SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2106.11959 · NeurIPS · 2021

Revisiting Deep Learning Models for Tabular Data

Yandex, Yury Gorishniy, Ivan Rubachev, Valentin Khrulkov, Artem Babenko

arXiv · PDF · Open in the Atlas

Code that ran

18 of 22 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, 1 violated it and 14 have no contract verdict; 3 failures 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/2106.11959.json · badge: /api/ran/2106.11959.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
Yura52/tabular-dl-revisiting-models canonical (classifier label)
classifier basis: census
10 ran · 3 no recorded run
XavierSpycy/tabtransformers community (classifier label)
classifier basis not recorded on the link
6 ran · 1 no recorded run
lucidrains/tab-transformer-pytorch reimplementation (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
jrzaurin/pytorch-widedeep community (classifier label)
classifier basis not recorded on the link
1 ran · 0 no recorded run
FunctionStatusWhere it lives
FeedForward Ran
ran; our contract draft was wrong, not the code
lucidrains/tab-transformer-pytorch/tab_transformer_pytorch/ft_transformer.py
code served (permissive licence) · get_code("5bbdefbbff4e08d8")
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 Sequential. 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)

# FeedForward is a function that returns an nn.Sequential module
# Parameters: dim (required), mult (default=4), dropout (default=0.0)

ARGS = {
    'dim': 8,
    'mult': 4,
    'dropout': 0.1
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
f1_score_macro Ran
ran on a synthesized input
XavierSpycy/tabtransformers/tabtransformers/metrics.py
code served (permissive licence) · get_code("4d711eaf9f0bb383")
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 numpy as np
from sklearn.metrics import f1_score

def f1_score_macro(y_true: np.ndarray, y_pred: np.ndarray) -> float:
    return f1_score(y_true, y_pred, average='macro')

# Set up valid arguments
np.random.seed(0)

# Create sample ground truth labels (3 classes)
y_true = np.array([0, 1, 2, 0, 1, 2, 0, 1])

# Create sample predictions (same shape, same classes)
y_pred = np.array([0, 1, 2, 0, 2, 2, 1, 1])

ARGS = {
    'y_true': y_true,
    'y_pred': y_pred
}
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 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
geglu Ran
ran; our contract draft was wrong, not the code
Yura52/tabular-dl-revisiting-models/lib/deep.py
code served (permissive licence) · get_code("d7bace13fb369137")
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 Tensor, ndim 2, dtype family float, shape (2, 2) 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 torch.nn.functional as F

torch.manual_seed(0)

# The geglu function takes a single parameter x which is a Tensor
# It chunks x along the last dimension into 2 equal parts
# So x needs to have an even size in the last dimension
# Let's create a small tensor with batch size 2 and last dimension 4 (even)

x = torch.randn(2, 4)

ARGS = {
    "x": x
}
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 Tensor, ndim 2, dtype family float, shape (2, 2) — shape (2, 2) 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_activation_fn Ran
ran on a synthesized input
jrzaurin/pytorch-widedeep/pytorch_widedeep/models/_get_activation_fn.py
code served (permissive licence) · get_code("755bddefe05fb03e")
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 ReLU. 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
import torch.nn.functional as F

torch.manual_seed(0)

# Test with a simple activation string
ARGS = {
    'activation': 'relu'
}
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 ReLU — a ReLU output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ReLU 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_data Ran
ran on a synthesized input
XavierSpycy/tabtransformers/tabtransformers/tools.py
code served (permissive licence) · get_code("f8e126db3b9cc8e1")
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 os
import tempfile
import pandas as pd
from sklearn.model_selection import train_test_split

# Create temporary directory with sample CSV files
tmpdir = tempfile.mkdtemp()

# Create sample train.csv
train_df = pd.DataFrame({
    'feature1': [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
    'feature2': [10, 20, 30, 40, 50, 60, 70, 80, 90, 100],
    'target': [0, 1, 0, 1, 0, 1, 0, 1, 0, 1]
})
train_df.to_csv(os.path.join(tmpdir, 'train.csv'), index=False)

# Create sample test.csv
test_df = pd.DataFrame({
    'feature1': [11, 12, 13],
    'feature2': [110, 120, 130],
    'target': [1, 0, 1]
})
test_df.to_csv(os.path.join(tmpdir, 'test.csv'), index=False)

ARGS = {
    'data_path': tmpdir,
    'split_val': True,
    'data_files': {'train': 'train.csv', 'test': 'test.csv'},
    'val_params': {'test_size': 0.2, 'random_state': 42},
    'index_col': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
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
get_data_loader Ran
ran on a synthesized input
XavierSpycy/tabtransformers/tabtransformers/tools.py
code served (permissive licence) · get_code("d29f63635e0b6468")
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 torch
import numpy as np
from torch.utils.data import Dataset

torch.manual_seed(0)

class TabularDataset(Dataset):
    """Minimal stand-in for a tabular dataset"""
    def __init__(self, num_samples=8, num_features=4):
        self.num_samples = num_samples
        self.num_features = num_features
        self.X = torch.randn(num_samples, num_features)
        self.y = torch.randint(0, 2, (num_samples,))
    
    def __len__(self):
        return self.num_samples
    
    def __getitem__(self, idx):
        return self.X[idx], self.y[idx]

# Create datasets with distinct sizes for traceability
train_dataset = TabularDataset(num_samples=8, num_features=4)
test_dataset = TabularDataset(num_samples=6, num_features=4)
val_dataset = TabularDataset(num_samples=4, num_features=4)

ARGS = {
    'train_dataset': train_dataset,
    'test_dataset': test_dataset,
    'val_dataset': val_dataset,
    'train_batch_size': 2,
    'inference_batch_size': 2,
}
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 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_loss_function Ran
ran on a synthesized input
XavierSpycy/tabtransformers/tabtransformers/config.py
code served (permissive licence) · get_code("516fa25431b82747")
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 MSELoss. 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)

# The function takes a config dict with 'loss_function' and optional 'loss_kwargs'
# Test with a simple MSE loss configuration
ARGS = {
    'config': {
        'loss_function': 'mse',
        'loss_kwargs': {
            'reduction': 'mean'
        }
    }
}
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 MSELoss — a MSELoss output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MSELoss 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
glu Ran
ran on a synthesized input
Yura52/tabular-dl-revisiting-models/bin/tabnet.py
code served (permissive licence) · get_code("e7ecc99fa574de72")
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 EagerTensor, ndim 2, dtype family float, shape (2, 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 tensorflow as tf
import numpy as np

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

# Create a small tensor for testing
# The function expects act to be a 2D tensor where it splits along axis 1
# act[:, :n_units] and act[:, n_units:] must have compatible shapes
# So act should have shape (batch_size, 2*n_units) or larger

batch_size = 2
n_units = 4

# Create act tensor with shape (batch_size, 2*n_units)
# This allows the function to split it into two parts of size n_units each
act = tf.constant(np.random.randn(batch_size, 2 * n_units), dtype=tf.float32)

ARGS = {
    "act": act,
    "n_units": n_units
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 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
init_gbnn Ran
ran on a synthesized input
Yura52/tabular-dl-revisiting-models/bin/grownet.py
code served (permissive licence) · get_code("b71e1c227eac8b6f")
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 numpy as np

# Construct a valid argument for the init_gbnn function
# The function expects y to be iterable and supports indexing
# It checks if elements are > 0 to count positive/negative values
# It computes blind accuracy and log odds ratio

y = np.array([1.5, -0.3, 2.1, -1.2, 0.8, -0.5])

ARGS = {
    'y': y
}
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 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
load_config Ran
ran on a synthesized input
XavierSpycy/tabtransformers/tabtransformers/config.py
code served (permissive licence) · get_code("d5ca06455ae6a9a5")
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 yaml
import tempfile
import os

# Create a temporary YAML config file
temp_dir = tempfile.mkdtemp()
config_file_path = os.path.join(temp_dir, "test_config.yaml")

# Write a sample config to the temporary file
sample_config = {
    "model": "test_model",
    "learning_rate": 0.001,
    "batch_size": 32,
    "epochs": 10,
    "layers": [64, 128, 256]
}

with open(config_file_path, 'w') as f:
    yaml.dump(sample_config, f)

ARGS = {
    "config_path": config_file_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
yaml — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_json Ran
ran on a synthesized input
Yura52/tabular-dl-revisiting-models/lib/util.py
code served (permissive licence) · get_code("2f33af03ea909beb")
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 json
import typing as ty
from pathlib import Path
import tempfile

# Create a temporary JSON file with valid content
temp_dir = tempfile.TemporaryDirectory()
temp_file = Path(temp_dir.name) / "test.json"

test_data = {
    "key1": "value1",
    "key2": 42,
    "key3": [1, 2, 3],
    "key4": {"nested": "object"}
}

temp_file.write_text(json.dumps(test_data))

ARGS = {
    "path": temp_file
}
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
load_pickle Ran
ran on a synthesized input
Yura52/tabular-dl-revisiting-models/lib/util.py
code served (permissive licence) · get_code("ba8f59b208956c3b")
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 pickle
import typing as ty
from pathlib import Path
import tempfile
import os

# Create a temporary pickle file with some test data
test_data = {
    "key1": "value1",
    "key2": [1, 2, 3],
    "key3": {"nested": "dict"}
}

with tempfile.NamedTemporaryFile(mode='wb', delete=False, suffix='.pkl') as f:
    pickle.dump(test_data, f)
    temp_pickle_path = f.name

try:
    ARGS = {
        "path": temp_pickle_path
    }
finally:
    pass
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
make_summary Ran
ran on a synthesized input
Yura52/tabular-dl-revisiting-models/lib/metrics.py
code served (permissive licence) · get_code("9fb7a99bb9386bb6")
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 typing as ty

torch_seed = 0

# Construct a metrics dictionary that the function expects
# The function processes metrics with keys that may be remapped
# and values that can be floats or dicts with nested structure

metrics = {
    'score': 0.8765,
    'accuracy': 0.92341,
    'roc_auc': 0.87654,
    'macro avg': {
        'precision': 0.85432,
        'recall': 0.84321,
        'f1-score': 0.84876,
        'support': 128.0
    },
    'weighted avg': {
        'precision': 0.91234,
        'recall': 0.92341,
        'f1-score': 0.91876,
        'support': 128.0
    },
    '0': 0.95,  # This key is a digit string and will be skipped
    'class_0': {
        'precision': 0.93421,
        'recall': 0.94532,
        'f1-score': 0.93976,
        'support': 64.0
    }
}

ARGS = {
    'metrics': metrics
}
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
needs_wd Ran
ran, violated the contract we drafted
Yura52/tabular-dl-revisiting-models/bin/autoint.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from whatashot/excelformer, is pointer only (licence: NONE) · get_code("9e66eafce9e7d0c5")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from whatashot/excelformer, and GitHub detected no licence in that repository 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.

normalize Ran
ran on a synthesized input
Yura52/tabular-dl-revisiting-models/lib/data.py
code served (permissive licence) · get_code("e9e943ad18a8b419")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import sklearn.preprocessing

# Create small training and test data
np.random.seed(42)
X_train = np.random.randn(50, 4).astype(np.float32)
X_test = np.random.randn(10, 4).astype(np.float32)

X = {
    'train': X_train,
    'test': X_test
}

ARGS = {
    'X': X,
    'normalization': 'standard',
    'seed': 42,
    'noise': 0.001
}
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 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
reglu Ran
ran; our contract draft was wrong, not the code
Yura52/tabular-dl-revisiting-models/lib/deep.py
code served (permissive licence) · get_code("2fb665ac3ecd7c20")
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 Tensor, ndim 2, dtype family float, shape (2, 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 torch
import torch.nn.functional as F

torch.manual_seed(0)

# The function reglu takes a Tensor x, chunks it in half along the last dimension,
# and returns a * relu(b). So we need a tensor with an even last dimension.
x = torch.randn(2, 8)  # batch size 2, last dim 8 (will be chunked into two size-4 tensors)

ARGS = {
    "x": x
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 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
root_mean_squared_logarithmic_error Ran
ran on a synthesized input
XavierSpycy/tabtransformers/tabtransformers/metrics.py
code served (permissive licence) · get_code("a18290fdacb1226e")
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 float64, ndim 0, 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

# Create small sample data for y_true and y_pred
# The function expects numpy arrays and computes RMSLE
y_true = np.array([1.0, 2.0, 3.0, 4.0])
y_pred = np.array([1.1, 2.1, 2.9, 4.2])

ARGS = {
    "y_true": y_true,
    "y_pred": y_pred,
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
to_tensors Ran
ran on a synthesized input
Yura52/tabular-dl-revisiting-models/lib/data.py
code served (permissive licence) · get_code("33a6192b7abb8353")
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 typing as ty
import numpy as np
import torch

torch.manual_seed(0)

# Construct a simple ArrayDict with numpy arrays
data = {
    "input": np.array([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0]], dtype=np.float32),
    "target": np.array([[0, 1, 0, 1], [1, 0, 1, 0]], dtype=np.int64),
    "mask": np.array([1.0, 0.5], dtype=np.float32),
}

ARGS = {
    "data": data
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
calculate_metrics Not yet run
unverified (harvested, no recorded run)
Yura52/tabular-dl-revisiting-models/lib/metrics.py
code served (permissive licence) · get_code("12f448bfa8afc6b7")
How it ran

no invocation recorded.

get_model Not yet run
unverified (harvested, no recorded run)
XavierSpycy/tabtransformers/tabtransformers/config.py
code served (permissive licence) · get_code("6e2718cdfa77cc6a")
How it ran

no invocation recorded.

load_dataset_info Not yet run
unverified (harvested, no recorded run)
Yura52/tabular-dl-revisiting-models/lib/data.py
code served (permissive licence) · get_code("657558d2e8d19a32")
How it ran

no invocation recorded.

make_optimizer Not yet run
unverified (harvested, no recorded run)
Yura52/tabular-dl-revisiting-models/lib/deep.py
code served (permissive licence) · get_code("27e9ee3019bc8ff0")
How it ran

no invocation recorded.

Also present in other papers' code

Of 22 samples lifted from this paper's linked repositories, 22 carry a function identity and 0 do not. Of the 22 distinct functions with an identity, 6 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 5 later) and 16 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
geglu 3 other papers
3 later
2207.03208 (2022-07)
Revisiting Pretraining Objectives for Tabular Deep Learning
2211.16887 (2022-11)
T2G-Former: Organizing Tabular Features into Relation Graphs Promotes Heterogeneous Feature Interaction
2403.01570 (2024-03)
Small Models are LLM Knowledge Triggers for Medical Tabular Prediction
reglu 3 other papers
3 later
2207.03208 (2022-07)
Revisiting Pretraining Objectives for Tabular Deep Learning
2211.16887 (2022-11)
T2G-Former: Organizing Tabular Features into Relation Graphs Promotes Heterogeneous Feature Interaction
2403.01570 (2024-03)
Small Models are LLM Knowledge Triggers for Medical Tabular Prediction
needs_wd 2 other papers
2 later
2211.16887 (2022-11)
T2G-Former: Organizing Tabular Features into Relation Graphs Promotes Heterogeneous Feature Interaction
2301.02819 (2023-01)
ExcelFormer: A neural network surpassing GBDTs on tabular data
FeedForward 1 other paper
1 dated earlier
2012.06678 (2020-12)
TabTransformer: Tabular Data Modeling Using Contextual Embeddings
glu 1 other paper
1 later
2211.16887 (2022-11)
T2G-Former: Organizing Tabular Features into Relation Graphs Promotes Heterogeneous Feature Interaction
init_gbnn 1 other paper
1 later
2211.16887 (2022-11)
T2G-Former: Organizing Tabular Features into Relation Graphs Promotes Heterogeneous Feature Interaction

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

The existing literature on deep learning for tabular data proposes a wide range of novel architectures and reports competitive results on various datasets. However, the proposed models are usually not properly compared to each other and existing works often use different benchmarks and experiment protocols. As a result, it is unclear for both researchers and practitioners what models perform best. Additionally, the field still lacks effective baselines, that is, the easy-to-use models that provide competitive performance across different problems. In this work, we perform an overview of the main families of DL architectures for tabular data and raise the bar of baselines in tabular DL by identifying two simple and powerful deep architectures. The first one is a ResNetlike architecture which turns out to be a strong baseline that is often missing in prior works. The second model is our simple adaptation of the Transformer architecture for tabular data, which outperforms other solutions on most tasks. Both models are compared to many existing architectures on a diverse set of tasks under the same training and tuning protocols. We also compare the best DL models with Gradient Boosted Decision Trees and conclude that there is still no universally superior solution. The source code is available at https: //github.com/yandex-research/tabular-dl-revisiting-models.

For agents

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

get_harvested_code_for_paper("2106.11959")
get_code_for_paper("2106.11959")
have("2106.11959")

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

curl https://syntology.ai/api/ran/2106.11959.json

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

[![Syntology run record](https://syntology.ai/api/ran/2106.11959.svg)](https://syntology.ai/paper/2106.11959)

Connect an agent — have() is free.