Implement torch.tensordot #10025
Implement torch.tensordot #10025
Conversation
li-roy
added
the
ready for review (this tag is deprecated)
fmassa
left a comment
fmassa
left a comment
There was a problem hiding this comment.
Thanks for the PR!
From a first look and by comparing side by side against the numpy implementation this looks good.
Also, I'd prefer if we named the ATen functions without an underscore.
aten/src/ATen/native/Linear.cpp
Outdated
|
|
||
| // implements tensordot, a matrix-multiplication-like contraction, but the dimensions given | ||
| // in the two dimension lists | ||
| Tensor _tensordot(const Tensor& input1, const Tensor& input2, IntList dims1, IntList dims2) { |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
torch/functional.py
Outdated
| :attr`b` respectively | ||
| When called with an integer argument :attr:`dims` = :math:`d`, and the number of | ||
| dimensions of :attr:`a` and :attr:`b` is :math:`m` and :math`n`, repsectively, |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
Also fix typo in the docstring Thank you, fmassa for the review comments.
|
Nice! Just wondering if there is a good reason to differ from the numpy api regarding renaming the keyword argument |
|
PyTorch consistently uses dim instead of axis, so I'm trying to blend in.
|
|
So I don't know that the CI failures are from the patch. Could you have it retested, please? |
|
@pytorchbot retest this please |
torch/functional.py
Outdated
| if dim is None: | ||
| return torch.sort(input, -1, descending)[1] | ||
| return torch.sort(input, dim, descending)[1] | ||
| return torch.sort(input, dim, descending)[1] |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
|
@pytorchbot retest this please |
ssnl
left a comment
ssnl
left a comment
There was a problem hiding this comment.
LGTM other than small nits
torch/functional.py
Outdated
| dimensions of :attr:`a` and :attr:`b` is :math:`m` and :math`n`, respectively, | ||
| it computes | ||
| :math:`r_{i_0,...,i_{m-d}, i_d,...,i_n}` |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
torch/functional.py
Outdated
| [4796., 5162.], | ||
| [4928., 5306.]]) | ||
| >>> a = torch.randn(3, 4, 5, device=d) |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
torch/functional.py
Outdated
| :math:`= \sum_{k_0,...,k_{d-1}} a_{i_0,...,i_{m-d},k_0,...,k_{d-1}} * b_{k_0,...,k_{d-1}, i_d,...,i_n}`. | ||
| When called with :attr:`dims` of the list form, the given dimensions will be contracted | ||
| in place of the last :math:`d` of :attr:`a` and the first :math:`d` of `b`. The sizes |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
facebook-github-bot
left a comment
facebook-github-bot
left a comment
There was a problem hiding this comment.
yf225 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
|
Anything I can do to move this further? |
|
@t-vi i'll push this through |
Summary: Fixes: #8988 Pull Request resolved: pytorch/pytorch#10025 Reviewed By: ezyang Differential Revision: D9540967 Pulled By: yf225 fbshipit-source-id: 6ba2a7777162983977db884b693e6f4543b31aeb
resolve conflict in data parallel model * master: (201 commits) Add cost inference to ConvGradient and WeightedSum operators (pytorch#10744) Move collapse dims into a single place (pytorch#11272) Fix some more warnings (pytorch#11257) Fix the batchnorm onnx exporting when affine=False Improve error message to include return types too (pytorch#11245) Check doxygen output in travis (pytorch#11124) Accept more numpy scalars as doubles (pytorch#9659) Fixed log message (pytorch#10874) Fix to distribution.__repr__ with lazy attributes (pytorch#11263) Add import export step to end to end tests Add complex hooks for out of tree complex implementation. (pytorch#11216) Unify opt flag for cmake codegen (pytorch#11227) nomnigraph - fix memory error in NN subgraph matchOp (pytorch#11127) Port PackedSequences functions to C++ (pytorch#11224) Treat numerical differences as warnings instead of errors when tracing (pytorch#11246) add a Float16UniformFill (pytorch#11123) Implement torch.tensordot (pytorch#10025) keep net type info when generating model complete net (pytorch#11032) Get rid of some uses of type() (pytorch#11215) Reorganize methods in Type, add CPUTypeDefault/CUDATypeDefault (pytorch#11205) ...
Summary: Fixes: pytorch#8988 Pull Request resolved: pytorch#10025 Reviewed By: ezyang Differential Revision: D9540967 Pulled By: yf225 fbshipit-source-id: 6ba2a7777162983977db884b693e6f4543b31aeb
9 participants
Fixes: #8988