Summary

Fine-tuning is the most technical piece of this book. With this little glimpse into this world, there is much to cover. Your data scientists and engineers will go deeper. When building production-ready systems, mix and match fine-tuned and generic models with internal software and third-party tools to balance speed of delivery, price, and performance (recall the saying, cheap, fast, or good, choose two). Innovative solutions have workflow steps that allow the solution to bail out if the AI isn’t performing, use a function to solve or address a specific problem, or use a more deterministic element to provide a robust solution. Injecting the suitable model and prompts for the correct part of a use case is one of the most critical decisions. Do this work before embarking on the fine-tuning approach.

Contribute to the process by helping define and improve these task flows through use case expertise, editing and improving prompts, creating, verifying, and editing fine...