1. Project Overview

Ask Anything

An open-source, enterprise-grade, on-premises knowledge base question-answering solution designed to support Q&A over files in any format or databases.

What Is QAnything?

QAnything (Question and Answer based on Anything) is an on-premises knowledge base question-answering system designed to support files in any format or databases. It can be installed and used offline.

You can upload local files in any format to get accurate, fast, and reliable question-answering experiences.

Currently supported formats:  PDF(pdf), Word(docx), PPT(pptx), XLS(xlsx), Markdown(md), Email(eml), TXT(txt), Images(jpg, jpeg, png), CSV(csv), and Web Links(html).

2. Open Source License

Licensed under Apache-2.0

3. Screenshots

4. Feature Overview

Features

• Your data is secure. Installation and usage are fully supported without an internet connection.

• Supports cross-language question answering. You can switch freely between Chinese and English, regardless of the language of your files.

• Supports question answering over massive datasets. Two-stage vector retrieval and reranking solve the retrieval degradation problem associated with large-scale data. The more data you have, the better the results.

• A high-performance, production-grade system ready for direct deployment in enterprise applications.

• Easy to use, with no complicated configuration required. Install and deploy with one click, then start using it immediately.

• Supports question answering across multiple selected knowledge bases.

5. Tech Stack

Architecture

Why use two-stage retrieval?

The advantages of two-stage retrieval are particularly evident when working with large knowledge bases. Using only one-stage embedding retrieval leads to retrieval degradation as the dataset grows, as illustrated by the green line in the figure below. The second-stage reranking process stabilizes and improves accuracy, meaning the more data you have, the better the results.

LLM

The open-source version of QAnything uses Tongyi Qianwen as its underlying large language model and has been fine-tuned on extensive professional question-answering datasets. It significantly enhances question-answering capabilities based on Qianwen. For commercial use, please comply with Qianwen's license. For details, please refer to: Tongyi Qianwen.

Start

System Requirements

For Linux

System	Required item	Minimum Requirement	Note
Linux	NVIDIA GPU Memory	>= 4GB (use OpenAI API)	Minimum: GTX 1050Ti (use OpenAI API) Recommended: RTX 3090
	NVIDIA Driver Version	>= 525.105.17
	Docker version	>= 20.10.5	Docker install
	docker compose version	>= 2.23.3	docker compose install
	git-lfs		git-lfs install

For Windows with WSL Ubuntu subsystem

System	Required item	Minimum Requirement	Note
Windows with WSL Ubuntu subsystem	NVIDIA GPU Memory	>= 4GB (use OpenAI API)	Minimum: GTX 1050Ti (use OpenAI API) Recommended: RTX 3090
	GEFORCE EXPERIENCE	>= 546.33	GEFORCE EXPERIENCE download
	Docker Desktop	>= 4.26.1（131620）	Docker Desktop for Windows
	git-lfs		git-lfs install

More content, please refer to README.md document