Freedom251 › AI Synthetic Data Generators

PluRel

Paid

What PluRel is

PluRel is a free, open-source framework for synthesising relational and tabular data, aimed at relational foundation-model research and complex multi-table databases. It models table schemas with directed graphs, inter-table links with bipartite graphs, and feature distributions through conditional causal mechanisms. Configuration controls table layouts, table and row counts, column counts, and structural causal-model parameters. An optional schema_file setting allows generation from an existing SQL schema. Generated columns can be numeric, categorical, or boolean, with trends, cycles, and fluctuations available for temporal patterns. Dataset objects can be made compatible with RelBench. A multiprocessing script can generate databases in parallel, with database and process counts configurable. Installation uses pip install plurel and requires Python 3.12 or later. The public repository lists an MIT license. The project reports that greater synthetic database diversity can improve generalisation, but also states synthetic pretraining alone is insufficient for robust zero-shot transfer and continued pretraining on real data is critical.

Who it suits

PluRel suits researchers and data practitioners working with synthetic relational databases and relational foundation models. It is intended for structured multi-table data rather than unstructured data.

What it does well

Open-source framework with MIT license
Supports SQL schema-based generation
Numeric, categorical, and boolean features
Temporal trends, cycles, and fluctuations
Parallel database generation available

What to know first

Requires Python 3.12 or later
Synthetic pretraining alone lacks robust zero-shot transfer
Focused on relational and tabular data
Unstructured data is not supported

Getting started

Open the PluRel project website
Install the library with pip install plurel
Use Python 3.12 or later
Configure table, row, column, and structural causal settings
Optionally provide an existing SQL schema through schema_file
Set database and process counts for parallel generation

Verdict

PluRel provides configurable synthetic relational data generation for research and data-driven work. Its own project notes that real-data continued pretraining remains important for distribution alignment.

Questions

Is PluRel free?

Yes. PluRel is free and open source, and its public GitHub repository lists an MIT license.

What data types can it generate?

Generated feature columns support numeric, categorical, and boolean values. Temporal behavior can include trends, cycles, and fluctuations.

Can it work from an existing SQL schema?

Yes. The optional schema_file setting lets PluRel generate data from an existing SQL schema.

Is the output compatible with RelBench?

Generated dataset objects can be made compatible with RelBench.

Does synthetic pretraining transfer robustly without real data?

The project states that synthetic pretraining alone is insufficient for robust zero-shot transfer and that continued pretraining on real data is critical for distribution alignment.

Facts from PluRel's own pages

Compatibility: Generated datasets can be made compatible with RelBench. · star-project.stanford.edu, 4 Oct 2026
Configurable output: Its configuration controls table layouts, table and row counts, column counts, and structural causal model parameters. · star-project.stanford.edu, 8 Oct 2026
Data types and patterns: Feature generation supports numeric, categorical, and boolean values, with temporal trends, cycles, and fluctuations. · star-project.stanford.edu, 4 Oct 2026
Feature types: Generated feature columns support numeric, categorical, and boolean types. · star-project.stanford.edu, 8 Oct 2026
Generation stages: It models database schemas with directed graphs, inter-table primary–foreign key links with bipartite graphs, and feature distributions with conditional causal mechanisms. · star-project.stanford.edu, 8 Oct 2026
Installation: The library is installed with pip install plurel and requires Python 3.12 or later. · github.com, 8 Oct 2026
Installation requirement: The library is installed with pip and requires Python 3.12 or later. · github.com, 4 Oct 2026
Intended context: The project targets relational foundation model research and data-driven work with complex multi-table databases. · star-project.stanford.edu, 4 Oct 2026
Intended use: The project describes PluRel as a framework for research on relational foundation models and synthetic relational database generation. · arxiv.org, 8 Oct 2026
Large-scale generation: A multiprocessing script can generate databases in parallel, with the number of databases and processes configurable. · github.com, 4 Oct 2026
License: The GitHub repository identifies the project license as MIT. · github.com, 8 Oct 2026
Parallel generation: A multiprocessing-based script can generate databases in parallel, with the process count configurable. · github.com, 8 Oct 2026
Purpose: PluRel is an open-source library for synthesizing relational and tabular data. · github.com, 8 Oct 2026
RelBench compatibility: The project says its generated dataset objects are compatible with RelBench. · github.com, 8 Oct 2026
Research results: The project reports that synthetic database diversity and pretraining token count both show power-law scaling in relational foundation model pretraining loss. · star-project.stanford.edu, 8 Oct 2026
Schema input: It can generate data from an existing SQL schema using the optional schema_file setting. · star-project.stanford.edu, 4 Oct 2026
SQL schemas: The configuration can optionally use an existing SQL schema as input. · star-project.stanford.edu, 8 Oct 2026
Temporal patterns: Feature generation can model temporal correlations using trend, cycle, and fluctuation components. · star-project.stanford.edu, 8 Oct 2026
Transfer limitation: The project states that synthetic pretraining alone is insufficient for robust zero-shot transfer and that continued pretraining on real data is critical for distribution alignment. · star-project.stanford.edu, 8 Oct 2026

Where PluRel stands on Freedom251

AI Synthetic Data Generators#23 of 25