Skip to content
All videos
newsDatabricks·November 16, 2026

Distill a 120B LLM into a 0.8B Model on Databricks

Summary

LLM distillation teaches a smaller student model on answers from a larger teacher model using a narrow task at high volume, saving cost and improving speed while maintaining accuracy. This Databricks tutorial uses Foundation Model APIs, Serverless GPUs, MLflow evaluate, and Model Serving to build and deploy the smaller model end to end.

Summary generated by brickster.ai from the video transcript.

More from Databricks