| Commit message (Collapse) | Author | Age |
|
|
|
| |
Change: 143142615
|
|
|
|
| |
Change: 143138639
|
|
|
|
| |
Change: 143136270
|
|
|
|
| |
Change: 143135973
|
|
|
|
| |
Change: 143135690
|
|
|
|
| |
Change: 143134173
|
|
|
|
| |
Change: 143132977
|
|
|
|
| |
Change: 143131671
|
|
|
|
|
|
|
| |
Some users do not want or need this support and it significantly increases
their dependencies and binary sizes. After this change, users must explicitly
depend on :gcs_file_system to enable this support.
Change: 143129406
|
|
|
|
| |
Change: 143129041
|
|
|
|
|
| |
Also make some touch-ups on the relevant doc strings.
Change: 143128831
|
|
|
|
| |
Change: 143121948
|
|
|
|
| |
Change: 143115584
|
|
|
|
| |
Change: 143104273
|
|
|
|
| |
Change: 143080146
|
|
|
|
|
|
|
| |
Google Cloud Platform support is now available for everyone by default.
Fixes #5349
Change: 143077241
|
|
|
|
| |
Change: 143061298
|
|
|
|
|
|
| |
initialization issues before inference is attempted (e.g. decodeJpeg nodes not being stripped from graph).
Change: 143060930
|
|
|
|
|
|
|
|
| |
The bug was related to getting a string representation of feed_dict keys when the keys are str/unicode (in cli_shared.py: get_run_short_description()). Unit tests added.
Another hitchhiking change:
Add command mouse-click shortcut to help output on TensorFlow runtime errors (in cli_shared.py: get_error_intro())
Change: 143055926
|
|
|
|
| |
Change: 143055602
|
|
|
|
|
|
| |
reference a null pointer).
Change: 143054450
|
|
|
|
| |
Change: 143054031
|
|
|
|
|
|
| |
targets and expose Android libs at http://ci.tensorflow.org/view/Nightly/job/nightly-matrix-android/lastBuild/TF_BUILD_CONTAINER_TYPE=ANDROID,TF_BUILD_IS_OPT=OPT,TF_BUILD_IS_PIP=NO_PIP,TF_BUILD_PYTHON_VERSION=PYTHON2,label=android-slave/
Change: 143046399
|
|
|
|
|
| |
large until it is fixed.
Change: 143046132
|
|
|
|
|
|
| |
surprising if invoked via the batching signature code.
Change: 143045795
|
|
|
|
| |
Change: 143044398
|
|
|
|
|
|
|
|
|
|
| |
The previous version was ignoring the timeout (if set) in the
`CallOptions*` passed to the server method, and blocking indefinitely
on the completion notification, instead of cancelling the call if the
notification was not set before the timeout.
Fixes #6742.
Change: 143042703
|
|
|
|
| |
Change: 143007748
|
|
|
|
|
| |
delete it entirely.
Change: 142996097
|
|
|
|
|
| |
Also put more emphasis on mouse event support.
Change: 142989939
|
|
|
|
|
| |
delete it entirely.
Change: 142980063
|
|
|
|
|
|
|
|
| |
(Leaving it as an error for now while I check some of the invocations
to see if there's really a need for the diagnostic message).
Current fatal prevents some helpful error messages from propagating
back to the caller.
Change: 142870362
|
|
|
|
|
|
| |
Also rework log calls during initialization to be more informative/less misleading.
Change: 142863423
|
|
|
|
| |
Change: 142863172
|
|
|
|
| |
Change: 142850910
|
|
|
|
| |
Change: 142849285
|
|
|
|
| |
Change: 142843471
|
|
|
|
|
|
|
| |
NO_TEST_ON_INSTALL to make it more useful
Previously, if NO_TEST_ON_INSTALL is set to 1, it'll skip all the following: python unit tests on install, user ops test, tutorial tests and ffmpeg integration test. But most frequently, we want to be able to skip the python unit tests on install, but still run user ops tests, tutorial tests and/or ffmpeg integration test. This change enables that.
Change: 142820160
|
|
|
|
|
|
|
| |
Also use `Tensor::AsProtoTensorContent()` when populating the fetched
values from a gRPC worker service, as this is more efficient for larger
values. This should improve #6256 slightly.
Change: 142813084
|
|
|
|
| |
Change: 142809066
|
|
|
|
| |
Change: 142807011
|
|
|
|
| |
Change: 142806683
|
|
|
|
| |
Change: 142805270
|
|
|
|
|
| |
SavedModel export formats.
Change: 142803996
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This change modifies the GrpcWorkerCache so that, when a master
attempts to communicate with the worker in the same process, it does
so by direct method calls on a `WorkerInterface*`, without making a
loopback RPC call.
This change is another incremental step towards addressing issue
#6256. There are further improvements possible, and we will continue
to investigate them, including:
* Avoiding the protobuf encoding/decoding for request/response objects
where this affects performance. The zero-copy `TensorResponse` class
is an example of how we could improve performance here, for
`RunGraphRequest` and `RunGraphResponse` objects.
* Profiling the closure creation/context switch overhead for
interactions with the local worker.
Change: 142793965
|
|
|
|
| |
Change: 142791193
|
|
|
|
| |
Change: 142790604
|
|
|
|
|
|
| |
more imports to this file.
Change: 142784743
|
|
|
|
| |
Change: 142783149
|
|
|
|
| |
Change: 142781367
|