Original r architecture using tiny domain-specialized language models for efficient CPU-level AI. XENTARA:A Modular Mixture-of-Experts Architecture of Domain-Specialised Tiny Transformers with Neural Query Routing.
language-model ai-research mixture-of-experts neuro-symbolic-ai llm-architecture tiny-llm sparse-routing domain-specialization
-
Updated
Apr 17, 2026