Datasets:

Modalities:
Text
Formats:
json
ArXiv:
Libraries:
Datasets
pandas
License:

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

PLEASE NOTE THAT THIS DATASET CONTAINS OFFENSIVE AND SENSITIVE CONTENT. WE HAVE TAGGED CONTENT FOR RATED G, PG, R, AND X CONTENT. OUR RATINGS COULD BE WRONG.*

Description of Dataset

This is a subset of the Falcon Refinedweb dataset, with tagging performed using the taggers in the ontocord/riverbed repository. This subset includes approximately 100K items (485MB or approximately 100M tokens). The items have been filtered to create a relatively diverse dataset of various tags, and content ratings, while keeping quality and textbook ratings relatively moderate. Tags are added to the beginning of some of the items for ratings (PG, R, X) and for content when they have been detected to be science based, how-to, legal, etc. The original tagging was performed by Ontocord for the Aurora-M project, and subsequently further tagged for textbook quality by @kenhktsui here. It was further filtered to create this subset.

Original Refinedweb Dataset

The authors of Refinedweb writes:

RefinedWeb is built through stringent filtering and large-scale deduplication of CommonCrawl; we found models trained on RefinedWeb to achieve performance in-line or better than models trained on curated datasets, while only relying on web data. See the 📓 paper on arXiv for more details on Refinedweb itself.

Purpose

The purpose of the dataset is to investigate filtering of data and creating performant mixtures of data for model training.

Disclaimer

This dataset is provided for AI safety research purposes only. We make no claims to ownership of the data and release any of our annotations under ODC-By because that was the original RefinedWeb license. We believe usage of web crawled data is permitted under fair use, but you use this dataset at your own risks, and we disclaim all warranties and liabilities including warranties for non-infringement. We tried to rate the items based on flagged words detection, and we may be wrong in the detection.

NO PERSONALLY IDENTIFIABLE INFORMATION REDACTION WAS PERFORMED. ALL SUCH INFORMATION ARE FROM THE ORIGINAL DATASET AND THE CORRESPONDING WEB PAGE. YOU MUST BE AT LEAST 18 YEARS OLD AND ACKNOWLEDGE THE SENSITIVE NATURE OF THIS DATASET IN ORDER TO ACCESS AND USE THIS DATASET.

Contact Us

Please reach out to us in the discussion link above if you have questions.

Downloads last month
50