Environment:Gretelai Gretel synthetics Python Base Environment

Knowledge Sources	gretel-synthetics
Domains	Infrastructure, Data_Science
Last Updated	2026-02-14 19:00 GMT

Overview

Python 3.9+ base environment with core data science libraries (NumPy, Pandas, SDV, RDT, SentencePiece) shared across all gretel-synthetics workflows.

Description

This environment defines the common Python runtime and base dependencies required by all gretel-synthetics workflows. It includes data manipulation libraries (NumPy, Pandas), the SDV/RDT ecosystem for tabular data handling, SentencePiece for tokenization, and utility libraries. This base layer is extended by the TensorFlow or PyTorch environments for specific model types.

Usage

This environment is a prerequisite for all workflows. The TensorFlow GPU Environment and PyTorch CUDA Environment both build on top of this base. Install this as the minimum requirement to use any gretel-synthetics functionality.

System Requirements

Category	Requirement	Notes
OS	Linux, macOS, Windows	Per setup.py classifiers (POSIX Linux, MacOS, Microsoft Windows)
Python	>= 3.9	Specified in setup.py `python_requires=">=3.9"`
Disk	Varies	Depends on training data size; model checkpoints stored locally

Dependencies

Python Packages (Core)

`category-encoders` == 2.2.2
`joblib` == 1.4.2
`numpy` >= 1.18.0, < 1.24
`packaging` < 22.0
`pandas` >= 1.1.0, < 2
`rdt` >= 1.2, < 1.3
`sdv` >= 0.17, < 0.18
`sentencepiece` == 0.2.0
`smart_open` >= 2.1.0, < 6.0
`tqdm` < 5.0

Python Packages (Utils Extra)

`dython` == 0.7.0
`scikit-learn` >= 0.24
`scipy` >= 1.4.1, < 1.11

Python Packages (Testing)

`pytest` == 6.1.2
`pytest-cov` == 2.11.1
`faker` == 20.1.0
`flake8` == 4.0.1
`pylint` == 2.14.3

Credentials

No API keys or credentials are required. All data is loaded from local files or in-memory DataFrames.

Quick Install

# Base install (core dependencies only)
pip install gretel-synthetics

# With utils extras (statistical evaluation tools)
pip install gretel-synthetics[utils]

# Full install with all model backends
pip install gretel-synthetics[all]

Code Evidence

Python version requirement from `setup.py:52`:

python_requires=">=3.9",

Base requirements loaded from `setup.py:32-37`:

base_reqs = reqs("requirements/base.txt")
utils_reqs = reqs("requirements/utils.txt")
torch_reqs = reqs("requirements/torch.txt")
tf_reqs = reqs("requirements/tensorflow.txt")

Extras defined in `setup.py:54-61`:

extras_require={
    "all": [req for reqs in all_reqs for req in reqs],
    "utils": utils_reqs,
    "test": test_reqs,
    "torch": torch_reqs,
    "tensorflow": tf_reqs,
    "docs": doc_reqs,
},

Version-pinned base dependencies from `requirements/base.txt`:

category-encoders==2.2.2
joblib==1.4.2
numpy>=1.18.0,<1.24
packaging<22.0
pandas>=1.1.0,<2
rdt>=1.2,<1.3
sdv>=0.17,<0.18
sentencepiece==0.2.0
smart_open>=2.1.0,<6.0
tqdm<5.0

Common Errors

Error Message	Cause	Solution
`ImportError: No module named 'sentencepiece'`	SentencePiece not installed	`pip install sentencepiece==0.2.0`
NumPy compatibility errors	NumPy >= 1.24 breaking changes	Pin NumPy: `pip install numpy>=1.18.0,<1.24`
Pandas API deprecation warnings	Pandas >= 2.0 breaking changes	Pin Pandas: `pip install pandas>=1.1.0,<2`
`ImportError: No module named 'sdv'`	SDV ecosystem not installed	`pip install sdv>=0.17,<0.18 rdt>=1.2,<1.3`

Compatibility Notes

NumPy: Strictly pinned to < 1.24 due to API changes in newer versions.
Pandas: Strictly pinned to < 2.0 due to breaking API changes.
SDV/RDT: Pinned to specific minor version ranges (SDV 0.17.x, RDT 1.2.x) due to tight coupling with internal APIs. The codebase includes monkey-patches for RDT FloatFormatter rounding bugs.
SentencePiece: Pinned to exact version 0.2.0 for consistent tokenizer behavior.
scipy: Utils extra requires scipy < 1.11 for statistical evaluation tools.

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment