SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2306.09035 · 2023

Improving Explainability of Disentangled Representations using Multipath-Attribution Mappings

arXiv · PDF · Open in the Atlas

Code that ran

5 of 8 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 5 have no contract verdict; 0 failures of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2306.09035.json · badge: /api/ran/2306.09035.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
iml-dkfz/m-pax_lib canonical (classifier label)
classifier basis: census
5 ran · 3 no recorded run
FunctionStatusWhere it lives
arr_im_convert Ran
ran on a synthesized input
iml-dkfz/m-pax_lib/src/evaluation/vis_helper.py
code served (permissive licence) · get_code("b2e93f279afbf340")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 3, dtype family int, shape (4, 3, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
from PIL import Image

# Create a small test image array
# The function converts an image array using PIL
# We need a numpy array that represents an image with valid pixel values

# Create a simple 4x3 RGB image (height=4, width=3, channels=3)
# Pixel values should be in range [0, 255] for uint8
arr = np.random.randint(0, 256, size=(4, 3, 3), dtype=np.uint8)

ARGS = {
    'arr': arr,
    'convert': 'RGBA'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (4, 3, 4) — shape (4, 3, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
concatenate_pad Ran
ran on a synthesized input
iml-dkfz/m-pax_lib/src/evaluation/vis_helper.py
code served (permissive licence) · get_code("ed74ccb76b1faf95")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 3, dtype family int, shape (22, 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 numpy as np

# Create small test arrays with distinct dimensions
# These will represent "gif arrays" that need padding between them
array1 = np.random.randint(0, 256, size=(4, 3, 3), dtype=np.uint8)  # 4x3x3
array2 = np.random.randint(0, 256, size=(4, 3, 3), dtype=np.uint8)  # 4x3x3
array3 = np.random.randint(0, 256, size=(4, 3, 3), dtype=np.uint8)  # 4x3x3

arrays = [array1, array2, array3]
pad_size = 2
pad_values = 128
axis = 0

ARGS = {
    "arrays": arrays,
    "pad_size": pad_size,
    "pad_values": pad_values,
    "axis": axis,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (22, 3, 3) — shape (22, 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
make_grid_img Ran
ran on a synthesized input
iml-dkfz/m-pax_lib/src/evaluation/vis_helper.py
code served (permissive licence) · get_code("7faac0f48cc9f483")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 3, dtype family int, shape (8, 16, 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 torchvision.utils import make_grid

torch.manual_seed(0)

# Create a small 4D tensor (B x C x H x W)
# B=2 (batch size), C=3 (RGB channels), H=4 (height), W=5 (width)
tensor = torch.rand(2, 3, 4, 5)

# Normalize to [0, 1] range as expected by make_grid
tensor = (tensor - tensor.min()) / (tensor.max() - tensor.min() + 1e-8)

ARGS = {
    "tensor": tensor,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (8, 16, 3) — shape (8, 16, 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
make_weights_for_balanced_classes Ran
ran on a synthesized input
iml-dkfz/m-pax_lib/src/datamodules/isic_datamodule.py
code served (permissive licence) · get_code("c133dbaa8240dd1d")
How it ran

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

# Construct valid arguments for make_weights_for_balanced_classes
# The function expects:
# - images: list of [image, label] pairs
# - nclasses: int, number of classes

# Create a simple dataset with imbalanced classes
# We'll have 3 samples of class 0, 1 sample of class 1, 2 samples of class 2
images = [
    ["img1", 0],
    ["img2", 0],
    ["img3", 0],
    ["img4", 1],
    ["img5", 2],
    ["img6", 2],
]

nclasses = 3

ARGS = {
    "images": images,
    "nclasses": nclasses,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list 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
make_weights_for_balanced_classes Ran
ran on a synthesized input
iml-dkfz/m-pax_lib/src/datamodules/oct_datamodule.py
code served (permissive licence) · get_code("4a4666cfd3de39ba")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, 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 expects:
# - images: a list of tuples where each tuple is (image_data, label)
# - nclasses: an integer representing the number of classes

# Create a simple list of image-label pairs
# Each item should be a tuple of (image, label)
images = [
    (torch.randn(3, 4), 0),  # image with label 0
    (torch.randn(3, 4), 1),  # image with label 1
    (torch.randn(3, 4), 0),  # image with label 0
    (torch.randn(3, 4), 2),  # image with label 2
    (torch.randn(3, 4), 1),  # image with label 1
    (torch.randn(3, 4), 2),  # image with label 2
]

nclasses = 3

ARGS = {
    "images": images,
    "nclasses": nclasses,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list 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
main Not yet run
unverified (harvested, no recorded run)
iml-dkfz/m-pax_lib/run_eval.py
code served (permissive licence) · get_code("1fbb2f377c3568f1")
How it ran

no invocation recorded.

main Not yet run
unverified (harvested, no recorded run)
iml-dkfz/m-pax_lib/run_head.py
code served (permissive licence) · get_code("9f547bdc38fad1b9")
How it ran

no invocation recorded.

main Not yet run
unverified (harvested, no recorded run)
iml-dkfz/m-pax_lib/run_tcvae.py
code served (permissive licence) · get_code("00510feab827eb87")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 7 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.

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

Explainable AI aims to render model behavior understandable by humans, which can be seen as an intermediate step in extracting causal relations from correlative patterns. Due to the high risk of possible fatal decisions in image-based clinical diagnostics, it is necessary to integrate explainable AI into these safety-critical systems. Current explanatory methods typically assign attribution scores to pixel regions in the input image, indicating their importance for a model's decision. However, they fall short when explaining why a visual feature is used. We propose a framework that utilizes interpretable disentangled representations for downstream-task prediction. Through visualizing the disentangled representations, we enable experts to investigate possible causation effects by leveraging their domain knowledge. Additionally, we deploy a multi-path attribution mapping for enriching and validating explanations. We demonstrate the effectiveness of our approach on a synthetic benchmark suite and two medical datasets. We show that the framework not only acts as a catalyst for causal relation extraction but also enhances model robustness by enabling shortcut detection without the need for testing under distribution shifts.

For agents

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

get_harvested_code_for_paper("2306.09035")
get_code_for_paper("2306.09035")
have("2306.09035")

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

curl https://syntology.ai/api/ran/2306.09035.json

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

[![Syntology run record](https://syntology.ai/api/ran/2306.09035.svg)](https://syntology.ai/paper/2306.09035)

Connect an agent — have() is free.