CLI Reference
Core commands
vaq run <model_id>vaq psvaq stop <model_id> [--purge-cache]vaq listvaq rm <model_id>vaq doctorvaq infervaq ui
Run options highlights
--device gpu|cpu--max-num-seqs--max-model-len--quantization--kv-cache-dtype
Manual overrides:
--gpu-utilization <ratio>in(0, 1]--cpu-utilization <ratio>in(0, 1]
When manual overrides are provided, automatic estimation/optimization is bypassed.