early_stopping Class Reference

Public Member Functions
	__init__ (self)
	check (self, cost)

Public Attributes
int	counter = 0
	counts how many times training is not getting better
	best_result = np.inf
	saves best training result

Detailed Description

 Using class to stop training early if it's not getting better

Definition at line 20 of file relations.py.

Constructor & Destructor Documentation

__init__()

__init__

(

self

)

 init class 

Definition at line 23 of file relations.py.

    def __init__(self):
        """ init class """
        
        self.counter = 0
        
        self.best_result = np.inf
 

Member Function Documentation

check()

check	(		self,
			cost )

Check if validation result is better than the best validation result.
Decide if training should be continued.

Definition at line 30 of file relations.py.

    def check(self, cost):
        """
        Check if validation result is better than the best validation result.
        Decide if training should be continued.
        """
        if cost < self.best_result:
            self.counter = 0
            self.best_result = cost
        else:
            self.counter += 1
            if self.counter >= 20:
                return False
        return True
 
 

Member Data Documentation

best_result

best_result = np.inf

saves best training result

Definition at line 28 of file relations.py.

counter

int counter = 0

counts how many times training is not getting better

Definition at line 26 of file relations.py.

---

The documentation for this class was generated from the following file:

• mva/examples/tensorflow/relations.py