SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2412.02527 · NeurIPS · 2024

The Multimodal Universe: Enabling Large-Scale Machine Learning with 100 TB of Astronomical Scientific Data The Multimodal Universe Collaboration

Miles Cranmer, Shirley Ho, François Lanusse, Peter Melchior, Mike Walmsley, Micah Bowles, Connor Stone, Helen Qu, Jeff Shen, Lucas Meyer, Jeroen Audenaert, David Chemaly, and 17 more

arXiv · PDF · Open in the Atlas

Code that ran

6 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 0 have no contract verdict; 6 failures of our instrument; 2 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/2412.02527.json · badge: /api/ran/2412.02527.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
multimodaluniverse/multimodaluniverse canonical (classifier label)
classifier basis: census
5 ran · 2 no recorded run
MultimodalUniverse/MultimodalUniverse canonical (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
FunctionStatusWhere it lives
conv1x1 Ran
ran; our contract draft was wrong, not the code
multimodaluniverse/multimodaluniverse/experimental_benchmark/galaxy_properties/modules.py
code served (permissive licence) · get_code("ff88353af1a886e1")
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 Conv1d. 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 torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "in_planes": 3,
    "out_planes": 64,
    "stride": 1,
}
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 Conv1d — a Conv1d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Conv1d 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
conv3x3 Ran
ran; our contract draft was wrong, not the code
multimodaluniverse/multimodaluniverse/experimental_benchmark/galaxy_properties/modules.py
code served (permissive licence) · get_code("db485754880e0747")
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 Conv1d. 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 torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'in_planes': 3,
    'out_planes': 64,
    'stride': 1,
    'groups': 1,
    'dilation': 1
}
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 Conv1d — a Conv1d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Conv1d 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
create_attention_mask Ran
ran; our contract draft was wrong, not the code
multimodaluniverse/multimodaluniverse/baselines/plasticc/preprocess.py
code served (permissive licence) · get_code("9fab1dddd51579d7")
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

# Create a minimal example dict that satisfies the function's requirements
example = {
    'transposed_target': np.zeros((2, 4)),  # Shape (batch=2, features=4)
    'transposed_times_wv': np.array([[1, 0, 1, 0]]),  # Shape (1, 4) - will be indexed with [0]
}

ARGS = {
    'example': example
}
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 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
extract_acknowledgements Ran
ran; our contract draft was wrong, not the code
MultimodalUniverse/MultimodalUniverse/scripts/info.py
code served (permissive licence) · get_code("f0a3007609b0f577")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned str. 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.

# The function extracts acknowledgements from a citation string
# It checks if the citation starts with specific prefixes and processes the text accordingly

# Test case 1: Citation with acknowledgements starting with '% % ACKNOWLEDGEMENTS'
citation_with_ack_double = """% % ACKNOWLEDGEMENTS
% This work was supported by XYZ Foundation
% We thank ABC for their contributions
% CITATION
Some other citation text here"""

# Test case 2: Citation with acknowledgements starting with '% ACKNOWLEDGEMENTS'
citation_with_ack_single = """% ACKNOWLEDGEMENTS
% This work was supported by XYZ Foundation
% We thank ABC for their contributions
% CITATION
Some other citation text here"""

# Test case 3: Citation without acknowledgements
citation_without_ack = """Some regular citation text
that does not start with the acknowledgements prefix"""

# Use the first test case as the default argument
ARGS = {
    'citation': citation_with_ack_double
}
Setup block drafted by
a language model, Claude Sonnet 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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
normalize_all Ran
ran; our contract draft was wrong, not the code
multimodaluniverse/multimodaluniverse/baselines/plasticc/preprocess.py
code served (permissive licence) · get_code("59c3d5e2ae6e0a8d")
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

# Create a simple example dictionary with a field containing numeric values
example = {
    'field_name': np.array([1.0, 2.0, 3.0, 4.0, 5.0])
}

# Define ARGS mapping all parameters to values
ARGS = {
    'example': example,
    'field_name': 'field_name'
}
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 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
normalize_by_channel Ran
ran; our contract draft was wrong, not the code
multimodaluniverse/multimodaluniverse/baselines/plasticc/preprocess.py
code served (permissive licence) · get_code("5f393d809299394f")
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

# Create a small example dict with a field containing data
# The function accesses example[field_name] and iterates through rows
# It performs min/max normalization on each row, handling zero values

example = {
    'data': np.array([
        [0, 1, 2, 3],
        [0, 5, 10, 15],
        [0, 0, 4, 8]
    ], dtype=np.float32)
}

field_name = 'data'

ARGS = {
    'example': example,
    'field_name': field_name
}
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 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
get_config Not yet run
unverified (harvested, no recorded run)
multimodaluniverse/multimodaluniverse/baselines/plasticc/plasticc.py
code served (permissive licence) · get_code("dace7f18ecd6d09e")
How it ran

no invocation recorded.

gz10_cli Not yet run
unverified (harvested, no recorded run)
multimodaluniverse/multimodaluniverse/experimental_benchmark/galaxy10_decals/trainer.py
code served (permissive licence) · get_code("33252f39c1ee9922")
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 8 distinct functions with an identity, 4 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 2 later) and 4 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
conv1x1 2 other papers
1 dated earlier, 1 later
2408.05178 (2024-08)
ECG-FM: An Open Electrocardiogram Foundation Model
2602.04266 (2026-02)
Aortic Valve Disease Screening from PPG via Physiology-Guided Self-Supervised Learning
conv3x3 1 other paper
1 later
2602.04266 (2026-02)
Aortic Valve Disease Screening from PPG via Physiology-Guided Self-Supervised Learning
normalize_all 1 other paper
1 dated earlier
2402.03325 (2024-02)
Connect Later: Improving Fine-tuning for Robustness with Targeted Augmentations
normalize_by_channel 1 other paper
1 dated earlier
2402.03325 (2024-02)
Connect Later: Improving Fine-tuning for Robustness with Targeted Augmentations

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

We present the Multimodal Universe, a large-scale multimodal dataset of scientific astronomical data, compiled specifically to facilitate machine learning research. Overall, the Multimodal Universe contains hundreds of millions of astronomical observations, constituting 100 TB of multi-channel and hyper-spectral images, spectra, multivariate time series, as well as a wide variety of associated scientific measurements and "metadata". In addition, we include a range of benchmark tasks representative of standard practices for machine learning methods in astrophysics. This massive dataset will enable the development of large multi-modal models specifically targeted towards scientific applications. All codes used to compile the Multimodal Universe and a description of how to access the data is available at https://github.com/MultimodalUniverse/MultimodalUniverse Pretraining Model Top-1 Accuracy EfficientNetB0 80.9 ±0.1 % No pretraining ConvNext-nano [144] 75.6 ±1.8 % ResNet18 73.9 ±0.9 % DenseNet121 73.5 ±2.4 % Galazy Zoo ConvNext-nano [144] 89.3 ±0.1 % ImageNet-12k ConvNext-nano [144] 83.9 ±0.3 % Table 3: Top-1 Accuracy on the Galaxy10 DECaLS morphology classification dataset. Morphology Classification Galaxy shape or morphology is a first-order tracer of the history of a galaxy. As such, classifying images of galaxies based on their apparent morphologies and visual features is a common task. Labeling these features is usually done by visual inspection from experts or using citizen science, such as [10,145]. Here, we use the Galaxy10 DECaLS catalog of human-labeled annotations spanning ten broad morphological classes paired with the Multimodal Universe's RGB-converted images from the Legacy Surveys and report the top-1 accuracy of supervised baselines on the held-out test set in Table 3. We include state of the art results from the literature from [144] as a point of reference and compare different common architectures for this task (EfficientNetB0[136], ResNet18[66], and DenseNet121[75]). Physical Property Inference Astrophysicists often seek to predict fundamental properties of galaxies from observational data. Accurate predictions of these properties provide key insights into galaxy formation and evolution. Here, we consider the following properties: (AEIMCINN) of the Spanish Ministry of Science and Innovation under the grants "Galaxy Evolution with Artificial Intelligence" with reference PGC2018-100852-A-I00 and "BASALT" with reference PID2021-126838NB-I00. For data specific acknowledgments see each individual dataset in Appendix A. The contribution categories listed below are broadly indicative of the ways in which authors contributed to the project, although we note these categories are not exhaustive.

For agents

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

get_harvested_code_for_paper("2412.02527")
get_code_for_paper("2412.02527")
have("2412.02527")

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

curl https://syntology.ai/api/ran/2412.02527.json

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

[![Syntology run record](https://syntology.ai/api/ran/2412.02527.svg)](https://syntology.ai/paper/2412.02527)

Connect an agent — have() is free.