Definition·
Optimization

Distillation

A process that trains a small "student" model to mimic a large "teacher" model, keeping quality at lower cost.

Detailed explanation

Distillation transfers knowledge from a large model to a smaller one. The student is trained on the teacher's outputs (logits, responses). The result: faster, cheaper-to-serve models, perfect for production.

Examples

DistilBERT derived from BERT
Small 'Instruct' models derived from GPT-4
Open-source models distilled from proprietary ones

Frequently asked questions

Distillation or fine-tuning?

Distillation = transfer from big to small. Fine-tuning = adapt a model to a specific task/data. The two combine.

Related terms

Last updated: 7/15/2026

Talent AI

Turn theory into practice

Post a mission or join the community of top AI, Data and Machine Learning experts.