A Streamlit-based web application to visualize GPU inventory
and AI capabilities across orchestrators in the Livepeer AI network using the Gateway
/capabilities
endpoint.
- GPU Type Distribution: View the distribution of GPU types across orchestrators.
- Orchestrator GPU Distribution: Analyze GPU availability per orchestrator.
- Capabilities Distribution: Explore AI capabilities provided by the orchestrators.
- Interactive Filters: Filter data by GPU models and AI models.
- Data Table: View detailed GPU and orchestrator data in a tabular format.
-
Clone the repository:
git clone https://github.com/rickstaa/ai-compute-visualizer cd ai-compute-visualizer
-
Install dependencies:
pip install -r requirements.txt
-
Set the required environment variable:
export CAPABILITIES_DATA_URL="<GATEWAY_CAPABILITIES_URL>"
-
Run the application:
streamlit run gpu_dashboard.py
-
Open your web browser and navigate to
http://localhost:8501
to view the dashboard.
To deploy the application on Streamlit Community Cloud:
- Push the code to a GitHub repository.
- Link the repository to Streamlit Community Cloud.
- Set the environment variable
CAPABILITIES_DATA_URL
in the Streamlit Cloud settings to the appropriate gateway capabilities URL.
- Data represents a snapshot of orchestrators and capabilities at the time of parsing.
- Only GPUs ready to take jobs are shown.
- Data is collected from a single gateway and may not include all orchestrators and GPUs.
- Realtime AI GPUs are excluded as they are still in beta and not discoverable on chain.