载入模型出现编码错误

Question

# 0~2π角度对应(X,Y)拟合圆 并保存训练结果
import numpy as np
import tensorflow as tf
from tensorflow import keras
import matplotlib.pyplot as plt

x = np.random.rand(1000) * 2 * np.pi
x = x.reshape([1000, 1])
y = np.hstack((np.cos(x).reshape([1000, 1]), np.sin(x).reshape([1000, 1])))

model = tf.keras.Sequential()
# 神经元个数越多,效果会越好,收敛越快,太少的话难以收敛到所需曲线
model.add(keras.layers.Dense(1000, activation='relu'))
# linear,relu,tanh, sigmoid, softmax
model.add(keras.layers.Dense(2, activation='tanh'))
model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.001),
              loss=tf.keras.losses.mse)
model.fit(x, y, batch_size=10, epochs=100, shuffle=True, verbose=2)

plt.plot(y[:, 0], y[:, 1], "b^", label="f(x)")
py = model.predict(x)
plt.plot(py[:, 0], py[:, 1], "r.", label="predict")
plt.legend(loc=0)
plt.show()

# Save the model
# serialize model to JSON

model_digit_json = model.to_json()
with open("model_circle.json", "w") as json_file:
    json_file.write(model_digit_json)
# serialize weights to HDF5
model.save("model_circle.h5")
model.save_weights("model_circle_weights.h5")

# 第二种载入方法：同时载入模型及训练好的参数
import numpy as np
from tensorflow import keras as tk
import matplotlib.pyplot as plt

x = np.random.rand(1000) * 2 * np.pi
x = x.reshape([1000, 1])

model = tk.models.load_model("model_circle.h5")
py = model.predict(x)
plt.plot(py[:, 0], py[:, 1], "r.", label="predict")
plt.legend(loc=0)
plt.show()

Answer 1

将模型的位置，放置在无中文的目录下就可解决这个问题，
来源：

tensorflow加载模型keras.models.load_model出错的问题_一级泡芙爱好者的博客-CSDN博客 TensorFlow正常保存模型，但是加载模型失败。将模型的位置，放置在无中文的目录下就可解决这个问题。保存模型的时候，保存位置出现中文。 https://blog.csdn.net/xiameimei_win/article/details/131056021