Welcome

Phinity is a synthetic data generation SDK designed to help you create high-quality, verifiable datasets for LLM development and evaluation.

Github

Jump right in

Quick Start

Generate your first dataset

Quickstart

In-domain SFT

Generate QA pairs grounded in your domain context and documents

In-domain SFT

RAG benchmark

Generate single-hop and multi-hop retrieval test sets with QA pairs from your documents.

Building a custom RAG benchmark

Read about synthetic data best practices (coming soon)

How to generate diverse, high quality data for your domain without collapse

Synthetic data guide (coming soon)