This is a pretty straight forward problem and a good fit for a standard text classifier as well.
Here is an example of fine-tuning a model with txtai: https://colab.research.google.com/github/neuml/txtai/blob/ma...