Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

NumpyOps: Better type-casting in asarray #656

Merged
merged 6 commits into from
May 17, 2022
Merged

NumpyOps: Better type-casting in asarray #656

Changes from 1 commit
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
b1b86c4
`NumpyOps`: Better type-casting in `asarray`
shadeMe Apr 26, 2022
132a582
Simplify `dtype` check
shadeMe May 9, 2022
b3d689c
Update thinc/backends/numpy_ops.pyx
shadeMe May 9, 2022
47fe5f8
Simplify casting further, avoid copies if possible
shadeMe May 11, 2022
c564256
Merge branch 'fix/numpyops-asarray-type-casting' of github.com:shadeM…
shadeMe May 11, 2022
2ccf2bb
Remove no-op
shadeMe May 12, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
25 changes: 16 additions & 9 deletions thinc/backends/numpy_ops.pyx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -61,20 +61,27 @@ class NumpyOps(Ops):
raise ValueError("BLIS support requires blis: pip install blis")

def asarray(self, data, dtype=None):
dtype = {"dtype": dtype} if dtype is not None else {}

array = None
shadeMe marked this conversation as resolved.
Show resolved Hide resolved
cast_array = True

if isinstance(data, self.xp.ndarray):
if dtype is not None:
return self.xp.asarray(data, dtype=dtype)
else:
return self.xp.asarray(data)
array = self.xp.asarray(data, **dtype)
shadeMe marked this conversation as resolved.
Show resolved Hide resolved
cast_array = False
elif hasattr(data, 'numpy'):
# Handles PyTorch Tensor
return data.numpy()
array = data.numpy()
elif hasattr(data, "get"):
return data.get()
elif dtype is not None:
return self.xp.array(data, dtype=dtype)
array = data.get()
else:
return self.xp.array(data)
array = self.xp.array(data)

if cast_array and dtype != {}:
array = array.astype(dtype["dtype"])

return array


def alloc(self, shape: Shape, *, dtype: Optional[DTypes] = "float32", zeros: bool = True) -> ArrayXd:
if zeros:
Expand Down