add option to pass in absolute tolerance

modules/deeplearning/mc_dropout_regression.py

@@ -8,15 +8,15 @@ def create_mnist_dataset():
     return tf.keras.datasets.mnist.load_data()
 
 
-def create_toy_regression_dataset(xmin=-10., xmax=10, noise_std=.2):
-    x_trn = np.linspace(-2.0, 2.0, 1000, dtype=np.float32)
-    y_trn = np.sin(x_trn) + np.random.normal(0, noise_std, size=1000).astype(np.float32)
+def create_toy_regression_dataset(xmin=-10., xmax=10, num_points=1000, noise_std=.05):
+    x_trn = np.linspace(-2.0, 2.0, num_points, dtype=np.float32)
+    y_trn = np.sin(x_trn) + np.random.normal(0, noise_std, size=num_points).astype(np.float32)
 
-    x_gt = np.linspace(xmin, xmax, 1000, dtype=np.float32)
+    x_gt = np.linspace(xmin, xmax, num_points, dtype=np.float32)
     y_gt = np.sin(x_gt)
 
-    x_tst = np.linspace(xmin, xmax, 1000, dtype=np.float32)
-    y_tst = np.sin(x_tst) + np.random.normal(0, noise_std, size=1000).astype(np.float32)
+    x_tst = np.linspace(xmin, xmax, num_points, dtype=np.float32)
+    y_tst = np.sin(x_tst) + np.random.normal(0, noise_std, size=num_points).astype(np.float32)
     return x_gt, y_gt, x_trn, y_trn, x_tst, y_tst
 
 
@@ -199,18 +199,18 @@ def predict(model, x, samples=20):
     y_std = np.sqrt(y_variance)
     return y_mean, y_std
 
-def run():
+def run(num_points=1000, samples=20):
     xmin = -10.
     xmax = 10.
 
-    x_gt, y_gt, x_trn, y_trn, x_tst, y_tst = create_toy_regression_dataset(xmin=xmin, xmax=xmax, noise_std=0.05)
+    x_gt, y_gt, x_trn, y_trn, x_tst, y_tst = create_toy_regression_dataset(xmin=xmin, xmax=xmax, num_points=num_points, noise_std=0.05)
 
     model = define()
     model = train(x_trn[:, np.newaxis],
                   y_trn[:, np.newaxis],
                   model)
 
-    yhat_mean, yhat_std = predict(model, x_tst[:, np.newaxis], samples=20)
+    yhat_mean, yhat_std = predict(model, x_tst[:, np.newaxis], samples=samples)
 
     plot_regression_model_analysis(gt=(x_gt, y_gt),
                                    trn=(x_trn, y_trn),