Knowledge Tree
faithfulness
Facts (12)
Sources
EdinburghNLP/awesome-hallucination-detection - GitHub github.com 3 facts
claimResearch presented at EMNLP 2025 found that alignment-tuned Large Language Models produce more faithful explanations than base models, and that faithfulness and plausibility are positively correlated.
measurementEvaluation of faithfulness for knowledge-grounded response generation on the FaithDial dataset uses FaithCritic, CoLA (Fluency), Dialog Engagement, and Length-penalised TF-IDF Diversity as metrics.
measurementEvaluation of faithfulness between predicted responses and ground-truth knowledge uses Critic, Q², BERT F1, and F1 as metrics, and utilizes datasets including Wizard-of-Wikipedia (WoW), DSTC9 and DSTC11 extensions of MultiWoZ 2.1, and FaithDial.
Evaluating RAG applications with Amazon Bedrock knowledge base ... aws.amazon.com Mar 14, 2025 2 facts
claimAmazon Bedrock Knowledge Bases evaluation measures generation quality using metrics for correctness, faithfulness (to detect hallucinations), and completeness.
referenceAmazon Bedrock knowledge base evaluation assesses quality through four dimensions: technical quality (context relevance and faithfulness), business alignment (correctness and completeness), user experience (helpfulness and logical coherence), and responsible AI metrics (harmfulness, stereotyping, and answer refusal).
Detect hallucinations for RAG-based systems - AWS aws.amazon.com May 16, 2025 2 facts
referenceThe RAGAS (Retrieval Augmented Generation Assessment) framework provides metrics to evaluate RAG pipelines, specifically focusing on faithfulness, answer relevance, context precision, and context recall.
claimFaithfulness in the RAGAS framework measures whether the generated answer is derived solely from the retrieved context, helping to detect hallucinations.
Detecting hallucinations with LLM-as-a-judge: Prompt ... - Datadog datadoghq.com Aug 25, 2025 2 facts
procedureDetermining faithfulness in RAG systems requires three components: a user-posed question, context retrieved from a knowledge base, and an answer generated by the LLM.
claimFaithfulness in the context of retrieval-augmented generation (RAG) is defined as the requirement that an LLM-generated answer agrees with the provided context, which is assumed to be the ground truth.
Benchmarking Hallucination Detection Methods in RAG - Cleanlab cleanlab.ai Sep 30, 2024 1 fact
referenceRAGAS is a RAG-specific, LLM-powered evaluation suite that provides various scores used to detect hallucination, specifically Faithfulness and Answer Relevancy.
LLM Hallucination Detection and Mitigation: State of the Art in 2026 zylos.ai Jan 27, 2026 1 fact
claimThe taxonomy of hallucination detection distinguishes between factuality, which is absolute correctness against real-world truth, and faithfulness, which is adherence to provided input or context.
Survey and analysis of hallucinations in large language models frontiersin.org Sep 29, 2025 1 fact
referenceMaynez et al. (2020) investigated faithfulness and factuality in abstractive summarization at the 58th Annual Meeting of the Association for Computational Linguistics.