NLP and model based systems engineering

LLMs have demonstrated a remarkable ability to generate text taking as input images, text, audio and video. They are able to achieve higher performance than traditional neural methods and pre-trained language models without the need of supervised training.
The project will examine different approaches for multimodal LLM-based NLP to address
complex and fine-grained tasks such as reasoning in model-based systems engineering. The
PhD will delve into LLM architectures, data augmentation methods, multi-task and domainspecific LLMs, prompting engineering and interpretability.