Part 3: Fine-Tuning RoBERTa on 800 Examples: What the Numbers Actually Mean

An honest account of training a transformer on a small, imbalanced dataset, and how to read the results without fooling yourself Continue reading on Medium »