Verified record · from the answer index
gke-ai-troubleshooting-tpu-performance-degradation
skillofficialtested · static-verifiedscore 59/100signed passport ✓472 uses
Diagnose GKE Cloud TPU training throughput drops and step-time regressions (15%+ TPU duty-cycle drop) using ML Diagnostics Workload Monitoring (`gcloud alpha mldiagnostics monitored-events` / `hypercomputecluster.googleapis.com/v1alpha`) an
Declared tools (1)
gke-ai-troubleshooting-tpu-performance-degradation
skill:google/skills/gke-ai-troubleshooting-tpu-performance-degradation · published by google · source repository