domains / families / data / online-resource-scheduling
online-resource-scheduling
separate
Design deterministic online scheduling policies from current observations. Use when assigning arriving work to limited resources without seeing future requests.
When to use
- building online schedulers
- assigning arriving work to limited resources
- making deterministic decisions from current observation
- scoring feasible actions by weighted marginal cost
- replaying final action lists
When not to use
- using fixed rules like always first-fit
- using fixed rules like always minimize fragmentation
- using fixed rules like always use the tightest slot
- guessing when no feasible action exists
Skill page
Source
| Repository | Path |
|---|
| benchflow-ai/skillsbench | tasks-extra/gpu-cluster-online-scheduling/environment/skills/online-resource-scheduling/SKILL.md |