| # Student & Developer Documentation |
|
|
| ## Overview |
| Welcome to the RemiAI Framework! This document is designed to help you understand how to customize, configure, and make this application your own. This framework is built to be "Plug-and-Play"—meaning you don't need to know Python or complex AI coding to use it. |
|
|
| ## 🛠️ Setup & How to Customize |
|
|
| ### 0. Quick Setup (Important!) |
| Before running the app, you **must** ensure the AI engine files are downloaded correctly. GitHub does not store large files directly, so we use **Git LFS**. |
|
|
| 1. **Install Git LFS**: |
| * Download and install from [git-lfs.com](https://git-lfs.com). |
| * Open a terminal and run: `git lfs install` |
| 2. **Pull Files**: |
| * Run: `git lfs pull` inside the project folder. |
| * *Why?* Without this, the app will say **"RemiAI Engine Missing"** or "Connection Refused". |
|
|
| ### 1. Changing the AI Name |
| Want to name the AI "Jarvis" or "MyBot"? |
| 1. Open `index.html` in any text editor (VS Code, Notepad, etc.). |
| 2. Search for "RemiAI" or "Bujji". |
| 3. Replace the text with your desired name. |
| 4. Save the file. |
| 5. Restart the app (`npm start`), and your new name will appear! |
|
|
| ### 2. Replacing the AI Model |
| This application is powered by a **GGUF** model file. You can swap this "brain" for a smarter one, a faster one, or one specialized in coding/storytelling. |
|
|
| **Steps to Change the Model:** |
| 1. **Download a Model**: Go to [Hugging Face](https://huggingface.co/models?library=gguf) and search for GGUF models (e.g., `Llama-3-8B-GGUF`, `Mistral-7B-GGUF`). |
| 2. **Select File**: Download the `.gguf` file (Q4_K_M or Q5_K_M are good balances of speed and intelligence). |
| 3. **Rename**: Rename your downloaded file to exactly: |
| > **`model.gguf`** |
| 4. **Replace**: |
| * Go to the `engine` folder in your project directory. |
| * Paste your new `model.gguf` there, replacing the old one (or place it one level up depending on your specific folder setup—check `main.js` which looks for `../model.gguf` relative to the engine binary). *Note: Standard setup is usually placing `model.gguf` in the root or `engine` folder as configured.* |
| 5. **Restart**: Run `npm start`. The app will now use the new intelligence! |
| |
| **Note**: Make sure your laptop have good health don't use laptop more then 5 years old because running an entire Gen AI model weights or Neural Network will damage the laptop - your laptop may stucks, over heat, shutdown automatically and some it will make your laptop or device dead so be carefule |
| **GOOD CONFIGURATION NO DAMAGE** (i3 processor, 8GB RAM) - if you are using the PC. |
| (i5 processor, 16GB RAM) - if you are using the laptop. |
| new i3 with 8GB RAM laptop's will easily runs but the laptop want to be new and good heavy if the laptop was too old it will not work even you have i5 processor and 16GB RAM |
|
|
| ### 3. Customizing the UI |
| All styles are in `styles.css` (or within `index.html`). |
| * **Colors**: Change the background colors or chat bubble colors in the CSS. |
| * **Icons**: Replace `remiai.ico` with your own `.ico` file to change the app icon. |
|
|
| ## ❓ Frequently Asked Questions (FAQ) |
|
|
| **Q: Do I need Python?** |
| A: **No.** The application comes with a pre-compiled engine (`bujji_engine.exe` / `llama-server.exe`) that runs the model directly. |
|
|
| **Q: Why does it say "AVX2"?** |
| A: AVX2 is a feature in modern CPUs that makes the AI run faster. The app automatically detects if you have it. If not, it switches to a slower but compatible mode (AVX). |
|
|
| **Q: The app opens but doesn't reply / "RemiAI Engine Missing" Error.** |
| A: |
| 1. **Git LFS Issue**: This usually means you downloaded "pointers" (tiny files) instead of the real engine. Open a terminal in the folder and run `git lfs pull`. |
| 2. **Model Issue**: Check if `model.gguf` exists in the `engine` folder. |
| 3. **Console Check**: Open Developer Tools (Ctrl+Shift+I) to see errors. |
|
|
| **Q: I see "Content Security Policy" warnings in the console.** |
| A: We have configured safeguards (`index.html` meta tags) to block malicious scripts. If you see warnings about `unpkg` or `jsdelivr`, ensure your CSP matches the latest trusted domains in our code. |
|
|
| **Q: How do I build it into an .exe file?** |
| A: Run the command: |
| ```bash |
| npm run dist |
| ``` |
| This will create an installer in the `release` folder that you can share with friends! |
|
|
| `if you are facing errors while building open the power shell as an administrator and run the above command then it will works 100%` |
|
|
