when I build a class funtion about StopMyCallbacks in model.fit() then I see it stop in the first steps 1/15 epochs and resulted greater than 72% so I want to know how many percent accuracy will fit of model and invoid overfitting
Perhaps you could plot out the accuracy/loss for both training sets and validation sets.
If you find an obvious deviation between training behaviour and validation behaviour, your model may be overfitted.
Just FYI. I’m also not sure, it’s hard to specify.