SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.04075 · 2026

Thermodynamic assessment of machine learning models for solid-state synthesis prediction

Jane Schlesinger, Simon Hjaltason, Nathan Szymanski, Christopher Bartel

arXiv · PDF · Open in the Atlas

Code that ran

17 of 23 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; 2 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 14 failures of our instrument; 6 no recorded run), measured 2026-11-21; 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/2602.04075.json · badge: /api/ran/2602.04075.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
BAMeScience/SynCoTrainMP canonical (classifier label)
classifier basis: census
10 ran · 4 no recorded run
kaist-amsg/Synthesizability-PU-CGCNN canonical (classifier label)
classifier basis: census
5 ran · 2 no recorded run
antoniuk1/SynthNN canonical (classifier label)
classifier basis: census
1 ran · 0 no recorded run
materialsproject/reaction-network canonical (classifier label)
classifier basis: census
1 ran · 0 no recorded run
FunctionStatusWhere it lives
ProbnPred Ran
ran; our contract draft was wrong, not the code
BAMeScience/SynCoTrainMP/syncotrainmp/pu_schnet/pu_learn/schnet_funcs.py
code served (permissive licence) · get_code("e615c78bc31bd209")
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 torch
from torch import nn

torch.manual_seed(0)

# The function expects a score parameter that:
# - Can be passed to nn.Sigmoid()
# - Has a .item() method (so it's a tensor)
# - Can be compared with 0 and 0.5 (scalar or single-element tensor)

score = torch.tensor([0.3])

ARGS = {
    'score': score
}
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 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
collate_pool Ran
ran; our contract draft was wrong, not the code
kaist-amsg/Synthesizability-PU-CGCNN/cgcnn/data_PU_learning.py
pointer only (licence: NONE) · get_code("44a7cad9874db8c5")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, kaist-amsg/Synthesizability-PU-CGCNN when it was checked. 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. A run is recorded.

compute_aggregate_df Ran
ran; our contract draft was wrong, not the code
BAMeScience/SynCoTrainMP/syncotrainmp/pu_alignn_analyze.py
code served (permissive licence) · get_code("2b90b0bbf3ce9e9d")
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 DataFrame. 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 pandas as pd
import numpy as np

# Create a small results DataFrame with material_id, prediction, and target columns
resdf = pd.DataFrame({
    'material_id': ['mat_1', 'mat_1', 'mat_2', 'mat_2', 'mat_3'],
    'prediction': [0.2, 0.3, 0.8, 0.9, 0.5],
    'target': [0, 0, 1, 1, 1]
})

# Create a small properties DataFrame with material_id and a property column
propDF = pd.DataFrame({
    'material_id': ['mat_1', 'mat_2', 'mat_3', 'mat_4'],
    'band_gap': [1.5, 2.3, 1.8, 2.1]
})

# Define the property name to merge on
prop = 'band_gap'

# Define the pseudo-label threshold
pseudo_label_threshold = 0.5

ARGS = {
    'resdf': resdf,
    'propDF': propDF,
    'prop': prop,
    'pseudo_label_threshold': pseudo_label_threshold
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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
current_setup Ran
ran; our contract draft was wrong, not the code
BAMeScience/SynCoTrainMP/syncotrainmp/experiment_setup.py
code served (permissive licence) · get_code("9db5db8d68238ccc")
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.

# Setup code for current_setup function

ARGS = {
    'small_data': False,
    'experiment': 'alignn0',
    'ehull015': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 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
directory_setup Ran
ran; our contract draft was wrong, not the code
BAMeScience/SynCoTrainMP/syncotrainmp/pu_schnet/pu_learn/schnet_funcs.py
code served (permissive licence) · get_code("dfdff3ee29de84d8")
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 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.

import os
import tempfile
from pathlib import Path

# Create temporary directories for testing
temp_base = tempfile.mkdtemp()

# Define arguments for directory_setup
ARGS = {
    'res_dir': os.path.join(temp_base, 'results'),
    'save_dir': os.path.join(temp_base, 'save'),
    'dataPath': os.path.join(temp_base, 'data', 'data.pkl'),
    'bestModelPath': os.path.join(temp_base, 'models', 'best_model.pth'),
    'split_file_name': 'split.npz'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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
get_res_dir Ran
ran; our contract draft was wrong, not the code
BAMeScience/SynCoTrainMP/syncotrainmp/pu_schnet_train.py
code served (permissive licence) · get_code("407c0bba2c01f9a0")
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 tuple, 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 argparse
import os

# Create args using argparse.Namespace
args = argparse.Namespace(
    output_dir="/tmp/output",
    experiment="exp001",
    ehull015=False
)

# Create config dictionary with required keys
config = {
    'schnetDirectory': 'schnet_models'
}

# Create cs (configuration settings) dictionary with required keys
cs = {
    'dataPrefix': 'test_data_'
}

# Define ARGS mapping all parameters
ARGS = {
    'args': args,
    'config': config,
    'cs': cs
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 1, 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_train_val_test_loader Ran
ran on a synthesized input
kaist-amsg/Synthesizability-PU-CGCNN/cgcnn/data_PU_learning.py
pointer only (licence: NONE) · get_code("4a371f59c58bd519")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, kaist-amsg/Synthesizability-PU-CGCNN when it was checked. 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. A run is recorded.

get_weighted_F1 Ran
ran, honoured the contract we drafted
antoniuk1/SynthNN/Figure_data/figure_reproduction_utils.py
code served (permissive licence) · get_code("2d448e18e1e0cbd8")
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 float, 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
import numpy as np

torch.manual_seed(0)

# Analyzing the function: it takes 6 scalar parameters
# TP, FP, TN, FN are used in arithmetic operations (division, addition, multiplication)
# num_positive and num_negative are used in arithmetic operations
# All parameters should be numeric scalars or small tensors that support arithmetic

# Create realistic confusion matrix values
TP = 45  # True Positives
FP = 15  # False Positives
TN = 50  # True Negatives
FN = 10  # False Negatives
num_positive = 55  # Total positives (TP + FN)
num_negative = 65  # Total negatives (TN + FP)

ARGS = {
    "TP": TP,
    "FP": FP,
    "TN": TN,
    "FN": FN,
    "num_positive": num_positive,
    "num_negative": num_negative
}
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 float, 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
is_pareto_efficient Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
materialsproject/reaction-network/src/rxn_network/costs/pareto.py
pointer only (licence: NOASSERTION) · get_code("3dc7d61f74771f89")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, materialsproject/reaction-network (NOASSERTION). 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. A run is recorded.

load_and_prepare_data Ran
ran; our contract draft was wrong, not the code
BAMeScience/SynCoTrainMP/syncotrainmp/data_selection.py
code served (permissive licence) · get_code("c4dade724c0cfd73")
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 DataFrame. 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 pandas as pd
import tempfile
import os

# Create a temporary pickle file with sample data
temp_dir = tempfile.mkdtemp()
data_path = os.path.join(temp_dir, "test_data.pkl")

# Create sample DataFrame
sample_data = {
    'material_id': ['mat_1', 'mat_2', 'mat_3', 'mat_4'],
    'band_gap': [1.5, 2.3, 1.8, 2.1],
    'formation_energy': [-0.5, -0.3, -0.7, -0.4],
}
df_sample = pd.DataFrame(sample_data)
df_sample.to_pickle(data_path)

ARGS = {
    'data_path': data_path,
    'prop': 'band_gap',
    'TARGET': 'formation_energy',
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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_crysdf Ran
ran; our contract draft was wrong, not the code
BAMeScience/SynCoTrainMP/syncotrainmp/pu_schnet_train.py
code served (permissive licence) · get_code("8bd67f0fff816e5d")
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 DataFrame. 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 pandas as pd
import tempfile
import os

# Create a temporary pickle file with mock crystal data
temp_dir = tempfile.mkdtemp()
pickle_path = os.path.join(temp_dir, "crysdf.pkl")

# Create mock crystal DataFrame with target data
mock_data = {
    'structure_id': ['crystal_1', 'crystal_2'],
    'synthesizability': [np.array([0.5, 0.3, 0.2]), np.array([0.1, 0.4, 0.5])],
}
mock_df = pd.DataFrame(mock_data)
mock_df.to_pickle(pickle_path)

# Define configuration dictionary
cs = {
    'prop': 'synthesizability',
    'TARGET': 'synthesizability',
    'propDFpath': pickle_path,
}

ARGS = {
    'cs': cs,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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_experiment_results Ran
ran; our contract draft was wrong, not the code
BAMeScience/SynCoTrainMP/syncotrainmp/pu_alignn_analyze.py
code served (permissive licence) · get_code("26dce5e046a7f86b")
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 tuple. 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 tempfile
import os
import pandas as pd

# Create a temporary directory structure with mock experiment results
temp_dir = tempfile.mkdtemp()

# Create subdirectories for iterations with prediction results
for i in range(3):
    iter_dir = os.path.join(temp_dir, f"PUiter_{i}")
    os.makedirs(iter_dir, exist_ok=True)
    
    # Create mock prediction results CSV
    mock_data = {
        'id': [f'CAR-material_{j}.vas' for j in range(2)],
        'prediction': [0.5, 0.7],
        'score': [0.8, 0.9]
    }
    df = pd.DataFrame(mock_data)
    df.to_csv(os.path.join(iter_dir, 'prediction_results_test_set.csv'), index=False)

ARGS = {
    'output_dir': temp_dir,
    'max_iter': 2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is 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
mae Ran
ran; our contract draft was wrong, not the code
kaist-amsg/Synthesizability-PU-CGCNN/main_PU_learning.py
pointer only (licence: NONE) · get_code("8d28ce13cc82e55f")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, kaist-amsg/Synthesizability-PU-CGCNN when it was checked. 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. A run is recorded.

preload Ran
ran; our contract draft was wrong, not the code
kaist-amsg/Synthesizability-PU-CGCNN/main_PU_learning.py
pointer only (licence: NONE) · get_code("e6d64749b78d6c11")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, kaist-amsg/Synthesizability-PU-CGCNN when it was checked. 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. A run is recorded.

preload Ran
ran; our contract draft was wrong, not the code
kaist-amsg/Synthesizability-PU-CGCNN/predict_PU_learning.py
pointer only (licence: NONE) · get_code("82414c78d42e17d4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, kaist-amsg/Synthesizability-PU-CGCNN when it was checked. 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. A run is recorded.

score_function Ran
ran, honoured the contract we drafted
BAMeScience/SynCoTrainMP/syncotrainmp/pu_schnet_analyze.py
code served (permissive licence) · get_code("fdcbc97232868349")
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 tuple, 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 numpy as np
import pandas as pd

def score_function(x):
    """Calculate score as a percentage of positive predictions and count the number of trials."""
    trial_num = sum(x.notna())
    if trial_num == 0:
        return (np.nan, trial_num)
    else:
        return (x.sum() / trial_num, trial_num)

# Create a pandas Series with some boolean/numeric values
x = pd.Series([True, False, True, True, False, np.nan])

ARGS = {
    'x': x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, 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
setup_output_directory Ran
ran; our contract draft was wrong, not the code
BAMeScience/SynCoTrainMP/syncotrainmp/data_selection.py
code served (permissive licence) · get_code("f31b0b0139dc29d5")
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 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.

import os
import tempfile

# Create a temporary directory to use as the base data path
temp_dir = tempfile.mkdtemp()

# Create a dummy data file path
data_file = os.path.join(temp_dir, "data.csv")

# Create the file so the dirname exists
with open(data_file, 'w') as f:
    f.write("dummy")

ARGS = {
    "data_path": data_file,
    "data_prefix": "experiment_",
    "TARGET": "target_col",
    "prop": "property_col"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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
load_config Not yet run
unverified (harvested, no recorded run)
BAMeScience/SynCoTrainMP/syncotrainmp/pu_schnet_analyze.py
code served (permissive licence) · get_code("1f9a04b438844a37")
How it ran

no invocation recorded.

load_configuration Not yet run
unverified (harvested, no recorded run)
BAMeScience/SynCoTrainMP/syncotrainmp/pu_schnet_train.py
code served (permissive licence) · get_code("1223870eb25f61c7")
How it ran

no invocation recorded.

predProb Not yet run
unverified (harvested, no recorded run)
BAMeScience/SynCoTrainMP/syncotrainmp/pu_schnet/pu_learn/schnet_funcs.py
code served (permissive licence) · get_code("04bf20d50cb7a07f")
How it ran

no invocation recorded.

prepare_experiment_data Not yet run
unverified (harvested, no recorded run)
BAMeScience/SynCoTrainMP/syncotrainmp/data_selection.py
code served (permissive licence) · get_code("e6f663d19ec26922")
How it ran

no invocation recorded.

validate Not yet run
unverified (harvested, no recorded run)
kaist-amsg/Synthesizability-PU-CGCNN/main_PU_learning.py
pointer only (licence: NONE) · get_code("6eb60bbb295782d8")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, kaist-amsg/Synthesizability-PU-CGCNN when it was checked. 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.

validate Not yet run
unverified (harvested, no recorded run)
kaist-amsg/Synthesizability-PU-CGCNN/predict_PU_learning.py
pointer only (licence: NONE) · get_code("b6b3efa5231edeb4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, kaist-amsg/Synthesizability-PU-CGCNN when it was checked. 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.

Also present in other papers' code

Of 23 samples lifted from this paper's linked repositories, 23 carry a function identity and 0 do not. Of the 23 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 0 later) and 20 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
mae 5 other papers
5 dated earlier
1710.10324 (2017-10)
Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties
2003.10580 (2020-03)
Meta Pseudo Labels
2206.04968 (2022-06)
What Information is Necessary and Sufficient to Predict Materials Properties using Machine Learning?
and 2 more not named here
collate_pool 2 other papers
2 dated earlier
2003.10580 (2020-03)
Meta Pseudo Labels
2104.10869 (2021-04)
CrysXPP:An Explainable Property Predictor for Crystalline Materials
get_train_val_test_loader 1 other paper
1 dated earlier
2003.10580 (2020-03)
Meta Pseudo Labels

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 2026-11-21.

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 have recently emerged to predict whether hypothetical solid-state materials can be synthesized. These models aim to circumvent direct first-principles modeling of solid-state phase transformations, instead learning from large databases of successfully synthesized materials. Here, we assess the alignment of several recently introduced synthesis prediction models with material and reaction thermodynamics, quantified by the energy with respect to the convex hull and a metric accounting for thermodynamic selectivity of enumerated synthesis reactions. A dataset of successful synthesis recipes was used to determine the likely bounds on both quantities beyond which materials can be deemed unlikely to be synthesized. With these bounds as context, thermodynamic quantities were computed using the CHGNet foundation potential for thousands of new hypothetical materials generated using the Chemeleon generative model. Four recently published machine learning models for synthesizability prediction were applied to this same dataset, and the resultant predictions were considered against computed thermodynamics. We find these models generally overpredict the likelihood of synthesis, but some model scores do trend with thermodynamic heuristics, assigning lower scores to materials that are less stable or do not have an available synthesis recipe that is calculated to be thermodynamically selective. In total, this work identifies existing gaps in machine learning models for materials synthesis and introduces a new approach to assess their quality in the absence of extensive negative examples (failed syntheses).

For agents

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

get_harvested_code_for_paper("2602.04075")
get_code_for_paper("2602.04075")
have("2602.04075")

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

curl https://syntology.ai/api/ran/2602.04075.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.04075.svg)](https://syntology.ai/paper/2602.04075)

Connect an agent — have() is free.