metadata

configs:
  - config_name: crag_task_1_and_2_subset_1
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_1.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_2
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_2.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_3
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_3.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_4
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_4.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_5
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_5.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_6
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_6.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_7
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_7.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_8
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_8.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_9
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_9.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_10
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_10.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_11
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_11.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_12
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_12.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_13
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_13.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_14
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_14.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_15
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_15.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_16
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_16.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_17
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_17.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_18
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_18.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_19
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_19.jsonl.bz2
  - config_name: crag_task_1_and_2_subset_20
    data_files: subset/crag_task_1_and_2/crag_task_1_and_2_dev_v4_subset_20.jsonl.bz2
license: cc-by-nc-4.0
task_categories:
  - question-answering
  - summarization
language:
  - en
tags:
  - music
  - finance
size_categories:
  - 1K<n<10K
pretty_name: CRA

Datasets are taken from Facebook's CRAG: Comprehensive RAG Benchmark, see their arXiv paper for details about the dataset construction.

CRAG Sampler

We have added a simple Python tool for performing stratified sampling on CRAG data.

Installation

Local Development Install (Recommended)

git clone https://huggingface.co/Quivr/CRAG.git
cd CRAG
pip install -r requirements.txt  # Install dependencies
pip install -e .  # Install package in development mode

Quick Start

Running the example

python -m examples.basic_sampling

CRAG dataset

CRAG (Comprehensive RAG Benchmark) is a rich and comprehensive factual question answering benchmark designed to advance research in RAG. The public version of the dataset includes:

2706 Question-Answer pairs
5 domains: Finance, Sports, Music, Movie, and Open domain
8 types of questions (see image below): simple, simple with condition, set, comparison, aggregation, multi-hop, post-processing heavy, and false premise

The datasets crag_task_1_and_2_dev_v4_subsample_*.json.bz2 have been created from the dataset crag_task_1_and_2_dev_v4.jsonl.bz2 available on CRAG's GitHub repository. For an easier handling and download of the dataset, we have used our CRAG sampler to split the 2706 rows of the original file in 5 subsamples, following the procedure below:

We have created a new label answer_type, classifying the answers in 3 categories:
- invalid for any answer == "invalid question"
- no_answer for any answer == "i don't know"
- valid for any other answer
We have considered the labels answer_type, domain, question_type and static_or_dynamic and performed stratified sampling, splitting the datasets in 5 subsamples. Each subsample has thus the same statistical properties of the full dataset.

We report below the data schema as provided in CRAG's GitHub repository.

Data Schema

Field Name	Type	Description
`interaction_id`	string	A unique identifier for each example.
`query_time`	string	Date and time when the query and the web search occurred.
`domain`	string	Domain label for the query. Possible values: "finance", "music", "movie", "sports", "open". "Open" includes any factual queries not among the previous four domains.
`question_type`	string	Type label about the query. Possible values include: "simple", "simple_w_condition", "comparison", "aggregation", "set", "false_premise", "post-processing", "multi-hop".
`static_or_dynamic`	string	Indicates whether the answer to a question changes and the expected rate of change. Possible values: "static", "slow-changing", "fast-changing", and "real-time".
`query`	string	The question for RAG to answer.
`answer`	string	The gold standard answer to the question.
`alt_ans`	list	Other valid gold standard answers to the question.
`split`	integer	Data split indicator, where 0 is for validation and 1 is for the public test.
`search_results`	list of JSON	Contains up to `k` HTML pages for each query (`k=5` for Task #1 and `k=50` for Task #3), including page name, URL, snippet, full HTML, and last modified time.

Search Results Detail

Key	Type	Description
`page_name`	string	The name of the webpage.
`page_url`	string	The URL of the webpage.
`page_snippet`	string	A short paragraph describing the major content of the page.
`page_result`	string	The full HTML of the webpage.
`page_last_modified`	string	The time when the page was last modified.