Optimal Expert-Attention Allocation in Mixture-of-Experts: A Scalable Law for Dynamic Model Design

source_context

{
  "surface": "signal_canvas",
  "mode": "paper",
  "query": "Optimal Expert-Attention Allocation in Mixture-of-Experts: A Scalable Law for Dynamic Model Design",
  "normalized_query": "2603.10379",
  "route": "/signal-canvas/optimal-expert-attention-allocation-in-mixture-of-experts-a-scalable-law-for-dynamic-model-design",
  "paper_ref": "optimal-expert-attention-allocation-in-mixture-of-experts-a-scalable-law-for-dynamic-model-design",
  "topic_slug": null,
  "benchmark_ref": null,
  "dataset_ref": null
}