Note (May 2025): This post is an updated version of my original 2020 walkthrough of using BERT for masked word prediction. I’ve refreshed the code to use the latest version of Hugging Face Transformers with PyTorch instead of TensorFlow, and clarified the explanation and examples.

Note (May 2025): This post is from 2020 and focuses on applied statistics in R. My recent work centers on large language models and neural architectures, but I’m keeping this post up as a snapshot of earlier work in my ML journey.