Fix bug in monitoring for DDP mode and other typos

ml-energy / zeus

Deep Learning Energy Measurement and Optimization

https://ml.energy/zeus

Apache License 2.0

180 stars 24 forks source link

Fix bug in monitoring for DDP mode and other typos #10

Closed Rosie-m closed 1 year ago

Rosie-m commented 1 year ago

This PR fixes an important bug in monitoring at DDP mode: device_id is not specified when spawning Zeus monitors. This bug will cause all the monitors to keep polling the same GPU.