Talk · International Conference on Future Networks and Distributed Systems · December 8, 2025
Development of Uzbek Question-Answering Models through Fine-Tuning Approaches
Ilyos Khujayorov, Mannon Ochilov, Malika Abdullayeva, Orzimurod Kholmatov, Rashid Nasimov
Abstract
In this article, experiments were conducted on modern models such as T5, Flan-T5, mT5 and Llama for use in Retrieval-Augmented Generation (RAG) technologies. During the research, each model was fine-tuned based on a special domain data set consisting of Context, question, and answer columns, and the results were evaluated using evaluation metrics such as BLEU, ROUGE and METEOR. The use of RAG technology allows for searching for the necessary information from external knowledge bases and generating answers to questions. The results obtained are of great importance in the development of question-answer systems in the Uzbek language, including the development of auxiliary systems in areas such as call centers and public services.
More by Ilyos Khujayorov
- PadSink-Track: A Calibrated Framework for Padding-Sink-Preserving, Alignment-Tracked Selective Cross-Attention Key–Value Cache Access in CPU-Based Whisper Speech RecognitionTechnologies · 2026
- A Cache-Aware Compression Framework: Deriving the Compression Target and Stopping Rule for Transformers on CPU-Class HardwareAI · 2026
- Developing A Question-Answering System In The Uzbek Language Based On The Xlm-Roberta ModelAmerican Journal Of Applied Science And Technology · 2025
- An Ensemble of Convolutional Neural Networks for Sound Event DetectionMathematics · 2025 · Citations: 10