kun-lab: a native lightweight AI dialog client based on Ollama

General Introduction

kun-lab is a program based on the Ollama is an open source AI conversation app focused on providing a lightweight, fast, localized and intelligent conversation experience. It supports Windows, macOS, and Linux (Windows is currently the main focus), and requires no complex configuration to use. Users can have smooth multi-round conversations with AI, parse documents, recognize images, and even search for answers online. All data is stored locally for privacy and security. kun-lab also offers code rendering, prompt templates, and a multi-language interface for developers, students, or those who need an efficient AI tool.

Function List

• intelligent dialog: Supports multiple rounds of real-time conversations, smooth AI response, and networkable search for more comprehensive answers.
• document resolution: Upload PDF, DOC, PPT, TXT files, AI Understand the content and answer the questions.
• image recognition: Recognize JPG, PNG images, extract text or analyze scenes, support multiple rounds of dialog.
• model management: Easily switch between Ollama or Hugging Face models, supporting GGUF and safetensors formats.
• Cue word templates: Built-in templates and support for customization make it easy to inspire AI creativity.
• Code Rendering: Automatically highlight multiple programming language codes for clear presentation.
• Quick Notes: Support Markdown syntax, real-time preview and one-click export.
• Multi-user support: Allows multiple people to log in at the same time, each with their own individual conversation space.
• multilingual interface: Support Chinese, English and other languages, the operation is more friendly.

Using Help

kun-lab is a feature-rich AI dialog tool, running on Ollama, easy to operate and suitable for local use. Below is a detailed description of the installation process, core functions and getting started steps to help users quickly master it.

Install kun-lab

kun-lab provides both desktop application and source code deployment. Currently, the desktop application only supports Windows, and the source code deployment supports Windows, macOS and Linux.

Mode 1: Desktop application (recommended)

1. Visit the GitHub release page (https://github.com/bahamutww/kun-lab/releases).
2. Download the appropriate installation package for your system:
   • Windows:.exe Documentation.
   • macOS:.dmg Documentation (future support).
   • Linux:.AppImage maybe .deb Documentation (future support).
3. Double-click the installation package and follow the prompts to complete the installation.
4. After installation, click the desktop icon to run kun-lab without additional configuration.
5. After startup, select the language (Chinese is supported by default) and enter the main interface.

Approach 2: Source Code Deployment

If you want to customize or develop, you need to install the environment and run the code. Below are the detailed steps:

1. Preparing the environment::
   • Make sure the system is Windows, macOS, or Linux.
   • Install Python 3.10 or later (download: https://www.python.org).
   • Install Node.js 20.16.0 or later (download: https://nodejs.org).
   • Install Ollama and start the service (reference: https://ollama.com).
2. clone warehouse::

   git clone https://github.com/bahamutww/kun-lab.git
   cd kun-lab
   

3. Creating a Virtual Environment::

   python -m venv venv
   .\venv\Scripts\activate  # Windows
   # source venv/bin/activate  # macOS/Linux
   

4. Installing back-end dependencies::

   cd backend
   pip install -r requirements.txt
   

5. Installing front-end dependencies::

   cd frontend
   npm install
   

6. Configuring Environment Variables::

   cp .env.example .env
   

   • Open with a text editor .env file, modify the configuration (e.g., port number) as needed.
7. launch an application::

   python run_dev.py
   

8. Open your browser and visit http://localhost:5173 to get started.

Core Function Operation

The following describes the main functions of kun-lab and the specific operation procedures to ensure that users can easily get started.

1. Intelligent AI dialogues

• Starting a conversation::
  1. Open kun-lab and click on "Chat Conversation" or "New Conversation".
  2. Select a model in the model list (by default there are models provided by Ollama).
  3. Enter a question and the AI responds in real time.
• Internet search::
  • If the question requires up-to-date information, check the box "Enable web search".
  • The AI will combine the web page data to answer with a more comprehensive answer.
• Managing History::
  • Conversations are automatically saved and can be viewed by clicking "History" in the sidebar.
  • A conversation can be deleted or exported.
• Code Support::
  • Enter a code-related question and AI displays the code in a highlighted format.
  • Support for Python, JavaScript, and many other languages.

2. Document parsing

• Upload a document::
  1. Click the Document Dialog screen.
  2. Click the "Upload" button and select PDF, DOC, PPT or TXT file.
  3. After the document is parsed, the AI displays a summary of the document.
• ask questions::
  • Enter a question related to the document in the dialog box.
  • AI answers based on content and supports contextualization.
• Search content::
  • Enter keywords and AI quickly locates the relevant part of the document.
  • Click on the results to jump to the location of the original article.

3. Image recognition

• Upload a picture::
  1. Go to the "Picture dialog" page.
  2. Click "Upload Image" and select JPG, PNG or JPEG file.
  3. AI automatically recognizes scenes or extracts text.
• many rounds of dialogue::
  • Ask questions based on pictures, such as "What's in this picture?" .
  • After the AI answers, you can continue to ask deeper questions.
• OCR Functions::
  • If the image contains text, AI will extract and display it.
  • Text can be copied or text-based questions can be asked.

4. Model management

• pull model::
  1. Go to the Model Library page.
  2. Click on "Pull Model".
  3. Enter commands such as ollama run qwq:32b maybe ollama run hf.co/Qwen/QwQ-32B-GGUF:Q2_KThe
  4. Wait for the download to complete and the model to be ready to use.
• Switching Models::
  • On the dialog page, click the Model drop-down menu.
  • Select the downloaded model to switch immediately.
• Custom Models::
  1. Click the "Customize" button.
  2. Enter a model name and a system prompt (e.g., "Play Math Teacher").
  3. Select the base model and click "Create".

5. Cue management

• Using templates::
  1. Open the "Prompts" page.
  2. Browse the built-in templates, such as "Write Article Outline" or "Code Debugging".
  3. Click on the template to apply it directly to the dialog.
• Customized Cues::
  • Click on "New Cue".
  • Enter the name and content, save it, and then you can categorize and manage it.
• Quick Apps::
  • During a dialog, select the prompt word and the AI responds as set.

6. Quick notes

• Creating Notes::
  1. Click the Notes page.
  2. Enter content in Markdown format, such as a title, list, or code.
  3. Real-time preview of the effect on the right side.
• Exporting notes::
  • Click on the "Export" button and save as .md Documentation.
  • Can be shared or imported into other tools.

caveat

• Ensure that the Ollama service is running, otherwise the AI function is not available.
• It may take time to pull the model for the first time and it is recommended to check the internet connection.
• Locally stored data takes up a lot of space, clean up your history regularly.
• If you run into problems, check out GitHub's issues page or submit feedback.

By following these steps, you can easily install and utilize the functions of kun-lab. Whether it is dialog, document parsing or image analysis, the operation is intuitive and convenient.

application scenario

1. Personal Learning Assistant
   Students can use kun-lab to parse courseware or textbooks, ask questions, and the AI will answer them in detail. Upload math handouts and the AI can explain formulas step by step.
2. Developer Tools
   Programmers can use kun-lab to debug code or learn a new language. Enter a code snippet and the AI provides optimization suggestions and highlights them.
3. Documentation
   Career professionals can quickly summarize reports with the document parsing feature. Upload long PPTs, AI extracts key points and generates concise notes.
4. Creative Exploration
   Creators can generate stories or design inspiration with prompt word templates. Upload sketches that AI analyzes and suggests improvements.

QA

1. Does kun-lab require an internet connection?
   Core functions run locally and do not require networking. Network search function is optional and needs to be turned on manually.
2. What document formats are supported?
   Supports PDF, DOC, PPT, TXT files, and may be expanded for more formats in the future.
3. How do I add a new model?
   On the Model Library screen, type ollama run command to pull an Ollama or Hugging Face model.
4. Is the data safe?
   All data is stored locally and not uploaded to the cloud to ensure privacy and security.