spacy在windows 10和Python 3.5.3:：Anaconda custom（64位）上找不到模型“en_core_web_sm”

Question

{}和{}之间有什么区别This link解释了不同的模型大小。但我仍然不清楚spacy.load('en_core_web_sm')和spacy.load('en')的区别

spacy.load('en')对我来说很好。但是spacy.load('en_core_web_sm')抛出错误

我已经安装了spacy，如下所示。当我转到jupyter笔记本并运行命令nlp = spacy.load('en_core_web_sm')时，我得到以下错误

---------------------------------------------------------------------------
OSError                                   Traceback (most recent call last)
<ipython-input-4-b472bef03043> in <module>()
      1 # Import spaCy and load the language library
      2 import spacy
----> 3 nlp = spacy.load('en_core_web_sm')
      4 
      5 # Create a Doc object

C:\Users\nikhizzz\AppData\Local\conda\conda\envs\tensorflowspyder\lib\site-packages\spacy\__init__.py in load(name, **overrides)
     13     if depr_path not in (True, False, None):
     14         deprecation_warning(Warnings.W001.format(path=depr_path))
---> 15     return util.load_model(name, **overrides)
     16 
     17 

C:\Users\nikhizzz\AppData\Local\conda\conda\envs\tensorflowspyder\lib\site-packages\spacy\util.py in load_model(name, **overrides)
    117     elif hasattr(name, 'exists'):  # Path or Path-like to model data
    118         return load_model_from_path(name, **overrides)
--> 119     raise IOError(Errors.E050.format(name=name))
    120 
    121 

OSError: [E050] Can't find model 'en_core_web_sm'. It doesn't seem to be a shortcut link, a Python package or a valid path to a data directory.

我是如何安装Spacy的--

(C:\Users\nikhizzz\AppData\Local\conda\conda\envs\tensorflowspyder) C:\Users\nikhizzz>conda install -c conda-forge spacy
Fetching package metadata .............
Solving package specifications: .

Package plan for installation in environment C:\Users\nikhizzz\AppData\Local\conda\conda\envs\tensorflowspyder:

The following NEW packages will be INSTALLED:

    blas:           1.0-mkl
    cymem:          1.31.2-py35h6538335_0    conda-forge
    dill:           0.2.8.2-py35_0           conda-forge
    msgpack-numpy:  0.4.4.2-py_0             conda-forge
    murmurhash:     0.28.0-py35h6538335_1000 conda-forge
    plac:           0.9.6-py_1               conda-forge
    preshed:        1.0.0-py35h6538335_0     conda-forge
    pyreadline:     2.1-py35_1000            conda-forge
    regex:          2017.11.09-py35_0        conda-forge
    spacy:          2.0.12-py35h830ac7b_0    conda-forge
    termcolor:      1.1.0-py_2               conda-forge
    thinc:          6.10.3-py35h830ac7b_2    conda-forge
    tqdm:           4.29.1-py_0              conda-forge
    ujson:          1.35-py35hfa6e2cd_1001   conda-forge

The following packages will be UPDATED:

    msgpack-python: 0.4.8-py35_0                         --> 0.5.6-py35he980bc4_3 conda-forge

The following packages will be DOWNGRADED:

    freetype:       2.7-vc14_2               conda-forge --> 2.5.5-vc14_2

Proceed ([y]/n)? y

blas-1.0-mkl.t 100% |###############################| Time: 0:00:00   0.00  B/s
cymem-1.31.2-p 100% |###############################| Time: 0:00:00   1.65 MB/s
msgpack-python 100% |###############################| Time: 0:00:00   5.37 MB/s
murmurhash-0.2 100% |###############################| Time: 0:00:00   1.49 MB/s
plac-0.9.6-py_ 100% |###############################| Time: 0:00:00   0.00  B/s
pyreadline-2.1 100% |###############################| Time: 0:00:00   4.62 MB/s
regex-2017.11. 100% |###############################| Time: 0:00:00   3.31 MB/s
termcolor-1.1. 100% |###############################| Time: 0:00:00 187.81 kB/s
tqdm-4.29.1-py 100% |###############################| Time: 0:00:00   2.51 MB/s
ujson-1.35-py3 100% |###############################| Time: 0:00:00   1.66 MB/s
dill-0.2.8.2-p 100% |###############################| Time: 0:00:00   4.34 MB/s
msgpack-numpy- 100% |###############################| Time: 0:00:00   0.00  B/s
preshed-1.0.0- 100% |###############################| Time: 0:00:00   0.00  B/s
thinc-6.10.3-p 100% |###############################| Time: 0:00:00   5.49 MB/s
spacy-2.0.12-p 100% |###############################| Time: 0:00:10   7.42 MB/s

(C:\Users\nikhizzz\AppData\Local\conda\conda\envs\tensorflowspyder) C:\Users\nikhizzz>python -V
Python 3.5.3 :: Anaconda custom (64-bit)

(C:\Users\nikhizzz\AppData\Local\conda\conda\envs\tensorflowspyder) C:\Users\nikhizzz>python -m spacy download en
Collecting en_core_web_sm==2.0.0 from https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-2.0.0/en_core_web_sm-2.0.0.tar.gz#egg=en_core_web_sm==2.0.0
  Downloading https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-2.0.0/en_core_web_sm-2.0.0.tar.gz (37.4MB)
    100% |################################| 37.4MB ...
Installing collected packages: en-core-web-sm
  Running setup.py install for en-core-web-sm ... done
Successfully installed en-core-web-sm-2.0.0

    Linking successful
    C:\Users\nikhizzz\AppData\Local\conda\conda\envs\tensorflowspyder\lib\site-packages\en_core_web_sm
    -->
    C:\Users\nikhizzz\AppData\Local\conda\conda\envs\tensorflowspyder\lib\site-packages\spacy\data\en

    You can now load the model via spacy.load('en')


(C:\Users\nikhizzz\AppData\Local\conda\conda\envs\tensorflowspyder) C:\Users\nikhizzz>

Answer 1

您误解的答案是Unix概念，软链接，我们可以说，在Windows中，软链接类似于快捷方式。让我们来解释一下

当您spacy download en时，spaCy会尝试找到与您的spaCy分布相匹配的最佳小型模型。我所说的小模型默认为en_core_web_sm，可以在不同的变体中找到，这些变体对应于不同的spaCy版本（例如spacy，spacy-nightly具有不同大小的en_core_web_sm）

当spaCy找到最适合您的模型时，它会下载该模型，然后将名称en链接到它下载的包，例如en_core_web_sm。这基本上意味着，每当你提到en时，你就会提到en_core_web_sm。换句话说，链接后的en不是一个“真正的”包，只是en_core_web_sm的一个名称

然而，它不工作的另一种方式。您不能直接引用en_core_web_sm，因为您的系统不知道您已经安装了它。当您执行spacy download en时，您基本上执行了一个pip安装。因此，pip知道您已经为python发行版安装了一个名为en的包，但对该包en_core_web_sm一无所知。当您导入此包时，它只是替换包en，这意味着包en只是到en_core_web_sm的软链接

当然，您可以直接下载en_core_web_sm，使用命令：python -m spacy download en_core_web_sm，或者您甚至可以将名称en链接到其他模型。例如，您可以执行python -m spacy download en_core_web_lg，然后执行python -m spacy link en_core_web_lg en。那将使 en是{}的名称，它是英语的一个大空间模型

希望现在一切都清楚了：）