Spacy does not return vector if GPU is enabled

I’m using a GPU-enabled Google Colab notebook.

After installing the requisite libraries and models

!pip install spacy[cuda100]~=2.2 scispacy~=0.2.4

!pip install https://s3-us-west-2.amazonaws.com/ai2-s2-scispacy/releases/v0.2.4/en_core_sci_lg-0.2.4.tar.gz

import spacy
spacy.prefer_gpu()
nlp = spacy.load("en_core_sci_lg")

text = """
Myeloid derived suppressor cells (MDSC) are immature 
myeloid cells with immunosuppressive activity. 
They accumulate in tumor-bearing mice and humans 
with different types of cancer, including hepatocellular 
carcinoma (HCC).
"""

doc = nlp(text)

Running

doc.ents

produces

(Myeloid,
 suppressor cells,
 MDSC,
 immature,
 myeloid cells,
 immunosuppressive activity,
 accumulate,
 tumor-bearing mice,
 humans,
 cancer,
 hepatocellular 
 carcinoma,
 HCC)

as expected, but

doc.vector

produces this error:

TypeError                                 Traceback (most recent call last)
<ipython-input-19-40a48203c66b> in <module>()
----> 1 doc.vector

doc.pyx in __iter__()

cupy/core/core.pyx in cupy.core.core.ndarray.__array_ufunc__()

cupy/core/_kernel.pyx in cupy.core._kernel.ufunc.__call__()

cupy/core/_kernel.pyx in cupy.core._kernel._preprocess_args()

TypeError: Unsupported type <class 'numpy.ndarray'>

Commenting out

spacy.prefer_gpu()

solves the issue. This means that getting token vectors will not be able when the GPU is enabled?

I’m not sure if this is related to #81 or #3431 in spaCy.

[josesho]

Hi @danielkingai2,

I also see your problem.

For full documentation’s sake:

import spacy
spacy.prefer_gpu()
nlp_core = spacy.load("en_core_web_lg")

text1 = "The effect of anxiogenic treatments on three rodent models of anxiety: \
         the open field test, the elevated plus-maze, and the light-dark box."

doc1 = nlp_core(text1)

doc1.vector

produces

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-3-531ef58ab65a> in <module>()
      1 doc1 = nlp_core(text1)
      2 
----> 3 doc1.vector

doc.pyx in __iter__()

cupy/core/core.pyx in cupy.core.core.ndarray.__add__()

cupy/core/_kernel.pyx in cupy.core._kernel.ufunc.__call__()

cupy/core/_kernel.pyx in cupy.core._kernel._preprocess_args()

TypeError: Unsupported type <class 'numpy.ndarray'>

Relevant package versions:

print(spacy.__version__)
print(cupy.__version__)

2.2.3
7.2.0

Closing this for now as it seems it’s a spacy issue and not scispacy’s.

[dakinggg]

Any update on this?