SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.14622 · 2026

Tabular Foundation Models Can Learn Association Rules

Daniel Daza, Paul Groth, Victoria Degeler, Erkan Karabulut, Martijn Schut

arXiv · PDF · Open in the Atlas

Code that ran

13 of 14 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; 5 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 8 failures of our instrument; 1 no recorded run), measured 2026-11-22; 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/2602.14622.json · badge: /api/ran/2602.14622.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
DiTEC-project/tabprobe community
Papers with Code archive
13 ran · 1 no recorded run
FunctionStatusWhere it lives
_calculate_base_metrics Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("b9ef3c69d24b2048")
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 numpy as np

# Create sample data with 5 rows and 3 features
data = np.array([
    [1, 2, 3],
    [1, 2, 4],
    [2, 2, 3],
    [1, 3, 3],
    [2, 2, 4]
], dtype=float)

# Define feature names
feature_names = ['feature_a', 'feature_b', 'feature_c']

# Create a rule with antecedents and consequent
rule = {
    'antecedents': [
        {'feature': 'feature_a', 'value': 1},
        {'feature': 'feature_b', 'value': 2}
    ],
    'consequent': {'feature': 'feature_c', 'value': 3}
}

ARGS = {
    'rule': rule,
    'data': data,
    'feature_names': feature_names
}
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 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
_calculate_interestingness Ran
ran, honoured the contract we drafted
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("f18468bfcdf9b8a2")
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

# Analyze the function to understand what it needs:
# - base_metrics: Dict[str, float] with keys 'confidence', 'support', 'rhs_support'
# - n_samples: int

# Construct base_metrics with the required keys
base_metrics = {
    'confidence': 0.85,
    'support': 0.3,
    'rhs_support': 0.5
}

# Construct n_samples as an integer
n_samples = 100

# Define ARGS mapping all parameters
ARGS = {
    'base_metrics': base_metrics,
    'n_samples': n_samples
}
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
_calculate_zhangs_metric Ran
ran, honoured the contract we drafted
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("ab8e41a3b44cdef6")
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)

# Construct base_metrics dictionary with the required keys
# confidence: the confidence of the rule (between 0 and 1)
# rhs_support: the support of the right-hand side (between 0 and 1)
base_metrics = {
    'confidence': 0.75,
    'rhs_support': 0.5
}

ARGS = {
    'base_metrics': base_metrics
}
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
_get_feature_index Ran
ran, honoured the contract we drafted
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("938b4eb50be744de")
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 int, ndim 0, dtype family int, 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
from typing import List, Dict, Optional, Tuple

torch.manual_seed(0)

# Test case 1: feature is an integer
feature_1 = 0
feature_names_1 = None
args_1 = {
    'feature': feature_1,
    'feature_names': feature_names_1
}

# Test case 2: feature is a string and feature_names is provided
feature_2 = 'age'
feature_names_2 = ['name', 'age', 'salary', 'department']
args_2 = {
    'feature': feature_2,
    'feature_names': feature_names_2
}

# Test case 3: feature is a string found in feature_names
feature_3 = 'salary'
feature_names_3 = ['col1', 'col2', 'salary', 'col4']
args_3 = {
    'feature': feature_3,
    'feature_names': feature_names_3
}

# Use the first test case as the primary ARGS
ARGS = args_1
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 int, ndim 0, dtype family int, 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_mask_for_conditions Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("b4ac846d95eaded2")
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 ndarray, ndim 1, dtype family float, shape (4,) 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 numpy as np

# Set up test data
data = np.array([
    [1.0, 2.0, 3.0, 4.0],
    [1.0, 5.0, 6.0, 7.0],
    [2.0, 2.0, 8.0, 9.0],
    [2.0, 5.0, 10.0, 11.0],
], dtype=float)

# Set up conditions - test with feature names and by index
conditions = [
    {'feature': 'col_a', 'value': 1.0},
    {'feature': 'col_b', 'value': 2.0},
]

feature_names = ['col_a', 'col_b', 'col_c', 'col_d']

ARGS = {
    'conditions': conditions,
    'data': data,
    'feature_names': feature_names,
}
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 ndarray, ndim 1, dtype family float, shape (4,) — shape (4,) 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
_initialize_input_vectors Ran
ran, honoured the contract we drafted
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("783a4b571ba8ebd0")
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 ndarray, ndim 1, dtype family float, shape (10,) 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 numpy as np

# Construct valid arguments for _initialize_input_vectors
ARGS = {
    "input_vector_size": 10,
    "categories": [
        {"start": 0, "end": 3},
        {"start": 3, "end": 7},
        {"start": 7, "end": 10}
    ],
    "use_zeros": False
}
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 ndarray, ndim 1, dtype family float, shape (10,) — shape (10,) 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
add_gaussian_noise Ran
ran, honoured the contract we drafted
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("da4ecd76c60aee17")
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 ndarray, ndim 2, dtype family float, shape (5, 8) 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 numpy as np

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

# Create sample data: numpy array of shape (n_samples, n_features)
# Using distinct dimensions: 5 samples, 8 features
data = np.random.rand(5, 8).astype(np.float32)

# Define arguments dictionary
ARGS = {
    "data": data,
    "noise_factor": 0.5
}
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 ndarray, ndim 2, dtype family float, shape (5, 8) — shape (5, 8) 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
calculate_rule_metrics Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("fa23c081527a33d8")
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 numpy as np

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

# Create sample data: 10 samples, 4 features
data = np.array([
    [1, 2, 3, 4],
    [1, 2, 3, 5],
    [1, 2, 4, 4],
    [2, 2, 3, 4],
    [2, 3, 3, 4],
    [1, 2, 3, 4],
    [2, 2, 3, 5],
    [1, 3, 4, 4],
    [2, 2, 4, 4],
    [1, 2, 3, 4],
], dtype=np.float32)

# Create feature names
feature_names = ['feature_0', 'feature_1', 'feature_2', 'feature_3']

# Create sample rules in PyAerial format
rules = [
    {
        'antecedents': [
            {'feature': 'feature_0', 'value': 1},
            {'feature': 'feature_1', 'value': 2}
        ],
        'consequent': {'feature': 'feature_2', 'value': 3}
    },
    {
        'antecedents': [
            {'feature': 'feature_1', 'value': 2}
        ],
        'consequent': {'feature': 'feature_3', 'value': 4}
    },
    {
        'antecedents': [
            {'feature': 'feature_0', 'value': 2},
            {'feature': 'feature_2', 'value': 3}
        ],
        'consequent': {'feature': 'feature_1', 'value': 2}
    }
]

# Metric names to calculate
metric_names = ['support', 'confidence', 'rule_coverage', 'zhangs_metric', 'interestingness']

ARGS = {
    'rules': rules,
    'data': data,
    'feature_names': feature_names,
    'metric_names': metric_names
}
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 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
discretize_numerical Ran
ran; our contract draft was wrong, not the code
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("8ef44af6fb4f9aa8")
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 pandas as pd
import numpy as np

# Create a sample DataFrame with mixed numerical and categorical data
np.random.seed(0)
data = pd.DataFrame({
    'continuous_1': np.random.normal(loc=100, scale=15, size=50),
    'continuous_2': np.random.uniform(0, 1, size=50),
    'low_cardinality': np.random.randint(1, 5, size=50),
    'categorical_col': np.random.choice(['A', 'B', 'C'], size=50),
})

ARGS = {
    'data': data,
    'n_bins': 5,
    'categorical_threshold': 10,
}
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 — 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
extract_rules_from_reconstruction Ran
ran; our contract draft was wrong, not the code
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("41a73ebc34947803")
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 list, 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
from sklearn.preprocessing import OneHotEncoder

# Create a small probability matrix (n_test_vectors=2, total_dim=10)
prob_matrix = np.array([
    [0.9, 0.1, 0.85, 0.2, 0.5, 0.75, 0.3, 0.88, 0.2, 0.6],
    [0.8, 0.3, 0.9, 0.15, 0.6, 0.82, 0.25, 0.85, 0.1, 0.7]
], dtype=np.float32)

# Test descriptions: list of tuples describing antecedents
# Each tuple contains (feature_idx, class_idx) pairs
test_descriptions = [
    ((0, 0), (1, 1)),  # First test vector: feature 0 class 0, feature 1 class 1
    ((1, 0),)           # Second test vector: feature 1 class 0
]

# Feature value indices: list of dicts with 'start', 'end', 'feature' for each feature
# Feature 0: indices 0-1 (2 classes)
# Feature 1: indices 2-4 (3 classes)
# Feature 2: indices 5-7 (3 classes)
# Feature 3: indices 8-9 (2 classes)
feature_value_indices = [
    {'start': 0, 'end': 2, 'feature': 0},
    {'start': 2, 'end': 5, 'feature': 1},
    {'start': 5, 'end': 8, 'feature': 2},
    {'start': 8, 'end': 10, 'feature': 3}
]

# Create a mock encoder with categories for each feature
class MockEncoder:
    def __init__(self):
        self.categories_ = [
            np.array(['cat_a', 'cat_b']),           # Feature 0: 2 classes
            np.array(['dog_a', 'dog_b', 'dog_c']),  # Feature 1: 3 classes
            np.array(['bird_a', 'bird_b', 'bird_c']),  # Feature 2: 3 classes
            np.array(['fish_a', 'fish_b'])          # Feature 3: 2 classes
        ]

encoder = MockEncoder()

# Feature names
feature_names = ['Animal', 'Color', 'Size', 'Habitat']

# Thresholds
ant_similarity = 0.5
cons_similarity = 0.8

ARGS = {
    'prob_matrix': prob_matrix,
    'test_descriptions': test_descriptions,
    'feature_value_indices': feature_value_indices,
    'ant_similarity': ant_similarity,
    'cons_similarity': cons_similarity,
    'feature_names': feature_names,
    'encoder': encoder
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list 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
generate_test_matrix Ran
ran; our contract draft was wrong, not the code
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("c3585a8836102c72")
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 numpy as np
from itertools import combinations, product

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

ARGS = {
    'n_features': 3,
    'classes_per_feature': [2, 3, 2],
    'max_antecedents': 2,
    'use_zeros_for_unmarked': False,
    'min_antecedents': 1,
    'allowed_items': None,
}
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 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
get_significant_single_items Ran
ran; our contract draft was wrong, not the code
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("ac4ecad2d1c6e8c5")
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 set. 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 numpy as np

torch.manual_seed(0)

# Build prob_matrix: shape (num_test_descriptions, num_feature_values)
# We'll have 2 test descriptions and 4 total feature values across all features
prob_matrix = torch.tensor([
    [0.6, 0.3, 0.2, 0.1],
    [0.1, 0.7, 0.4, 0.2]
], dtype=torch.float32)

# Build test_descriptions: list of lists where each inner list contains tuples of (feat_idx, class_idx)
# First description has 1 item, second description has 1 item
test_descriptions = [
    [(0, 1)],  # Feature 0, class index 1
    [(1, 0)]   # Feature 1, class index 0
]

# Build feature_value_indices: dict mapping feature index to a dict with 'start' key
# This maps each feature to the starting index in prob_matrix columns
# Feature 0 starts at 0, Feature 1 starts at 2 (has 2 values)
feature_value_indices = {
    0: {'start': 0},  # Feature 0 starts at column 0
    1: {'start': 2}   # Feature 1 starts at column 2
}

# ant_similarity threshold
ant_similarity = 0.5

ARGS = {
    'prob_matrix': prob_matrix,
    'test_descriptions': test_descriptions,
    'feature_value_indices': feature_value_indices,
    'ant_similarity': ant_similarity
}
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 set — a set output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a set 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
prepare_categorical_data Ran
ran; our contract draft was wrong, not the code
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("22341c3f03ca019d")
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 pandas as pd
import numpy as np
from sklearn.preprocessing import OneHotEncoder

# Create a small pandas DataFrame with categorical features
data = {
    'color': ['red', 'blue', 'green', 'red'],
    'size': ['small', 'large', 'small', 'medium'],
    'shape': ['circle', 'square', 'circle', 'triangle']
}
X = pd.DataFrame(data)

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, sklearn — 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
TabProbe Not yet run
unverified (harvested, no recorded run)
DiTEC-project/tabprobe/src/wrapper/tabprobe.py
code served (permissive licence) · get_code("79598edb1498588d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 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 14 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

Association Rule Mining (ARM) is a fundamental task for knowledge discovery in tabular data and is widely used in high-stakes decision-making. Classical ARM methods rely on frequent itemset mining, leading to rule explosion and poor scalability, while recent neural approaches mitigate these issues but suffer from degraded performance in low-data regimes. Tabular foundation models (TFMs), pretrained on diverse tabular data with strong in-context generalization, provide a basis for addressing these limitations. We introduce a model-agnostic association rule learning framework that extracts association rules from any conditional probabilistic model over tabular data, enabling us to leverage TFMs. We then introduce TabProbe, an instantiation of our framework that utilizes TFMs as conditional probability estimators to learn association rules out-of-the-box without frequent itemset mining. We evaluate our approach on tabular datasets of varying sizes based on standard ARM rule quality metrics and downstream classification performance. The results show that TFMs consistently produce concise, high-quality association rules with strong predictive performance and remain robust in low-data settings without task-specific training. Source code is available at https://github.com/DiTEC-project/tabprobe.

For agents

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

get_harvested_code_for_paper("2602.14622")
get_code_for_paper("2602.14622")
have("2602.14622")

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

curl https://syntology.ai/api/ran/2602.14622.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.14622.svg)](https://syntology.ai/paper/2602.14622)

Connect an agent — have() is free.