RADIANT-LLM: an Agentic Retrieval Augmented Generation Framework for Reliable Decision Support in Safety-Critical Nuclear Engineering

This document is an arXiv RSS batch (multiple new papers) focused on retrieval-augmented generation (RAG), agentic LLM systems, retrieval/benchmarking, and domain-specific deployments. Key items: RADIANT-LLM — a multi-modal, local-first RAG + agent framework tailored for nuclear safety/decision-support with provenance, human-in-the-loop validation, and domain-aware metrics (Context Precision, Hallucination Rate, Visual Recall); StratRAG — an open multi-hop retrieval evaluation dataset and retrieval benchmarks; Quantifying Divergence in Inter-LLM Communication — analysis exposing hidden, domain