安装TensorFlow时出现numpy值错误

Question

导入此时遇到此错误

from gensim.models import KeyedVectors

错误是

File "c:Usersfraseeg1.py", line 11, in <module>
from gensim.models import KeyedVectors
File "C:UsersfraseAppDataLocalProgramsPythonPython39libsite-packagesgensim__init__.py", line 11, in <module>
from gensim import parsing, corpora, matutils, interfaces, models, similarities, utils  # noqa:F401
File "C:UsersfraseAppDataLocalProgramsPythonPython39libsite-packagesgensimcorpora__init__.py", line 6, in <module>
from .indexedcorpus import IndexedCorpus  # noqa:F401 must appear before the other classes
File "C:UsersfraseAppDataLocalProgramsPythonPython39libsite-packagesgensimcorporaindexedcorpus.py", line 14, in <module>
from gensim import interfaces, utils
File "C:UsersfraseAppDataLocalProgramsPythonPython39libsite-packagesgensiminterfaces.py", line 19, in <module>
from gensim import utils, matutils
File "C:UsersfraseAppDataLocalProgramsPythonPython39libsite-packagesgensimmatutils.py", line 1024, in <module>
from gensim._matutils import logsumexp, mean_absolute_difference, dirichlet_expectation
File "gensim/_matutils.pyx", line 1, in init gensim._matutils
#!/usr/bin/env cython
ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

我以前曾使用KeyedVectors在模型中加载，但当我安装"TensorFlow"时，这种情况就开始出现了。

我查阅了这个问题，并将Numpy的版本降低到1.19.2(这是Tensorflow的最大值(，但这并没有奏效。我也试过简单地卸载和重新安装，但也没有成功。

发生这种情况的原因是什么？

用于的代码示例(注意：我无法测试它，因此可能存在错误(：

model = KeyedVectors.load_word2vec_format(
'Word2vecGoogleNews-vectors-negative300.bin.gz', binary=True)
# create the embedding matrix
embedding_matrix = np.zeros((vocab_size, 300))  # 300 is dim size
for word, vector in model.key_to_vectors():
if(word in tokenizer.word_index):
idx = tokenizer.word_index[word]
embedding_matrix[idx] = np.array(vector, dtype=np.float32)

Answer 1

解决方案。

1. 将python环境降级到3.8/3.7

2. 卸载Numpy并升级到最新的Numpy版本

引用-值错误：numpy.ndarray大小已更改，可能表示二进制不兼容。预期88来自C标头，得到80来自PyObject