Tag
This paper presents an end-to-end pipeline for adapting LLMs to low-resource programming languages, using Pharo as a case study, achieving superior code completion with small models suitable for real-time IDE support.