Dataset Card for WikiBio GPT-3 Hallucination Dataset

GitHub repository: https://github.com/potsawee/selfcheckgpt Paper: SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models

  Dataset Summary

We generate Wikipedia-like passages using GPT-3 (text-davinci-003) using the prompt: This is a Wikipedia passage about {concept} where concept represents an individual from the WikiBio dataset. We split the generated passages into… See the full description on the dataset page: https://huggingface.co/datasets/potsawee/wiki_bio_gpt3_hallucination.

Dataset contains 728,321 biographies from wikipedia. For each article, it provides the first paragraph and the infobox (both tokenized).

WikiBio is constructed using Wikipedia biography pages, it contains the first paragraph and the infobox tokenized. The dataset follows a standarized table format.

To use this dataset:

import tensorflow_datasets as tfds

ds = tfds.load('wiki_bio', split='train')
for ex in ds.take(4):
 print(ex)

See the guide for more informations on tensorflow_datasets.

Text summarization and data-to-text generation datasets

This is the data-to-text generation datasets collected by TextBox, including:

WebNLG v2.1 (webnlg) WebNLG v3.0 (webnlg2) WikiBio (wikibio) E2E (e2e) DART (dart) ToTTo (totto) ENT-DESC (ent) AGENDA (agenda) GenWiki (genwiki) TEKGEN (tekgen) LogicNLG (logicnlg) WikiTableT (wikit) WEATHERGOV (wg).

The detail and leaderboard of each dataset can be found in TextBox page.

WikiBioCTE is a dataset for controllable text edition based on the existing dataset WikiBio (originally created for table-to-text generation). In the task of controllable text edition the input is a long text, a question, and a target answer, and the output is a minimally modified text, so that it fits the target answer. This task is very important in many situations, such as changing some conditions, consequences, or properties in a legal document, or changing some key information of an event in a news text.

Dataset Card for ICE

This dataset is uesd for our work: In-Context Editing: Learning Knowledge from Self-Induced Distributions, and our code has been released on GitHub here.

  Dataset Sources

Our dataset is constructed based on KnowEdit, and we generate contexts for each data using GPT-4o.

  Dataset Structure

We evaluate our method using four datasets, WikiDatarecent, ZsRE, WikiBio, WikiDatacounter fact. These datasets encompass two knowledge editing tasks… See the full description on the dataset page: https://huggingface.co/datasets/Yofuria/ICE.

This is the WikiBio dataset released as part of IndicNLG Suite. Each example has four fields: id, infobox, serialized infobox and summary. We create this dataset in nine languages including as, bn, hi, kn, ml, or, pa, ta, te. The total size of the dataset is 57,426.

KnowEdit: A Benchmark of Knowledge Editing for LLMs

This README is about reproducing the paper A Comprehensive Study of Knowledge Editing for Large Language Models. You can use EasyEdit to load and use this benchmark.

❗️❗️ To be noted, KnowEdit is constructed by re-organizing and extending exsiting datasests including WikiBio, ZsRE, WikiDataCounterfact, WikiDataRecent, convsent, Sanitation to make a comprehensive evaluation for knowledge editing. Special thanks to the builders and… See the full description on the dataset page: https://huggingface.co/datasets/zjunlp/KnowEdit.

Dataset Card for "raw-wiki-bio-filtered"

More Information needed