SlidingServe achieves up to 30% higher service capacity and 16-53% fewer SLO violations in LLM inference by using dynamic chunking and priority-based batch construction.
2025.PolyServe: Efficient Multi-SLO Serving at Scale
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Beyond Greedy Chunking: SLO-Aware Sliding-Window Scheduling for LLM Inference
SlidingServe achieves up to 30% higher service capacity and 16-53% fewer SLO violations in LLM inference by using dynamic chunking and priority-based batch construction.