CodeXEmbed: A Generalist Embedding Model Family for Multiligual and Multi-task Code Retrieval

By Ye Liu, Rui Meng, Shafiq Joty, S. Savarese, Cai-Ming Xiong, Ying-Bo Zhou, Semih Yavuz. 2024

Subfields: Information Retrieval & Search, Representation & Self-Supervised Learning, Code Generation & Program Synthesis

Citations in the corpus

• Recall Before Rerank: Benchmarking Deep Learning Models for Large-Scale Code-to-Code Retrieval (2026, unknown): “Both omit newer architectures (CodeSage-V2 [36], CodeXEmbed [16], Nomic Embed [23], Code Llama [28], Qwen [35]), cover at most two datasets per language, and disregard throughput.”
• SweRank: Software Issue Localization with Code Ranking (2025, unknown): “Liu et al. (2024b) incorporate multi-task contrastive data that includes code contest generation (Billah et al., 2024), code summarization (Son-takke et al., 2022), code completion (Liu et al., 2024a), code translation (Pan et al., 2024) and code agent conversation (Jin et al., 2024).”
• LLMs Are Also Effective Embedding Models: An In-depth Overview (2024, unknown): “They also used multi-query attention for efficient long-context embedding (8K tokens).”
• jina-embeddings-v4: Universal Embeddings for Multimodal Multilingual Retrieval (2025, unknown): “There are text embedding models designed for programming code [Liu et al., 2024], legal texts [Voyage AI, 2024], and other special domains.”