Skip to content

Commit 2c746f1

Browse files
committed
trim doc for brevity and fix display error
1 parent c616b20 commit 2c746f1

File tree

2 files changed

+2
-12
lines changed

2 files changed

+2
-12
lines changed

cuda_core/cuda/core/_memoryview.pyx

Lines changed: 0 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -91,17 +91,6 @@ cdef class StridedMemoryView:
9191
If the view is created with :meth:`from_buffer`,
9292
it will be the Buffer instance passed to the method.
9393
94-
Notes
95-
-----
96-
The :attr:`dtype` property supports narrow data types (e.g., ``bfloat16``) when the optional
97-
`ml_dtypes <https://github.com/jax-ml/ml_dtypes>`_ package is installed. This enables
98-
interoperability with libraries that use narrow dtype tensors, such as PyTorch with
99-
``torch.bfloat16`` or CuPy with ``"bfloat16"`` dtype. If ``ml_dtypes`` is not available
100-
and such a tensor is encountered, a :obj:`NotImplementedError` will be raised.
101-
102-
Currently supported narrow data types:
103-
- ``bfloat16``
104-
10594
"""
10695
cdef readonly:
10796
intptr_t ptr

cuda_core/docs/source/interoperability.rst

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -86,7 +86,8 @@ If ``ml_dtypes`` is not available and such a tensor is encountered, a
8686
:obj:`NotImplementedError` will be raised.
8787

8888
Currently supported narrow data types:
89-
- ``bfloat16``
89+
90+
* ``bfloat16``
9091

9192
.. rubric:: Footnotes
9293

0 commit comments

Comments
 (0)