Investigation of the performance and interpretability of two models, a large language models (LLM) and a small-scale model, trained on low-resource language pairs Xhosa Zulu and Tswana-Zulu
Domaine:
natural language processingType de record:
datasetmodel
Créateur:
Gom
Éditeur:
Hôte:
This submission contains images and datasets used in the research for a dissertation "Assessing interpretability in machine translation models for low-resource languages".The images include machine translation model-generated heatmaps and machine translation model-generated translations.The datasets include the following:BLEU scores from model training and graphsPost model evaluation results for MQM and graphsPost model evaluation results for ESS and resultsSmall-scale model training results comparisons with generated graphs [to evaluate early stopping]
Similaires
Scaling Performance of Models Trained on Artificially Code-Switched Data for Unseen Low-Resource Language PairsEmploying large language models in Swahili, a low-resource languagePerformance of Zero-Shot Cross-Lingual Retrieval Models on Low-Resource Language PairsAdapting Large Language Models to Low-Resource Tibetan: A Two-Stage Continual and Supervised Fine-Tuning Studyshruthi0429/BLOOMZ-and-mT5-Fine-Tuning-Optimizing-Large-Language-Models-for-a-Low-Resource-LanguageInvestigation of Language Impairment in Zulu