{"ID":2876588,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2508.21290","arxiv_id":"2508.21290","title":"Efficient Code Embeddings from Code Generation Models","abstract":"jina-code-embeddings is a novel code embedding model suite designed to retrieve code from natural language queries, perform technical question-answering, and identify semantically similar code snippets across programming languages. It makes innovative use of an autoregressive backbone pre-trained on both text and code, generating embeddings via last-token pooling. We outline the training recipe and demonstrate state-of-the-art performance despite the relatively small size of the models, validating this approach to code embedding model construction.","short_abstract":"jina-code-embeddings is a novel code embedding model suite designed to retrieve code from natural language queries, perform technical question-answering, and identify semantically similar code snippets across programming languages. It makes innovative use of an autoregressive backbone pre-trained on both text and code,...","url_abs":"https://arxiv.org/abs/2508.21290","url_pdf":"https://arxiv.org/pdf/2508.21290v1","authors":"[\"Daria Kryvosheieva\",\"Saba Sturua\",\"Michael Günther\",\"Scott Martens\",\"Han Xiao\"]","published":"2025-08-29T01:18:15Z","proceeding":"cs.CL","tasks":"[\"cs.CL\",\"cs.AI\",\"cs.IR\"]","methods":"[]","has_code":false}