def report_function(e: float, params: np.array, X: spa.csr_matrix,
Z: spa.csr_matrix, X_val: spa.csr_matrix,
Z_val: spa.csr_matrix, penalty: float):
'''Reports loss and accuracy given parameters, data and labels'''
D = X.shape[1]
K = Z.shape[1]
b = params[:K]
W = params[K:].reshape((K, D))
loss = val_func(params, X, Z, penalty)
class_probs = logisticClassProbability(X, b, W)
Y_pred = class_probs.argmax(axis=1)
Y = Z.argmax(axis=1)
train_accuracy = np.mean(Y_pred == Y)
print("\t", e, "Loss =", loss, "Train_Accuracy", train_accuracy, end=" ")
if not (X_val is None): # if we have a valuation set we can report
# how well we are doing out of sample
val_e = val_func(params, X_val, Z_val, penalty)
class_probs = logisticClassProbability(X_val, b, W)
Y_pred = class_probs.argmax(axis=1)
Y_val = Z_val.argmax(axis=1)
val_accuracy = np.mean(Y_pred == Y_val)
print("Evaluation Loss =", val_e, "Accuracy =", val_accuracy)
else:
print()
