SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2302.12254 · ICML · 2023

Change is Hard: A Closer Look at Subpopulation Shift

Yuzhe Yang, Dina Katabi, Marzyeh Ghassemi, Haoran Zhang

arXiv · PDF · Open in the Atlas

Code that ran

3 of 7 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; 1 honoured the contract we drafted, 1 violated it and 0 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/2302.12254.json · badge: /api/ran/2302.12254.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
yyzharry/subpopbench canonical (classifier label)
classifier basis: classifier_2025
2 ran · 2 no recorded run
skyve2012/DBA extension (classifier label)
classifier basis: classifier_2025
1 ran · 1 no recorded run
copy not recorded community
not in the paper's declared repositories
0 ran · 1 no recorded run
FunctionStatusWhere it lives
binary_metrics Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
skyve2012/DBA/subpopbench/utils/eval_helper.py
code served (permissive licence) · get_code("e67d7e380e66761c")
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) 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

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

# Create simple binary classification targets and predictions
targets = np.array([0, 1, 0, 1, 1, 0, 1, 0])
preds = np.array([0, 1, 0, 0, 1, 0, 1, 1])

# Create sample weights
sample_weights = np.array([1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0])

# Define ARGS dictionary with all parameters
ARGS = {
    'targets': targets,
    'preds': preds,
    'label_set': [0, 1],
    'return_arrays': False,
    'sample_weights': sample_weights
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, sklearn — 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
block_until_running Ran
ran, violated the contract we drafted
yyzharry/subpopbench/subpopbench/command_launchers.py
code served (permissive licence) · get_code("9d4328ab806dde75")
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) 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 subprocess
import time
import sys

def get_slurm_jobs(user):
    out = subprocess.run(['squeue -u ' + user], shell=True, stdout=subprocess.PIPE).stdout.decode(sys.stdout.encoding)
    a = list(filter(lambda x: len(x) > 0, map(lambda x: x.split(), out.split('\n'))))
    queued, running = ([], [])
    for i in a:
        if i[0].isnumeric():
            if i[4].strip() == 'PD':
                queued.append(int(i[0]))
            else:
                running.append(int(i[0]))
    return (queued, running)

def block_until_running(n, user):
    while True:
        queued, running = get_slurm_jobs(user)
        if len(queued) + len(running) < n:
            time.sleep(0.2)
            return True
        else:
            time.sleep(10)

ARGS = {
    'n': 1,
    'user': 'testuser'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 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
get_slurm_jobs Ran
ran, honoured the contract we drafted
yyzharry/subpopbench/subpopbench/command_launchers.py
code served (permissive licence) · get_code("f2059a864ab0dc16")
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) and returned tuple, ndim 2, 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 subprocess
import sys

def get_slurm_jobs(user):
    out = subprocess.run(['squeue -u ' + user], shell=True, stdout=subprocess.PIPE).stdout.decode(sys.stdout.encoding)
    a = list(filter(lambda x: len(x) > 0, map(lambda x: x.split(), out.split('\n'))))
    queued, running = ([], [])
    for i in a:
        if i[0].isnumeric():
            if i[4].strip() == 'PD':
                queued.append(int(i[0]))
            else:
                running.append(int(i[0]))
    return (queued, running)

ARGS = {
    'user': 'testuser'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 2, 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
get_algorithm_class Not yet run
unverified (harvested, no recorded run)
this paper's copy was not recorded; identical code first harvested from facebookresearch/domainbed
pointer only · get_code("b0bc80b1655a6802")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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_dataset_class Not yet run
unverified (harvested, no recorded run)
yyzharry/subpopbench/subpopbench/dataset/datasets.py
code served (permissive licence) · get_code("5b0092cb7767b24c")
How it ran

no invocation recorded.

num_environments Not yet run
unverified (harvested, no recorded run)
yyzharry/subpopbench/subpopbench/dataset/datasets.py
code served (permissive licence) · get_code("38ace74bf6716b77")
How it ran

no invocation recorded.

predict_on_set Not yet run
unverified (harvested, no recorded run)
skyve2012/DBA/subpopbench/utils/eval_helper.py
code served (permissive licence) · get_code("aa7d2110b04bb70f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 5 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 1 the same month, 5 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.

FunctionAlso present inEarliest-datedOthers named
get_algorithm_class 39 other papers
19 dated earlier, 3 the same month, 17 later
1907.02893 (2019-07)
Invariant Risk Minimization
2007.01434 (2020-07)
In Search of Lost Domain Generalization
2007.02931 (2020-07)
Adaptive Risk Minimization: Learning to Adapt to Domain Shift
and 36 more not named here
num_environments 21 other papers
10 dated earlier, 11 later
1907.02893 (2019-07)
Invariant Risk Minimization
2007.01434 (2020-07)
In Search of Lost Domain Generalization
2106.02266 (2021-06)
SAND-mask: An Enhanced Gradient Masking Strategy for the Discovery of Invariances in Domain Generalization
and 18 more not named here
get_dataset_class 3 other papers
3 later
2403.07241 (2024-03)
Calibrating Multi-modal Representations: A Pursuit of Group Robustness without Annotations
2603.25839 (2026-03)
A Compression Perspective on Simplicity Bias
2606.00082 (2026-06)
Hoeffding Concept Bottleneck Models with Applications to Overhead Images
block_until_running 2 other papers
1 dated earlier, 1 later
2210.10769 (2022-10)
"Why did the Model Fail?": Attributing Model Performance Changes to Distribution Shifts
2401.06091 (2024-01)
A Closer Look at AUROC and AUPRC under Class Imbalance
get_slurm_jobs 2 other papers
1 dated earlier, 1 later
2210.10769 (2022-10)
"Why did the Model Fail?": Attributing Model Performance Changes to Distribution Shifts
2401.06091 (2024-01)
A Closer Look at AUROC and AUPRC under Class Imbalance

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

Machine learning models often perform poorly on subgroups that are underrepresented in the training data. Yet, little is understood on the variation in mechanisms that cause subpopulation shifts, and how algorithms generalize across such diverse shifts at scale. In this work, we provide a fine-grained analysis of subpopulation shift. We first propose a unified framework that dissects and explains common shifts in subgroups. We then establish a comprehensive benchmark of 20 stateof-the-art algorithms evaluated on 12 real-world datasets in vision, language, and healthcare domains. With results obtained from training over 10,000 models, we reveal intriguing observations for future progress in this space. First, existing algorithms only improve subgroup robustness over certain types of shifts but not others. Moreover, while current algorithms rely on group-annotated validation data for model selection, we find that a simple selection criterion based on worst-class accuracy is surprisingly effective even without any group information. Finally, unlike existing works that solely aim to improve worst-group accuracy (WGA), we demonstrate the fundamental tradeoff between WGA and other important metrics, highlighting the need to carefully choose testing metrics.

For agents

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

get_harvested_code_for_paper("2302.12254")
get_code_for_paper("2302.12254")
have("2302.12254")

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

curl https://syntology.ai/api/ran/2302.12254.json

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

[![Syntology run record](https://syntology.ai/api/ran/2302.12254.svg)](https://syntology.ai/paper/2302.12254)

Connect an agent — have() is free.