Local or Escalate? Post-Training Expert Routing for Real-Time Duplex Speech Models
cs.AI, eess.AS
Submitted: 2026-11-01
Updated: 2026-11-01
Terminology
Sources
- Understanding intermediate layers using linear classifier probes
- The Internal State of an LLM Knows When It's Lying
- Discovering Latent Knowledge in Language Models Without Supervision
- FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance
- VoiceBench: Benchmarking LLM-Based Voice Assistants
- MoshiRAG: Asynchronous Knowledge Retrieval for Full-Duplex Speech Language Models
- MiniCPM-o 4.5: Towards Real-Time Full-Duplex Omni-Modal Interaction
- Moshi: a speech-text foundation model for real-time dialogue
- Hybrid LLM: Cost-Efficient and Quality-Aware Query Routing
- SD-QA: Spoken Dialectal Question Answering for the Real World
- BayLing-Duplex: Native Full-Duplex Speech Dialogue with a Single Autoregressive LLM
- DuplexOmni: Real-Time Listening, Seeing, Thinking, and Speaking for Full-Duplex Interaction
- Language Models (Mostly) Know What They Know
- Semantic Entropy Probes: Robust and Cheap Hallucination Detection in LLMs
- Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generation
- Baichuan-Audio: A Unified Framework for End-to-End Speech Interaction
- Teaching Models to Express Their Uncertainty in Words
- LLMs Encode Their Failures: Predicting Success from Pre-Generation Activations
- The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets
- RouteLLM: Learning to Route LLMs with Preference Data
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents