Cacheable by Design? Training Mixture-of-Experts Routers for Locality Against th

arXiv:2605.00254v1 Announce Type: cross Abstract: Mixture-of-experts (MoE) architectures have turned LLM serving into a cluster-scale workload in which communication consumes a

arXiv:2603.18567v1 Announce Type: cross Abstract: Large language models incur high inference latency due to sequential autoregressive decoding. Speculative decoding alleviates

arXiv:2605.00422v1 Announce Type: cross Abstract: Large language models (LLMs) have driven major progress in NLP, yet their substantial memory and compute demands still hinder

arXiv:2604.10152v1 Announce Type: new Abstract: The Mixture-of-Experts (MoE) architecture has emerged as a promising approach to mitigate the rising computational costs of large

arXiv:2610.02762v1 Announce Type: new Abstract: Dynamic LLM routers promise to cut inference costs by sending each query to the cheapest model that can answer it correctly. We

arXiv:2608.11212v1 Announce Type: new Abstract: Top-k Mixture-of-Experts (MoE) routing is discontinuous, so a deployment-motivated numerical disturbance -- simulated 4-bit

<h1>Mixture of Experts Architecture: A Deep Dive into Sparse Models and Scaling</h1> <p>Traditional large language models have hit a massive hardware wall. Every time you run a

arXiv:2603.22324v1 Announce Type: cross Abstract: We introduce Delta-Aware Quantization (DAQ), a data-free post-training quantization framework that preserves the knowledge