
Nonprofit · Tehran, Iran
Hooshvare Research Lab
A small non-profit Persian NLP lab that published ParsBERT, the first widely used monolingual BERT for Persian, along with Persian GPT-2 and NER models.
Tehran, Iran
- Type
- Nonprofit
- Based in
- Tehran, Iran
- Key papers
- 1
- Featured projects
- 8
About
Hooshvare Research Lab is a Tehran-based group that publishes open models and datasets for Persian natural language processing. Its GitHub organization describes it as a platform for sharing information about artificial intelligence concepts with enthusiasts. The lab's work is published through the Hooshvare organizations on GitHub and Hugging Face.
Its best-known release is ParsBERT, a monolingual BERT-style model for Persian described in a May 2020 preprint by Mehrdad Farahani, Mohammad Gharachorloo, Marzieh Farahani and Mohammad Manthouri. The paper reports gains over multilingual BERT on sentiment analysis, text classification and named entity recognition. The lab later added a Persian GPT-2 model, NER models, a news summarization dataset and workshop materials.
Code and weights are released under permissive licenses such as MIT and Apache 2.0, which has made the models a common baseline for Persian-language research and applications.
Mission and ambitionsThe lab aims to make capable language models and datasets for Persian freely available, so that researchers and developers working in the language are not limited to multilingual models built for other languages.
Known for
- 01
ParsBERT, a monolingual Persian BERT
- 02
Persian GPT-2 and NER models
- 03
The pn-summary Persian summarization dataset
Milestones
- 2020
ParsBERT preprint
Introduced a Persian-specific BERT-Base model that outperformed multilingual BERT on three downstream tasks.
- 2020
pn-summary dataset
Released a Persian news summarization dataset alongside ParsBERT-based summarization work.
- 2021
ParsBERT in Neural Processing Letters
Peer-reviewed journal version of the ParsBERT work published in volume 53.
- 2021
Persian GPT-2 and NER models
Published ParsGPT and ParsNER repositories to extend the Persian model family beyond encoders.
- 2021
ParsBERT workshop
Published workshop material covering ParsBERT's architecture and applications.
Open source
Persian BERT.
Persian summarization dataset.
Persian GPT2.
Persian PDF text extraction.
Persian BERT.
Persian BERT.
ParsBERT.
- HooshvareLab/pn_summarydataset
Summarization.
Research papers
Notable releases
Openness
Open models.
People
- Mehrdad FarahaniLead author of ParsBERT
Public member of the Hooshvare GitHub organization
- Mohammad ManthouriCo-author of the ParsBERT paper
Public member of the Hooshvare GitHub organization
- Mohammad GharachorlooCo-author of the ParsBERT paper
- Marzieh FarahaniCo-author of the ParsBERT paper
Listed from the lab's own pages and public profiles.
Contact and community
The GitHub org page is the lab's public home; use issues on its repos.
Sources
Compiled from the lab's public pages, Hugging Face and GitHub. Last checked 2026-11-17. Something wrong or out of date? Tell us on Discord.



