datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
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 crafted questions that some humans would answer falsely due to a false belief or misconception. To perform well, models must avoid generating false answers… See the full description on the dataset page: https://huggingface.co/datasets/domenicrosati/TruthfulQA.the-truthtrilemma-of-truth
Dataset Card for Trilemma of Truth (ToT) Dataset
🧾 Dataset Summary
The Trilemma of Truth (ToT) dataset serves as a benchmark for evaluating veracity probes across three distinct statement types:
Factually true statements.
Factually false statements.
Neither-valued statements are defined as those for which the language model lacks sufficient evidence to assign a truth value (see formal definition below).
The dataset includes three domain configurations:… See the full description on the dataset page: https://huggingface.co/datasets/carlomarxx/trilemma-of-truth.bfsi-bench
BFSI-Bench
BFSI-Bench is a benchmark for testing how well language models answer questions about India’s banking, financial services, and insurance (BFSI) rules.
In this domain, the correct answer often depends on circulars and regulations that change frequently, and the official sources (sites like RBI, SEBI, and IRDAI) can be hard to find, parse, and keep current. BFSI-Bench measures five capability areas:
Jurisdiction-Aware Compliance: Disambiguate to the Indian context, or… See the full description on the dataset page: https://huggingface.co/datasets/ground-truth/bfsi-bench.TruthfulQA_CoT_GPT4SciBench-TruthfulQA-RAGTruthfulQA_zhTruthfulQA dataset csv with question and answer field translated into Chinese by requesting GPT-4.
TruthfulQA_LLMstruthfulqa_indicOriginal Repository
Tasks (from original repository)
Generation (main task):
Task: Given a question, generate a 1-2 sentence answer.
Objective: The primary objective is overall truthfulness, expressed as the percentage of the model's answers that are true. Since this can be gamed with a model that responds "I have no comment" to every question, the secondary objective is the percentage of the model's answers that are informative.
Future Work:
Validate… See the full description on the dataset page: https://huggingface.co/datasets/vakyansh/truthfulqa_indic.TruthGen
Dataset Card for TruthGen
TruthGen is a dataset of generated political statements, created to assess the relationship between truthfulness and political bias in reward models and language models. It consists of non-repetitive, non-political factual statements paired with false statements, designed to evaluate models for their ability to distinguish true from false information while minimizing political content. The dataset was generated using GPT-3.5, GPT-4 and Gemini, with a focus… See the full description on the dataset page: https://huggingface.co/datasets/wwbrannon/TruthGen.ground-truth-ob
Ground Truth OB
This repository contains ground_truth_kb.csv, a tabular ground-truth or knowledge-base resource. The current repository is deliberately small and contains no executable training or evaluation script.
Recommended use
Load the CSV, inspect its column names and encoding, validate identifiers and labels, and record the provenance of every ground-truth field before joining it with model outputs. Keep an immutable copy of the raw file and create derived… See the full description on the dataset page: https://huggingface.co/datasets/MR-CODESPIKE/ground-truth-ob.StockSensei_Ground_Truth
Financial Advice Finetuning Ground Truth Dataset
Georgia Institute of Technology, College of Computing
Authors: Hersh Dhillon, Mathan Mahendran, Will Ferguson, Ayushi Mathur, Dorsa Ajami
December 2024
Motivation
Given the unprecendented rise of day trading, social-media based financial advice, and trading apps, more people then ever are buying and selling stocks
without proper financial literacy. Oftentimes, people make high-risk trades with little more quantitative… See the full description on the dataset page: https://huggingface.co/datasets/iamwillferguson/StockSensei_Ground_Truth.nodesHere is the processed node data from PubTator used for knowledge-enhanced hypothesis generation in the TruthHypo benchmark (IJCAI 25), which is designed to evaluate the capabilities of LLMs in generating truthful scientific hypothesis. The paper is available at https://arxiv.org/abs/2505.14599.
GMO-Myths-and-Truths
Dataset Card for GMO Myths and Truths (NLP Classification)
Dataset Summary
This dataset contains a structured collection of claims and evidence-based findings regarding Genetically Modified Organisms (GMOs). The data was extracted and adapted from the technical report: "GMO Myths and Truths: An evidence-based examination of the claims made for the safety and efficacy of genetically modified crops" (Version 1.3a, June 2012).
It is designed for binary text classification… See the full description on the dataset page: https://huggingface.co/datasets/siddqamar/GMO-Myths-and-Truths.edges_testHere is the test data of the TruthHypo benchmark (IJCAI 25), which is designed to evaluate the capabilities of LLMs in generating truthful scientific hypothesis. The paper is available at https://arxiv.org/abs/2505.14599.
edges_trainHere is the processed relation data from PubTator used for knowledge-enhanced hypothesis generation in the TruthHypo benchmark (IJCAI 25), which is designed to evaluate the capabilities of LLMs in generating truthful scientific hypothesis. The paper is available at https://arxiv.org/abs/2505.14599.
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 crafted questions that some humans would answer falsely due to a false belief or misconception. To perform well, models must avoid generating false answers… See the full description on the dataset page: https://huggingface.co/datasets/jethalal23/TruthfulQA.truthfulqa_trclinical-structural-similarity-scoring-against-ground-truth-v0.1What this dataset tests
Whether a model can match the later-discovered explanationby structural logic, not by diagnosis label.
Input
pre-explanation case summary and data
predicted structure
ground truth structure
Required outputs
structural_similarity_score_0_100
alignment_strengths
divergence_points
Representation format
Predicted and ground truth structures use this schema text
systems A B C
nodes n1 n2 n3
edges n1->n2 n2->n3
phases p1 p2 p3
failure_modes f1 f2
Typical… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-structural-similarity-scoring-against-ground-truth-v0.1.truthful_qa_TrueFalse_Feedback
Dataset Card for Dataset Name
This is a reduced variation of the truthful_qa dataset (https://huggingface.co/datasets/truthful_qa), modified to associate boolean values with the given answers, with a correct answer as a reference, and a feedback.
Dataset Details
Dataset Description
Curated by: [More Information Needed]
Funded by [optional]: [More Information Needed]
Shared by [optional]: [More Information Needed]
Language(s) (NLP): [More Information… See the full description on the dataset page: https://huggingface.co/datasets/nmarafo/truthful_qa_TrueFalse_Feedback.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 crafted questions that some humans would answer falsely due to a false belief or misconception. To perform well, models must avoid generating false answers… See the full description on the dataset page: https://huggingface.co/datasets/M1STERPERFECT/TruthfulQA.Palestinian_Truth_arPalestinian_Truth_engTruthfulQA
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 crafted questions that some humans would answer falsely due to a false belief or misconception. To perform well, models must avoid generating false answers… See the full description on the dataset page: https://huggingface.co/datasets/Kavya5705/TruthfulQA.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 crafted questions that some humans would answer falsely due to a false belief or misconception. To perform well, models must avoid generating false answers… See the full description on the dataset page: https://huggingface.co/datasets/hamesh05/TruthfulQA.truthfulQA-boolPalestinian_Truth_EnglishTruthSeekerwhisper_8_avg_ground_truth_scores_2columnsTruthfulQA-Audited
TruthfulQA-Audited
Datasets accompanying an anonymous NeurIPS 2026 Evaluations & Datasets
Track submission on surface-form leakage in binary-choice truth
benchmarks. The release contains three related artifacts:
TruthfulQA-476
Cleaned subset of binary-choice TruthfulQA, with surface-form leakage
removed via an audit-and-prune procedure.
canonical_label: TruthfulQA-476
theta: 0.53
n_pairs: 476
audit AUC: 0.528
derived from: binary-choice TruthfulQA (790 pairs)… See the full description on the dataset page: https://huggingface.co/datasets/AnonymNeurIPS2026submission/TruthfulQA-Audited.