hash stringlengths 40 40 | msg stringlengths 1 131k | author stringlengths 1 33 | email stringlengths 0 57 | date int64 1,447B 1,698B |
|---|---|---|---|---|
07fa55268fc63a04cab81508e5468ee1b1914448 | TensorFlow: fix missing GetCudaVersion() for libcuda.so. h/t to @3XX0.
Change: 114983764 | Vijay Vasudevan | vrv@google.com | 1,455,823,321,000 |
e4ebc457c9d8aa47bf8295d158c16b19482c230a | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,455,823,840,000 |
cd35bb1e9fb8daa9c3fbc93c854337700cf8a896 | pip.sh: die on pip install failure | Shanqing Cai | cais@google.com | 1,455,810,763,000 |
30628d68ce37738db6b65da4b0b46148b640601c | Merge pull request #1177 from vrv/master. | Vijay Vasudevan | vrv@google.com | 1,455,824,733,000 |
555e73da8f171992085c68614f74b23b8180292c | Fix dataset encoding in MNIST example for Python 3 | Danijar Hafner | mail@danijar.com | 1,455,802,066,000 |
99813c26f9fae11b2d6a6f8d12e68493cb06f73c | Merge pull request #1170 from danijar/python-3-mnist-encoding. | Vijay Vasudevan | vrv@google.com | 1,455,824,802,000 |
af99f717ad5e0dbecc8abe509d910d317fc1f618 | Create tensorboard/lib/js/backend, a typed, convenient interface for talking to tensorboard.
It loads the data from the tensorboard backend, and also presents a slightly cleaner
abstraction. It's typed and tested.
Change: 114984720 | A. Unique TensorFlower | nobody@tensorflow.org | 1,455,823,824,000 |
edf89c878524742fd89ee156c28df668ae1780b2 | Avoid use of std::function in GPUBFCAllocator deallocation path.
Speeds up allocation microbenchmarks by 8% to 15%
Run on REDACTED (40 X 2801 MHz CPUs); 2016/02/17-16:56:24
CPU: Intel Ivybridge with HyperThreading (20 cores) dL1:32KB dL2:256KB dL3:25MB
Benchmark Base (ns) New (ns) Improvement
------------------------------------------------------------------
BM_Allocation 184 164 +10.9%
BM_AllocationThreaded/1 185 169 +8.6%
BM_AllocationThreaded/4 1966 1771 +9.9%
BM_AllocationThreaded/16 9989 9197 +7.9%
BM_AllocationDelayed/1 204 183 +10.3%
BM_AllocationDelayed/10 171 146 +14.6%
BM_AllocationDelayed/100 152 130 +14.5%
BM_AllocationDelayed/1000 155 131 +15.5%
Change: 114984794 | A. Unique TensorFlower | nobody@tensorflow.org | 1,455,823,861,000 |
67b2abcd673f466d32c4017ad8bab0d0b9ac86c5 | Context manager should return self.
Change: 114985229 | A. Unique TensorFlower | nobody@tensorflow.org | 1,455,824,105,000 |
16f46d932c4e25773003c580924f8c718f2ad9c1 | Move iron-component-page from dev-deps to deps in bower.json since it is used by tf-regex-group, tf-categorizer and tf-collapsable-pane components. Also make the version number exact instead of ^1.0.0, and make it match the version inside google.
Change: 114987970 | Dan Smilkov | dsmilkov@gmail.com | 1,455,825,632,000 |
d0a822fbcb04d95a643d8efe65699a8d1cdce98b | TensorFlow: switch dsoloader path for libcuda to end in .so.1
Change: 114990321 | Vijay Vasudevan | vrv@google.com | 1,455,827,123,000 |
ea083742206761139f734c04b354b84c0cb027de | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,455,827,397,000 |
2f756013f5bcc39ee288eed9da8d27cf4ec16f7e | Enable deserializing protos > 64MB in length.
Update protobuf commit
Change: 114990608 | Manjunath Kudlur | keveman@gmail.com | 1,455,827,325,000 |
2797af42ae7d0c6e794a3b51eefb5688f2feb6c0 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,455,829,020,000 |
2b5d76976001dcd0982871b299ad1ae03558e96b | upgrade build to cuda 7.5 and cudnn 4 | Jan Prach | jendap@google.com | 1,455,786,412,000 |
1f54e2c70683ecf70c5811b8d0d7c3f05ced17f7 | upgrade docker images to cuda 7.5 and cudnn 4 | Jan Prach | jendap@google.com | 1,455,787,998,000 |
17af61be2c171d0ff0166b090d59cd8f2d24ac34 | Merge pull request #1166 from jendap/r0.7-cudnn4. | Vijay Vasudevan | vrv@google.com | 1,455,837,334,000 |
648e2654e1c14ee965493749f1d3620498c44ddb | Tracks some basic statistics in cpu and gpu allocator.
The basic stats is basicly free in gpu allocator.
The cpu stats collection can be optionally turned on.
Change: 115000479 | A. Unique TensorFlower | nobody@tensorflow.org | 1,455,833,187,000 |
0aa874f254f9eb2fe0015f5e40423a2f8359a1e0 | Added documentation for import_meta_graph and export_meta_graph.
Change: 115005379 | Sherry Moore | sherrym@google.com | 1,455,835,726,000 |
5fe957f055c5f138e012782ab0d28de9ca78e55b | Support more than two session factories.
This CL changes how the session factory is chosen. Previous an empty
target would always use DIRECT_SESSION, and a non-empty target would
always REMOTE_SESSION. In preparation for multiple distributed session
implementations (issue #23), we now delegate to the SessionFactory to
see whether it accepts a given SessionOptions. Existing programs
should continue to work unmodified.
NOTE: This CL assumes that the domains of the registered session factories
do not overlap. We may need to revisit this in future.
Change: 115008046 | Derek Murray | mrry@google.com | 1,455,837,068,000 |
8174381077c92bfbcaff3d06ccfbc13083d91b1f | TensorFlow: change cuda-diagnostics to search for so.1
Change: 115010103 | Vijay Vasudevan | vrv@google.com | 1,455,838,173,000 |
094aa778c0a027390b5bb45c0ee461392a85d8ad | TensorFlow: no such thing as LOG(INFO) << endl
Change: 115013578 | Vijay Vasudevan | vrv@google.com | 1,455,840,248,000 |
122806f60258bb5439963bca99cf10a2821f1bb0 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,455,840,877,000 |
5151459cc691f5d7fe8e84f5aa9577f7d233b3d8 | Enable GPU reduction for doubles | Igor Babuschkin | igor@babuschk.in | 1,455,733,078,000 |
ac418ba56ad1a06740aeaf7f283a50fbca684d49 | Merge pull request #1089 from ibab/gpu-reduce-more-types. | Vijay Vasudevan | vrv@google.com | 1,455,842,189,000 |
cc53ecf5a91ae796cef823d5b9e5982a7bcc597e | Added _MetaGraphFilename() function to return the name of meta_graph file.
Remove the correct file in the case of sharded checkpoints.
Change: 115015294 | Sherry Moore | sherrym@google.com | 1,455,841,202,000 |
b74d4b48ee16051fbd6f27bd7518b7a1b5d18ebe | Fixed new lines in URLS in file format docs for bug #1169.
Change: 115017744 | Pete Warden | pete@petewarden.com | 1,455,842,846,000 |
46089c1a765e44ae5eb6206c93367b0afb370a54 | Fixed new lines in URLS in file format docs for bug #1169.
Change: 115018272 | Pete Warden | pete@petewarden.com | 1,455,843,233,000 |
de7b961da6cea7692d9f36eeec1c7f220c15af3f | Create tf.image.central_crop() to output a centrally cropped version of an image.
Change: 115023941 | Jon Shlens | jon.from.california@gmail.com | 1,455,847,880,000 |
76989f9839f4824aa9f5c1c1907fba3a02c1c83a | TensorFlow: make pip wheel depend on 3.0.0b2 protobuf,
since that's roughly what we package too.
Change: 115027171 | Vijay Vasudevan | vrv@google.com | 1,455,851,506,000 |
bd29f655457d250f67cf28d846d69506c4e2e76c | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,455,851,758,000 |
369ca60ee6f025ecd29ab21268bb1e4aaeaa3995 | Loosen test tolerances which seem to fail on Mac.
Change: 115027725 | Vincent Vanhoucke | vanhoucke@google.com | 1,455,852,241,000 |
3c900a20f6866af515cf75e05c6f28039064cd78 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,455,855,873,000 |
1c91e5118162070c6cf9ebee2afb76d0df7dfef0 | TensorFlow: disable proto_test for now, since we don't know how to
support it in pip installed form.
Change: 115034582 | Vijay Vasudevan | vrv@google.com | 1,455,861,783,000 |
3ff0f0d17b4eec946a2874da0cf896776c01a80a | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,455,862,134,000 |
cd9389e26fb3d2219afc203b9705138af1c4bfc3 | update release version to 0.7.1 | Jan Prach | jendap@google.com | 1,455,871,197,000 |
97bde2c459bcb170b9d138f43a45a52cce80537c | Merge pull request #1190 from jendap/release-0.7.1
update release version to 0.7.1 | Vijay Vasudevan | vrv@google.com | 1,455,901,617,000 |
b44abb88bdc00cda5dcf8178cbc3896e776090ad | Fix dataset encoding in MNIST example for Python 3 | Danijar Hafner | mail@danijar.com | 1,455,802,066,000 |
431ad4be41718a84a1483fd1687f15e092e3108c | TensorFlow: switch dsoloader path for libcuda to end in .so.1
Change: 114990321 | Vijay Vasudevan | vrv@google.com | 1,455,827,123,000 |
f600f4d82de7feded78087b4edf7295eea64dae3 | TensorFlow: change cuda-diagnostics to search for so.1
Change: 115010103 | Vijay Vasudevan | vrv@google.com | 1,455,838,173,000 |
af88ce5e2c657749670471a13c9b9b29069bf11f | TensorFlow: make pip wheel depend on 3.0.0b2 protobuf,
since that's roughly what we package too.
Change: 115027171 | Vijay Vasudevan | vrv@google.com | 1,455,851,506,000 |
21bdc9ed001b98f4ec4820686d876c1e9cefd4ac | Merge pull request #1184 from vrv/r0.7. | Vijay Vasudevan | vrv@google.com | 1,455,901,704,000 |
716b7713e4c8b2d8f093f639ca41816cf4e1c696 | Fix tensorboard import path | Vesnica | hsdpsch@gmail.com | 1,455,784,611,000 |
fb1065e8ffe92d17eb50a4ce942136649b917ecd | Merge pull request #1164 from Vesnica/master. | Vijay Vasudevan | vrv@google.com | 1,455,904,985,000 |
4b3069401fde5ba8a6a01d1aab95cc31e2169626 | Fix tensorboard import path | Vesnica | hsdpsch@gmail.com | 1,455,784,611,000 |
76056aba2bff5a3f87f4ea4114cdcb34168d3d38 | Merge pull request #1200 from vrv/r0.7. | Vijay Vasudevan | vrv@google.com | 1,455,907,717,000 |
0dca2febe451dca7db3cdcb2160a2afb55353e16 | remove ld_library_path from docker image | Jan Prach | jendap@google.com | 1,455,869,197,000 |
2b27e881036f6ee150673dd91729960f94319e3a | Merge pull request #1188 from jendap/remove-ld_library_path-from-docker. | Vijay Vasudevan | vrv@google.com | 1,455,915,120,000 |
8be1a1f2f70e5ae08e14b812c7aa114db9f11c9c | + config addon | terrytangyuan | terrytangyuan@gmail.com | 1,455,926,808,000 |
b0240486be5a8c4286961d1751fe8560e9c6970e | Exposes the memory limit in the allocator's stats.
Change: 115036211 | A. Unique TensorFlower | nobody@tensorflow.org | 1,455,864,296,000 |
898d26a1cef46c0848188e22ba4b435846649b6d | Support NCHW in forward and backward convolution op.
Test both layouts in tests.
Change: 115096872 | Xiaoqiang Zheng | zhengxq@google.com | 1,455,919,074,000 |
7e8f71b23c7774a857c52b337b93ff7ae1de4221 | Add tag option to summarize_tensor.
Change: 115102931 | A. Unique TensorFlower | nobody@tensorflow.org | 1,455,922,788,000 |
875a67f3d298ab7f7438a3004e4597c71d92b529 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,455,927,269,000 |
45b5e60db248fdfb6f824e0c4a9959d72d9272d6 | + config addon added to estimators | terrytangyuan | terrytangyuan@gmail.com | 1,455,928,828,000 |
fa13ecb63a545d6ae1813bbe52e545ede2e4da9e | * fix in restore/saver | terrytangyuan | terrytangyuan@gmail.com | 1,455,931,701,000 |
4d998154a31f186fbf154a39372315b225ccb826 | * lint fixes | terrytangyuan | terrytangyuan@gmail.com | 1,455,931,977,000 |
9aa213772e4bff20f33a270e9547c4fe6e1bda23 | + allow reconfigure config_addon for restored model | terrytangyuan | terrytangyuan@gmail.com | 1,455,932,597,000 |
ae6f9b20f07427fca44cf2e13efb419c2fee0d5e | + example of custom GPU configurations | terrytangyuan | terrytangyuan@gmail.com | 1,455,933,089,000 |
376bb0e94ccf74936ea3362d1b9a162384a9bea8 | TensorFlow r0.7: change pip dependency for numpy to 1.10.1 on OS X
pip3 install numpy==1.8.2 results in failures to find Python.h
during our wheel building process, and it appears to be a bug
in numpy.distutils at 1.8.2. | Vijay Vasudevan | vrv@google.com | 1,455,934,113,000 |
aae445a02cd2bd26f695f6257159e101d0e2e3c6 | * allow reconfigurable_params to be reconfigurable | terrytangyuan | terrytangyuan@gmail.com | 1,455,934,396,000 |
712229d477021699618296134d22a8a8a5e129ea | Removing --upgrade and numpy-1.8.2 from pip.sh
Consolidating the two steps into one. | Shanqing Cai | cais@google.com | 1,455,942,659,000 |
d281609da7548cf05c4a79c56416e8ed9387b337 | Updated image recognition docs to point to new tutorial.
Change: 115111428 | Pete Warden | pete@petewarden.com | 1,455,928,120,000 |
d032feb7ff54289c4d1e6aaa1322f8a7429be151 | Fixing the backward compatibility test.
Change: 115111581 | Xiaoqiang Zheng | zhengxq@google.com | 1,455,928,241,000 |
232ff15e1d0e1e73c1a7e3e4e1d9cb08b2c9f172 | Adopt TensorFlow's standard variable regularization mechanism.
Support for variable regularization was recently added to tf.get_variable(),
and I've modified layers.py to take advantage of it. This CL changes the
semantics of tf.fully_connected() to match TensorFlow's standard conventions:
a regularization function will only be applied at the time that a variable is
first created.
Change: 115111632 | A. Unique TensorFlower | nobody@tensorflow.org | 1,455,928,270,000 |
28117ca16f60501feb1716f7c4b05cd3f20bee31 | Fixed a small mistake that prevents the gpu_bfc_allocator_test from compiling
Change: 115113975 | Benoit Steiner | benoit.steiner.goog@gmail.com | 1,455,930,037,000 |
fbf652193590d56ae556fbc1a3ea0ab5c7d5e154 | Flush denormals to zero on both CPU and GPU
Two different mechanisms are required. On the CPU, we push and pop the
appropriate processor flags in the executor (for the master thread) *and*
in each threadpool thread, since the processor flags are thread local. On
the GPU, we set -ftz=true for both nvcc and gcudacc so that kernels that we
build flush denormals to zero using instruction flags.
Caveat: On GPU, only single precision denormals are flushed to zero; double
precision is unchanged.
Change: 115114845 | Geoffrey Irving | geoffreyi@google.com | 1,455,930,609,000 |
6a98a74997bfd9d1855a7f7fbb5fd94ed8c30b51 | Add assert_scalar_int to contrib.
Change: 115121755 | A. Unique TensorFlower | nobody@tensorflow.org | 1,455,939,400,000 |
5ecbc0817bd97135b9f4cf8458cf8659bfb5d643 | Merge pull request #1211 from caisq/r0.7-dev2. | Vijay Vasudevan | vrv@google.com | 1,455,942,787,000 |
93f257bcb8639f27bbffab765ba9775508a320df | Using ln -sf in python_config.sh | Shanqing Cai | cais@google.com | 1,455,943,614,000 |
c9ad1dfebee7c0bd963b706ed8d8f014e8d2941b | Merge pull request #1212 from caisq/r0.7-dev2. | Vijay Vasudevan | vrv@google.com | 1,455,943,758,000 |
f43cea0c6b10873c893929a8a7ea2d5176da5183 | Revert "Fix tensorboard import path"
This reverts commit 4b3069401fde5ba8a6a01d1aab95cc31e2169626.
This cherry-pick was not necessary -- the change to tensorboard
paths was done after 0.7.0. | Vijay Vasudevan | vrv@google.com | 1,455,953,079,000 |
028d0b46004c921acd48fdd0ec18128d79e18bf4 | Merge pull request #1213 from vrv/r0.7
Revert "Fix tensorboard import path" | Vijay Vasudevan | vrv@google.com | 1,455,954,134,000 |
a147270d9f53af6e7142b4aa8b86284d9e1ad599 | * Fixed broken dask io test | terrytangyuan | terrytangyuan@gmail.com | 1,455,981,023,000 |
2b1712945630d42be42910168973739b5e78f66c | Update CHANGES.md | terrytangyuan | terrytangyuan@gmail.com | 1,455,981,945,000 |
d3352e7257dc9f95068a22933c5b0b6fdad79890 | Temporarily disable denormal test for open source
The test is breaking for OSS, and I can't look at it in detail right now. The
fact that it breaks is harmless (it's equivalent to yesterday), so disable the
test in the open source case for the moment.
Change: 115128265 | Geoffrey Irving | geoffreyi@google.com | 1,455,952,974,000 |
39b332ba07db732ff8e0a07459e74995b92aaa83 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,455,993,875,000 |
543e38cd14fd6d24d5df1421aa64f42744b88c94 | * restore the correct ConfigAddon | terrytangyuan | terrytangyuan@gmail.com | 1,456,004,287,000 |
0447a324ec1ee6df14300a31a2f6da77d1174103 | upgrade pip links for 0.7.1 | Jan Prach | jendap@google.com | 1,456,021,408,000 |
8cbc3dd738d950041b9df8fc821ee056536f746f | Merge pull request #1217 from jendap/update-pip-links-for-0.7.1
upgrade pip links for 0.7.1 | Vijay Vasudevan | vrv@google.com | 1,456,029,662,000 |
f0eb9ce3a7d0405674fe4e8b4571741a97dc1fba | add samples to docker and start jupyter by default | Jan Prach | jendap@google.com | 1,455,870,377,000 |
98623127830bb6a937edbb1ee0dd3bac2d0b331a | Merge pull request #1189 from jendap/improve-docker. | Vijay Vasudevan | vrv@google.com | 1,456,030,136,000 |
a163fe171a5f9f46f9012ef3e010a06ddc79d80f | Using ln -sf in python_config.sh | Shanqing Cai | cais@google.com | 1,455,943,614,000 |
e79290da197b2a5e63975d8f0044c50a9870efd7 | Fix a wrong link to random_crop() | Johnny Lim | izeye@naver.com | 1,455,938,912,000 |
ec3b8d3fe684e5c4f522fa017e5eeb352913f444 | Merge pull request #1210 from izeye/patch-4. | Martin Wicke | wicke@google.com | 1,456,033,024,000 |
e5ca7cce7de65e546310908d5d13f1773c621389 | Fix typos in comments. | Dongjoon Hyun | dongjoon@apache.org | 1,456,035,927,000 |
ccd255ad53b955432e57567591c00527e1e0b5b8 | Merge pull request #1209 from dongjoon-hyun/fix_typos. | Martin Wicke | wicke@google.com | 1,456,035,943,000 |
715fc5aa1a2e465ddcf0ae45dcb25d8e9fa09fef | Rename `testDoubleRandomTranposeBoth` to `testDoubleRandomTransposeBoth`.
This PR renames a test case in `matmul_op_test.py` to fix a typo. | Dongjoon Hyun | dongjoon@apache.org | 1,455,926,465,000 |
7e3b8b23835ab0ac55d390aed2349af6e05dbe3b | Merge pull request #1208 from dongjoon-hyun/rename_testcase. | Martin Wicke | wicke@google.com | 1,456,037,505,000 |
b8caa301299a3ca743c97bb9299386467b39044e | fix path to allow for restoring saved models | Scott Graham | git@gramhagen.com | 1,456,114,476,000 |
598eeeec928c1de1c089d4617774a6de9207818a | Merge pull request #108 from gramhagen/restore_fix
fix path to allow for restoring saved models | Illia Polosukhin | ilblackdragon@gmail.com | 1,456,116,984,000 |
e370bc2a188a978240de4632dd610ed3a384bff2 | Removing --upgrade and numpy-1.8.2 from pip.sh
Consolidating the two steps into one. | Shanqing Cai | cais@google.com | 1,455,942,659,000 |
8584086f11f5b741c4cb3dc167fd50c401c60bcb | Add community projects to the resources page
This supercedes some pull requests requesting to put these on the from page. We will only include what's included in the repo itself in the top-level readme, and add a pointer here instead.
Closes #995, closes #1223. | Martin Wicke | martin.wicke@gmail.com | 1,456,116,987,000 |
2b4004c7006fd1a06d1508bce121f95be7d33949 | Point to community resources from README | Martin Wicke | martin.wicke@gmail.com | 1,456,118,511,000 |
43f54865ba39acd021eae1aaf08f291a3a50206b | Merge pull request #1232 from tensorflow/community-resources. | Vijay Vasudevan | vrv@google.com | 1,456,163,609,000 |
a2ba0d3b89b4f51fa7d7ab091dd17235adbedfc3 | Clean up unused variables in a example | Johnny Lim | izeye@naver.com | 1,455,843,554,000 |
7c869aaf081c5e671cd74431c8a1838aa70e9393 | Merge pull request #1183 from izeye/polish-20160219. | Vijay Vasudevan | vrv@google.com | 1,456,163,677,000 |
8c861dd7095ef637a68e89dac4cad6d24f5de34c | Fix training of decoder embeddings.
Now gradients could propagate through embeddings of decoders when feed_previous=True. | manipopopo | shushukk123@gmail.com | 1,453,737,610,000 |
d599ad036f27ac9f18d1cec0cd3eae2587414f52 | Add a test unit.
Test the model behavior when feed_previous is True. | manipopopo | shushukk123@gmail.com | 1,454,866,473,000 |
d671f9153506e2b536beaabd17aaabb33e0543c8 | Merge pull request #882 from manipopopo/fix_train_embedding. | Vijay Vasudevan | vrv@google.com | 1,456,163,720,000 |
9eb7b0f137cb3dd546c735275da874ab4d2ac156 | Fixed bug in example: s/len(num_steps)/num_steps/g
num_steps is a value (an integer), not a list, so it was a typo to take `len(num_steps)`.
This fixes that typo. | Nathan Daly | NHDaly@gmail.com | 1,456,042,041,000 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.