tablevault 0.2.0

Dataframe and File Transformation Execution and Versioning Tool.

Welcome to TableVault

TableVault is designed to manage data tables and artifacts in complex and dynamic data workflows. It promotes data reusability by capturing the full context of data transformations and ensuring atomic and transactional data states—transformations either clearly succeed or fail. It enhances data interoperability by easily connecting previous results with the configuration and input variables of subsequent transformations.

TableVault integrates with Python and can be used with popular data science libraries and tools, including Jupyter Notebooks, Pandas, NumPy, Transformers, and many others. The tool is particularly effective for workflows involving multiple dataframes, external artifacts (e.g., images, videos, documents), and large language model executions. TableVault is suited for agentic pipelines since one data-generating process can easily spin additional subprocesses.

Explore: Detailed Docs and Use Cases

Installation via pip:

pip install tablevault

This library is fully compatible with Python>=3.11.

Quick Start

Check out Basic Workflow for a simple generic setup, and our CoLab examples (1) Short Stories Q&A with OpenAI and (2) GritLM Embeddings from Scientific Abstracts for concrete use cases.