spider-dataset

Text-to-SQL in the Wild: A Naturally-Occurring Dataset Based on Stack Exchange Data

执行验证驱动的 Text-to-SQL DPO：零人工标注，Spider test 留出集 82.25% → 85.61% (+3.35, p=0.0000)

Schema-aware Text-to-SQL Transformer

🧠 Schema-Aware Natural Language to SQL Agent with Fine-tuned T5 Models

Multi-agent NL→SQL pipeline built on LangGraph. Converts natural language questions into validated SQL queries using parallel discovery, generation, and validation agents. Supports multi-model mixing (GPT-4o + Gemini) via OpenRouter. Tested on Spider dataset.

🍂 Text-to-SQL with multi-step agent reasoning on the SPIDER dataset — CS175 NLP project @ UCI

Change ur natural language queries about a database to actual SQL queries with 60% and 33% execution accuracy.

Natural language → Entity Framework (ORM) queries via a fine-tuned LLaMA 3, evaluated on the Spider benchmark. MSc thesis.

Fine-tuned T5-base transformer that translates natural language questions into SQL queries, trained on the Spider dataset. Includes schema-aware input formatting, full training/eval pipeline, and exact-match evaluation beyond token accuracy.