Trinity-RFT: A General-Purpose and Unified Framework for Reinforcement Fine-Tuning of Large Language Models

Xuchen Pan, Yanxi Chen, Yushuo Chen, yuchang, Daoyuan Chen, garyzhang, Yuexiang Xie, Dylan Huang, Yilei Zhang, Dawei Gao, Yaliang Li, Bolin Ding, Jingren Zhou

Trinity-RFT: A General-Purpose and Unified Framework for Reinforcement Fine-Tuning of Large Language Models: 10 upvotes on Hugging Face Daily Papers, #27 of 47 papers on 2025-05-26. Day-by-day upvote history.

Trinity-RFT is a general-purpose, flexible and scalable framework designed for reinforcement fine-tuning (RFT) of large language models. It is built with a decoupled design, consisting of (1) an RFT-core that unifies and generalizes synchronous/asynchronous, on-policy/off-policy, and online/offline modes of RFT, (2) seamless integration for agent-environment interaction with high efficiency and robustness, and (3) systematic data pipelines optimized for RFT. Trinity-RFT can be easily adapted for diverse application scenarios, and serves as a unified platform for exploring advanced reinforcement learning paradigms. This technical report outlines the vision, features, design and implementations of Trinity-RFT, accompanied by extensive examples demonstrating the utility and user-friendliness of the proposed framework.

Paper page on Hugging Face · arXiv

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.