DAY 60 / 210
Fundamentals of LLM Fine-Tuning
This first day of phase-2-finetune establishes why parameter-efficient adaptation matters for production models at StartupTribunal scale. It creates the conceptual baseline needed before any code-level work on the existing Maku pipeline.
⏱ 45 min target📝 3 quiz Qs
Resources
- 25 minreadingHugging FacePEFT: Parameter-Efficient Fine-Tuning of Billion-Scale Models
Overview + Quicktour sections
- 20 min
Deliverable
200-word journal entry comparing full fine-tuning vs LoRA on the Maku brief-generation task
Quiz · 3 questions
1. Which statement best describes the primary goal of LoRA?
2. Name one concrete risk of full fine-tuning that parameter-efficient methods mitigate.
3. In one sentence, explain why a startup building legal tools might prefer LoRA over full fine-tuning.