#Import files you need
    from deepNets import Net as nt
    from deepNets import Trainer as trn
    #Create a list of dictionaries to include your layers
    make_layers = []
    #Remember to specify the first layer as input, also note the layer type syntax
    make_layers.append({"layer_type":"input","inp":x})
    #You can add relu non linearities with hidden units accounting to any valid number
    make_layers.append({"layer_type":"relu","hidden_layers":100})
    #Normalization layers like batchnorm and layernorm are supported
    make_layers.append({"layer_type":"batchnorm"})
    #Always specify the last layer as loss
    make_layers.append({"layer_type":"loss","num_classes":10})
    #Create a object for neural net
    net = nt.Net()
    #This function constructs your neural net
    #You can also specify weight_scale = 1e-2 . Default weight_scale = 1e-3
    net.makeLayers(make_layers)
    #The Trainer function has net object as input
    #Specify training data(x,y) and validation data(x_val,y_val)
    #Set the update rules to sgd,rmsprop,adam etc and learning_rate and batch_size
    trainer = trn.Trainer(net,x,y,(x_val,y_val),update_rule="sgd",optim_config={'learning_rate':0.001},batch_size=100,verbose=True)
    #Setting verbose to true will print values of training accuracy and val accuracy
    trainer.train()

    #Import files you need
    from deepNets import Net as nt
    from deepNets import Trainer as trn
    make_layers = []
    make_layers.append({"layer_type":"input","inp":x})
    #Add Convolutional layers you can specify padding,stride,filter_size etc
    make_layers.append({"layer_type":"conv","filters":16,"filter_size":5,"padding":2})
    #You can also pooling layers currently maxpooling is done
    make_layers.append({"layer_type":"pool","filter_size":2,"stride":2})
    #No need for flatten layers , the below layer flattens and uses a dense layer
    make_layers.append({"layer_type":"relu","hidden_layers":50})
    #Make sure the last layer is loss
    make_layers.append({"layer_type":"loss","num_classes":10})
    net = nt.Net()
    #You can add regularization as below
    net.reg = 0.5
    net.makeLayers(make_layers)
    trainer = trn.Trainer(net,x,y,(x_val,y_val),update_rule="adam",optim_config={'learning_rate':0.01},batch_size=50)
    #Train your network
    trainer.train()