RouteNLP: Closed-Loop LLM Routing with Conformal Cascading and Distillation Co-Optimization
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866914507797823488 |
|---|---|
| author | Guo, Dongxin Wu, Jikun Yiu, Siu Ming |
| author_facet | Guo, Dongxin Wu, Jikun Yiu, Siu Ming |
| contents | Serving diverse NLP workloads with large language models is costly: at one enterprise partner, inference costs exceeded $200K/month despite over 70% of queries being routine tasks well within the capability of smaller models. We present RouteNLP, a closed-loop framework that routes queries across a tiered model portfolio to minimize cost while satisfying per-task quality constraints. The framework integrates three components: a difficulty-aware router with shared task-conditioned representations trained on preference data and quality signals; confidence-calibrated cascading that uses conformal prediction for distribution-free threshold initialization; and a distillation-routing co-optimization loop that clusters escalation failures, applies targeted knowledge distillation to cheaper models, and automatically retrains the router, yielding over twice the cost improvement of untargeted distillation. In an 8-week pilot deployment processing ~5K queries/day at an enterprise customer-service division, RouteNLP reduced inference costs by 58% while maintaining 91% response acceptance and reducing p99 latency from 1,847 ms to 387 ms. On a six-task benchmark spanning finance, customer service, and legal domains, the framework achieves 40-85% cost reduction while retaining 96-100% quality on structured tasks and 96-98% on generation tasks, with human evaluation confirming that 74.5% of routed generation outputs match or exceed frontier-model quality. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_23577 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | RouteNLP: Closed-Loop LLM Routing with Conformal Cascading and Distillation Co-Optimization Guo, Dongxin Wu, Jikun Yiu, Siu Ming Computation and Language Machine Learning I.2.7; I.2.6; H.3.3; C.4 Serving diverse NLP workloads with large language models is costly: at one enterprise partner, inference costs exceeded $200K/month despite over 70% of queries being routine tasks well within the capability of smaller models. We present RouteNLP, a closed-loop framework that routes queries across a tiered model portfolio to minimize cost while satisfying per-task quality constraints. The framework integrates three components: a difficulty-aware router with shared task-conditioned representations trained on preference data and quality signals; confidence-calibrated cascading that uses conformal prediction for distribution-free threshold initialization; and a distillation-routing co-optimization loop that clusters escalation failures, applies targeted knowledge distillation to cheaper models, and automatically retrains the router, yielding over twice the cost improvement of untargeted distillation. In an 8-week pilot deployment processing ~5K queries/day at an enterprise customer-service division, RouteNLP reduced inference costs by 58% while maintaining 91% response acceptance and reducing p99 latency from 1,847 ms to 387 ms. On a six-task benchmark spanning finance, customer service, and legal domains, the framework achieves 40-85% cost reduction while retaining 96-100% quality on structured tasks and 96-98% on generation tasks, with human evaluation confirming that 74.5% of routed generation outputs match or exceed frontier-model quality. |
| title | RouteNLP: Closed-Loop LLM Routing with Conformal Cascading and Distillation Co-Optimization |
| topic | Computation and Language Machine Learning I.2.7; I.2.6; H.3.3; C.4 |
| url | https://arxiv.org/abs/2604.23577 |