Arctic-SnowCoder: Demystifying High-Quality Data in Code Pretraining
Research paper by Yuxiang Wei and collaborators. Published 2024-09-03. Topics: code domain, data-efficient, pretraining, BERT-style quality annotator, synthetic data.
Resource information
- Authors
- Yuxiang Wei, Hojae Han, Rajhans Samdani
- Research topics
- code domain, data-efficient, pretraining, BERT-style quality annotator, synthetic data, BigCodeBench, StarCoderBase-3B, StarCoder2-3B, HumanEval+, function-level code generation
- Publication date
- 2024-09-03
- Publication
- arXiv
- Last record update
- Nov 3, 2026
