A course on practical natural language processing (NLP) was conceived and taught for the benefit of MTech students of IISc, from the Artificial Intelligence programme, Centre for Computational and Data Sciences (CDS), and Centre for Computer Science and Automation (CSA). The instructors were Monojit Choudhury (from Microsoft Turing), and Kabir Ahuja and Harshita Diddee (from Microsoft Research).
The course aimed to give students an accelerated understanding of building practical engineering NLP systems. By demonstrating concepts of practical use through widely adopted frameworks like HuggingFace and PyTorch, the course illustrated the end-end functioning of language models, particularly neural machine translation systems. A hands-on approach was adopted to demonstrate the development of these systems under data constraints typically seen in low-resource languages and specific domains, and also give a brief overview on utilising the developed models in user-facing applications like ChatBots. Six classes, each of 3 hours duration, were scheduled between February to March 2023. During the last class, the students presented their projects.