tabqa 0.1.3

This Python package provides functions to convert natural language to SQL queries using a pre-trained language model.

tabqa: Natural Language to SQL Query Converter

Overview

This Python library enables the conversion of natural language queries into SQL queries. It simplifies the process of querying databases by allowing users to express their queries in everyday language.

Installation

You can install the package via pip:

pip install tabqa

Quick Start

from tabqa import sql_model, generate_schema

# Define the natural language question
question = "Count Number of products"

# Path to the SQL file containing the database schema
file_path = "schema.sql"  #

# Initialize the SQL model
model = sql_model()

# Generate the SQL schema from the natural language question
result = generate_schema(question, file_path, model)

# Print the generated SQL query
print(result)

Parameters for sql_model function :

1. model_name: str = 'defog/sqlcoder-7b',
2. cache_dir: str = './',
3. device_map: str = 'auto',
4. trust_remote_code: bool = True,
5. torch_dtype: torch.dtype = torch.float16,
6. use_cache: bool = True,
7. force_download: bool = False,
8. resume_download: bool = False,
9. output_loading_info: bool = False,
10. local_files_only: bool = False,

Model available :

1. defog/sqlcoder-7b (lowest Accuracy)
2. defog/sqlcoder-34b-alpha (Medium Accuracy)
3. defog/sqlcoder-70b-alpha (Highest Accuracy)

Warning: 'defog/sqlcoder-34b-alpha' and 'defog/sqlcoder-70b-alpha' demand high GPU and RAM resources. Please don't use on local machines.

How to Use

1. Import the necessary functions from tabqa.
2. Define your natural language question.
3. Provide the path to the SQL file containing your database schema.
4. Initialize the SQL model.
5. Use generate_schema() function to convert the natural language question into an SQL query.
6. Print or use the generated SQL query as needed.

Contribution

Contributions are welcome! If you encounter any issues or have suggestions for improvements, feel free to open an issue or submit a pull request.

License

This project is licensed under the MIT License.