hash stringlengths 40 40 | msg stringlengths 1 131k | author stringlengths 1 33 | email stringlengths 0 57 | date int64 1,447B 1,698B |
|---|---|---|---|---|
eb9dfe550d18fd8b898ca7d9090b0db10db98c00 | Implementation of immutable_const operator based on readonly memory region
Change: 117953780 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,759,475,000 |
f3c81dfc8d8f2c0c9c1e399ec3ed20c9ee70893d | Fix /dev/urandom-related hanging on some systems | Shanqing Cai | cais@google.com | 1,458,763,924,000 |
2e149258893b8b7d58e456ad3b1a54a4cf2ebb6b | Merge pull request #1605 from caisq/cloud-fix-1
Fix /dev/urandom-related hanging on some systems | caisq | cais@google.com | 1,458,765,393,000 |
62df5bef660a3b56fcabf7d8266b62b48fb4425b | Creating a new contrib directory for audio ops using FFmpeg.
Change: 117956472 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,761,062,000 |
bc0754d59dbb038a86e16cb592f9b716aaebfc0d | Print per-node step stats from Android demo for better benchmarking.
Change: 117958944 | Andrew Harp | andrew.harp@gmail.com | 1,458,762,706,000 |
bf4c66c2e7cfc4b35f06e95671993db675f2eb34 | Add missing framework header selective_registration.h to framework filegroup
and cuda library.
Change: 117960844 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,763,870,000 |
99e298a84775fcfe91ce0e2d40566d4bd00f53f9 | Split depthwise conv gradient ops out into their own file.
Change: 117961829 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,764,418,000 |
16454ee2679e1ee254a5d3995c7f673da09891dd | Fixed merge conflicts in skflow | Manjunath Kudlur | keveman@gmail.com | 1,458,766,332,000 |
f6e7205516eeb54d3e85b5d283c2d7a387ec81bc | Add dependency for contrib kernels.
Change: 117967381 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,767,589,000 |
ca62bbcf26932d4230f29005b344062aa113dcad | In backprop, we may create zero tensors for the gradient of cond. Before this CL, the zero tensors are created unconditionally. This CL changes it to create the tensors only when they are needed.
Change: 117970158 | Yuan Yu | yuanbyu@google.com | 1,458,768,998,000 |
650f649de46779edd102660e18293a59537ba52f | Update TensorBoard documentations.
Change: 117976198 | Dan Mané | danmane@gmail.com | 1,458,772,314,000 |
1a25c924a90ac806dd8b8036a9adcb7792442727 | Merge pull request #1608 from keveman/branch_117961829. | Manjunath Kudlur | keveman@gmail.com | 1,458,773,109,000 |
f8c5e693ed22fcab4418bcc94a091c73d2a66363 | Internal tooling cleanup.
Change: 117977121 | Zongheng Yang | zongheng.y@gmail.com | 1,458,772,866,000 |
8c7554997f005f4af37409eedce6c32050b87444 | Fix link in TensorBoard README
Change: 117978853 | Dan Mané | danmane@gmail.com | 1,458,774,026,000 |
82056ebef6b0cd69e3c0ff8b8609bc6d953f07f4 | Update 1_notmnist.ipynb | Vincent Vanhoucke | vanhoucke@google.com | 1,458,776,689,000 |
c5f37f464c22b590cbc8eddbb002858bf0852527 | Merge pull request #1611 from tensorflow/vincentvanhoucke-fix-notmnist. | Vincent Vanhoucke | vanhoucke@google.com | 1,458,776,806,000 |
61c252c4b412d4ffad437b14937006fc27ea31e8 | Adding ability for users to store `RunOutputs` information to the events file and serve it via the TensorBoard back-end.
The `RunOutputs` information contains execution statatistics, such as compute time and memory usage for each node in the subgraph executed by a particular `session.run()`.
A follow-up change will a... | Dan Smilkov | dsmilkov@gmail.com | 1,458,775,677,000 |
0240d776a35c72aabaa4ba30bec3b821854e1923 | Update 2_fullyconnected.ipynb | Vincent Vanhoucke | vanhoucke@google.com | 1,458,783,838,000 |
005386dc198220601293d4821e6de63246527b0c | Merge pull request #1614 from tensorflow/vincentvanhoucke-typo-2
Typo in 2_fullyconnected.ipynb | Vijay Vasudevan | vrv@google.com | 1,458,783,955,000 |
4778f782609ff44b06ecbc222ae5acdad0cb6dbe | The graph-level random seed of this graph.
Change: 117993641 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,786,653,000 |
98e4d78c324bda520bc37b690802be93d4983972 | Use this method with the `with` keyword to specify that ops created within the scope of a block should be added to this graph.
Change: 117993733 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,786,774,000 |
9afc524348d47d5678b6283209650b8f73c2ab6b | norm = tf.random_normal([2, 3], seed=1234)
sess = tf.Session()
print(sess.run(norm))
print(sess.run(norm))
sess = tf.Session()
print(sess.run(norm))
print(sess.run(norm))
Change: 117994542 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,787,733,000 |
0cc25f0b95af9a99dd228c75d411670831fd6cec | Allows the caller of a function to specify that a function is not
inlined into the call site regardless of the graph optimizer option
setting.
Change: 117994922 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,788,233,000 |
a075f1014dc9736d50b79ba0de62112030df2c5f | Add tests for hinge loss updater.
Change: 117998775 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,793,030,000 |
f4ebdaba60d2fb2698d14ee87960e3a6294be196 | Add missing license header
Change: 117999025 | Vijay Vasudevan | vrv@google.com | 1,458,793,402,000 |
5e06a231344f6324cc5054b867b0eb8cbeb37771 | Fix int64 implicit downcast
Change: 118000502 | David G. Andersen | dga@google.com | 1,458,795,537,000 |
0a55f6b5dba109d5d22e2595024b05a221bd87e9 | Fix pipeline example. | Dongjoon Hyun | dongjoon@apache.org | 1,458,801,108,000 |
131e465bab4a7e7d75ead3e4d2cc3750b603f27f | Fix arguments in docstrings. | Dongjoon Hyun | dongjoon@apache.org | 1,458,803,501,000 |
0d77f556297f0641b43b8fb5e972a33c3781bfc0 | fix for RNN/GRU | Vu Pham | phvu225@gmail.com | 1,458,805,015,000 |
7deb45a7197d66423b00b933a6e11caaa698f383 | Typo in comment, random weights init
According to the code and reference standard deviation would be 0.1 | amchercashin | amchercashin@gmail.com | 1,458,808,359,000 |
3bffedaeaaeb93254cf1c02953477ba4fc6356fb | added edge cases tests | Vu Pham | phvu225@gmail.com | 1,458,817,228,000 |
949043c0fc1b46e020a6fdc31bb65cca4d2371b6 | removed logs, added docs | Vu Pham | phvu225@gmail.com | 1,458,820,138,000 |
98ec1dee841b4ed7562c611a22e02846bd93cc88 | solve the boringssl problem in grpc
boringssl in grpc previous is held in googlesouce which block people under the wall,
upgrade to this commit version will change its source to https://github.com/google/boringssl.git
and this solve all the problems.
Detail can be found in https://github.com/tensorflow/tensorflow/i... | Wenjian Huang | nextrush@163.com | 1,458,824,737,000 |
38c2f5e5366629c19899bfdc4c5afe2bef1f992a | add matplotlib inline
to redirect the image of matplotlib to notebook | Jay Young | github@yjmade.net | 1,458,826,287,000 |
e61b7f7eed8d198d492ae82d7f9aae1a64d7181c | Fix nbviewer link to deepdream.ipynb | znah | moralex@google.com | 1,458,826,392,000 |
3febfae7a3603d0f21564289347dbfccb948ebfb | Merge pull request #1621 from yjmade/patch-1
add matplotlib inline | Vincent Vanhoucke | vanhoucke@google.com | 1,458,827,533,000 |
672a63d5c9255a75388dd18f63aaa87dbbb05bf7 | Delete tf.reduce_mean on scalar
Excuse me, have little to no exp. in python and tensorflow.
According to reference, tf.nn.l2_loss() always retuns a scalar, thus no
need to average it.
At last code block where loss function is reimplemented by hand,there is
no mean() operation. | amchercashin | amchercashin@gmail.com | 1,458,827,653,000 |
30e524e162a126b44e5a6a1062ebbdc4eed8b8d9 | Merge branch 'master' into del-tf.reduce_mean(scalar) | amchercashin | amchercashin@gmail.com | 1,458,828,078,000 |
a2a263921fb9a7f9e15e59f7741d0f56937aa47c | Merge pull request #1622 from znah/patch-2
Fix nbviewer link to deepdream.ipynb | Vincent Vanhoucke | vanhoucke@google.com | 1,458,828,165,000 |
46bd9dae34cac597f7264b21395970b60bb41393 | Merge pull request #1617 from amchercashin/master
Typo in comment, random weights init | Vincent Vanhoucke | vanhoucke@google.com | 1,458,828,280,000 |
d21108e9f3574abf3d7a6bb5c92150aa3bfed89d | Merge branch 'master' into del-tf.reduce_mean(scalar) | amchercashin | amchercashin@gmail.com | 1,458,828,396,000 |
61bfe8a817ad6d3fa62f075546691d889f8e8152 | Update 2_fullyconnected.ipynb | Vincent Vanhoucke | vanhoucke@google.com | 1,458,829,916,000 |
332fdef787824eea5ad5d7f0aa0e525bfb232e71 | Merge pull request #1625 from tensorflow/vincentvanhoucke-relu
Update relu link in 2_fullyconnected.ipynb | Vincent Vanhoucke | vanhoucke@google.com | 1,458,829,995,000 |
c404448d3b1e44fddc2d6e1c6da9862443112721 | Replace graph_def with graph when passed to the SummaryWriter and update tutorials to reflect the new API.
Change: 118033430 | Dan Smilkov | dsmilkov@gmail.com | 1,458,835,332,000 |
4da6ca28e2a2685872a2d01cc2a0ceb699f01606 | Fix for bool strictness in CIFAR multi-GPU trainer.
Fixes #1595. | Derek Murray | derek.murray@gmail.com | 1,458,836,028,000 |
728418b8f9ca75681a273a26cb150f4dd365cf45 | Merge pull request #1600 from mrry/pr-1595-fix. | Derek Murray | mrry@google.com | 1,458,836,047,000 |
f6800b8e1b95d81d828ddb34485fd6bebadaf071 | Merge pull request #1616 from dongjoon-hyun/fix_arguments_in_docstring
Fix arguments in docstrings. | Vijay Vasudevan | vrv@google.com | 1,458,836,178,000 |
672b969b8c033548df4cbec89b87694b6024bf12 | Merge pull request #1615 from dongjoon-hyun/fix_pipeline_example
Fix pipeline example. | Vijay Vasudevan | vrv@google.com | 1,458,836,830,000 |
5b5b8412f0684a548e1e9001421e5d095cda0142 | In tf.contrib, move framework and loss packages up a level.
Simplify loss ops, only return scalar in public API (see discussion at go/tf_contrib_ops).
Add softmax loss.
Change: 118034691 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,836,352,000 |
d859d6bd11aa70f645f3fea226997d2408fa8e7f | Merge branch 'master' into del-tf.reduce_mean(scalar) | amchercashin | amchercashin@gmail.com | 1,458,837,519,000 |
d782317f55a1fc4944cb321c5c8132a7514acadc | Make configure hard code path to swig
I've set it up to just try "swig" if configure hasn't been run since
syncing to this change, so users who sync and have no swig issues won't
have to do anything new.
Fixes #1329. | Geoffrey Irving | geoffreyi@google.com | 1,458,689,318,000 |
821920c1f25968f5dfcd2f8999b293ebedf85957 | Several changes to reduce TensorFlow code size (especially important
for mobile apps):
(1) Change many interfaces in node_def_builder.h, node_def_util.h,
op_kernel.h, node_builder.h, mirror_pad_mode.h, padding.h to use
'StringPiece', rather than 'const string&'. The interfaces that
were changed tend to be heavily use... | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,837,867,000 |
53543e040932b6589e2a805d312e26bd5c4b33cf | Remove a stray warning log message and CHECK failure when a device is missing.
Missing device errors are handled at a higher level in the distributed
runtime. Logging a warning when a task is not found in a particular
channel cache is unhelpful, because the runtime may search multiple
caches (e.g. one per job) and mi... | Derek Murray | mrry@google.com | 1,458,838,992,000 |
a21979e7c19764eff688b3c296648c21c66c0a64 | correct worker_service.proto link
Fixes up a link in the distributed_runtime's README.md | Jeff Hodges | jeff@somethingsimilar.com | 1,458,848,971,000 |
f9b3c9f9c8c0cbf1880b976b3619961333c07980 | Merge pull request #1627 from girving/swig
Make configure hard code path to swig | Vijay Vasudevan | vrv@google.com | 1,458,850,454,000 |
9765c2c99ec7e350bff218abcdafcf1be2c99915 | Improve LSTMCell's appearance in TensorBoard
This is a very small cosmetic change, right now the 'concat' operation
isn't carried out in the scope created by calling LSTMCell so
this makes it appear weirdly inside tensorboard. | Andy Kitchen | kitchen.andy@gmail.com | 1,458,850,191,000 |
5b1ebf1f34b9a638fba94716604fc07c76ba7f97 | Benchmark tests
To run the tests, do (for example):
export TF_BUILD_CONTAINER_TYPE=CPU
export TF_BUILD_PYTHON_VERSION=PYTHON2
export TF_BUILD_IS_OPT=OPT
export TF_BUILD_IS_PIP=NO_PIP
export TF_BUILD_RUN_BENCHMARKS=1
tensorflow/tools/ci_build/ci_parameterized_build.sh
Use the environment variable TF_BUILD_BENCHMARK_HO... | Shanqing Cai | cais@google.com | 1,458,747,828,000 |
ddbdc3a1d5638e4e012ad28ff709e7b8ba06a5d8 | Merge branch 'master' into benchmark-2 | caisq | cais@google.com | 1,458,854,253,000 |
b1aeb4495bd10b3f1f0d4aed64880f8896fe990b | Merge pull request #1603 from caisq/benchmark-2
Benchmark tests | Vijay Vasudevan | vrv@google.com | 1,458,859,047,000 |
3b1bf2254fbb7fa2ed10ee5470449e05d9336326 | Merge pull request #1629 from jmhodges/patch-1
correct worker_service.proto link | Vijay Vasudevan | vrv@google.com | 1,458,872,318,000 |
1cc26ee2d49ba38f24f0dbb51452a90dd054e0d3 | replace xrange with range for python3 compatibility | Jay Young | github@yjmade.net | 1,458,883,703,000 |
1e401c42465bd4a28b89fc80ecddca2ac77e337f | Plumbing to allow RunOptions in the remote_session:
- Add a RunOptions proto in RunStepRequest.
- Still need to pass it back to the grpc session.
Change: 118045453 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,842,651,000 |
f5a9af1b83530db6492c69aed5345fa84ca0c921 | Changes to reduce code size:
1. Add a BufferBase base class for Buffer, and define the non-type-dependent
functions there.
2. Pull common LOG(ERROR) in tensor.cc out from a templated function.
Change: 118046104 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,842,982,000 |
8cf6edc95b9e2a62289df2e550ad01604ad1e1b8 | TensorFlow: restore sharded tests to use shard_count of 50.
Change: 118047026 | Vijay Vasudevan | vrv@google.com | 1,458,843,484,000 |
8016e72bae635392b06598b74c6f87d452e07034 | fixing linting errors
Change: 118049024 | Olivia Nordquist | nolivia@umich.edu | 1,458,844,574,000 |
c5351a11e9ef46b59772a50dbd8c512a65f70474 | Add nic_compatible to the default attributes of Variable storage.
Change: 118058044 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,850,227,000 |
4ab4b3c4c4a5d88b82baef21103136198b8807fe | Bugfix in RNN python benchmark
Change: 118058606 | Eugene Brevdo | ebrevdo@gmail.com | 1,458,850,571,000 |
6d910661a54eac94dae7cc88b8d244d75470dba5 | Make testlib not depend on all_kernels and tensorflow
This allows individual tests to depend on a subset of kernels.
Also harden immutable_constant_op_test to not segfault if the unexpected
happens.
Change: 118059523 | Geoffrey Irving | geoffreyi@google.com | 1,458,851,111,000 |
2e9f0fb87d38d200406ba581258d29ac39a69c06 | TensorFlow: Speed up server_lib_test by reducing size of large constants by
3 (but enough to trigger grpc limits). Reduced runtime in non-opt mode
from 30s to 8s.
Change: 118059583 | Vijay Vasudevan | vrv@google.com | 1,458,851,149,000 |
22cfbd1cebda28c3b55c64519f4af23c462b0000 | Switch Docker instructions to always `--pull` on build.
This fixes situations like Vincent hit, where a stale base image would lead to
new packages based on old base packages.
Change: 118071412 | Craig Citro | craigcitro@google.com | 1,458,857,265,000 |
d4987d43287e90e82c59c764a3a2ba4afb26e112 | TensorFlow: make direct_session a publicly linkable library.
Also change DirectSessionTest to use the public Session constructor when
possible, to discourage use of DirectSession as anything but an opaque
object that implements the Session API. The one exception is the experimental
API, which requires a known down_Ca... | Vijay Vasudevan | vrv@google.com | 1,458,857,741,000 |
4735c21adea019b1fd2999b13a8043fa39f30114 | Change the order of notification before returning.
Change: 118075047 | Xiaoqiang Zheng | zhengxq@google.com | 1,458,859,446,000 |
c9b94f0f0eb029299f7615dc8bab5e697f165e2d | Mention OpKernel thread-safety requirements.
Change: 118078260 | RJ Ryan | rryan@alum.mit.edu | 1,458,861,424,000 |
9da6d3a3cb653dc66e22a7be6d04245528a9c3b8 | Fix tensorflow version (was missing a --pull)
Change: 118080472 | Vincent Vanhoucke | vanhoucke@google.com | 1,458,862,894,000 |
55863b1cb56ef9562d5295ba2a397aa7a86cd9c8 | Add local_variable constructor.
Change: 118080678 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,863,026,000 |
0a8ac4d0d977eb804099a65271572a8497af304e | Fix race in ExecutorBarrier::WhenDone between error thread and delete thread by
copying the rendez_ pointer while holding the lock.
Change: 118083489 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,865,004,000 |
4d5d4c0a64234c93ed5d21769fe4eb28ab3cc8da | Fix comment in boston.py to avoid abbreviation, capitalization.
Change: 118186721 | Vijay Vasudevan | vrv@google.com | 1,458,887,283,000 |
8838f1e71dfc2eef965e3781f99afa1ebd71c144 | Revert protobuf gitmodules back to github | Vijay Vasudevan | vrv@google.com | 1,458,888,529,000 |
fdb20bb9acc07e38a580b251526f3a206fc81b07 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,458,888,942,000 |
ef6e7364e1cbefdc94f009269ec9a438288d285c | Fix immutable constant op test compile failure | Vijay Vasudevan | vrv@google.com | 1,458,889,559,000 |
db367ce5ae44aa3f3e35dee2f97846c2446d20e6 | added gridGRU | Vu Pham | phvu225@gmail.com | 1,458,892,353,000 |
a5fa8a7bcd3357e22f88ba53462902a3572816b1 | Merge remote-tracking branch 'refs/remotes/tensorflow/master' | amchercashin | amchercashin@gmail.com | 1,458,895,519,000 |
c0eb1add7f1ea7187fbc663d08072139c8f6b1fc | rename var mu to learning_rate
New var mu defined as Learning rate, but later in the same code cell var
learning_rate is used instead, wich was defined in previous cell.
Suggest to rename becouse cell uses var which is not defined in it, but
defines var wich is not used.
Suggest to rename mu to learning_rate (not lear... | amchercashin | amchercashin@gmail.com | 1,458,897,091,000 |
6e88dd4bfeb55c5ba17b1b5626c6b18ab6c2fcff | Typo-descient-to-descent
Typo in text cell | amchercashin | amchercashin@gmail.com | 1,458,908,329,000 |
965fdf718d6a4099d5d1610dae6eee0f20e26a4e | gradient calc description change
I have found the previous explanation of manual gradient calculation a
little confusing because we compute a vector of gradients with the size
of number of neurons (with bias) and in the explanation it looks like we
compute a scalar.
Propose more detailed variant if community finds it... | amchercashin | amchercashin@gmail.com | 1,458,913,555,000 |
29352d126820e1233a5937a993603a88c0bf447b | Fix mktemp issue on Mac | Shanqing Cai | cais@google.com | 1,458,913,667,000 |
41d476938c1efe78076ec47af0c837db325aa2ad | Merge pull request #1640 from caisq/mktemp-fix
Fix mktemp issue on Mac | caisq | cais@google.com | 1,458,915,378,000 |
020fe191a0b436888d6d61d353540c3918a898ee | Merge pull request #1626 from fayeshine/fayeshine-patch-2
Updates to a more recent version of gRPC.
This is primarily useful because it obtains boringssl from github.com, rather than googlesource.com, which makes the build process work in more locations. | Derek Murray | derek.murray@gmail.com | 1,458,916,515,000 |
81c6bfada9c795609d5b39cff1efa1bd04de9b30 | TensorFlow: fix immutable_constant_op_test to assert pointer inequality
Change: 118207941 | Vijay Vasudevan | vrv@google.com | 1,458,920,506,000 |
76f32c4225582e5d9f7b8180fb5e5f2e1aef2858 | histogram_fixed_width now uses unsorted_segment_sum to create the
histogram, eliminating the need for internal variables.
Change: 118208659 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,921,181,000 |
afd500c39c242459921513b96ad4289e03db581e | Add check_futures_test to verify that everything uses __future__ correctly
Change: 118209139 | Geoffrey Irving | geoffreyi@google.com | 1,458,921,582,000 |
362525f1dcdb9f4f047a73bc20dd1a140647e6fc | Merge pull request #1637 from vrv/branch_118186722
Pushing changes from internal for March 24 | Vijay Vasudevan | vrv@google.com | 1,458,922,381,000 |
740d6a0822b044cf7ed39b413c9b1a4c63cf70a4 | Put edge label in the middle of edge only if it is thick enough.
`2.5` looks like the correct stroke width threshold. | Kanit Wongsuphasawat | kanitw@gmail.com | 1,458,875,147,000 |
f96dcc1584ada70a1e58513dab4af82fe54cb3fa | Make tf.group() a tf.no_op()
Change: 118211924 | Geoffrey Irving | geoffreyi@google.com | 1,458,924,006,000 |
19b607c86818c94e4c8a82a1bb2c28b6e5de84a6 | Adds documentation for distributed TensorFlow.
Change: 118215791 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,926,676,000 |
7913f95c428793df8274758b25ff4194d12e0f1c | Add some layers tests, and add reduce_sum_n.
Change: 118215873 | A. Unique TensorFlower | nobody@tensorflow.org | 1,458,926,732,000 |
73f3960f4ab93cd9a549ac38e2e0909f12ed4f2e | Merge pull request #1583 from girving/ignore
Add __pycache__ to .gitignore | Vijay Vasudevan | vrv@google.com | 1,458,928,104,000 |
903337734dca35879ac3e29a927c65762e2e7972 | Implement gradients for SparseAdd.
Change: 118216819 | Zongheng Yang | zongheng.y@gmail.com | 1,458,927,291,000 |
b440634ae326fa6ae1745a3212b4ab41e361690e | Improve core/kernels/BUILD so that no test is bigger than 25M
Tests now depend on a much finer grain set of kernels. | Geoffrey Irving | geoffreyi@google.com | 1,458,929,615,000 |
da767e1079dfc683b126c0ba69ed6ced925c3300 | Sped up the computation of the gradients of the convolution layer wrt the
inputs by 30% on GPU.
Change: 118221540 | Benoit Steiner | benoit.steiner.goog@gmail.com | 1,458,929,975,000 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.