deepsearchkit 1.0.1

A small, fast, and easy package for semantic searching using artificial intelligence.

Note: This project is still a work-in-progress. It works, but it doesn't have many features yet. Please suggest features, report problems (bugs should go in Issues), or just say hi in Discussions!

:mag: DeepSearchKit

A small, fast, and easy package that allows easy semantic searching using :brain: artificial intelligence. It is based on and similar to the closed-source DeepSearch system.

:computer: Installation

You can install the latest stable version from the registry:

pip3 install deepsearchkit

You can install the very latest version directly from the Git repository, however certain features may not work:

pip3 install git+https://github.com/fakerybakery/deepsearchkit

:white_check_mark: Features

• CPU, CUDA, and MPS support (enhanced GPU acceleration)!
• Simple usage

:newspaper: Usage

Documentation is available here.

:thought_balloon: Todo

• Integrate DeepSearch into DeepSearchKit
  • Open-source DeepSearch
• Add Web Interface (from DeepSearch)
• Add document search demo
  • Add document chat demo
• Add upsert feature (txtai#251)
• Add more data support, e.g. parquet, MySQL/hosted DBs
• Custom prompt/data format for multiple columns in JSON
• Custom progress callback for indexing
• Make some example projects
  • Chat with a folder using open-sourced conversational models
  • Search an entire directory
• Allow easy publishing with a .dskpkg file - compressed DeepSearchKit package that includes the index, the data, and some attributes (name, author, license, etc)
  • Maybe in the future: "DSK Hub" - hub for DSK packages

:memo: Credits

We would like to thank the authors of the following open source projects:

• sentence-transformers
• txtai

:briefcase: Disclaimer/Agreement (Read BEFORE you contribute!)

In every new issue/PR, make sure to include "I agree to the disclaimer!"

By using/contributing to this software, you agree to the agreement.