This paper presents an analysis of the application of the Low-Rank Adaptation method (LoRA) for the task of monolingual text generation in Uzbek. We used the T5-base, T5-Large and uzT5 models to determine which one shows the best results when using LoRA, and also compared their performance with traditional fine tuning. The text from 5,000 news items from the Kun.uz platform was used as a dataset of which 4,000 were used for training and 1,000 for testing. The performance of the models was evaluated using the metrics BLEU, ROUGE-1, ROUGE-2, ROUGE-L and ROUGE-LSUM. Our results showed that the uzT5-base model with LoRA r=256 and α=512 parameters demonstrate the highest performance among all the considered models, providing the best values of the ROUGE and BLEU metrics with a moderate number of training parameters, which makes it more computationally efficient compared to mT5-Large.
| Mualliflar | Адилова, Фатима, Давронов, Рифкат, Кушмуратов, Самариддин |
|---|---|
| Jurnal | Рақамли технологияларнинг назарий ва амалий масалалари |
| Nashr sanasi | 2024-10-09 |
| Jild | 7 |
| Son | 3 |
| Betlar | 112-116 |
| Til | Rus |
| DOI | 10.62132/ijdt.v7i3.204 |
DOI: 10.62132/ijdt.v7i3.204 · Maqolaning asl sahifasi
низкоранговые адаптации, T5-base, T5-Large, uzT5, сжатые модели
This article reviewed the methods and algorithms for processing and diagnosing X-ray images that are widely used in medicine today. In order to improve the implementation of deep learning in modern medicine and further…
While non-linear S-tables are important in the security of symmetric cryptosystems, their hardware implementation also requires a large number of logic elements compared to the rest of the mappings. This makes it…
This article is dedicated to studying the bending characteristics of thin plates under the influence of non-stationary stresses. The deformation process of plates under non-stationary loads is analyzed based on…
In the article, a study was conducted to comprehensively study the important factors affecting the process of migration and diffusion of harmful substances and dust particles in the atmosphere. The physical and…
In this study, yawning, which is considered one of the physiological deviations of humans, mouth opening state classification and several machine learning algorithms were used. The CNN model outperformed traditional…
Methods have been developed to optimize the identification of pollen grains using statistical, dynamic, textural and specific image characteristics. Mechanisms for point and nonlinear verification of the correspondence…
Magnetoelastic materials are gaining significant importance in research aimed at utilizing new technologies, such as smart materials and nanotechnology, which may lead to the development of new devices and systems in…
Exploration of an unknown environment by a mobile robot during SLAM processes is a fundamental challenge. In this study, an extended iteration of the Rmap algorithm is presented by modifying its Exploration submodule…
Mathematical modeling is the main method and means for understanding objects, processes, and phenomena occurring in nature. It is used in all areas of science, technology, and the national economy, including solving…
This research work aims to develop a model of the Uzbek language using artificial neural networks to improve natural language processing (NLP) capabilities. The article classifies various methods of creating language…
Рақамли технологияларнинг назарий ва амалий масалалари — barcha maqolalar