SQL-Encoder: Improving NL2SQL In-Context Learning Through a Context-Aware Encoder
This addresses the problem of selecting effective examples for in-context learning in NL2SQL tasks, offering incremental improvements over existing embedding models.
The paper tackles the challenge of detecting structural similarity between queries for in-context learning in NL2SQL by proposing a context-aware encoder, which improves downstream performance by 1-8% across models like GPT-3.5-turbo and CodeLlama.
Detecting structural similarity between queries is essential for selecting examples in in-context learning models. However, assessing structural similarity based solely on the natural language expressions of queries, without considering SQL queries, presents a significant challenge. This paper explores the significance of this similarity metric and proposes a model for accurately estimating it. To achieve this, we leverage a dataset comprising 170k question pairs, meticulously curated to train a similarity prediction model. Our comprehensive evaluation demonstrates that the proposed model adeptly captures the structural similarity between questions, as evidenced by improvements in Kendall-Tau distance and precision@k metrics. Notably, our model outperforms strong competitive embedding models from OpenAI and Cohere. Furthermore, compared to these competitive models, our proposed encoder enhances the downstream performance of NL2SQL models in 1-shot in-context learning scenarios by 1-2\% for GPT-3.5-turbo, 4-8\% for CodeLlama-7B, and 2-3\% for CodeLlama-13B.