SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2210.06280 · 2022

Language Models are Realistic Tabular Data Generators

arXiv · PDF · Open in the Atlas

Code that ran

6 of 6 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 6 have no contract verdict; 0 failures of our instrument; 0 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/2210.06280.json · badge: /api/ran/2210.06280.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
kathrinse/be_great canonical (classifier label)
classifier basis not recorded on the link
6 ran · 0 no recorded run
FunctionStatusWhere it lives
apply_null_probabilities Ran
ran on a synthesized input
kathrinse/be_great/be_great/great_mock_datasets.py
code served (permissive licence) · get_code("83e59932f7b41b20")
How it ran

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

import numpy as np
import pandas as pd
import typing as tp

# Create a sample DataFrame
df = pd.DataFrame({
    'age': [25, 30, 35, 40, 45],
    'salary': [50000.0, 60000.0, 70000.0, 80000.0, 90000.0],
    'category': ['A', 'B', 'A', 'C', 'B']
})

# Create col_stats dictionary with null probabilities
col_stats = {
    'age': {
        'type': 'numeric',
        'null_prob': 0.2
    },
    'salary': {
        'type': 'numeric',
        'null_prob': 0.1
    },
    'category': {
        'type': 'categorical',
        'null_prob': 0.15
    }
}

# Create a random generator for reproducibility
rng = np.random.default_rng(seed=42)

ARGS = {
    'df': df,
    'col_stats': col_stats,
    'rng': rng
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
build_effective_conditions Ran
ran on a synthesized input
kathrinse/be_great/be_great/great_mock_datasets.py
code served (permissive licence) · get_code("d463c7b56fcd4ea6")
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 typing as tp

_CAT_NO_MATCH_SENTINEL = '__GREAT_NO_MATCH__'

# Construct col_stats dict with schema information
col_stats = {
    'age': {
        'type': 'numerical',
        'min': 0,
        'max': 100
    },
    'income': {
        'type': 'numerical',
        'min': 20000,
        'max': 500000
    },
    'category': {
        'type': 'categorical',
        'values': ['A', 'B', 'C']
    },
    'status': {
        'type': 'categorical',
        'values': ['active', 'inactive']
    }
}

# Construct user_conditions dict with some constraints
user_conditions = {
    'age': '>= 18',
    'category': "== 'A'"
}

ARGS = {
    'col_stats': col_stats,
    'user_conditions': user_conditions
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
build_few_shot_prefix Ran
ran on a synthesized input
kathrinse/be_great/be_great/great_mock_datasets.py
code served (permissive licence) · get_code("47258ece5fbeda33")
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 tp

# Construct valid arguments for build_few_shot_prefix

# The function expects:
# 1. examples: tp.Sequence[tp.Mapping[str, tp.Any]] - a sequence of dicts
# 2. columns: tp.Sequence[str] - a sequence of column names

examples = [
    {
        'name': 'Alice',
        'age': 30,
        'city': 'New York'
    },
    {
        'name': 'Bob',
        'age': 25,
        'city': 'London'
    },
    {
        'name': 'Charlie',
        'age': None,  # Test None value handling
        'city': 'Paris'
    }
]

columns = ['name', 'age', 'city']

ARGS = {
    'examples': examples,
    'columns': columns
}
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
build_trie Ran
ran on a synthesized input
kathrinse/be_great/be_great/great_constrained.py
code served (permissive licence) · get_code("7618dd48226e8d77")
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 TokenPrefixTrie. 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 tp

class MockTokenizer:
    """A minimal tokenizer that encodes strings to token IDs."""
    
    def __init__(self):
        self.vocab = {
            ' ': 0,
            'a': 1,
            'b': 2,
            'c': 3,
            'd': 4,
            'e': 5,
            'f': 6,
            'r': 7,
            'e': 5,
            'd': 4,
        }
        self.token_counter = 100
    
    def encode(self, text: str, add_special_tokens: bool = True) -> tp.List[int]:
        """Simple encoding: map each character to a unique token ID."""
        tokens = []
        for char in text:
            if char == ' ':
                tokens.append(0)
            elif char == 'a':
                tokens.append(1)
            elif char == 'b':
                tokens.append(2)
            elif char == 'c':
                tokens.append(3)
            elif char == 'd':
                tokens.append(4)
            elif char == 'e':
                tokens.append(5)
            elif char == 'f':
                tokens.append(6)
            elif char == 'r':
                tokens.append(7)
            else:
                # Assign new token ID for unseen characters
                self.token_counter += 1
                tokens.append(self.token_counter - 1)
        return tokens

valid_values = ["red", "blue", "green"]
tokenizer = MockTokenizer()

ARGS = {
    "valid_values": valid_values,
    "tokenizer": tokenizer,
}
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 TokenPrefixTrie — a TokenPrefixTrie output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a TokenPrefixTrie 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
enumerate_valid_values Ran
ran on a synthesized input
kathrinse/be_great/be_great/great_constrained.py
code served (permissive licence) · get_code("68f0043fac80bd8a")
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 typing as tp

# Test case 1: numeric column with >= operator
numeric_col_stats = {
    'type': 'numeric',
    'min': 0.0,
    'max': 100.0,
    'precision': 2
}

ARGS = {
    'col_name': 'age',
    'op': '>=',
    'threshold': 50.0,
    'col_stats': numeric_col_stats,
    'float_precision': 2
}
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 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
parse_condition Ran
ran on a synthesized input
kathrinse/be_great/be_great/great_constrained.py
code served (permissive licence) · get_code("0781cc4ea48a2608")
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 re
import typing as tp

_CONDITION_RE = re.compile('^\\s*(>=|<=|!=|==|>|<)\\s*(\'([^\']*)\'|\\"([^\\"]*)\\"|(.+?))\\s*$')

def parse_condition(condition_str: str) -> tp.Tuple[str, tp.Union[float, str]]:
    """Parse a condition string into (operator, threshold).

    Args:
        condition_str: e.g. ``">= 30"``, ``"!= 'New York'"``

    Returns:
        Tuple of (operator_string, threshold_value).
        threshold_value is float for numeric, str for quoted strings.

    Raises:
        ValueError: If the condition string cannot be parsed.
    """
    m = _CONDITION_RE.match(condition_str)
    if not m:
        raise ValueError(f"""Cannot parse condition: {condition_str!r}. Expected format like ">= 30", "!= 'New York'", "< 50000".""")
    op = m.group(1)
    value_str = m.group(3) or m.group(4) or m.group(5)
    value_str = value_str.strip()
    try:
        threshold = float(value_str)
    except ValueError:
        threshold = value_str
    return (op, threshold)


ARGS = {
    "condition_str": ">= 30"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 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 6 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

Tabular data is among the oldest and most ubiquitous forms of data. However, the generation of synthetic samples with the original data's characteristics remains a significant challenge for tabular data. While many generative models from the computer vision domain, such as variational autoencoders or generative adversarial networks, have been adapted for tabular data generation, less research has been directed towards recent transformer-based large language models (LLMs), which are also generative in nature. To this end, we propose GReaT (Generation of Realistic Tabular data), which exploits an auto-regressive generative LLM to sample synthetic and yet highly realistic tabular data. Furthermore, GReaT can model tabular data distributions by conditioning on any subset of features; the remaining features are sampled without additional overhead. We demonstrate the effectiveness of the proposed approach in a series of experiments that quantify the validity and quality of the produced data samples from multiple angles. We find that GReaT maintains state-of-the-art performance across numerous real-world and synthetic data sets with heterogeneous feature types coming in various sizes.

For agents

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

get_harvested_code_for_paper("2210.06280")
get_code_for_paper("2210.06280")
have("2210.06280")

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

curl https://syntology.ai/api/ran/2210.06280.json

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

[![Syntology run record](https://syntology.ai/api/ran/2210.06280.svg)](https://syntology.ai/paper/2210.06280)

Connect an agent — have() is free.