AI Model Catalog | Microsoft Foundry Models

microsoft-biomednlp-pubmedbert-large-uncased-abstract

Version: 4

Hugging Face•Last updated December 2025

MSR BiomedBERT-large (abstracts only)

This model was previously named "PubMedBERT large (abstracts)".
You can either adopt the new model name "microsoft/BiomedNLP-BiomedBERT-large-uncased-abstract" or update your transformers library to version 4.22+ if you need to refer to the old name.

Pretraining large neural language models, such as BERT, has led to impressive gains on many natural language processing (NLP) tasks. However, most pretraining efforts focus on general domain corpora, such as newswire and Web. A prevailing assumption is that even domain-specific pretraining can benefit by starting from general-domain language models. Recent work shows that for domains with abundant unlabeled text, such as biomedicine, pretraining language models from scratch results in substantial gains over continual pretraining of general-domain language models. Followup work explores larger model sizes and the impact of these on performance on the BLURB benchmark. This BiomedBERT is pretrained from scratch using abstracts from PubMed .

Citation

If you find BiomedBERT useful in your research, please cite the following paper:

@misc{https://doi.org/10.48550/arxiv.2112.07869,
  doi = {10.48550/ARXIV.2112.07869},
  url = {https://arxiv.org/abs/2112.07869},
  author = {Tinn, Robert and Cheng, Hao and Gu, Yu and Usuyama, Naoto and Liu, Xiaodong and Naumann, Tristan and Gao, Jianfeng and Poon, Hoifung},
  keywords = {Computation and Language (cs.CL), Machine Learning (cs.LG), FOS: Computer and information sciences, FOS: Computer and information sciences},
  title = {Fine-Tuning Large Neural Language Models for Biomedical Natural Language Processing},
  publisher = {arXiv},
  year = {2021},
  copyright = {arXiv.org perpetual, non-exclusive license}
}

microsoft/BiomedNLP-PubMedBERT-large-uncased-abstract is a pre-trained language model available on the Hugging Face Hub. It's specifically designed for the fill-mask task in the transformers library. If you want to learn more about the model's architecture, hyperparameters, limitations, and biases, you can find this information on the model's dedicated Model Card on the Hugging Face Hub . Here's an example API request payload that you can use to obtain predictions from the model:

{
  "inputs": "[MASK] is a tyrosine kinase inhibitor."
}

Model Specifications

LicenseMit

Last UpdatedDecember 2025

ProviderHugging Face

Quick Start