Objective

By the end of the course. students will be able to:

• Understand what knowledge is captured by self-supervised models for NLP
• Understand the challenges of finding relevant information and updating models
• Build NLP systems that can combine multiple sources of information for reasoning
• Chose an information retrieval system appropriate for the task

Format

• 4x4 hour sessions
• Start: 14:00, End: 18:00
• 2 hour lecture
• 2 hour hands-on lab (all labs will involve training transformer models with huggingface library)

Lab

• (if not installed) Install python (e.g. Miniconda)
  • https://docs.conda.io/en/latest/miniconda.html
• (Conda example) Create environment for summer school. (virtual env also OK)
  • conda create -n summerschool python=3.8
  • conda activate summerschool
• Install dependencies
  • pip install torch transformers datasets jupyter accelerate evaluate scikit-learn
• Make a jupyer environment
  • python -m ipykernel install --user --name summerschool --display-name "Summer School”
• Open jupyter notebok
  • jupyter notebook