我建立了LSTM网络，现在我想用Keras进行网格搜索。我在try... except的文档中看到，输入形状必须是二维的。
由于我的LSTM网络，我具有3维输入。我像这样重塑张量：
GridSearchCV当我进行网格搜索时，我还想改变输入的时间步长（look_back）。如何将其适合我的网格搜索？
更新：
更新scikit-learn的版本后，我的3D输入不再是问题。但是我仍然不知道如何更改nb_samples_train = x_train.shape[0] - look_back
num=nb_samples_train
x_train_reshaped = np.zeros((nb_samples_train, look_back, dim_x))
y_train_reshaped = np.zeros((nb_samples_train,dim_y))
for i in range(nb_samples_train):
y_position = i + look_back
x_train_reshaped[i] = x_train[i:y_position]
y_train_reshaped[i] = y_train[y_position]
nb_samples_test = x_test.shape[0] - look_back
x_test_reshaped = np.zeros((nb_samples_test, look_back, dim_x))
y_test_reshaped = np.zeros((nb_samples_test, dim_y))
for i in range(nb_samples_test):
y_position = i + look_back
x_test_reshaped[i] = x_test[i:y_position]
y_test_reshaped[i] = y_test[y_position]
x_train_r=x_train_reshaped
x_test_r=x_test_reshaped
y_train_r=y_train_reshaped
y_test_r=y_test_reshaped
。