threadingdata

Training,
in parallel.

A forthcoming platform for analytics workloads that don't fit one core. The six jobs below are real. The launch date is still in training.

// cluster · us-west-2
8×B200 192GB HBM3e  ·  NVLink 5  ·  16 vCPU · 256 GB  ·  6 active jobs
cpu pool0%
// active training runs uptime 0s jobs 6/6 steps 0