Enhancing repository-level code completion with selective retrieval
70% faster. Amazon just showed how selective retrieval cuts code completion times while boosting accuracy.

Why it matters
Amazon Science's self-supervised method for repository-level code completion demonstrates a meaningful efficiency gain in AI-assisted development tools—a critical competitive advantage as enterprises scale AI coding assistants across engineering teams.
The key facts
5 to know70% speed improvement in code completion times
Increased accuracy alongside performance gains
Self-supervised learning approach for selective retrieval
Repository-level context utilization
Published by Amazon Science on October 17, 2024
Go to the source
Amazon Scienceamazon.science
Publisher excerpt: Self-supervised method for learning when to retrieve contextual information from a code repository speeds up code completion times by 70% while increasing accuracy.
