Unable to load the spacy model 'en_core_web_lg' on Google colab

Question

I am using spacy in google colab to build an NER model for which I have downloaded the spaCy 'en_core_web_lg' model using

import spacy.cli
spacy.cli.download("en_core_web_lg")

and I get a message saying

✔ Download and installation successful
You can now load the model via spacy.load('en_core_web_lg')

However then when i try to load the model

nlp = spacy.load('en_core_web_lg')

the following error is printed:

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

Could anyone help me with this problem?

Answer 1

Running

import spacy.cli
spacy.cli.download("en_core_web_lg")
nlp = spacy.load("en_core_web_lg")

shouldn't yield any errors anymore with recent spaCy versions.

If running the code still gives errors, you should be all set with running in one cell (takes a while, but gives you visual feedback about progress, differently from spacy.cli)

!python -m spacy download en_core_web_lg

Then, *** restart the colab runtime *** via

• the colab menu Runtime > Restart runtime, or
• use the keyboard shortcut Ctrl+M .

After that, executing

import spacy
nlp = spacy.load('en_core_web_lg')

should work flawlessly.

Answer 2

In Google Colab Notebooks, you should import the model as a package.

However you download and install the model:

!pip install <model_s3_url> # tar.gz file e.g. from release notes like https://github.com/explosion/spacy-models/releases//tag/en_core_web_lg-2.3.1
!pip install en_core_web_lg
import spacy

you don't have permission in Colab to load the model with normal spacy usage:

nlp = spacy.load("en_core_web_lg") # not via packages
nlp = spacy.load("/path/to/en_core_web_lg") #not via paths
nlp = spacy.load("en") # nor via shortcut links
spacy.load()

Instead, import the model and load it directly:

import en_core_web_lg
nlp = en_core_web_lg.load()

Then use as directed:

doc = nlp("This is a sentence. Soon, it will be knowledge.")

Answer 3

It seems the best answer is on this thread: How to install models/download packages on Google Colab?

import spacy.cli
spacy.cli.download("en_core_web_lg")
import en_core_web_lg
nlp = en_core_web_lg.load()