Trigger Word Detection PreTrained Model

The pre-trained model as mentioned has been trained on 4000 training examples. If we wish to fit such a large data,what should be the appropriate values of parameters in i.e. the values of batch_size and number_of_epochs for achieving such high accuracy?

Those values are typically set by experimentation.

You want the training to give “good enough” results, but you don’t want the training to take longer than your are willing to wait, or require a computer larger than you an afford to use.