📚 Documentation

I only know of two places where DeviceStatsMonitor() is mentioned in the docs: here and here. Neither place documents what any of the metrics it logs mean!

• e.g. "active.all.current" what does active mean? is this memory or compute? what units is it in?
• or "active.large_pool.current" what is the large vs small pool?
• Furthermore here it says "ensure that you’re using the full capacity of your accelerator (GPU/TPU/HPU). This can be measured with the DeviceStatsMonitor()" this implies it can measure GPU utilization but it does not say which metric records that.

cc @lantiga @Borda