BWhat changed
Trained and released code models on deduplicated project-level repositories with dependency-aware code organization and fill-in-the-middle training, documenting a reusable code-generation and infilling pipeline.
Scientific scope
Trained and released code models on deduplicated project-level repositories with dependency-aware code organization and fill-in-the-middle training, documenting a reusable code-generation and infilling pipeline. Project-level processing and infilling have predecessors; not invention of transformers or all code completion. Original models appeared in2023 and this report is2024. V2 continues an existing MoE/GRPO pipeline and is not a separate acquisition merely for more tokens, languages or context.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Project-level processing and infilling have predecessors; not invention of transformers or all code completion. Original models appeared in2023 and this report is2024. V2 continues an existing MoE/GRPO pipeline and is not a separate acquisition merely for more tokens, languages or context.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
- DeepSeek-Coder: When the Large Language Model Meets Programming -- The Rise of Code Intelligence
- Guo et al.: DeepSeek-Coder project-level code research resource — author/institution implementation
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction