DASC7606
Assignment II: Machine Translation
Optional Assignment:
• This assignment is optional due to the coming final exams
• The final assignment grades will be the maximum between Assignment 1 and 2
Objectives:
• Use the Hugging Face Hub for datasets: discover, download, and prepare translation datasets
• Load pretrained models and tokenizers from the community
• Build training pipelines using transformers, datasets, accelerate, and evaluate
• Fine-tune a translation model end-to-end for a chosen language pair (zh-sim->en)
• Evaluate translation quality using BLEU (via sacrebleu) and report results
• Develop debugging skills for identifying and fixing common deep learning issues in HF version