7b60c60c")
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 module. 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 tensorflow as tf
import numpy as np

class DummyModel(tf.Module):
    def __init__(self):
        super().__init__()
    
    def __call__(self, x_test):
        batch_size = tf.shape(x_test)[0]
        h, w, c = 32, 24, 3
        
        x_recon = tf.random.uniform([batch_size, h, w, c], 0, 1)
        z_what = tf.random.normal([batch_size, 4, 4, 8])
        z_what_mean = tf.random.normal([batch_size, 4, 4, 8])
        z_what_sigma = tf.random.uniform([batch_size, 4, 4, 8], 0.1, 1)
        z_where = tf.random.normal([batch_size, 4, 4, 4])
        z_where_mean = tf.random.normal([batch_size, 4, 4, 4])
        z_where_sigma = tf.random.uniform([batch_size, 4, 4, 4], 0.1, 1)
        z_depth = tf.random.normal([batch_size, 4, 4, 1])
        z_depth_mean = tf.random.normal([batch_size, 4, 4, 1])
        z_depth_sigma = tf.random.uniform([batch_size, 4, 4, 1], 0.1, 1)
        z_pres = tf.random.uniform([batch_size, 16, 1], 0, 1)
        z_pres_logits = tf.random.normal([batch_size, 16, 1])
        z_pres_pre_sigmoid = tf.random.normal([batch_size, 16, 1])
        all_glimpses = tf.random.uniform([batch_size, 4, 4, 8, 8, 3], 0, 1)
        obj_recon_unnorm = tf.random.uniform([batch_size, 4, 4, 8, 8, 3], 0, 1)
        obj_recon_alpha = tf.random.uniform([batch_size, 4, 4, 8, 8, 1], 0, 1)
        obj_full_recon_unnorm = tf.random.uniform([batch_size, h, w, 3], 0, 1)
        obj_bbox_mask = tf.random.uniform([batch_size, 16, 4], 0, 1)
        
        return (x_recon, z_what, z_what_mean, z_what_sigma, z_where, z_where_mean, 
                z_where_sigma, z_depth, z_depth_mean, z_depth_sigma, z_pres, 
                z_pres_logits, z_pres_pre_sigmoid, all_glimpses, obj_recon_unnorm, 
                obj_recon_alpha, obj_full_recon_unnorm, obj_bbox_mask)


def create_test_dataset():
    batch_size = 2
    h, w, c = 32, 24, 3
    
    images = tf.random.uniform([batch_size, h, w, c], 0, 1)
    labels = tf.random.uniform([batch_size], 0, 10, dtype=tf.int32)
    
    dataset = tf.data.Dataset.from_tensor_slices((images, labels))
    dataset = dataset.batch(batch_size)
    
    return dataset


model = DummyModel()
test_dataset = create_test_dataset()
filename = "_test"
filepath = "/tmp/"
label = True
n = 2

ARGS = {
    "model": model,
    "test_dataset": test_dataset,
    "filename": filename,
    "filepath": filepath,
    "label": label,
    "n": n
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type module — a module output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a module output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_confirm_token Not yet run
unverified (harvested, no recorded run)
51616/split-vae/vae/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from med-air/FedBN, is pointer only (licence: NONE) · get_code("135b3dc835ffe6ad")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from med-air/FedBN, 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. No invocation recorded.

get_model Not yet run
unverified (harvested, no recorded run)
51616/split-vae/spair/spair.py
code served (permissive licence) · get_code("5438d8d9576107a0")
How it ran

no invocation recorded.

glimpses_reconstruction_test Not yet run
unverified (harvested, no recorded run)
51616/split-vae/spair/visualizer.py
code served (permissive licence) · get_code("98bc49e1bae2904c")
How it ran

no invocation recorded.

reconstruction_test Not yet run
unverified (harvested, no recorded run)
51616/split-vae/spair/visualizer.py
code served (permissive licence) · get_code("8d20e3820ec4c18d")
How it ran

no invocation recorded.

reconstruction_test_lg_vae Not yet run
unverified (harvested, no recorded run)
51616/split-vae/vae/visualizer.py
code served (permissive licence) · get_code("32c56d8020d4ae89")
How it ran

no invocation recorded.

style_transfer_celeba Not yet run
unverified (harvested, no recorded run)
51616/split-vae/vae/visualizer.py
code served (permissive licence) · get_code("fa6264be29eb01f9")
How it ran

no invocation recorded.

style_transfer_test Not yet run
unverified (harvested, no recorded run)
51616/split-vae/vae/visualizer.py
code served (permissive licence) · get_code("9f8ebb6945e3f4ee")
How it ran

no invocation recorded.

Also present in other papers' code

Of 18 samples lifted from this paper's linked repositories, 18 carry a function identity and 0 do not. Of the 18 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 1 the same month, 1 later) and 17 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
get_confirm_token 38 other papers
18 dated earlier, 1 the same month, 19 later
1409.8230 (2014-09)
RENOIR - A Dataset for Real Low-Light Image Noise Reduction
1503.03832 (2015-03)
FaceNet: A Unified Embedding for Face Recognition and Clustering
1506.03134 (2015-06)
Pointer Networks
and 35 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

Visual data can be understood at different levels of granularity, where global features correspond to semantic-level information and local features correspond to texture patterns. In this work, we propose a framework, called SPLIT, which allows us to disentangle local and global information into two separate sets of latent variables within the variational autoencoder (VAE) framework. Our framework adds generative assumption to the VAE by requiring a subset of the latent variables to generate an auxiliary set of observable data. This additional generative assumption primes the latent variables to local information and encourages the other latent variables to represent global information. We examine three different flavours of VAEs with different generative assumptions. We show that the framework can effectively disentangle local and global information within these models leads to improved representation, with better clustering and unsupervised object detection benchmarks. Finally, we establish connections between SPLIT and recent research in cognitive neuroscience regarding the disentanglement in human visual perception. The code for our experiments is at https://github.com/51616/split-vae .

For agents

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

get_harvested_code_for_paper("2001.08957")
get_code_for_paper("2001.08957")
have("2001.08957")

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

curl https://syntology.ai/api/ran/2001.08957.json

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

[![Syntology run record](https://syntology.ai/api/ran/2001.08957.svg)](https://syntology.ai/paper/2001.08957)

Connect an agent — have() is free.