Fine-Tune, or Don't
A Practical Decision Process for Customizing AI Models
A book by Alpesh Nakrani
Subscribe to get the complete book. Your PDF downloads immediately.
- Pages
- 152
- Chapters
- 17
- Format
- PDF · 29.4 MB
What's in the book
Fine-tuning is reached for too early and dismissed too fast. This deep dive is a decision procedure: when prompting, retrieval, or routing solves it cheaper, and the narrow cases where changing the weights is genuinely the right tool.
Fine-tuning is the first instinct and usually the wrong one. A sober procedure for deciding when weights should change.
Table of contents
PDF page- FMFront Matter: Fine-Tune, or Don't—
- INTIntroduction: The Wrong Operation, Performed Well16
- 01The Support Bot That Knew the Old Product22
- 02What Fine-Tuning Actually Changes30
- 03Five False Diagnoses36
- 04The Customization Menu and a Decision Tree43
- 05Format, Behavior, and the Shape of a Repeated Task50
- 06House Style, Domain Phrasing, and Tool Discipline57
- 07Specializing Small Models and Distilling Down63
- 08Demonstrations, Corrections, and Preferences70
- 09Labels, Disagreement, and Coverage77
- 10Contamination, Leakage, and the Splits That Save You84
- 11Synthetic Data: When It Helps, When It Poisons90
- 12SFT, LoRA, and QLoRA in Practical Terms97
- 13Preference Tuning, DPO, and Distillation104
- 14What to Fine-Tune: Generators, Retrievers, and Routers111
- 15Baselines, Regression Walls, and the Release Gate118
- 16Versioning, Lineage, Drift, and Retirement125
- 17Ten Playbooks for the Decision Meeting133
- AAppendix A: Back Matter142
The full chapters, illustrations, and reference material are included in the downloadable PDF.
Fine-Tune, or Don't
The complete edition,
ready to read offline.
Get your free copy.
Subscribe to Alpesh’s book and writing updates. Unsubscribe anytime.
Your download is ready.
You’re subscribed. Your PDF download should start automatically.
Download PDF againIf it doesn’t start, use the button above.
