Quality-Constrained Routing over a Fixed Pool of Quantized Mixture-of-Experts Instances

arXiv:2609.12550v1 Announce Type: new Abstract: Quantized Mixture-of-Experts (MoE) services can hold several pre-materialized instances of one base model, but quantization damage varies sharply across requests and bitwidths. Because instance materialization and replica counts consume memory and require slow reconfiguration, we treat them as upstream provisioning decisions and study routing within a fixed resident pool. Within this fixed-pool boundary, we route each request to maximize modeled throughput under a class-level expected quality-degradation budget and measured instance capacities. To predict this request-specific risk, we introduce FWP (Fragility-Weighted Perplexity), computed from prompt tokens on a reference-instance prefill and calibrated to candidate-instance degradation. Underlying FWP is an exact two-expert affinity--fragility decomposition and a conditional multi-layer top-k expansion whose bias, interaction, route-change, separability, and higher-order terms remain explicit. Using these calibrated risks, a window-level linear program yields a signed reduced-reward score that is KKT-consistent with the LP optimum under optimal prices and primal-feasible tie allocation. On 88 extended Qwen prompts, complete W2, W3, and W4 instances quantizing all 6,144 expert blocks incur mean DeltaNLL of 0.9437, 0.1832, and 0.0513. Under the same population and au=0.1513, FWP allocation reaches a 1.284imes offline model-based multiplier versus 1.253imes for request-agnostic mixing and 1.000imes for static W4, an incremental 2.5% relative FWP gain.

Related

• Spend Experts Where You Are Unsure: Confidence-Adaptive Routing for Mixture-of-Experts LoRA
• Beyond Routing: Decoupling Expert Dispatch and Aggregation in Sparse Mixture-of-Experts
• Quality Is Not a Safety Proxy Under Quantization

Source: arXiv cs.LG | 2026-09-14