FastVideo Performance Dashboard

Local FastAPI + React dashboard for records stored in the Hugging Face performance tracking dataset.

Data Source

The dashboard reads the same normalized JSON records used by fastvideo/tests/performance/compare_baseline.py.

Defaults:

Set one of HF_API_KEY, HUGGINGFACE_HUB_TOKEN, or HF_TOKEN if the configured dataset repo requires authenticated access:

export HF_TOKEN=hf_...

If Hugging Face returns 401 Unauthorized, confirm that HF_REPO_ID points to the dataset repo you expect and that your token has access to it.

Run the API:

python -m fastvideo.performance_dashboard --host 127.0.0.1 --port 8000 --reload

Run the React dev server:

cd performance_dashboard/frontend
npm install
npm run dev

Open http://127.0.0.1:5173. Vite proxies /api/* to the FastAPI server on port 8000.

Build the frontend:

cd performance_dashboard/frontend
npm install
npm run build

Serve API and built frontend from one FastAPI process:

python -m fastvideo.performance_dashboard --host 0.0.0.0 --port 8000

Expose it:

ngrok http 8000

The ngrok URL will serve the dashboard UI and all /api/performance/* endpoints from the same local port.

The current v1 grouping key is (model_id, gpu_type). Baselines are computed from the latest five previous successful records in each group.