| Commit message (Collapse) | Author | Age |
|\
| |
| |
| | |
PiperOrigin-RevId: 216217509
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This cleanup will make the future CL implementing lazy compilation simpler.
Includes some supporting changes:
- Teach NewInternalScope to create a scope that doesn't do shape inference. We
need this because we don't have a ShapeRefiner that has been run over the
entire graph available in the build_xla_ops pass.
- Add a WithAssignedDevice modifier to tensorflow::Scope.
- Make cc_op_gen write out an Operation field for nodes which may not
necessarily have any outputs. We already did this in most cases, but we
weren't doing it for nodes that have possibly-empty list outputs.
- Minor change renaming ops/xla_jit_op.cc to ops/xla_jit_ops.cc, now that we
have more than one XLA JIT op.
PiperOrigin-RevId: 215293817
|
| |\
| |/
|/| |
|
| | |
|
|\ \
| | |
| | |
| | | |
PiperOrigin-RevId: 213343364
|
| | |\
| |_|/
|/| | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Used to prepare all the header files so they can easily be installed
into /usr/include when packaging TF.
Signed-off-by: Jason Zaman <jason@perfinion.com>
|
| | | |
|
| | | |
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
When building a CPP API application with "-Wall -Werror" option ,
`error: comparison between signed and unsigned integer expressions'
occurs since return type of num_elements() is 'int64' instead of
'size_t' in ops.h to express -1. This patch fixes this bug by explicit
type casting.
* related issue: https://github.com/tensorflow/tensorflow/issues/20428
Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
|
| |
| |
| |
| |
| | |
1. Add C++ gradient of gradient definition of LeakyReLu and revalant UT.
2. Using forward compatibility layer for python code changes.
|
| |
| |
| |
| | |
PiperOrigin-RevId: 210596417
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
conversions.
That is, instances of sp.ToString() are replaced with string(sp).
This will allow tensorflow::StringPiece::ToString to be removed, which is necessary before it can be replaced with absl::string_view.
PiperOrigin-RevId: 210394878
|
| |
| |
| |
| |
| |
| |
| |
| | |
That is, instances of sp.ToString() are replaced with string(sp).
This will allow tensorflow::StringPiece::ToString to be removed, which is necessary before it can be replaced with absl::string_view.
PiperOrigin-RevId: 209806694
|
|\ \
| | |
| | |
| | | |
PiperOrigin-RevId: 209508149
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
references.
This method is currently duplicated in Python code and will be used for TypeScript generation.
PiperOrigin-RevId: 209207139
|
| |/
|/|
| |
| |
| |
| |
| |
| | |
LeakyRelu, defined as 'y = { x (x>=0) or alpha*x (x<0) }', was computed
by combined Ops 'max(x, alpha*x)' in current codes. Hence its gradient
calculation for back propagation would contain a serial of element-wise
Ops. This looks really unnecessary for such a simple op and it could be
done within just one Op with less memory accesses.
|
| |\
| |/
|/| |
|
| | |
|
|\ \
| | |
| | |
| | | |
PiperOrigin-RevId: 208565050
|
|\ \ \
| | | |
| | | |
| | | | |
PiperOrigin-RevId: 208352779
|
|\ \ \ \
| | | | |
| | | | |
| | | | | |
PiperOrigin-RevId: 208266242
|
| | | |/
| | |/| |
|
|\ \ \ \
| | | | |
| | | | |
| | | | | |
PiperOrigin-RevId: 207959776
|
| | | | | |
|
|\ \ \ \ \
| | | | | |
| | | | | |
| | | | | | |
PiperOrigin-RevId: 207735371
|
| | | | |\ \
| |_|_|_|/ /
|/| | | | | |
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This change returns NaN as the maximum error if any gradient
entry is NaN. This avoids masking gradient implementation
errors in tests that don't expect to see NaN results.
PiperOrigin-RevId: 207551265
|
|\ \ \ \ \ \
| | | | | | |
| | | | | | |
| | | | | | | |
PiperOrigin-RevId: 207350813
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
PiperOrigin-RevId: 207323298
|
| | | | | | | |
|
| | | | | | | |
|
| |_|/ / / /
|/| | | | | |
|
| | | | |\ \
| |_|_|_|/ /
|/| | | | | |
|
| | | | | | |
|
|\ \ \ \ \ \
| | | | | | |
| | | | | | |
| | | | | | | |
PiperOrigin-RevId: 206341656
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
instead of Run(), to avoid leaving behind non-GC'ed state after model initialization.
PiperOrigin-RevId: 206266841
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
we remove duplicated code paths by mapping legacy_init_op into main_op in the SavedModelBuilder, and we deprecate the legacy_init_op arg. Note that the loader will still look for both, so old SavedModels will still load without trouble.
PiperOrigin-RevId: 206026743
|
|\ \ \ \ \ \ \
| | | | | | | |
| | | | | | | |
| | | | | | | | |
PiperOrigin-RevId: 205880828
|
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
PiperOrigin-RevId: 205746329
|
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
we remove duplicated code paths by mapping legacy_init_op into main_op in the SavedModelBuilder, and we deprecate the legacy_init_op arg. Note that the loader will still look for both, so old SavedModels will still load without trouble.
PiperOrigin-RevId: 205728344
|
| |_|/ / / / /
|/| | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
Added gradients and tests for
- ResizeBilinear
- ResizeBicubic
- ResizeNearestNeighbor
Note: Some of the tests are for the operator itself rather than
the gradient, paralleling existing tests in image_grad.py
See https://github.com/tensorflow/tensorflow/issues/21019
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
mobile.
PiperOrigin-RevId: 205248073
|
| |_|/ / / /
|/| | | | |
| | | | | |
| | | | | | |
See https://github.com/tensorflow/tensorflow/issues/20926
|
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
PiperOrigin-RevId: 204468340
|
|/ / / / /
| | | | |
| | | | |
| | | | | |
Committing in a new PR as the old PR has too many commit files
|
| |_|_|/
|/| | | |
|
|/ / /
| | |
| | |
| | | |
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
|
|\ \ \
| | | |
| | | | |
Branch 202152026
|
| | | |
| | | |
| | | | |
See https://github.com/tensorflow/tensorflow/issues/9645
|