Capstone: fine-tune a model for your own task, evaluate it and write the report
A complete project brief for the Fine-tuning course: choose a real task, build the baseline, create and clean data, train with LoRA, evaluate with intervals and a regression suite, count the cost, and deliver a short report. Includes a rubric, a week-by-week plan, deliverables and a break-even calculator for deciding whether the fine-tune is worth running.
You now have the pieces: when to fine-tune, how to prepare data, how LoRA and the training recipe work, how much GPU you need, and how to run and evaluate a real fine-tune. The capstone ties them together on a task that matters to you. A good capstone proves a point with evidence, so the deliverable is a short, honest report as much as a model.
Pro content
Keep reading with Inference Academy Pro
Unlock the rest of this lesson, including 1 animated diagram, plus every advanced lesson, hands-on lab, the larger playground models and completion certificates.