SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2202.13341 · 2022

Overlooked Implications of the Reconstruction Loss for VAE Disentanglement

arXiv · PDF · Open in the Atlas

Code that ran

2 of 3 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 2 have no contract verdict; 0 failures of our instrument; 1 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/2202.13341.json · badge: /api/ran/2202.13341.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
nmichlo/disent canonical (classifier label)
classifier basis not recorded on the link
2 ran · 1 no recorded run
FunctionStatusWhere it lives
compute_axis_score Ran
ran on a synthesized input
nmichlo/disent/disent/metrics/_factored_components.py
code served (permissive licence) · get_code("a7e61db586268ded")
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)

# zs_traversal should be a 2D tensor where:
# - dim=0 is the batch dimension
# - dim=1 is the latent dimension
# The function computes variance over dim=0 and returns a 1D tensor of shape (latent_dim,)
zs_traversal = torch.randn(2, 4)

ARGS = {
    "zs_traversal": zs_traversal,
    "use_std": True,
    "top_2": False,
    "norm": True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 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
compute_flatness Ran
ran on a synthesized input
nmichlo/disent/disent/metrics/_flatness.py
code served (permissive licence) · get_code("353304daf1a3fbec")
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)

# Analyze the function:
# - compute_flatness takes: widths, lengths, factor_sizes
# - widths and lengths are tensors passed to filter_inactive_factors
# - filter_inactive_factors expects a tensor and factor_sizes list
# - factor_sizes is converted to a tensor and must have all elements >= 1
# - active_factors filters based on factor_sizes - 1 (nonzero means factor_size > 1)
# - After filtering, widths and lengths are indexed by active_factors
# - Assertions check: widths >= 0, lengths >= 0, widths==0 iff lengths==0
# - widths and lengths are compared element-wise and have .mean() called

# Create small tensors with distinct dimensions
# factor_sizes determines which factors are "active" (size > 1)
# Let's say we have 4 factors
widths = torch.tensor([2.0, 4.0, 0.0, 3.0])  # 4 factors
lengths = torch.tensor([5.0, 8.0, 0.0, 6.0])  # 4 factors, matching widths pattern
factor_sizes = [2, 3, 1, 2]  # size 1 means inactive; others are active

# Verify the logic:
# factor_sizes - 1 = [1, 2, 0, 1]
# nonzero gives indices where this is nonzero: [0, 1, 3]
# So after filtering: widths[0,1,3] = [2.0, 4.0, 3.0], lengths[0,1,3] = [5.0, 8.0, 6.0]
# Check: widths >= 0? Yes. lengths >= 0? Yes.
# Check: widths == 0 iff lengths == 0? Yes (no zeros in filtered tensors)
# All conditions satisfied

ARGS = {
    "widths": widths,
    "lengths": lengths,
    "factor_sizes": factor_sizes,
}
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
filter_inactive_factors Not yet run
unverified (harvested, no recorded run)
nmichlo/disent/disent/metrics/_flatness.py
code served (permissive licence) · get_code("4ae1225700e757f0")
How it ran

no invocation recorded.

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 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 3 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

Learning disentangled representations with variational autoencoders (VAEs) is often attributed to the regularisation component of the loss. In this work, we highlight the interaction between data and the reconstruction term of the loss as the main contributor to disentanglement in VAEs. We show that standard benchmark datasets have unintended correlations between their subjective ground-truth factors and perceived axes in the data according to typical VAE reconstruction losses. Our work exploits this relationship to provide a theory for what constitutes an adversarial dataset under a given reconstruction loss. We verify this by constructing an example dataset that prevents disentanglement in state-of-the-art frameworks while maintaining human-intuitive ground-truth factors. Finally, we re-enable disentanglement by designing an example reconstruction loss that is once again able to perceive the ground-truth factors. Our findings demonstrate the subjective nature of disentanglement and the importance of considering the interaction between the ground-truth factors, data and notably, the reconstruction loss, which is under-recognised in the literature.

For agents

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

get_harvested_code_for_paper("2202.13341")
get_code_for_paper("2202.13341")
have("2202.13341")

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

curl https://syntology.ai/api/ran/2202.13341.json

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

[![Syntology run record](https://syntology.ai/api/ran/2202.13341.svg)](https://syntology.ai/paper/2202.13341)

Connect an agent — have() is free.