Performance Comparison of T5 and Marian Machine Translation Models for English to French and German Translation
Performance Comparison of T5 and Marian Machine Translation Models for English to French and German Translation
sarinakasaiyan1 Reyhaneh SafariNejad Meybodi2 nava eslami3
1) Master s student in Artificial Intelligence and Robotics, Islamic Azad University, North Tehran Branch
2) Bachelor in Mechanical Engineering, Politecnico di Torino.Italy
3) University Professor, Islamic Azad University, North Tehran Branch
IOI Code
:
XKBA-AAAFFC
Publication :
3rd International Conference on Recent Advances in Engineering, Innovation and Technology - Belgium(eitconf.com/3rd)
Abstract :
This paper presents a comparative analysis of two prominent machine translation models, T5 and Marian, focusing on their performance in translating English text into French and German. The primary objective is to evaluate and contrast the quality of translations generated by these models, employing the METEOR score as a key metric for assessing translation accuracy. The methodology involves loading pre-trained models, defining input texts, generating translations using both T5 and Marian, and subsequently computing METEOR scores to quantify translation quality. Experimental results indicate that the Marian model exhibits superior performance in English-to-French translation compared to T5. However, T5 demonstrates competitive results in translating English into German. Furthermore, the inference time of each model is assessed and reported. The findings underscore that the optimal choice between T5 and Marian is contingent upon the target language and specific application requirements, highlighting the significance of employing standardized evaluation metrics such as METEOR to enhance machine translation systems.
Keywords : machine translation t5 marian meteor score english to french translation english to german translation