Datasets:
license: apache-2.0
dataset_info:
features:
- name: title
dtype: string
- name: text
dtype: string
- name: added
dtype: string
- name: created
dtype: string
- name: id
dtype: string
- name: source
dtype: string
- name: version
dtype: string
splits:
- name: pes2o_abstract
num_bytes: 46073557231
num_examples: 30569017
download_size: 26787550136
dataset_size: 46073557231
configs:
- config_name: default
data_files:
- split: pes2o_abstract
path: data/pes2o_abstract-*
task_categories:
- question-answering
- text-classification
- feature-extraction
- sentence-similarity
language:
- en
tags:
- biology
- medical
- chemistry
Introducing Pes2o-X, also known as Pes2o-Abstract-X, a derived dataset from the original Pes2o dataset released by Allen AI. The Pes2o dataset aimed to provide a large corpus of open-access research papers, including both abstracts and full text. However, it required pre-processing before the abstracts could be used for training or fine-tuning machine learning models.
At LAION AI, we initiated a project called X, focusing on developing high-quality training corpora from scratch, reorganising, and maintaining high-quality research data dumps (e.g., Wikipedia). As part of this initiative, we reorganized the original Pes2o dataset version 2, extracted the abstracts, and compiled our Pes2o-Abstract-X.
It's important to note that Pes2o-X and our entire X project stem from an ongoing commitment to developing high-quality abstractions from these datasets. Our goal is to release them to the public to support the development of advanced artificial intelligence models and enhance existing large language model pipelines with modular features.
Dataset information
Index: Pes2o v2 dataset 2023
How many research papers’ abstract present?
- 30.57M
WE PRESERVED ALL THE ORIGINAL METADATA.