live-high/my-text2sql

Install

pip install .

Setup database preprocessing based on CHESS

1. Run CHESS to preprocess the databases.
2. Copy the preprocessed databases to ./data/dev_databases

Setup Bird eval data

1. Copy the eval data to ./data/dev/dev.json
2. Copy the gold sql to ./data/dev/dev_gold.sql

Run pipeline

pipeline_type=basic
model=gpt-4o
temperature=1.0
bash run_main.sh  -p $pipeline_type -m $model [-t temperature]

Evaluation

pipeline_type=basic
model=gpt-4o
bash run_evaluation.sh  -p $pipeline_type -m $model