biostat-285-final-proj

This project is designed to run on Lambda Labs GPU instances. It includes AI-assisted medical image diagnosis using BLIP-2 + MedLLaMA 2 and a medical chatbot powered by BioMistral-7B.

---

1. Connect to the Lambda Labs Server

Before running the project, SSH into your Lambda Labs instance:

ssh -i ~/.ssh/id_rsa ubuntu@YOUR_INSTANCE_IP

2. Install Dependencies

Check Python Environment

Ensure that you are using the correct Python version (Python 3.10 or higher):

python --version

Install Required Packages

Run the following command to install all required dependencies:

pip install -r requirements.txt

If pip install fails due to version conflicts, try:

pip install --upgrade --force-reinstall -r requirements.txt

3. Login to Hugging Face

The project uses Llama 2 and BioMistral-7B, which are gated models on Hugging Face. You must authenticate before running the app.

Step 1: Request Model Access

Go to Hugging Face - Llama 2 and request access to the model.

Step 2: Login to Hugging Face on the Server

Run:

pip install huggingface_hub  # If not installed
huggingface-cli login

Enter your Hugging Face access token when prompted. You can generate a token from: Hugging Face Tokens

If authentication is successful, your credentials will be saved for future model downloads.

4. Run the Application

Start the Web Interface Once all dependencies are installed, start the app:

python app.py

If everything is set up correctly, you should see an output like:

* Running on local URL: http://127.0.0.1:7860
* Running on public URL: https://xxxxx.gradio.live

Click on the public Gradio link to access the application from your browser.