42 entries matching the current filters. Sorted by recently added.
Scheduling with predicted reasoning depth. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Sharing KV caches across reasoning paths. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Scheduling expansion and aggregation in LLM reasoning. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Queueing and job-shop models for multi-stage LLM inference. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Scheduling parallel and dependent multi-LLM tasks. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Queueing tradeoffs between LLM accuracy and time. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Reconciling predictions with live inference feedback. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Routing requests among LLMs under cost and latency limits. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Tandem queue models of speculative decoding. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Load balancing API-augmented inference with quality tradeoffs. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Scheduling guarantees for API-augmented LLM requests. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Adapting allocation to the bottleneck LLM phase. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
When many smaller GPUs outperform one larger GPU. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Tandem queue analysis of split-phase LLM serving. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Heterogeneous GPUs across prefill and decode. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Combining pooled and dedicated GPU allocation. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Prompt sharing without delaying small standalone requests. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Joint scheduling of interactive and background requests. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Cost and latency targets in adaptive LLM scheduling. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Limited-preemption SPRPT and response-time tails. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Adaptive thresholds for costly LLM preemption. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Ranking LLM requests across prefill and decode. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Scheduling jobs with predicted speedups across server counts. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Fluid limits for power-of-two routing with richer predictions. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.
Multi-server queueing theory with predictions. Current research status is unassessed. The dated source poses this question; inclusion does not certify that it remains open today.