Renaissance Labs

Atlas / Middle East

Independent company · Tehran, Iran

Part AI

A Tehran-based developer of Persian-language models, known for the Dorna decoder LLMs and the TookaBERT and Tooka-SBERT encoders.

Persianlanguage modelsembeddings

Tehran, Iran

Type
Independent company
Based in
Tehran, Iran
Key papers
1
Featured projects
7

About

Part AI publishes open Persian-language models on Hugging Face. Its Dorna family consists of decoder-only models trained or fine-tuned on Persian data, including Dorna2 variants built on Llama 3.1.

Its encoder line includes TookaBERT in base and large sizes, described in an arXiv paper as a step forward for Persian language understanding, and the Tooka-SBERT sentence embedding models in small and large sizes.

The models serve Persian NLP developers and researchers who need open baselines for a language that is underrepresented in multilingual models.

Mission and ambitions

Part AI works to strengthen Persian language technology by releasing models and benchmarks that others can build on.

Known for

  1. 01

    Dorna Persian LLMs

  2. 02

    TookaBERT Persian encoders

  3. 03

    Tooka-SBERT embeddings

  4. 04

    Open Persian NLP models

Milestones

  1. 2024

    TookaBERT

    Published a family of Persian encoder models in base and large sizes, with an accompanying paper.

  2. 2024

    Dorna

    Released Persian-tuned decoder models starting with Dorna-Llama3-8B-Instruct.

  3. 2024

    Dorna2

    Followed with Dorna2 models built on Llama 3.1.

  4. 2025

    Tooka-SBERT

    Released small and large Persian sentence embedding models.

Open source

GitHubgithub.com/partdpai

Research papers

Notable releases

Openness

Weights on Hugging Face.

People

Listed from the lab's own pages and public profiles.

Contact and community

partdp.ai has a contact page.

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 Something wrong or out of date?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.

Labs nearby