epoch 1부터 loss가 너무 낮게 나와 학습이 안되네요

ratio = 0.2
x_train = tf.random.normal(shape = (300,), dtype = tf.float32)
y_train = 3 * x_train + 1 + ratio * tf.random.normal(shape = (300, ), dtype = tf.float32)

x_test = tf.random.normal(shape = (30,), dtype = tf.float32)
y_test = 3 * x_test + 1 + ratio * tf.random.normal(shape = (30, ), dtype = tf.float32)

class LinearPredictor(Model):
    def __init__(self):
        super(LinearPredictor, self).__init__()

        self.d1 = Dense(1, activation = "linear")

    def call(self, x):
        x = self.d1(x)
        return x

model = LinearPredictor()

loss_object = tf.keras.losses.MeanSquaredError()
optimizer = SGD(learning_rate=0.01)

for epoch in range(5):
    for x, y in zip(x_train, y_train):
        x = tf.reshape(x, (1, 1))
        with tf.GradientTape() as tape:
            predictions = model(x)
            loss = loss_object(y, predictions)
        gradients = tape.gradient(loss, model.trainable_variables)
        optimizer.apply_gradients(zip(gradients, model.trainable_variables))
    print(f"Epoch: {epoch + 1}")
    print(f"Train Loss: {loss:.4f}")
    

강사님이랑 똑같이 코드를 짠것 같은데 train loss가 너무 낮게 나와 학습이 안되네요. 틀린곳이 있는건지 데이터가 너무 심플하게 생성되서 그런건지 잘 모르겠습니다.