METR-LA • ST-GraphMamba
Traffic Forecasting
Real inference from your trained ST-GraphMamba model through the Render API.
Inference API
The Space calls GET /health and GET /predict/latest.
The model itself stays on the Python inference service.
Run real prediction
Uses the latest 24 observations from the real METR-LA dataset loaded by the backend and predicts 12 future steps for 207 sensors.
Model information
ModelST-GraphMamba
DatasetMETR-LA
Input24 × 207
Output207 × 12
Prediction matrix—
Forecast horizon—
Sensors—
Traffic speed forecast
Forecast values
Predicted speed in mph for the selected sensor.
First 20 sensors
| Sensor | +5m | +10m | +15m | +20m | +25m | +30m | +35m | +40m | +45m | +50m | +55m | +60m |
|---|
Deployment architecture
METR-LA→Render FastAPI→
ST-GraphMamba→207 × 12→
This Space
This is a static frontend. It does not contain the .pt checkpoint.