使用Tensorflow给出错误答案的多项式回归

import tensorflow as tf
import numpy as np

# 生成一些随机数据
x_train = np.linspace(-1, 1, 100)
y_train = 2 * x_train**2 + 3 * x_train + 1 + np.random.randn(*x_train.shape) * 0.3

# 定义模型的输入和输出
x = tf.placeholder(tf.float32)
y = tf.placeholder(tf.float32)

# 定义模型的参数
w = tf.Variable(tf.random_normal([3, 1]))
b = tf.Variable(tf.random_normal([1]))

# 定义模型
y_pred = tf.matmul(tf.pow(x, tf.constant([2, 1, 0], dtype=tf.float32)), w) + b

# 定义损失函数
loss = tf.reduce_mean(tf.square(y_pred - y))

# 定义优化器
optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.1)
train_op = optimizer.minimize(loss)

# 创建会话并训练模型
with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    for epoch in range(100):
        sess.run(train_op, feed_dict={x: x_train, y: y_train})
        if epoch % 10 == 0:
            curr_loss = sess.run(loss, feed_dict={x: x_train, y: y_train})
            print("Epoch %d, Loss: %f" % (epoch, curr_loss))

    # 使用训练好的模型进行预测
    x_test = np.linspace(-1, 1, 10)
    y_pred_test = sess.run(y_pred, feed_dict={x: x_test})

    print("Predictions:", y_pred_test)