| Commit message (Collapse) | Author | Age |
... | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
* Fix some typos in the documation (tensorflow/docs_src)
* update another minor typo fix
* revert typo correction
It seems "infed" was being used as a past-tense verb here and so was correct.
|
| | | |
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183737063
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183735457
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183734695
|
| | | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This avoids potential deadlock, see the warnings in
https://docs.python.org/2/library/subprocess.html#popen-objects. I
found that enabling the C API caused us to deadlock without this
change.
PiperOrigin-RevId: 183730170
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The default signal handling for Python delivers the signal to the thread for
processing; if broad exceptions are being caught, this can often result in the
interrupt being swallowed.
PiperOrigin-RevId: 183729637
|
| | |
| | |
| | |
| | | |
android_tensorflow_lib_selective_registration (#16117)
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183728562
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This fix is a follow up to 11875 so that MaxPool in Python
use v2 version. As 11875 has been merged some time ago,
this fix conforms to the deprecation policy.
This fix is realted to 11875 and 4746.
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
PiperOrigin-RevId: 183727668
|
| | |
| | |
| | |
| | |
| | |
| | | |
decorator information appear in the AST.
PiperOrigin-RevId: 183723446
|
| | |
| | |
| | |
| | |
| | | |
This fix fixes incorrect docs for DecodeVideoOp
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183718481
|
| | |
| | |
| | |
| | |
| | |
| | | |
* contrib/learn: Typo in variable name x_exrta --> x_extra
flake8 testing of https://github.com/tensorflow/tensorflow
|
| | |
| | |
| | |
| | |
| | |
| | | |
* Allow step callback for scipy SLSQP
* Add test_callbacks() for scipy optimizer
|
| | | |
|
| | | |
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183709796
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183709590
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183705898
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183705773
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
types of function calls by only failing on unresolved symbols if they're needed.
Simplify code structure all around. Remove the awkward activity analysis that deemed a function parameter as "modified". Consolidate activity analysis by tracking function parameters and returned symbols separately. Strengthen the type inference a little by using more interpret-like constructs.
PiperOrigin-RevId: 183705547
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183705008
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183701716
|
| | |
| | |
| | |
| | |
| | |
| | | |
representation of axis from an attribute to a tensor.
PiperOrigin-RevId: 183701017
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
xla_dump_hlo_proto_to and xla_dump_prepass_hlo_proto_to.
Renamed:
xla_dump_hlo_proto_to -> xla_dump_optimized_hlo_proto_to
xla_dump_prepass_hlo_proto_to -> xla_dump_unoptimized_hlo_proto_to
xla_dump_per_pass_hlo_proto_to takes a directory from which to dump the serialized HLO
module protos after each HLO pass. This will help us compare HLO modules as
they change during HLO passes, such as what they evaluate to, or how fast they
perform.
The directory passed will contain multiple protos with the filename format:
module_<module id>.<pass number>.<pipeline_name>.after_<pass name>.pb
example:
module_0000.0009.simplification.after_dce.pb
such that pass number 0000_* is the first pass run, followed immediately by
pass number 0001_*.
PiperOrigin-RevId: 183700119
|
|\ \ \
| | | |
| | | | |
Branch 183683856
|
| | | |
| | | |
| | | |
| | | | |
PiperOrigin-RevId: 183698184
|
| | | |
| | | |
| | | |
| | | | |
PiperOrigin-RevId: 183696425
|
| | | |
| | | |
| | | |
| | | | |
PiperOrigin-RevId: 183689499
|
| | | |
| | | |
| | | |
| | | | |
PiperOrigin-RevId: 183687487
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Fixes: #15882
PiperOrigin-RevId: 183685645
|
| | | |
| | | |
| | | |
| | | | |
PiperOrigin-RevId: 183685209
|
|/| | |
| | |/
| |/| |
|
| | |
| | |
| | |
| | |
| | |
| | | |
cache fill completes.
PiperOrigin-RevId: 183683856
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183681594
|
| | |
| | |
| | |
| | |
| | |
| | | |
This ensures that operations nested under an `init_scope` do not ignore name scopes that were opened in a function-building graph. In light of cl/183320576, which causes `tfe.defun` to automatically hoist variables out of its function-building graph, this change also eliminates logic in make_template that special-cased the first call when `create_graph_function` was True.
PiperOrigin-RevId: 183680794
|
|\ \ \
| | | |
| | | | |
Add `tf.unravel_index` as an equivalent of `np.unravel_index`
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
contains SparseTensor.
PiperOrigin-RevId: 183679584
|
| | | |
| | | |
| | | |
| | | | |
PiperOrigin-RevId: 183679060
|
| | | | |
|
| | | |
| | | |
| | | |
| | | | |
PiperOrigin-RevId: 183676663
|
| | | |
| | | |
| | | |
| | | | |
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
|
| | | |
| | | |
| | | |
| | | | |
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
|
| | | |
| | | |
| | | |
| | | | |
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
|
| | | |
| | | |
| | | |
| | | | |
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
|
| | | |
| | | |
| | | |
| | | | |
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
|
| | | |
| | | |
| | | |
| | | | |
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
|
| | | |
| | | |
| | | |
| | | | |
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
|