AI engineering glossary · LLM training alignment
What is safety fine-tuning?
Safety fine-tuning (also called Safety tuning): Post-training that teaches a model to refuse or safely handle harmful requests, using examples of unsafe prompts paired with appropriate refusals or safer alternatives, and often followed by preference tuning to refine the balance.
Safety fine-tuning: an example
Training on prompts asking for phishing scripts, each paired with a polite refusal and safe alternative.
Related terms
More LLM training alignment terms
- Scalable oversight
- Scaling laws
- Self-instruct
- Self-rewarding language models
- Self-Taught Reasoner
- Sequence packing
- Sequence-level distillation
- Simple preference optimisation
Where you learn it
Safety fine-tuning is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.