A simple tutorial on how to use distilBERT or BERT for a NLP classification problem on Google Colab using TensorFlow and Hugging Face pre-trained models. The used data are reviews scraped from the web and pulled through Google Translate to obtain an English review dataset together with the corresponding ratings. Two Colab notebooks are provided. The first notebook uses the simple "fit" command to train the model. The second notebook is more advanced and uses a custom training loop, while also incorporating TensorBoard to log experiments.