Python NumpyBackend'；对象没有属性'；塔克'；

我有一些使用svd进行图像压缩的代码，还有一些用于实验的代码

所以我有以下实验代码：

plt.figure(figsize=(20,12))
for i,rank in enumerate([2,10,20]):
    plt.subplot(2,3,i+1)
    plt.imshow(svd_compression(image1,rank))
    plt.title("rank:%i - compression: %.2f%%" %(rank,svd_compress_ratio(image1,rank)*100))
for i,rank in enumerate([2,10,20]):
    plt.subplot(2,3,i+4)
    plt.imshow(svd_compression(image2,rank))
    plt.title("rank:%i - compression: %.2f%%" %(rank,svd_compress_ratio(image2,rank)*100))

plt.show()

但我犯了以下错误：

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
/usr/local/lib/python3.6/dist-packages/tensorly/backend/__init__.py in _get_backend_method(key)
     91     try:
---> 92         return getattr(_LOCAL_STATE.backend, key)
     93     except AttributeError:

AttributeError: 'NumpyBackend' object has no attribute 'tucker'

During handling of the above exception, another exception occurred:

AttributeError                            Traceback (most recent call last)
3 frames
/usr/local/lib/python3.6/dist-packages/tensorly/backend/__init__.py in _get_backend_method(key)
     92         return getattr(_LOCAL_STATE.backend, key)
     93     except AttributeError:
---> 94         return getattr(_LOADED_BACKENDS[_DEFAULT_BACKEND], key)
     95 
     96 def _get_backend_dir():

AttributeError: 'NumpyBackend' object has no attribute 'tucker'

压缩方法如下：

!pip install tensorly
import numpy as np
import pylab as plt
import scipy.sparse.linalg 
import PIL.Image
import tensorly as tl
import matplotlib.pyplot as plt2
from scipy.sparse.linalg import svds

def svd_compression(image, rank):


    unfolded = tl.unfold(image, mode=0)
    u, s, vt = svds(unfolded, rank)
    return tl.fold(np.round(np.dot(u*s,vt),1), 0, image.shape) # need to round to have a color lvl