Do not copy and paste! Rewriting strategies for code retrieval
📰 ArXiv cs.AI
Learn rewriting strategies for code retrieval to improve embedding-based code search by mitigating overfitting to surface syntax
Action Steps
- Apply stylistic rephrasing to queries and corpora using LLMs to normalize style
- Use NL-enriched PseudoCode to add semantic information to code snippets
- Implement full Natural-Language transcription to capture high-level intent
- Evaluate the representational shift of each rewriting strategy to determine its effectiveness
- Compare the performance of different rewriting strategies to justify the use of per-query LLM calls
Who Needs to Know This
ML engineers and researchers working on code retrieval tasks can benefit from this knowledge to improve the accuracy of their models
Key Insight
💡 Rewriting strategies can help mitigate overfitting to surface syntax in embedding-based code retrieval
Share This
🚀 Improve code retrieval with rewriting strategies! 🤖
Key Takeaways
Learn rewriting strategies for code retrieval to improve embedding-based code search by mitigating overfitting to surface syntax
Full Article
Title: Do not copy and paste! Rewriting strategies for code retrieval
Abstract:
arXiv:2605.08299v1 Announce Type: cross Abstract: Embedding-based code retrieval often suffers when encoders overfit to surface syntax. Prior work mitigates this by using LLMs to rephrase queries and corpora into a normalized style, but leaves two questions open: how much representational shift helps, and when is the per-query LLM call justified? We study a hierarchy of three rewriting strategies: stylistic rephrasing, NL-enriched PseudoCode, and full Natural-Language transcription, under joint
Abstract:
arXiv:2605.08299v1 Announce Type: cross Abstract: Embedding-based code retrieval often suffers when encoders overfit to surface syntax. Prior work mitigates this by using LLMs to rephrase queries and corpora into a normalized style, but leaves two questions open: how much representational shift helps, and when is the per-query LLM call justified? We study a hierarchy of three rewriting strategies: stylistic rephrasing, NL-enriched PseudoCode, and full Natural-Language transcription, under joint
DeepCamp AI