aiqbench analysis
This study evaluates parameter-efficient fine-tuning (PEFT) and prompt engineering techniques for hate speech detection in Roman Urdu, a low-resource language variant. By comparing zero-shot inference against methods like LoRA and custom prompts, the research addresses critical challenges in handling informal grammar and spelling variations with limited data. The findings offer practical insights for deploying efficient AI models to mitigate toxic content in specific linguistic contexts without requiring massive computational resources.
Hate Speech Classification In Roman Urdu: A Comparative Study On Parameter Efficient Fine-Tuning And Prompt Engineering
arXiv:2608.21408v1 Announce Type: new Abstract: Due to the widespread accessibility of the internet and social media, toxic and hateful con-tent has grown exponentially, causing significant distress and negative societal impacts. Ro-man Urdu, a low-resource language used in Pakistan and among Urdu-speaking communities worldwide, presents additional challenges because of its informal grammar, inconsistent sen-tence structures, and multiple variations in word spellings. This research aims to identify the most effective techniques for hate speech classification in such low-resource settings with limited data. To address this, the study investigates and compares the latest approaches, in-cluding prompt tuning, parameter-efficient fine-tuning (PEFT) using LoRA, and prompt en-gineering, under various experimental configurations. To achieve this objective, four exper-iments were designed. The first experiment involved direct inferencing with LLMs without any fine-tuning, to evaluate how well these models understand Roman Urdu in a zero-shot setting, especially given limited data. The second experiment utilized parameter-efficient fine-tuning (PEFT) with LoRA, which updates only a small subset of parameters, thereby reducing computational cost. The third experiment explored prompt tuning with both mixed and manually crafted prompts, using very small sets of training examples relative to the entire dataset, making it computationally efficient as well. Finally, the fourth experiment applied prompt engineering through zero-shot and few-shot learning, relying solely on care-fully designed instruction prompts for classification without further training.
Abstract:Due to the widespread accessibility of the internet and social media, toxic and hateful con-tent has grown exponentially, causing significant distress and negative societal impacts. Ro-man Urdu, a low-resource language used in Pakistan and among Urdu-speaking communities worldwide, presents additional challenges because of its informal grammar, inconsistent sen-tence structures, and multiple variations in word spellings. This research aims to identify the most effective techniques for hate speech classification in such low-resource settings with limited data. To address this, the study investigates and compares the latest approaches, in-cluding prompt tuning, parameter-efficient fine-tuning (PEFT) using LoRA, and prompt en-gineering, under various experimental configurations. To achieve this objective, four exper-iments were designed. The first experiment involved direct inferencing with LLMs without any fine-tuning, to evaluate how well these models understand Roman Urdu in a zero-shot setting, especially given limited data. The second experiment utilized parameter-efficient fine-tuning (PEFT) with LoRA, which updates only a small subset of parameters, thereby reducing computational cost. The third experiment explored prompt tuning with both mixed and manually crafted prompts, using very small sets of training examples relative to the entire dataset, making it computationally efficient as well. Finally, the fourth experiment applied prompt engineering through zero-shot and few-shot learning, relying solely on care-fully designed instruction prompts for classification without further training.
Artificial Intelligence (cs.AI)
arXiv:2608.21408 [cs.AI]
(or arXiv:2608.21408v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2608.21408 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 305 words
Related coverage
Efficient Adaptation of LLMs for Hate Speech Detection in Low-Resource Languages: A Comparative Study on Roman Urdu
arXiv · Aug 21, 2026
Bodhan AI Releases Four Indic Models for OCR, Translation and Speech
Analytics Vidhya · Sep 10, 2026
Joint Text-Audio Alignment for EEG-to-Text Decoding in Chinese Speech Production and Perception
arXiv · Jul 29, 2026
TeamHerald@CHIPSAL 2026: Hate Speech Detection and Sentiment Analysis of Nepali Memes using Transformer-based Architectures and Ensemble Learning
arXiv · Jun 9, 2026