Persian-Phi: Efficient Cross-Lingual Adaptation of Compact LLMs via Curriculum Learning

2025-12-08 · Amir Mohammad Akhlaghi, Amirhossein Shabani, Mostafa Abdolmaleki, Saeed Reza Kheradpisheh arxiv

The democratization of AI is currently hindered by the immense computational costs required to train Large Language Models (LLMs) for low-resource languages. This paper presents Persian-Phi, a 3.8B parameter model that challenges the assumption that robust multilingual capabilities require massive model sizes or multilingual baselines. We demonstrate how Microsoft Phi-3 Mini -- originally a monolingual English model -- can be effectively adapted to Persian through a novel, resource-efficient curriculum learning pipeline. Our approach employs a unique "warm-up" stage using bilingual narratives (Tiny Stories) to align embeddings prior to heavy training, followed by continual pretraining and instruction tuning via Parameter-Efficient Fine-Tuning (PEFT). Despite its compact size, Persian-Phi achieves competitive results on Open Persian LLM Leaderboard in HuggingFace. Our findings provide a validated, scalable framework for extending the reach of state-of-the-art LLMs to underrepresented languages with minimal hardware resources. The Persian-Phi model is publicly available at https://huggingface.co/amirakhlaghiqqq/PersianPhi.

📄 PDF Abstract BibTeX arXiv:2512.07454

Code (0)

등록된 구현이 없습니다.

Tasks

parameter-efficient fine-tuning Continual Pretraining

Extending LLMs to New Languages: A Case Study of Llama and Persian Adaptation

2024-12-17 · Samin Mahdizadeh Sani, Pouya Sadeghi, Thuy-Trang Vu, Yadollah Yaghoobzadeh 외

Large language models (LLMs) have made great progress in classification and text generation tasks. However, they are mainly trained on English data and often struggle with low-resource languages. In this study, we explor…

Classification parameter-efficient fine-tuning Text Generation Transfer Learning

PersianMedQA: Language-Centric Evaluation of LLMs in the Persian Medical Domain

2025-05-30 · Mohammad Javad Ranjbar Kalahroodi, Amirhossein Sheikholselami, Sepehr Karimi, Sepideh Ranjbar Kalahroodi 외

Large Language Models (LLMs) have achieved remarkable performance on a wide range of NLP benchmarks, often surpassing human-level accuracy. However, their reliability in high-stakes domains such as medicine, particularly…

Instruction Following Multiple-choice

PARSE: An Open-Domain Reasoning Question Answering Benchmark for Persian

2026-02-01 · Jamshid Mozafari, Seyed Parsa Mousavinasab, Adam Jatowt arxiv

Reasoning-focused Question Answering (QA) has advanced rapidly with Large Language Models (LLMs), yet high-quality benchmarks for low-resource languages remain scarce. Persian, spoken by roughly 130 million people, lacks…

Question Answering

Cross-lingual Few-shot Learning for Persian Sentiment Analysis with Incremental Adaptation

2025-07-15 · Farideh Majidi, Ziaeddin Beheshtifard arxiv

This research examines cross-lingual sentiment analysis using few-shot learning and incremental learning methods in Persian. The main objective is to develop a model capable of performing sentiment analysis in Persian us…

Incremental Learning Sentiment Analysis Few-Shot Learning

Winning with Less for Low Resource Languages: Advantage of Cross-Lingual English_Persian Argument Mining Model over LLM Augmentation

2025-11-25 · Ali Jahan, Masood Ghayoomi, Annette Hautli-Janisz arxiv

Argument mining is a subfield of natural language processing to identify and extract the argument components, like premises and conclusions, within a text and to recognize the relations between them. It reveals the logic…

Argument Mining