TruthfulQA

Dataset Card for TruthfulQA Dataset Summary TruthfulQA: Measuring How Models Mimic Human Falsehoods We propose a benchmark to measure whether a language model is truthful in generating answers to questions. The benchmark comprises 817 questions that span 38 categories, including health, law, finance and politics. We cr

Papers using TruthfulQA (27)

MechELK: A Mechanistic Interpretability Framework for Eliciting Latent Knowledge in Large Language Models2026SERC: LDPC-Inspired Semantic Error Correction for Retrieval-Augmented Generation2026CausalGaze: Unveiling Hallucinations via Counterfactual Graph Intervention in Large Language Models2026DeCoVec: Building Decoding Space based Task Vector for Large Language Models via In-Context Learning2026Mitigating LLM Hallucinations through Domain-Grounded Tiered Retrieval2026ROAST: Rollout-based On-distribution Activation Steering Technique2026Dr.LLM: Dynamic Layer Routing in LLMs2025KatotohananQA: Evaluating Truthfulness of Large Language Models in Filipino2025Too Helpful, Too Harmless, Too Honest or Just Right?2025Hallucination Detection with the Internal Layers of LLMs2025We Think, Therefore We Align LLMs to Helpful, Harmless and Honest Before They Go Wrong2025Counterfactual Probing for Hallucination Detection and Mitigation in Large Language Models2025Steering When Necessary: Flexible Steering Large Language Models with Backtracking2025GrAInS: Gradient-based Attribution for Inference-Time Steering of LLMs and VLMs2025MALM: A Multi-Information Adapter for Large Language Models to Mitigate Hallucination2025Selective Self-to-Supervised Fine-Tuning for Generalization in Large Language Models2025Sample, Don't Search: Rethinking Test-Time Alignment for Language Models2025Temporal Self-Rewarding Language Models: Decoupling Chosen-Rejected via Past-Future2025Steering When Necessary: Flexible Steering Large Language Models with Backtracking2025Test-Time Scaling in Diffusion LLMs via Hidden Semi-Autoregressive Experts2025When Persuasion Overrides Truth in Multi-Agent LLM Debates: Introducing a Confidence-Weighted Persuasion Override Rate (CW-POR)2025More is Less: The Pitfalls of Multi-Model Synthetic Preference Data in DPO Safety Alignment2025Teuken-7B-Base & Teuken-7B-Instruct: Towards European LLMs2024 · 2 citesBenchmark Inflation: Revealing LLM Performance Gaps Using Retro-Holdouts2024 · 1 citesEvaluating Consistencies in LLM responses through a Semantic Clustering of Question Answering2024 · 1 citesMaintaining Informative Coherence: Migrating Hallucinations in Large Language Models via Absorbing Markov Chains2024Mitigating Adversarial Attacks in LLMs through Defensive Suffix Generation2024