PyTorch
Attach training runs, steps, duration, model metadata, and experiment context.
Overview
PyTorch integration helps Matcha connect training and fine-tuning activity with GPU energy telemetry. This allows teams to compare energy across experiments, models, steps, and training runs instead of only looking at cluster-level power.
Configuration Steps
Add the Matcha Python SDK to your training script.
Log run ID, model name, step timing, and metadata.
Sync timestamps with GPU telemetry.
Attach experiment tags or tenant labels.
View energy by training run in Matcha.
