haive.agents.rag¶

RAG (Retrieval-Augmented Generation) agents for the Haive framework.

Available agents: - BaseRAGAgent: Foundation retriever with HuggingFace embeddings - SimpleRAGAgent: Retriever + answer agent in sequence - DynamicRAGAgent: Multi-source dynamic retrieval - LLMRAGAgent: LLM-based RAG - AgenticRAGAgent: ReactAgent with retrieval tools - CorrectiveRAGAgent: Self-correcting retrieval - SelfReflectiveRAGAgent: Reflective retrieval with grading - SelfRouteRAGAgent: Query-aware retrieval routing - StepBackRAGAgent: Abstract query generation for better retrieval - HyDERAGAgent: Hypothetical document embeddings - FLARERAGAgent: Forward-looking active retrieval - MultiQueryRAGAgent: Multiple query variants - RAGFusionAgent: Reciprocal rank fusion - SpeculativeRAGAgent: Hypothesis + parallel verification - DocumentGradingRAGAgent: Document relevance grading - AdaptiveRAGAgent: Adaptive strategy selection - QueryDecomposerAgent: Hierarchical query decomposition - HallucinationGraderAgent: Output hallucination detection - MemoryAwareRAGAgent: RAG with memory context

Submodules¶