SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2107.08221 · 2021

Visual Representation Learning Does Not Generalize Strongly Within the Same Domain

arXiv · PDF · Open in the Atlas

Code that ran

1 of 2 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 1 has 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/2107.08221.json · badge: /api/ran/2107.08221.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
bethgelab/InDomainGeneralizationBenchmark canonical (classifier label)
classifier basis: census
1 ran · 1 no recorded run
FunctionStatusWhere it lives
evaluate_model Ran
ran on a synthesized input
bethgelab/InDomainGeneralizationBenchmark/src/lablet_generalization_benchmark/evaluate_model.py
code served (permissive licence) · get_code("84a0f2cd3abb6e6f")
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 numpy as np
import torch
from torch import nn
from torch.utils.data import Dataset, DataLoader

torch.manual_seed(0)

# Create a simple model function
def simple_model_fn(image_batch):
    """Mock model that takes images and returns predictions"""
    # image_batch shape: (batch_size, ...)
    # Return predictions with shape (batch_size, num_factors)
    batch_size = image_batch.shape[0]
    num_factors = 3
    # Simple deterministic prediction based on input
    return np.random.randn(batch_size, num_factors).astype(np.float32)

# Create a mock dataset class
class MockDataset(Dataset):
    def __init__(self, num_samples=8, num_factors=3):
        self.num_samples = num_samples
        self.num_factors = num_factors
        # Normalized targets in range [0, 1] for R² calculation
        self.normalized_targets = np.random.rand(num_samples, num_factors).astype(np.float32)
        self._factor_names = ['factor_0', 'factor_1', 'factor_2']
    
    def __len__(self):
        return self.num_samples
    
    def __getitem__(self, idx):
        # Return (image, target) as tensors
        image = torch.randn(4, 5).float()  # 4x5 image
        target = torch.from_numpy(self.normalized_targets[idx]).float()
        return image, target

# Create a mock dataloader
dataset = MockDataset(num_samples=8, num_factors=3)
dataloader = DataLoader(dataset, batch_size=2, shuffle=False)

ARGS = {
    'model_fn': simple_model_fn,
    'dataloader': dataloader,
}
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 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
load_dataset Not yet run
unverified (harvested, no recorded run)
bethgelab/InDomainGeneralizationBenchmark/src/lablet_generalization_benchmark/load_dataset.py
code served (permissive licence) · get_code("07e887249f407dee")
How it ran

no invocation recorded.

Also present in other papers' code

Of 2 samples lifted from this paper's linked repositories, 2 carry a function identity and 0 do not. Of the 2 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 2 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

An important component for generalization in machine learning is to uncover underlying latent factors of variation as well as the mechanism through which each factor acts in the world. In this paper, we test whether 17 unsupervised, weakly supervised, and fully supervised representation learning approaches correctly infer the generative factors of variation in simple datasets (dSprites, Shapes3D, MPI3D) from controlled environments, and on our contributed CelebGlow dataset. In contrast to prior robustness work that introduces novel factors of variation during test time, such as blur or other (un)structured noise, we here recompose, interpolate, or extrapolate only existing factors of variation from the training data set (e.g., small and medium-sized objects during training and large objects during testing). Models that learn the correct mechanism should be able to generalize to this benchmark. In total, we train and test 2000+ models and observe that all of them struggle to learn the underlying mechanism regardless of supervision signal and architectural bias. Moreover, the generalization capabilities of all tested models drop significantly as we move from artificial datasets towards more realistic real-world datasets. Despite their inability to identify the correct mechanism, the models are quite modular as their ability to infer other in-distribution factors remains fairly stable, providing only a single factor is out-of-distribution. These results point to an important yet understudied problem of learning mechanistic models of observations that can facilitate generalization.

For agents

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

get_harvested_code_for_paper("2107.08221")
get_code_for_paper("2107.08221")
have("2107.08221")

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

curl https://syntology.ai/api/ran/2107.08221.json

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

[![Syntology run record](https://syntology.ai/api/ran/2107.08221.svg)](https://syntology.ai/paper/2107.08221)

Connect an agent — have() is free.