Ablation and Shapley analysis of NatSQL, GAZP synthetic fine-tuning, token preprocessing, and a novel beam reranker on SmBoP and RASAT show that component interactions—not full stacking—drive the best lightweight NL2SQL gains.
RASAT : Integrating Relational Structures into Pretrained S eq2 S eq Model for Text-to- SQL
1 Pith paper cite this work, alongside 78 external citations. Polarity classification is still indexing.
1
Pith paper citing it
78
external citations · OpenAlex
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The Nuts and Bolts of Natural Language to SQL Translation: A Systematic Analysis of Model Pipeline Optimisation Approaches and their Interactions
Ablation and Shapley analysis of NatSQL, GAZP synthetic fine-tuning, token preprocessing, and a novel beam reranker on SmBoP and RASAT show that component interactions—not full stacking—drive the best lightweight NL2SQL gains.