Early stopping the training

How do i stop training when my model has reached a desired training accuracy lets say 99 percent .

Please watch the lectures and the practice labs for explanation on keras callbacks.
Here’s the link on the implementation detail.

Is it a good practise to stop training when training accuracy has reached to 99 percent or should I stop training when it has reached to 100 percent

There is no one rule for when to stop training a model. Stopping criteria is specific to a problem which is often defined in collaboration with the consumer of the model.