metadata

task_categories:
  - visual-question-answering
tags:
  - MLLM
  - hallucination evaluation
  - VQA
  - perception
configs:
  - config_name: default
    data_files:
      - split: phd_ccs
        path: phd_webdataset/phd_ccs.tar
      - split: phd_sec
        path: phd_webdataset/phd_sec.tar
      - split: phd_icc
        path: phd_webdataset/phd_icc.tar
      - split: phd_base
        path: phd_webdataset/phd_base.tar

PhD Webdataset

This repository contains the packaged version of PhD. For a detailed introduction to PhD, please visit the official website.

Overview

The PhD Webdataset is designed to facilitate easy access and usage of the PhD dataset. It includes various fields in 'json' key. The data in this repo is totally the same as in PhD.

Installation

Ensure you have Hugging Face's datasets library installed. You can install it via pip:

pip install datasets

Usage

To load and use the PhD dataset, simply run the following code:

from datasets import load_dataset

phd = load_dataset("AIMClab-RUC/PhD-package")['phd_base']  # options include (phd_base, phd_sec, phd_icc, phd_ccs)

# see phd[0]['json']['question']

The loaded dataset includes the following fields:

question_id
question
answer
image
task
mode