Measuring RAG End to End: Retrieval, Evidence, and Answer QualityA practical framework for evaluating retrieval, evidence alignment, Corrective RAG, and LLM-generated answers9 min read·curious-pm.hashnode.devSep 27
Beyond Basic Retrieval: Improving Precision and Context in RAGExploring small-to-large, sentence-window, summary-to-detail retrieval, and LLM reranking9 min read·curious-pm.hashnode.devAug 2
How RAG Finds the Right Evidence: Dense, Sparse, and Hybrid RetrievalComparing semantic search, BM25, Reciprocal Rank Fusion, evidence grading, and Corrective RAG10 min read·curious-pm.hashnode.devAug 2
RAG From First Principles: How Documents Become Retrievable KnowledgeUnderstanding the complete RAG pipeline through document loading, metadata preservation, and chunking experiments7 min read·curious-pm.hashnode.devAug 2
Temperature, Top-k, and Top-p: How Decoding Changes What a GPT Model Says Experiments with temperature, greedy decoding, top-k, and top-p to understand the trade-offs between predictability, repetition, coherence, and creativity.10 min read·curious-pm.hashnode.devJul 22
