[DOC] Not 100% accurate statement in 10 min to cuDF and CuPy

[miguelusque]

Report incorrect documentation

Hi!

I think the following paragraph is not 100% accurate:

Because CuPy arrays have a single dtype, each column in our DataFrame must have the same dtype, regardless of which method we use.

The following code works just fine:

import cudf
import cupy

df = cudf.DataFrame({'a': 3, 'b': 3.0})
arr = cupy.asarray(df.as_gpu_matrix())

In the example above, int datatypes as casted to float64.

Hope it helps!
Miguel

Location of incorrect documentation
https://github.com/rapidsai/docs/blob/gh-pages/api/cudf/0.18/10min-cudf-cupy.ipynb

[ajschmidt8]

@miguelusque, the documentation you mentioned is automatically generated from cudf's source files, so I've transferred your issue into the cudf repo.