Can we do fine-tuning on NLP problems?

[ashikshafi08]

You can do fine-tuning in NLP problems very similar to how you can do fine-tuning in computer vision problems.

One option to do so is when downloading embedding from TensorFlow Hub, set trainable=True.

See an example of this in the code sample for the Albert TF Hub model: https://tfhub.dev/google/albert_base/3