System Metrics

wandb automatically logs system metrics every 2 seconds, averaged over a 30 second period. The metrics include:

• CPU Utilization
• System Memory Utilization
• Disk I/O Utilization
• Network traffic (bytes sent and received)
• GPU Utilization
• GPU Temperature
• GPU Time Spent Accessing Memory (as a percentage of the sample time)
• GPU Memory Allocated
• TPU Utilization

GPU metrics are collected on a per-device basis using nvidia-ml-py3. For more information on how to interpret these metrics and optimize your model's performance, see this helpful blog post from Lambda Labs.