Adapt and Decompose: Efficient Generalization of Text-to-SQL via Domain Adapted Least-To-Most Prompting. (arXiv:2308.02582v3 [cs.CL] UPDATED)

Cross-domain and cross-compositional generalization of Text-to-SQL semantic
parsing is a challenging task. Existing Large Language Model (LLM) based
solutions rely on inference-time retrieval of few-shot exemplars from the
training set to synthesize a run-time prompt for each Natural Language (NL)
test query. In contrast, we devise an algorithm which performs offline sampling
of a minimal set-of few-shots from the training data, with complete coverage of
SQL clauses, operators and functions, and maximal domain coverage within the
allowed token length. This …

arxiv contrast inference language language model large language large language model least llm natural natural language parsing prompt prompting query retrieval semantic set solutions sql test text text-to-sql training