SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.05556 · NeurIPS · 2022

On Embeddings for Numerical Features in Tabular Deep Learning

Yury Gorishniy, Yandex Ivan, Rubachev Hse, Yandex Artem, Babenko Yandex

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; 1 honoured the contract we drafted, 1 violated it and 8 have no contract verdict; 3 failures of our instrument; 1 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/2203.05556.json · badge: /api/ran/2203.05556.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-num-embeddings canonical (classifier label)
classifier basis not recorded on the link
8 ran · 0 no recorded run
aruberts/TabTransformerTF community (classifier label)
classifier basis not recorded on the link
5 ran · 1 no recorded run
FunctionStatusWhere it lives
build_categorical_prep Ran
ran on a synthesized input
aruberts/TabTransformerTF/tabtransformertf/utils/preprocessing.py
code served (permissive licence) · get_code("45b8b069db767fae")
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 pandas as pd
import tensorflow as tf
import numpy as np

# Create a small DataFrame with categorical features
data = pd.DataFrame({
    'color': ['red', 'blue', 'green', 'red', 'blue'],
    'size': ['small', 'large', 'medium', 'small', 'large'],
    'brand': ['A', 'B', 'A', 'C', 'B']
})

categorical_features = ['color', 'size', 'brand']

ARGS = {
    'data': data,
    'categorical_features': categorical_features
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, tensorflow, tqdm — 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
build_mlp Ran
ran on a synthesized input
aruberts/TabTransformerTF/tabtransformertf/utils/helper.py
code served (permissive licence) · get_code("6696b12816459682")
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 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 tensorflow as tf

ARGS = {
    "input_dim": 64,
    "factors": [2, 4, 8],
    "dropout": 0.5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — 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 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
build_numerical_prep Ran
ran on a synthesized input
aruberts/TabTransformerTF/tabtransformertf/utils/preprocessing.py
code served (permissive licence) · get_code("c326b3c38a495cec")
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 pandas as pd
import tensorflow as tf
from sklearn.tree import DecisionTreeClassifier, DecisionTreeRegressor

# Create a small pandas DataFrame with numerical data
np.random.seed(0)
n_samples = 10
n_features = 2

data = pd.DataFrame({
    'feature_1': np.random.randn(n_samples).astype(np.float32),
    'feature_2': np.random.randn(n_samples).astype(np.float32),
})

# Create target variable for classification
y = np.random.randint(0, 2, n_samples)

# Define numerical features
numerical_features = ['feature_1', 'feature_2']

ARGS = {
    'data': data,
    'numerical_features': numerical_features,
    'y': y,
    'n_bins': 5,
    'type': 'ple'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, sklearn, tensorflow — 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_rmse Ran
ran, violated the contract we drafted
Yura52/tabular-dl-num-embeddings/lib/metrics.py
code served (permissive licence) · get_code("6a428cefdcb71c15")
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

# Create small sample arrays for y_true and y_pred
np.random.seed(0)
y_true = np.array([1.0, 2.0, 3.0, 4.0, 5.0])
y_pred = np.array([1.1, 2.2, 2.9, 4.1, 4.8])

# Create a standard deviation value
std = 2.5

ARGS = {
    "y_true": y_true,
    "y_pred": y_pred,
    "std": std
}
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
compute_bins Ran
ran on a synthesized input
Yura52/tabular-dl-num-embeddings/package/rtdl_num_embeddings.py
code served (permissive licence) · get_code("8c3de6a19e7b2e83")
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 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import warnings

torch.manual_seed(0)

# Create minimal valid input data
# X must be 2D, have at least 2 rows, at least 1 column, all finite values, and at least 2 distinct values per column
X = torch.randn(100, 4)  # 100 samples, 4 features

# n_bins must be > 1 and < len(X)
n_bins = 48

# tree_kwargs is None for quantile-based binning
tree_kwargs = None

# y must be None when tree_kwargs is None
y = None

# regression must be None when tree_kwargs is None
regression = None

# verbose must be False when tree_kwargs is None
verbose = False

ARGS = {
    'X': X,
    'n_bins': n_bins,
    'tree_kwargs': tree_kwargs,
    'y': y,
    'regression': regression,
    'verbose': verbose,
}
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 list, ndim 2, 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
corrupt_dataset Ran
ran on a synthesized input
aruberts/TabTransformerTF/tabtransformertf/utils/helper.py
code served (permissive licence) · get_code("e4c14bee9f9e6324")
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 3, 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

np.random.seed(0)

# Create a small pandas DataFrame as input
# The function calls .mask() and .values on the input, indicating it expects a pandas DataFrame
x = pd.DataFrame({
    'col1': [1.0, 2.0, 3.0, 4.0],
    'col2': [5.0, 6.0, 7.0, 8.0],
    'col3': [9.0, 10.0, 11.0, 12.0]
})

ARGS = {
    'x': x,
    'p_replace': 0.2
}
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 3, 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
cos_sin Ran
ran; our contract draft was wrong, not the code
Yura52/tabular-dl-num-embeddings/lib/deep.py
code served (permissive licence) · get_code("61b240fd493b6733")
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, 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 torch
from torch import Tensor

torch.manual_seed(0)

# The function takes a single parameter x which is a Tensor
# It applies torch.cos and torch.sin to it, then concatenates along the last dimension
# So x should be a tensor of any shape

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, 8) — shape (2, 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
generate_mask Ran
ran on a synthesized input
aruberts/TabTransformerTF/tabtransformertf/utils/helper.py
code served (permissive licence) · get_code("d049416624d7d820")
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 (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 numpy as np

np.random.seed(0)

# The function generates a boolean mask based on an input array x
# and a probability p_replace. It uses:
# - x.shape to get dimensions
# - np.random.choice with size, p parameters
# - Returns a boolean array

# Create a small test array with distinct dimensions
x = np.random.randn(2, 4)

# p_replace is a probability, should be between 0 and 1
p_replace = 0.2

ARGS = {
    "x": x,
    "p_replace": p_replace,
}
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 (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
get_category_sizes Ran
ran on a synthesized input
Yura52/tabular-dl-num-embeddings/lib/data.py
code served (permissive licence) · get_code("7f2a902735fdb4d9")
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 int; 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

torch.manual_seed(0)

# Test with torch.Tensor input
X_torch = torch.randint(0, 3, (2, 4))

# Test with numpy.ndarray input
X_numpy = np.random.randint(0, 3, (2, 4))

ARGS = {
    "X": X_torch
}
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 list, ndim 1, dtype family int — 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
get_n_parameters Ran
ran, honoured the contract we drafted
Yura52/tabular-dl-num-embeddings/lib/deep.py
code served (permissive licence) · get_code("4bb318a745ffe343")
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 torch.nn as nn

torch.manual_seed(0)

# Build a small nn.Module for testing
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

model = SimpleModel()

# Construct ARGS dictionary
ARGS = {
    "m": model
}
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 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
normalize Ran
ran on a synthesized input
Yura52/tabular-dl-num-embeddings/lib/data.py
code served (permissive licence) · get_code("2a8f8becc7258f1f")
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
from typing import Optional

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

# Create small sample data for train, test, and validation sets
train_data = np.random.randn(60, 4).astype(np.float32)
test_data = np.random.randn(20, 4).astype(np.float32)
val_data = np.random.randn(15, 4).astype(np.float32)

# Create the ArrayDict with train/test/val splits
X = {
    'train': train_data,
    'test': test_data,
    'val': val_data
}

# Define ARGS
ARGS = {
    'X': X,
    'normalization': 'standard',
    'seed': 42
}
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
pack_config Ran
ran; our contract draft was wrong, not the code
Yura52/tabular-dl-num-embeddings/lib/util.py
code served (permissive licence) · get_code("eb10a2d0f65a0faf")
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
import numpy as np
from typing import Any, cast, get_args, get_origin

torch.manual_seed(0)

# Define the helper functions from the provided code
def _replace(data, condition, value):
    def do(x):
        if isinstance(x, dict):
            return {k: do(v) for k, v in x.items()}
        elif isinstance(x, list):
            return [do(y) for y in x]
        else:
            return value if condition(x) else x
    return do(data)

_CONFIG_NONE = '__none__'

def pack_config(config: dict[str, Any]) -> dict[str, Any]:
    config = cast(dict[str, Any], _replace(config, lambda x: x is None, _CONFIG_NONE))
    return config

# Create test config with various types including None values
test_config = {
    "learning_rate": 0.001,
    "batch_size": 32,
    "model_name": "bert",
    "dropout": None,
    "hidden_size": 768,
    "nested": {
        "param1": 1.5,
        "param2": None,
        "param3": "value"
    },
    "list_config": [1, 2, None, "item"],
    "deep_nested": {
        "level2": {
            "value": None,
            "other": 42
        }
    }
}

ARGS = {
    "config": test_config
}
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
unpack_config Ran
ran; our contract draft was wrong, not the code
Yura52/tabular-dl-num-embeddings/lib/util.py
code served (permissive licence) · get_code("4e3409f8a7cbd6d1")
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
import numpy as np
from typing import Any

torch.manual_seed(0)

# Create a simple config dictionary with some nested structures
# and include the special _CONFIG_NONE sentinel value
config_with_none = {
    'model_name': 'test_model',
    'learning_rate': 0.001,
    'batch_size': 32,
    'nested': {
        'param1': '__none__',
        'param2': 'value2',
        'deep': {
            'value': '__none__',
            'other': 42
        }
    },
    'list_param': [1, 2, '__none__', 'four'],
    'mixed_list': [
        {'inner': '__none__'},
        {'inner': 'data'}
    ],
    'normal_value': 3.14
}

ARGS = {
    'config': config_with_none
}
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
df_to_dataset Not yet run
unverified (harvested, no recorded run)
aruberts/TabTransformerTF/tabtransformertf/utils/preprocessing.py
code served (permissive licence) · get_code("ceddda2d1d949088")
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, 5 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 5 later) and 9 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
cos_sin 3 other papers
3 later
2209.15421 (2022-09)
TabDDPM: Modelling Tabular Data with Diffusion Models
Tiong_Flexible_Biometrics_Recognition_Bridging_the_Multimodality_Gap_through_Attention_Alignment_CVPR_2024_paper (2024 (year only))
Flexible Biometrics Recognition: Bridging the Multimodality Gap through Attention, Alignment and Prompt Tuning
2601.22816 (2026-01)
Cascaded Flow Matching for Heterogeneous Tabular Data with Mixed-Type Features
get_n_parameters 3 other papers
3 later
2209.15421 (2022-09)
TabDDPM: Modelling Tabular Data with Diffusion Models
2406.19380 (2024-06)
TabReD: Analyzing Pitfalls and Filling the Gaps in Tabular Deep Learning Benchmarks
2601.22816 (2026-01)
Cascaded Flow Matching for Heterogeneous Tabular Data with Mixed-Type Features
calculate_rmse 2 other papers
2 later
2209.15421 (2022-09)
TabDDPM: Modelling Tabular Data with Diffusion Models
2601.22816 (2026-01)
Cascaded Flow Matching for Heterogeneous Tabular Data with Mixed-Type Features
pack_config 2 other papers
2 later
2209.15421 (2022-09)
TabDDPM: Modelling Tabular Data with Diffusion Models
2601.22816 (2026-01)
Cascaded Flow Matching for Heterogeneous Tabular Data with Mixed-Type Features
unpack_config 2 other papers
2 later
2209.15421 (2022-09)
TabDDPM: Modelling Tabular Data with Diffusion Models
2601.22816 (2026-01)
Cascaded Flow Matching for Heterogeneous Tabular Data with Mixed-Type Features

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

Recently, Transformer-like deep architectures have shown strong performance on tabular data problems. Unlike traditional models, e.g., MLP, these architectures map scalar values of numerical features to high-dimensional embeddings before mixing them in the main backbone. In this work, we argue that embeddings for numerical features are an underexplored degree of freedom in tabular DL, which allows constructing more powerful DL models and competing with gradient boosted decision trees (GBDT) on some GBDT-friendly benchmarks (that is, where GBDT outperforms conventional DL models). We start by describing two conceptually different approaches to building embedding modules: the first one is based on a piecewise linear encoding of scalar values, and the second one utilizes periodic activations. Then, we empirically demonstrate that these two approaches can lead to significant performance boosts compared to the embeddings based on conventional blocks such as linear layers and ReLU activations. Importantly, we also show that embedding numerical features is beneficial for many backbones, not only for Transformers. Specifically, after proper embeddings, simple MLP-like models can perform on par with the attention-based architectures. Overall, we highlight embeddings for numerical features as an important design aspect with good potential for further improvements in tabular DL. The source code is available at https://github.com/yandex-research/tabular-dl-num-embeddings.

For agents

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

get_harvested_code_for_paper("2203.05556")
get_code_for_paper("2203.05556")
have("2203.05556")

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

curl https://syntology.ai/api/ran/2203.05556.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.05556.svg)](https://syntology.ai/paper/2203.05556)

Connect an agent — have() is free.