This GitHub repository serves as a comprehensive cookbook for building, deploying, and fine-tuning generative AI models using the Fireworks platform. It offers specialized training recipes for various optimization techniques and an automated agent to handle the full model lifecycle from data inspection to deployment.
Highlights
Provides ready-to-run training recipes for RL (GRPO, DAPO), preference optimization (DPO, ORPO), and supervised fine-tuning (SFT)
Features a Fireworks Agent that automates the end-to-end fine-tuning lifecycle via natural language instructions
Includes a structured repository with dedicated directories for training SDKs, utilities, and worked examples
Offers a verifier for correctness validation and a live React viewer