Code question and answering (QA) systems have demonstrated potential as practical tools for supporting students in introductory programming courses by providing automated assistance with understanding and debugging source code snippets. Although several studies have examined such systems, they primarily focus on a single language, typically English, resulting in limited investigation of other languages, such as Portuguese, or multilingual settings. This work investigates the performance of three Transformer-based models, CodeT5, Flan-T5, and T5, on a QA task involving the Java programming language. Experiments were conducted using the CodeQA dataset to fine-tune and evaluate the models across three scenarios: (i) the original English data, (ii) a Brazilian Portuguese translation developed in this work, and (iii) a bilingual version combining English and Portuguese data. Model performance was assessed using the ROUGE-L and BERTScore evaluation metrics. The results show that the base version of the CodeT5 model consistently outperformed the other models across all evaluated scenarios. The main contributions of this work are the creation of a Portuguese version of the CodeQA dataset and a comparative analysis of the multilingual performance of T5-based models in code-related QA tasks.

The ornamental rock industry in Brazil is distinguished by its diverse assortment of rock types, presenting a unique challenge in classification due to its inherent subjectivity and reliance on expert judgment. to address this problem, the present study introduces a publicly accessible database encompassing 12 distinct classes of ornamental rocks, including granite, marble, and quartzite. This database comprises 1,798 images sourced directly from entities within the stone sector. We employ and compare the performance of seven neural network models for ornamental rock image classification: VGG16, VGG19, ResNet50, ResNet101, Xception, InceptionV3, and the Vision Transformer (ViT). Our empirical analysis reveals that the ViT model outperforms conventional architectures, achieving an accuracy rate of 98.36%.