The process of further training a pre-trained machine learning model on a specific dataset to adapt its general capabilities to a particular task or domain.
Overview
- Cursor’s Dev Relations rep Lee Robinson confirmed that Composer 2 started from an open-source base, with only a quarter of the compute spent on the final model coming from the base and the rest from their own training. (Source: Lee Robinson (Cursor Dev Relations), via AI Daily Brief, 2026-08-24)