| Commit message (Collapse) | Author | Age |
|
|
|
| |
Change: 136067395
|
|
|
|
| |
Change: 135127376
|
|
|
|
|
|
| |
on CUDA devices.
Change: 135053757
|
|
|
|
| |
Change: 134721831
|
|
|
|
| |
Change: 134595813
|
|
|
|
|
| |
devices thread safe.
Change: 134321151
|
|
|
|
| |
Change: 134093881
|
|
|
|
|
| |
on MacOS
Change: 134008488
|
|
|
|
| |
Change: 133968335
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This change does the following:
- Always use {,new_}http_archive rather than git_repository
- Make liberal use of strip_prefix
- Clarify licenses() in BUILD files
- On POSIX include headers like a normal C/C++ program
This change accomplishes the following:
- Reduce download size >100MB: The biggest culprit is grpc which has
tens of thousands of commits in its GitHub repository.
- Reduce disk size >200MB: On disk, grpc takes up 250MB when cloned even
though the tarball of the git repo is 3.2MB. By never using git
externals, we save on network.
- Consume less cpu: Cloning git repositories is much slower than
downloading and extracting a tarball.
Change: 133895791
|
|
|
|
| |
Change: 133874452
|
|
|
|
|
| |
numbers
Change: 133723907
|
|
|
|
|
| |
support for computing the absolute value of complex numbers on GPU.
Change: 132940500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
3.0.2 is the latest release.
Add a level of indirection to the int64 type in
1) core/example/feature_util.h, feature_util.cc and the places where the type is used
2) tools/proto_text/gen_proto_text_functions_lib.cc and its tests.
This is for dealing with the variability of "int64" type definition among different platforms. On some systems, int64 is "long int", while on others it is "long long int".
See GitHub Issue for more details: https://github.com/tensorflow/tensorflow/issues/3626.
This change list fixes the issue and eliminates the need to stick to an old (3.0.0-beta-2) commit of protobuf.
Change: 132814372
|
|
|
|
|
| |
The existing code will become an error in future versions of Bazel.
Change: 132748055
|
|
|
|
| |
Change: 132733397
|
|
|
|
|
| |
doubles and fixes issue #4131
Change: 132114002
|
|
|
|
| |
Change: 131951691
|
|
|
|
|
|
| |
that break opensource_build test.
Change: 131452196
|
|
|
|
| |
Change: 131437429
|
|
|
|
| |
Change: 131310818
|
|
|
|
| |
Change: 131132846
|
|
|
|
| |
Change: 130451359
|
|
|
|
| |
Change: 129887348
|
|
|
|
|
|
| |
To address recent CA certificate issues with openswitch, see e.g. (Jenkins login required):
http://ci.tensorflow.org/view/Experimental/job/experimental-cais-tensorflow-mac-2/46/console
Change: 128622641
|
|
|
|
| |
Change: 128401884
|
|
|
|
|
| |
avoid creating a new one each time.
Change: 127624630
|
|
|
|
|
|
| |
improvements for fp16
Added SpecialFunctions to the list of eigen headers TensorFlow depends on
Change: 127264575
|
|
|
|
| |
Change: 127253427
|
|
|
|
|
| |
improvements for fp16
Change: 127233960
|
|
|
|
| |
Change: 127144397
|
|
|
|
| |
Change: 126335170
|
|
|
|
|
|
| |
handle per-thread buffer allocation for the tileable executor without resorting to thread_local that is not fully supported on Android.
Change: 126009029
|
|
|
|
|
|
| |
will enable the implementation of the cumsum operation in TensorFlow
Change: 125697517
|
|
|
|
|
| |
performance of the toy mnist training by 1 order of magnitude
Change: 124374286
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
NEW
BM_fullReduction/10 4591 4595 153149 20.8M items/s
BM_fullReduction/64 5073 5075 100000 770.0M items/s
BM_fullReduction/512 9067 9070 75263 26.9G items/s
BM_fullReduction/4k 243984 244125 2868 64.0G items/s
BM_fullReduction/5k 359125 359273 1951 64.8G items/s
OLD
BM_fullReduction/10 9085 9087 74395 10.5M items/s
BM_fullReduction/64 9478 9478 72014 412.1M items/s
BM_fullReduction/512 14643 14646 46902 16.7G items/s
BM_fullReduction/4k 260338 260384 2678 60.0G items/s
BM_fullReduction/5k 385076 385178 1818 60.5G items/s
Change: 124290852
|
|
|
|
|
| |
gradients, some variants etc.).
Change: 124197406
|
|
|
|
| |
Change: 124012080
|
|
|
|
|
|
|
|
|
|
|
| |
This takes previously generated code, and includes it in the
repository. The main advantage of doing this is that we can specialize
the deserialization routines for various protobuf types that tend to
be large, and thereby avoid the problem where we brush up against the
default protobuf limits.
Fixes #2233.
Change: 124007049
|
|
|
|
|
| |
gradients, some variants etc.).
Change: 123967787
|
|
|
|
|
| |
gradients, some variants etc.).
Change: 123967117
|
|
|
|
|
|
| |
trick work when TF is imported as a submodule.
Change: 123805260
|
|
|
|
| |
Change: 123659102
|
|
|
|
|
| |
tf.train.Example.
Change: 123445810
|
|
|
|
| |
Change: 123427036
|
|
|
|
| |
Change: 123238579
|
|
|
|
|
| |
Add string_to_hash_bucket_strong to assign hash buckets using the strong keyed hash function.
Change: 123080459
|
|
|
|
|
|
|
| |
Implements an authentication mechanism based on Application Default Credentials:
https://developers.google.com/identity/protocols/application-default-credentials
https://developers.google.com/identity/protocols/OAuth2ServiceAccount
Change: 122741738
|
|
|
|
|
|
|
|
| |
with many cpu cores
For example, the wall time for the following tutorial went down from 13m35 to 5m27:
bazel run -c opt --copt=-mavx tensorflow/examples/tutorials/word2vec/word2vec_basic
Change: 122462177
|
|
|
|
| |
Change: 122273744
|