magpie-align/magpie
Research-grade pipeline for generating high-quality synthetic alignment data by prompting aligned LLMs with their native pre-query templates.
Not currently ranked — collecting fresh signals.
star history
Magpie generates training data for LLM alignment and fine-tuning by exploiting the prompt templates of aligned LLMs to produce both user queries and model responses without manual annotation. It requires no seed questions or prompt engineering, making synthetic data generation more scalable. The project provides generated datasets (1M+ examples) from models like Llama-3.3, QwQ, and Skywork-o1, specifically targeting supervised fine-tuning workflows.
Frequently asked
- What is magpie-align/magpie?
- Research-grade pipeline for generating high-quality synthetic alignment data by prompting aligned LLMs with their native pre-query templates.
- Is magpie open source?
- Yes — magpie-align/magpie is open source, released under the MIT license.
- What language is magpie written in?
- magpie-align/magpie is primarily written in Python.
- How popular is magpie?
- magpie-align/magpie has 874 stars on GitHub.
- Where can I find magpie?
- magpie-align/magpie is on GitHub at https://github.com/magpie-align/magpie.