Kaalaman AI

An awesome A.I that you can use anytime with or without an internet!
Get Started »

View Demo · Report Bug · Request Feature

About the Project

Kaalaman AI is an open source web app that use GPT-Generated Unified Format (GGUF) such as:

Requires Higher Specs

• llama-2-7b-chat.Q8_0 Download URL
• codellama-7b.Q8_0 Download URL
• mistral-7b-instruct-v0.2.Q8_0 Download URL
• orca-2-7b.Q8_0.gguf Download URL

Low - Mid Specs

• llama-2-7b-chat.Q4_0 Download URL
• codellama-7b.Q4_0 Download URL
• mistral-7b-instruct-v0.2.Q4_0 Download URL
• orca-2-7b.Q4_0.gguf Download URL

Getting Started

Pre requisite:

Things you need to use and knowlegde you need to use this web app.

• GIT
• TYPESCRIPT & JAVASCRIPT
• PYTHON
• NODEJS & NPM
• NEXTJS
• REST API
• GGUF MODELS
• MODEL TRAINING (FINE-TUNING, PYTORCH)

(Back to top)

Installation

1. Download all of the Models, create "server/models" directory and put the Models inside the model directory

2. Clone the repo

   git clone https://github.com/Chysev/kaalaman-ai

3. Install NPM packages - both server and client

   npm install or yarn

4. Define the API Model - /server/server.js

    // Models
    const Models = {
      llama2_7b_chat_Q8: "llama-2-7b-chat.Q8_0.gguf", // Requires higher specs
      codellama_7b_Q8: "codellama-7b.Q8_0.gguf", // Requires higher specs
      mistral_7b_instruct_Q8: "mistral-7b-instruct-v0.2.Q8_0", // Requires higher specs
      orca2_7b_Q8: "orca-2-7b.Q8_0.gguf", // Requires higher specs
    };
    const Models2 = {
      llama2_7b_chat_Q4: "llama-2-7b-chat.Q4_0.gguf", // Mid-low higher specs
      codellama_7b_Q4: "codellama-7b.Q4_0.gguf", // Mid-low higher specs
      mistral_7b_instruct_Q4: "mistral-7b-instruct-v0.2.Q4_0", // Mid-low higher specs
      orca2_7b_Q4: "orca-2-7b.Q4_0.gguf", // Mid-low higher specs
    };
    // Model Path
    const model = new LlamaModel({
      modelPath: path.join(process.cwd(), "models", Models.llama2_7B_Chat),- Define here
    });

5. Start the Server

   cd server

   npm run server or yarn server

6. Start the Client

   npm run start or yarn start

(Back to top)

License

Distributed under the MIT License. See LICENSE for more information.

(Back to top)

Contact

John Layda - John Layda (Chysev) - Johnlayda92@gmail.com

Project Link: kaalaman-ai

(back to top)