| # ๐ Beginner's Guide: Using Your Desktop as an AI Brain |
|
|
| You want to use your powerful Desktop (for its GPU) to do the heavy thinking, while working comfortably on your Laptop. This guide explains how to connect them. |
|
|
| ## ๐งฉ The Parts |
|
|
| Imagine this system like a **Remote Control Car**: |
| 1. **The Controller (Your Laptop)**: It has the buttons and the map (AgentIC). It tells the car where to go. |
| 2. **The Car (Your Desktop)**: It has the engine (GPU). It does the actual driving (thinking). |
| 3. **The Signal (Network)**: The connection between them. |
|
|
| **Crucially:** You cannot use an HDMI cable for this. You must use **Ethernet (LAN cable)** or **Wi-Fi**. |
|
|
| --- |
|
|
| ## ๐ ๏ธ Step 1: Set Up The Desktop (The Car) |
|
|
| Go to your **Desktop computer** for these steps. |
|
|
| ### 1. Create a Folder |
| Open a terminal on your desktop and run: |
| ```bash |
| mkdir ~/agentic-server |
| cd ~/agentic-server |
| ``` |
|
|
| ### 2. Copy Files |
| You need to copy two files from your laptop's `AgentIC/scripts/` folder to this new folder on your desktop. You can email them to yourself, use a USB drive, or use `scp`. |
| - `launch_server.py` |
| - `setup_desktop.sh` |
|
|
| ### 3. Install Requirements |
| Run the setup script to install the AI software (`vllm`): |
| ```bash |
| bash setup_desktop.sh |
| ``` |
|
|
| ### 4. Start the Engine |
| Run the server. Replace the model name with your preferred model if needed. |
| ```bash |
| # First, activate the environment |
| source venv/bin/activate |
| |
| # Launch the server (this will download the model first, which takes time!) |
| python launch_server.py --model Qwen/Qwen2.5-Coder-32B-Instruct |
| ``` |
| *Wait until you see "Application startup complete".* |
|
|
| --- |
|
|
| ## ๐ Step 2: Find Desktop's Address |
|
|
| Your laptop needs to know *where* the desktop is on the network. |
| 1. Open a new terminal on the **Desktop**. |
| 2. Run: `hostname -I` |
| 3. You will see numbers like `192.168.1.50`. Write this down. This is your **Desktop IP**. |
|
|
| --- |
|
|
| ## ๐ฎ Step 3: Connect The Laptop (The Controller) |
|
|
| Go back to your **Laptop**. |
|
|
| 1. Open `AgentIC/src/agentic/config.py`. |
| 2. Find the `LOCAL_CONFIG` section. |
| 3. Change `base_url` to match your Desktop IP: |
|
|
| ```python |
| LOCAL_CONFIG = { |
| "model": "Qwen/Qwen2.5-Coder-32B-Instruct", # Must match what you ran on Desktop |
| "base_url": "http://192.168.1.50:8000/v1", # <--- REPLACE WITH DESKTOP IP |
| "api_key": "EMPTY" # Key is not needed for local network |
| } |
| ``` |
|
|
| ## โ
Step 4: Test It |
|
|
| On your laptop, run a simple build command: |
| ```bash |
| agentic build --prompt "A simple D-flip flop" --model Qwen/Qwen2.5-Coder-32B-Instruct |
| ``` |
|
|
| If it works, your laptop is successfully steering the desktop! |
|
|