datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
PMC
Data collected from PMC
Only CC-BY, CC-BY-SA licenses are included.
For all records, check the jsonl files in the data folder
PMC-Clinical-VQA
PMC-VQA: A Large-Scale Visual Question Answering Dataset for Clinical Figures
This dataset contains over 1,700,000 Visual Question Answering (VQA) samples derived from figures and charts in biomedical articles from PubMed Central (PMC).
This is a preliminary release. A full dataset card and an accompanying research paper are currently in preparation.
Raw version of this dataset with licenses and metadata can be found on Hugging Face: DermaVLM/pmc_clinical_VQA_raw
Dataset… See the full description on the dataset page: https://huggingface.co/datasets/DermaVLM/PMC-Clinical-VQA.PMC-VQA-1
PMC-VQA-1
This dataset is a streaming-friendly version of the PMC-VQA dataset, specifically containing the "Compounded Images" version (version-1). It is designed to facilitate efficient training and evaluation of Visual Question Answering (VQA) models in the medical domain, straight from the repository
Dataset Description
The original PMC-VQA dataset, available at https://huggingface.co/datasets/xmcmic/PMC-VQA, comprises Visual Question Answering pairs derived from… See the full description on the dataset page: https://huggingface.co/datasets/hamzamooraj99/PMC-VQA-1.PMC-VQA-refinedpmc_open_access_figure_commMedVLThinker-pmc_vqaCode: https://github.com/UCSC-VLAA/MedVLThinker
Project Page: https://ucsc-vlaa.github.io/MedVLThinker/
📊 Datasets
Available Datasets
Our project provides several curated datasets for medical vision-language understanding and training:
Dataset
Modality
Description
Download
MedVLThinker-m23k-tokenized
Text-only
Tokenized version of the m23k dataset
🤗 HF
MedVLThinker-pmc_vqa-gpt_4o_reasoning-tokenized
Image-Text
Tokenized PMC-VQA dataset with GPT-4o generated… See the full description on the dataset page: https://huggingface.co/datasets/UCSC-VLAA/MedVLThinker-pmc_vqa.soda-vec-data-full_pmc_title_abstract
SODA-VEC Clean Dataset
This is a cleaned and filtered version of the SODA-VEC dataset, containing high-quality biomedical title-abstract pairs from PubMed Central (PMC) articles.
Dataset Overview
Total examples: 26,573,900
Training set: 26,473,900 examples (99.6%)
Validation set: 50,000 examples (0.2%)
Test set: 50,000 examples (0.2%)
Quality Filtering Applied
This dataset has been processed with the following quality filters:
Abstract Length… See the full description on the dataset page: https://huggingface.co/datasets/EMBO/soda-vec-data-full_pmc_title_abstract.pmc_open_access_figure_noncommpmc_vqa_baseeuro_pmc_chemistry_paperstrain_valid_split_pmc_neuroscience_2002-2022_filtered_subsetData from PubMed for abstracts and PubMed Central Open Access Subset (PMC OAS) for full-text articles using the Entrez Programming Utilities (E-utilities) API
and the pubget Python package, respectively. The data span publication dates from 2002 to 2022. For science general journals, a keyword filter of ``Neuroscience" was applied (all sourced journals are below).
Data extraction efforts yielded 332,807 abstracts and 123,085 full-text articles, totaling 1.3 billion tokens.
Figures and tables… See the full description on the dataset page: https://huggingface.co/datasets/BrainGPT/train_valid_split_pmc_neuroscience_2002-2022_filtered_subset.pmc_vqa_train_c1_gl2_noshufflePMC-VQA_smallpmc-oa-markdown
PubMed Central (PMC) Open Access in Markdown
This is a subset filtered for the words: obesity, weight loss, and diabetes.
The purpose of this extraction is to further research in Biomedicine + NLP. While there are many biomedical datasets, most of them focus on the clinical space, which is often different from the early developments and discovery.
The Markdown format includes figure description, tables, and a YAML header with some metadata like original license. It is filtered to… See the full description on the dataset page: https://huggingface.co/datasets/casperhansen/pmc-oa-markdown.pmc_vqa_test_gl1pmc_vqa_nonegrab_pens_to_bin_3pmc_vqa_test_none_testeuro_pmc_chemistry_abstractsPMC-VQA
PMC-VQA - PubMed Central Visual Question Answering
Description
This dataset contains visual question answering data from PubMed Central medical literature. Questions are designed as multiple choice format requiring understanding of medical figures and images. 16 reasoning traces were collected for each example in this task by sampling with GPT-4o, available in the responses column. We greatly appreciate and build from the original data source available at… See the full description on the dataset page: https://huggingface.co/datasets/OctoMed/PMC-VQA.soda-vec-data-full_pmc_title_abstract_paired
SODA-VEC Paired Dataset for Negative Sampling
This is a paired version of the SODA-VEC dataset, specifically formatted for negative sampling training with MultipleNegativesRankingLoss.
Dataset Overview
Total examples: 26,573,900
Format: Paired (anchor-positive) for contrastive learning
Source: EMBO/soda-vec-data-full_pmc_title_abstract
Purpose: Training sentence transformers with negative sampling
Data Format
Each example contains:
anchor (string): The title… See the full description on the dataset page: https://huggingface.co/datasets/EMBO/soda-vec-data-full_pmc_title_abstract_paired.pmc-train-2600000-to-2700000-GemmaTokenspmc-vqa-robustnesspmc-train-100000-to-200000-GemmaTokenseurope_pmc_articles_part_1
Dataset Card for "europe_pmc_articles_part_1"
More Information needed
pmc_vqa_test_c1europe_pmc_articles_part_2
Dataset Card for "europe_pmc_articles_part_2"
More Information needed
pmc-train-4900000-to-5000000-GemmaTokensPMC-CaseReport_original
PMC-CaseReport_original Dataset
PMC-CaseReport_original Dataset
Daraset Structure
Sample
This is the text parts and the figure parts can be dowloaded from https://pan.baidu.com/s/1Src_rhXsaOFp8zJ_3zMFsQ?pwd=p3ne.
Dataset Structure
PMC-CaseReport (Original version: 884K VQA pairs for taining and of 240K for testing images).
The dataset can be loading following huggingface datasets rule:
from datasets import load_dataset
dataset =… See the full description on the dataset page: https://huggingface.co/datasets/chaoyi-wu/PMC-CaseReport_original.pmc-train-1200000-to-1300000-GemmaTokens