README.md

---
title: Transformer Model Structure Visualizer
emoji: 🧩
colorFrom: indigo
colorTo: green
sdk: docker
app_file: backend/app.py
pinned: false
license: mit
short_description: Visualize transformer architectures.
tags:
  - transformers
  - visualization
  - attention
  - model-architecture
  - huggingface
  - model-layers
  - model-structure
  - hidden-states

---


# Run the app locally
This app is developed using React + Fastapi. You can run this app locally with the following steps.

### Run Frontend

1. Install dependencies:
    ```
    npm i
    ```

2. Run the app:
    ```commandline
    npm run dev
    ```
3. Open the browser and access the service from http://localhost:5173/


### Run Backend

1. **Activate the virtual environment**

   Make sure you have your Python virtual environment set up and activated.

2. **Navigate to the backend directory**

   ```commandline
   cd backend
   ```

3. **Install dependencies**
   ```commandline
   pip install -r requirements.txt
   ```

4. **(Optional) Install GPU-related packages** 

   If you are running on a GPU-enabled device, you can install additional packages (support for more models):
   
   ```commandline
   python install_gpu_packages.py
   ```

5. **Start the backend server**
   
   Run the FastAPI app with Uvicorn:
   ```commandline
   uvicorn app:app --reload --host 0.0.0.0 --port 8000
   ```