Skip to content

Latest commit

 

History

History
40 lines (18 loc) · 1004 Bytes

whats_happening.md

File metadata and controls

40 lines (18 loc) · 1004 Bytes

number_classifier

This is a basic number classifier which learns number from various writing styles.

we use the data set mnist which avaible in tensorflow itself.

then we load the data in x,y_train and x,y_test.

the use a simple nueral net model which has:-

1)One Flatten layer

2)Three fully connected layers with a (relu) activation function.

3)Output layer with a (sigmoid) activation function.

the model was fitted by using optimizer='adam' and loss_function='sparse_categorical_crossentropy'and Epoch=10.

then saved it by using model.save("name_of_model.model")

this model has

           acc=0.9929
           
           loss=0.0212
       
       val_acc=0.9772
       
       val_loss=0.0531

For more info checkout my code at ("https://github.com/rustombhesania/number_classifier").

Thnx for watching.

the reference was taken from ("https://pythonprogramming.net/introduction-deep-learning-python-tensorflow-keras/")