层次嵌入融合用于检索增强的代码生成

Retrieval-augmented code generation often relies on large retrieved code snippets to condition the decoder, tying online inference costs to repository size and introducing noise from long contexts. A two-stage approach called Hierarchical Embedding Fusion (HEF) is proposed for repository representation in code completion. First, an offline cache compresses repository chunks into a reusable hierarchy of dense vectors using a small fuser model. Second, an online interface maps a small number of retrieved vectors into learned pseudo-tokens that are consumed by the code generator. This approach replaces thousands of retrieved tokens with a fixed pseudo-token budget while preserving access to repository-level information. HEF demonstrates strong performance on RepoBench and RepoEval with a 1.8B-parameter model.