datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
t2-ragbench
Dataset Card for T2-RAGBench
Project Page | Paper | Code
IMPORTANT NOTICE:
We deleted VQAonBD from the dataset due to low quality of the question reformulations. If you still want to use it you will find the data in the previous commit history.
Dataset Description
Dataset Summary
T2-RAGBench is a benchmark dataset designed to evaluate Retrieval-Augmented Generation (RAG) on financial documents containing both text and tables. It consists of 23,088… See the full description on the dataset page: https://huggingface.co/datasets/G4KMU/t2-ragbench.ragbench
RAGBench
Dataset Overview
RAGBEnch is a large-scale RAG benchmark dataset of 100k RAG examples.
It covers five unique industry-specific domains and various RAG task types.
RAGBench examples are sourced from industry corpora such as user manuals, making it particularly relevant for industry applications.
RAGBench comrises 12 sub-component datasets, each one split into train/validation/test splits
Usage
from datasets import load_dataset
# load… See the full description on the dataset page: https://huggingface.co/datasets/galileo-ai/ragbench.ragbench-sentence-relevance-balancedt2-ragbench
Dataset Card for T2-RAGBench
Project Page | Paper | Code
IMPORTANT NOTICE:
We deleted VQAonBD from the dataset due to low quality of the question reformulations. If you still want to use it you will find the data in the previous commit history.
Dataset Description
Dataset Summary
T2-RAGBench is a benchmark dataset designed to evaluate Retrieval-Augmented Generation (RAG) on financial documents containing both text and tables. It consists of 23,088… See the full description on the dataset page: https://huggingface.co/datasets/botay/t2-ragbench.rag-bench
Dataset card for RAG-BENCH
Data Summary
RAG-bench aims to provide results of many commonly used RAG datasets. All the results in this dataset are evaluated by the RAG evaluation tool Rageval, which could be easily reproduced with the tool.
Currently, we have provided the results of ASQA dataset,ELI5 dataset and HotPotQA dataset.
Data Instance
ASQA
{
"ambiguous_question":"Who is the original artist of sound of silence?",
"qa_pairs":[{… See the full description on the dataset page: https://huggingface.co/datasets/golaxy/rag-bench.t2-ragbench
Dataset Card for T2-RAGBench
Project Page | Paper | Code
IMPORTANT NOTICE:
We deleted VQAonBD from the dataset due to low quality of the question reformulations. If you still want to use it you will find the data in the previous commit history.
Dataset Description
Dataset Summary
T2-RAGBench is a benchmark dataset designed to evaluate Retrieval-Augmented Generation (RAG) on financial documents containing both text and tables. It consists of 23,088… See the full description on the dataset page: https://huggingface.co/datasets/grasson/t2-ragbench.t2-ragbench-splitst2-ragbench
Dataset Card for T2-RAGBench
Project Page | Paper | Code
IMPORTANT NOTICE:
We deleted VQAonBD from the dataset due to low quality of the question reformulations. If you still want to use it you will find the data in the previous commit history.
Dataset Description
Dataset Summary
T2-RAGBench is a benchmark dataset designed to evaluate Retrieval-Augmented Generation (RAG) on financial documents containing both text and tables. It consists of 23,088… See the full description on the dataset page: https://huggingface.co/datasets/tomsummerfield/t2-ragbench.ragbench-ru
New Dataset (Russian Translation)
This dataset is a translation of the original dataset from English to Russian.
Translated by model Qwen2.5-72B-Instruct.
License
The dataset is licensed under the CC BY 4.0.
Original Source
The original dataset can be found at Dataset Source Link.
LIT-RAGBench
LIT-RAGBench
LIT-RAGBench is a benchmark for evaluating generator capabilities in Retrieval-Augmented Generation (RAG). It focuses on whether a model can answer questions correctly given retrieved documents, independent of retrieval quality. The benchmark covers five categories: Integration, Reasoning, Logic, Table, and Abstention.
Dataset Summary
LIT-RAGBench contains:
114 human-constructed Japanese questions
An English version generated by machine translation with… See the full description on the dataset page: https://huggingface.co/datasets/neoai-inc/LIT-RAGBench.ragbench-corpus
RAGBench Corpus
A small, focused document corpus designed for evaluating Retrieval-Augmented Generation (RAG) systems and comparing different retrieval and document chunking strategies.
Dataset Description
RAGBench Corpus contains 20 short documents covering concepts related to modern information retrieval and RAG systems.
The corpus is designed to be used together with the RAGBench Queries dataset to benchmark retrieval performance.
Topics covered include:
Dense… See the full description on the dataset page: https://huggingface.co/datasets/Gul55555/ragbench-corpus.ragbench-queries
RAGBench Queries
A collection of evaluation queries designed for benchmarking Retrieval-Augmented Generation (RAG) systems.
Dataset Description
RAGBench Queries contains test queries used to evaluate different retrieval and chunking strategies in a RAG pipeline.
The dataset was created as part of the RAGBench project, which compares retrieval performance using different document chunking approaches.
Purpose
The queries are designed to evaluate… See the full description on the dataset page: https://huggingface.co/datasets/Gul55555/ragbench-queries.vectara_open_ragbench
Dataset Card for Vectara Open RAGBench
Dataset Description
Dataset Summary
Vectara Open RAGBench is a benchmark dataset designed to evaluate Retrieval-Augmented Generation (RAG) on open-source scientific documents containing text, tables & images. It consists of 3045 questions-answers-triples coming from in total 2301* documents. The dataset is split up into two main parts: text-table questions and image-based questions. The text-table part contains 2062… See the full description on the dataset page: https://huggingface.co/datasets/G4KMU/vectara_open_ragbench.RAGBench_Kazakh
RAGBench_Kazakh
Summary
RAGBench_Kazakh is a machine-translated Kazakh version of the original RAGBench benchmark. It is designed to evaluate retrieval-augmented generation (RAG) systems, focusing on how well models use retrieved context to produce grounded answers.
The dataset is built from the test splits of multiple RAGBench subsets covering domains such as biomedical research, general knowledge, legal documents, customer support, and finance. Each example… See the full description on the dataset page: https://huggingface.co/datasets/issai/RAGBench_Kazakh.rag-benchmark-qa-datasetragbench
RAGBench
Dataset Overview
RAGBEnch is a large-scale RAG benchmark dataset of 100k RAG examples.
It covers five unique industry-specific domains and various RAG task types.
RAGBench examples are sourced from industry corpora such as user manuals, making it particularly relevant for industry applications.
RAGBench comrises 12 sub-component datasets, each one split into train/validation/test splits
Usage
from datasets import load_dataset
# load… See the full description on the dataset page: https://huggingface.co/datasets/menggaotian/ragbench.AM-RAGBench
AM-RAGBench
Human-verified Arabic-Malay benchmark for evaluating retrieval-augmented generation (RAG) faithfulness. 1,140 question-answer pairs spanning a specialized domain (Quran, Arabic and Basmeih Malay translation) and a general domain (Arabic and Malay Wikipedia), each with a gold passage, a gold answer, and a verification decision made during construction.
Files
quran_verified.jsonl: specialized-domain records.
wiki_verified.jsonl: general-domain records.… See the full description on the dataset page: https://huggingface.co/datasets/Akram98/AM-RAGBench.ragbench_10row_tester_synthetic_mistakerag-bench-public-textsPublic RAG bench dataset with texts
ragbench-5dragbench_10row_tester_synthetic_mistake_evaluatedragbench-allrag-bench-public-questionsPublic RAG bench dataset with questions
RAG-Bench-LegalE2E-v2
Dataset Structure
This dataset contains two subsets:
QA: Question-answer pairs, spanning both single-turn and multi-turn interactions
conversation_id (string): A unique identifier for a conversation session. In multi-turn configurations, multiple rows share the same ID to represent a continuous dialogue.
turn_id (int32): The sequential order of messages within a session (0 represents the first user query).
question (string): The question text.
ground_truth (string): The… See the full description on the dataset page: https://huggingface.co/datasets/QomSSLab/RAG-Bench-LegalE2E-v2.ragbench_10row_tester_annotatedragbench_10row_testerQwen7b_ragbench_emanual_400row_mistake_addedrag-benchmark-datasetragbench_delucionqa_400row_mistake_addedragbench_emanual_400row_mistake_added