How you approach developing a language model with small amounts of data
Sigiloso
Start with a basic model, e.g Naïve Bayes as a baseline, choose to either increase model complexity or use pretrained models and fine tune them. For additional performance boosting create synthetic data via paraphrasing.