安装scikit-optimize与hyperopt后出现sklearn导入错误求助

之前代码运行完全正常，今天我用conda安装了scikit-optimize（skopt）和hyperopt，执行的命令是：

conda install -c conda-forge scikit-optimize
conda install -c conda-forge hyperopt

安装完成后运行代码，出现了以下导入错误：

51 # from plotly_resampler import FigureResampler, FigureWidgetResampler
     52 
     53 # from kneebow.rotor import Rotor
     54 import random 
---> 55 from sklearn import manifold   # for t-SNE
     57 from sklearn import tree
     58 import joblib

File c:\ProgramData\anaconda_envs\dash2\lib\site-packages\sklearn\manifold\__init__.py:5
      1 """
      2 The :mod:`sklearn.manifold` module implements data embedding techniques.
      3 """
----> 5 from ._locally_linear import locally_linear_embedding, LocallyLinearEmbedding
      6 from ._isomap import Isomap
      7 from ._mds import MDS, smacof

File c:\ProgramData\anaconda_envs\dash2\lib\site-packages\sklearn\manifold\_locally_linear.py:23
     21 from ..utils.validation import check_is_fitted
     22 from ..utils.validation import FLOAT_DTYPES
---> 23 from ..neighbors import NearestNeighbors
     26 def barycenter_weights(X, Y, indices, reg=1e-3):
     27     """Compute barycenter weights of X from Y along the first axis
...
     14     ArgKminClassMode32,
     15     ArgKminClassMode64,
     16 )

ImportError: cannot import name 'METRIC_MAPPING64' from 'sklearn.metrics._dist_metrics' (c:\ProgramData\anaconda_envs\dash2\lib\site-packages\sklearn\metrics\_dist_metrics.cp39-win_amd64.pyd)

我试过用pip更新scikit-learn，命令是：

pip install scikit-learn

但更新后错误依旧存在，求各位大佬帮忙解决这个问题。

内容的提问来源于stack exchange，提问作者roudan