How to create a good model with small dataset?

Hi. I am trying to make a toy model with ~ 600 commands(25 minutes). Hyperparameters are below:

  1. Hidden layer 64
  2. Dropout rate 0.3

After some epochs loss goings up. And the difference between train and validation losses is big.
But loss is too high.
How can I make an accurate model?

Ask the coqui guys. Read more in this post.

Thanks a lot. I will try to find the answer with the help of coqui forum members.