Welcome to ReaL’s documentation!

🚀 Get Started 🚀

For users new to ReaL, we recommend starting with the Quickstart section to learn how to run simple experiments on a local node. If you have multiple nodes available, please read the Setting Up Distributed Experiments section to learn how to run experiments on a cluster. These tutorials cover the basic usage of the implemented algorithms in ReaL, including SFT, Reward Modeling, DPO, and PPO, and do not require understanding the code structure.

For advanced users, we recommend proceeding to the Customization section to learn how to customize the algorithms and models in ReaL. This requires an understanding of how an algorithm and its experiment configuration are defined in ReaL (i.e., as a dataflow graph), but understanding the system-wide implementation (e.g., model workers) is not mandatory.

For potential developers, please refer to the Implementation Details and the Code Architecture sections for a deeper understanding of the system architecture.

Besides these illustrations, we present the reference manual of various configuration objects in the Configurations section, and a brief overview of the system architecture in the Introduction section.

⭐ Contents ⭐