tensorflow: add a few TensorFlow functions
#13364
Conversation
hoel-bagard
force-pushed
the
hoel/add_functions
branch
from
4497164 to
bb1e1ca
Compare
hoel-bagard
force-pushed
the
hoel/add_functions
branch
from
bb1e1ca to
169f785
Compare
add bad_indices_policy arg to gather_nd
hoel-bagard
force-pushed
the
hoel/add_functions
branch
from
169f785 to
43304b7
Compare
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
| @@ -403,4 +413,22 @@ def ones_like( | |||
| input: RaggedTensor, dtype: DTypeLike | None = None, name: str | None = None, layout: Layout | None = None | |||
| ) -> RaggedTensor: ... | |||
| def reshape(tensor: TensorCompatible, shape: ShapeLike | Tensor, name: str | None = None) -> Tensor: ... | |||
| def pad( | |||
| tensor: TensorCompatible, | |||
| paddings: Tensor | IntArray | Iterable[Iterable[int]], | |||
There was a problem hiding this comment.
Why are the last two necessary? The docs just specify A Tensor of type int32.
There was a problem hiding this comment.
The documentation is more restrictive than what is actually accepted by the function. Removing the last two will likely cause a lot of false positives.
I opened three of the example links, and all 3 of them use lists instead of Tensors.
| ) -> Tensor: ... | ||
| def gather_nd( | ||
| params: TensorCompatible, | ||
| indices: UIntTensorCompatible, |
There was a problem hiding this comment.
Should this be TensorCompatible?
A Tensor. Must be one of the following types: int32, int64. Index tensor.
https://www.tensorflow.org/api_docs/python/tf/gather_nd#args
There was a problem hiding this comment.
TensorCompatible contains str, which is not a valid type for indices.
|
According to mypy_primer, this change has no effect on the checked open source code. 🤖🎉 |
|
In that case this LGTM |
This PR adds the
tf.round,tf.pad,tf.shape,tf.where,tf.gather_ndandtf.signal.hamming_windowfunctions.