New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Go bindings tensors #1
base: go_bindings
Are you sure you want to change the base?
Go bindings tensors #1
Commits on Apr 11, 2016
-
Update bower dependencies. Also force urls to lowercase. Change: 119584968
Configuration menu - View commit details
-
Copy full SHA for 3c59c1e - Browse repository at this point
Copy the full SHA 3c59c1eView commit details -
Redo the TensorBoard tutorial code (mnist_with_summaries.py).
Goals: - Have enough of each summary type that tag grouping is useful. (Wound up recording e.g. mean and stddev and min/max for each variable) - Use every summary type (adds images) - Write to multiple directories so there are several "runs" Change: 119585022
Configuration menu - View commit details
-
Copy full SHA for a77499c - Browse repository at this point
Copy the full SHA a77499cView commit details -
Configuration menu - View commit details
-
Copy full SHA for f6a1d34 - Browse repository at this point
Copy the full SHA f6a1d34View commit details -
Working on the MarkDown auto documentation
Alonso Vidales committedApr 11, 2016 Configuration menu - View commit details
-
Copy full SHA for 6f9779d - Browse repository at this point
Copy the full SHA 6f9779dView commit details
Commits on Apr 12, 2016
-
Alonso Vidales committed
Apr 12, 2016 Configuration menu - View commit details
-
Copy full SHA for 8fca2f0 - Browse repository at this point
Copy the full SHA 8fca2f0View commit details -
Updated the distributed runtime how-to to use the new API in `tf.trai…
…n.*`. Change: 119589456
Configuration menu - View commit details
-
Copy full SHA for 76d3808 - Browse repository at this point
Copy the full SHA 76d3808View commit details -
Enable test_benchmark.* files for Android build.
Change: 119591021
Configuration menu - View commit details
-
Copy full SHA for ea6cdc0 - Browse repository at this point
Copy the full SHA ea6cdc0View commit details -
Merge pull request tensorflow#1863 from martinwicke/branch_119585022
Merge internal changes
Configuration menu - View commit details
-
Copy full SHA for 0ebbb99 - Browse repository at this point
Copy the full SHA 0ebbb99View commit details -
Add SyncReplicasOptimizer test in dist_test
Usage example: ./remote_test.sh --num-workers 3 --sync-replicas Also changed: 1) In local and remote tests, let different workers contact separate GRPC sessions. 2) In local and remote tests, adding the capacity to specify the number of workers. Before it was hard-coded at 2. Usage example: ./remote_test.sh --num-workers 2 --sync-replicas 3) Using device setter in mnist_replica.py Change: 119599547
Configuration menu - View commit details
-
Copy full SHA for 7c86e9c - Browse repository at this point
Copy the full SHA 7c86e9cView commit details -
Configuration menu - View commit details
-
Copy full SHA for a4cee9a - Browse repository at this point
Copy the full SHA a4cee9aView commit details -
Merge pull request tensorflow#1865 from martinwicke/branch_119599547
Merge internal changes
Configuration menu - View commit details
-
Copy full SHA for 4c50340 - Browse repository at this point
Copy the full SHA 4c50340View commit details -
Moved mnist reading from input_data in tutorial to learn/datasets (te…
…nsorflow#1864) * Moved mnist reading from input_data in tutorial to learn/datasets
Configuration menu - View commit details
-
Copy full SHA for 41ae528 - Browse repository at this point
Copy the full SHA 41ae528View commit details -
Fix a typo in Mechanics 101 tutorial to match recent code. (tensorflo…
…w#1866) * Update TF Mechanics 101 Tutorial to match recent code In the section of Inputs and Placeholders, the IMAGE_PIXELS should be mnist.IMAGE_PIXELS
Configuration menu - View commit details
-
Copy full SHA for dabcdc7 - Browse repository at this point
Copy the full SHA dabcdc7View commit details -
Applying editorial changes to the distributed how-to.
Change: 119605636
Configuration menu - View commit details
-
Copy full SHA for 16b856e - Browse repository at this point
Copy the full SHA 16b856eView commit details -
Configuration menu - View commit details
-
Copy full SHA for 766e79c - Browse repository at this point
Copy the full SHA 766e79cView commit details -
Configuration menu - View commit details
-
Copy full SHA for b2f6d50 - Browse repository at this point
Copy the full SHA b2f6d50View commit details -
Configuration menu - View commit details
-
Copy full SHA for f72d0d7 - Browse repository at this point
Copy the full SHA f72d0d7View commit details -
Removing lazy loading of sdca_ops.so since it is no longer necessary.
Change: 119643238
Configuration menu - View commit details
-
Copy full SHA for ca9cfc9 - Browse repository at this point
Copy the full SHA ca9cfc9View commit details -
Working on the autogenerated documentation / Some minor fixes on C me…
…mory allocation
Alonso Vidales committedApr 12, 2016 Configuration menu - View commit details
-
Copy full SHA for adbd4c9 - Browse repository at this point
Copy the full SHA adbd4c9View commit details -
Merge pull request tensorflow#1878 from caisq/update-version-1
Adding server/Dockerfile.test to versioning roll
Configuration menu - View commit details
-
Copy full SHA for 48dee4d - Browse repository at this point
Copy the full SHA 48dee4dView commit details -
Cleaning up the bash scripts a bit.
Fred Bertsch committedApr 12, 2016 Configuration menu - View commit details
-
Copy full SHA for 974042e - Browse repository at this point
Copy the full SHA 974042eView commit details -
Removing lazy loading of sdca_ops.so since it is no longer necessary.
Change: 119648305
Configuration menu - View commit details
-
Copy full SHA for f5258ed - Browse repository at this point
Copy the full SHA f5258edView commit details -
test_installation.sh: head command compatibility on Mac
Change: 119649248
Configuration menu - View commit details
-
Copy full SHA for 0a9f572 - Browse repository at this point
Copy the full SHA 0a9f572View commit details -
Add unused placeholder for readahead file path support.
Change: 119650236
Configuration menu - View commit details
-
Copy full SHA for 1b059ae - Browse repository at this point
Copy the full SHA 1b059aeView commit details -
Modified supervisor event writing behavior to be safer.
* Set supervisor fields for event writing to None for non-chiefs, so that non-chiefs do not accidentally write summaries. * Added RuntimeErrors for public API endpoints to alert users when they try to use non-chief supervisors to write events. Change: 119654303
Configuration menu - View commit details
-
Copy full SHA for 9b20d6a - Browse repository at this point
Copy the full SHA 9b20d6aView commit details -
Removes high rank and large tensors from the test.
Configuration menu - View commit details
-
Copy full SHA for bf4b58c - Browse repository at this point
Copy the full SHA bf4b58cView commit details -
Update generated Python Op docs.
Change: 119657458
Configuration menu - View commit details
-
Copy full SHA for cbef061 - Browse repository at this point
Copy the full SHA cbef061View commit details -
Bumping version: 0.7.1 --> 0.8.0rc0 (tensorflow#1884)
version_test.py: patch can be alphanumeric (e.g., 0rc0)
Configuration menu - View commit details
-
Copy full SHA for 1d5b211 - Browse repository at this point
Copy the full SHA 1d5b211View commit details -
skflow: trivial fixup (tensorflow#1873)
* skflow: remove .orig files Files with suffix `.orig` were added mistakenly by commit 16454ee ("Fixed merge conflicts in skflow") and removed by commit 641cb25 ("Refactor skflow"). However, few files are not removed yet. This commit removes the remaining .orig files. Signed-off-by: SeongJae Park <sj38.park@gmail.com> * skflow: fix trivial typos Signed-off-by: SeongJae Park <sj38.park@gmail.com>
Configuration menu - View commit details
-
Copy full SHA for 7a6a047 - Browse repository at this point
Copy the full SHA 7a6a047View commit details -
More tests and minor fixes for tf/contrib/tensor_forest.
Change: 119666432
Configuration menu - View commit details
-
Copy full SHA for cc3267f - Browse repository at this point
Copy the full SHA cc3267fView commit details -
Configuration menu - View commit details
-
Copy full SHA for ed6b912 - Browse repository at this point
Copy the full SHA ed6b912View commit details -
Make diag_op_test faster (tensorflow#1886)
Removes high rank and large tensors from the test.
Configuration menu - View commit details
-
Copy full SHA for bc5abfb - Browse repository at this point
Copy the full SHA bc5abfbView commit details -
Change: 119671640
Configuration menu - View commit details
-
Copy full SHA for 353bdea - Browse repository at this point
Copy the full SHA 353bdeaView commit details -
Make diag_op_test faster (tensorflow#1889)
Removes high rank and large tensors from the test.
Configuration menu - View commit details
-
Copy full SHA for bce8c55 - Browse repository at this point
Copy the full SHA bce8c55View commit details -
Fix chat link at start of README (tensorflow#1885)
Original formatting for this link included the pipe character at the end, so that clicking it would lead to a 404. The more explicit URL formatting makes it so that clicking the link leads to the correct chat room.
Configuration menu - View commit details
-
Copy full SHA for f66a4c4 - Browse repository at this point
Copy the full SHA f66a4c4View commit details -
Fix bugs in the indexing operator with latest NumPy (tensorflow#1895)
* Made the indexing operator accept objects that can be converted with `int()`. Previously, the `_SliceHelper()` would test for `isinstance(..., int)`, which is too restrictive. * Replace `np.random.random_integers()` with `np.random.randint()`
Configuration menu - View commit details
-
Copy full SHA for 1e2a17a - Browse repository at this point
Copy the full SHA 1e2a17aView commit details -
Configuration menu - View commit details
-
Copy full SHA for 2177658 - Browse repository at this point
Copy the full SHA 2177658View commit details -
Merge pull request tensorflow#1896 from martinwicke/master
Fixes for release
Configuration menu - View commit details
-
Copy full SHA for 9536379 - Browse repository at this point
Copy the full SHA 9536379View commit details -
Extends functionality of Tensor::flat_inner_dims() and Tensor::flat_o…
…uter_dims() by adding a template param to specify the desired rank of the output. This is useful, e.g., in order to turn a tensor into a (rank-3) batch of matrices. Change: 119685357
Configuration menu - View commit details
-
Copy full SHA for 919d72a - Browse repository at this point
Copy the full SHA 919d72aView commit details -
Configuration menu - View commit details
-
Copy full SHA for a110a16 - Browse repository at this point
Copy the full SHA a110a16View commit details -
Merge pull request tensorflow#1897 from martinwicke/master
Make sparse_tensor_dense_matmul_op_test medium
Configuration menu - View commit details
-
Copy full SHA for 47623eb - Browse repository at this point
Copy the full SHA 47623ebView commit details -
Configuration menu - View commit details
-
Copy full SHA for c9b6c5e - Browse repository at this point
Copy the full SHA c9b6c5eView commit details -
Added a unit test to batch normalization.
Change: 119691101
Configuration menu - View commit details
-
Copy full SHA for 9e8541b - Browse repository at this point
Copy the full SHA 9e8541bView commit details -
Remove lib/core/command_line_flags.h, which is superceded by util/com…
…mand_line_flags.h Change: 119694323
Configuration menu - View commit details
-
Copy full SHA for 8217a41 - Browse repository at this point
Copy the full SHA 8217a41View commit details
Commits on Apr 13, 2016
-
Avoid using functions from the std namespace on GPU.
Change: 119695799
Configuration menu - View commit details
-
Copy full SHA for c052df4 - Browse repository at this point
Copy the full SHA c052df4View commit details -
Adds option to pass callable initializer function to Variable constru…
…ctor to allow colocation of variable initialization with the device the variable is on, instead of always being on the chief supervisor. Also updates variable_scope.get_variable() and create_partitioned_variables() to take advantage of this when an initializer fn is passed in. Change: 119697860
Configuration menu - View commit details
-
Copy full SHA for 8b34de6 - Browse repository at this point
Copy the full SHA 8b34de6View commit details -
Update generated Python Op docs.
Change: 119698585
Configuration menu - View commit details
-
Copy full SHA for bbf32fc - Browse repository at this point
Copy the full SHA bbf32fcView commit details -
Configuration menu - View commit details
-
Copy full SHA for ec0552f - Browse repository at this point
Copy the full SHA ec0552fView commit details -
add reference for CBOW and skip-gram (tensorflow#1876)
add reference for CBOW and skip-gram in word-vec tutorial
Configuration menu - View commit details
-
Copy full SHA for 04cf898 - Browse repository at this point
Copy the full SHA 04cf898View commit details -
Fix typo in Mechanics 101 tutorial to match recent code. (tensorflow#…
…1867) * Fix Typo in TF Mechanics 101 Tutorial In the section of Inputs and Placeholders, the IMAGE_PIXELS should be mnist.IMAGE_PIXELS * Update TF Mechanics 101 Tutorial to match recent code * Update TF Mechanics 101 Tutorial to match recent code
Configuration menu - View commit details
-
Copy full SHA for 351b4b3 - Browse repository at this point
Copy the full SHA 351b4b3View commit details -
make the formula to be centralized (tensorflow#1880)
remove the period after the formula to make the formula centralized, so it can be consistent with other formulas
Configuration menu - View commit details
-
Copy full SHA for 3809ef3 - Browse repository at this point
Copy the full SHA 3809ef3View commit details -
Configuration menu - View commit details
-
Copy full SHA for 1b06f75 - Browse repository at this point
Copy the full SHA 1b06f75View commit details -
Standalone benchmark for Tensorflow models that runs on desktop and A…
…ndroid. Change: 119706066
Configuration menu - View commit details
-
Copy full SHA for dc62e16 - Browse repository at this point
Copy the full SHA dc62e16View commit details -
Configuration menu - View commit details
-
Copy full SHA for a65c94b - Browse repository at this point
Copy the full SHA a65c94bView commit details -
Change: 119712558
Configuration menu - View commit details
-
Copy full SHA for eddacad - Browse repository at this point
Copy the full SHA eddacadView commit details -
Configuration menu - View commit details
-
Copy full SHA for b4d76bb - Browse repository at this point
Copy the full SHA b4d76bbView commit details -
Configuration menu - View commit details
-
Copy full SHA for 84fcb39 - Browse repository at this point
Copy the full SHA 84fcb39View commit details -
Configuration menu - View commit details
-
Copy full SHA for e265e46 - Browse repository at this point
Copy the full SHA e265e46View commit details -
Configuration menu - View commit details
-
Copy full SHA for 51d8187 - Browse repository at this point
Copy the full SHA 51d8187View commit details -
Merge pull request tensorflow#1902 from martinwicke/branch_119712558
Merge internal changes
Configuration menu - View commit details
-
Copy full SHA for c84742c - Browse repository at this point
Copy the full SHA c84742cView commit details -
Configuration menu - View commit details
-
Copy full SHA for cb8ebef - Browse repository at this point
Copy the full SHA cb8ebefView commit details -
Merge pull request tensorflow#1903 from jendap/more-0.8-versions
update more places with version to 0.8
Configuration menu - View commit details
-
Copy full SHA for f819d48 - Browse repository at this point
Copy the full SHA f819d48View commit details -
Configuration menu - View commit details
-
Copy full SHA for 8dc5ede - Browse repository at this point
Copy the full SHA 8dc5edeView commit details -
Configuration menu - View commit details
-
Copy full SHA for 9ff62fc - Browse repository at this point
Copy the full SHA 9ff62fcView commit details -
Configuration menu - View commit details
-
Copy full SHA for bc5e961 - Browse repository at this point
Copy the full SHA bc5e961View commit details -
Configuration menu - View commit details
-
Copy full SHA for 31ea3db - Browse repository at this point
Copy the full SHA 31ea3dbView commit details -
Removing the automated flag processing from the FFmpeg test and repla…
…cing with simpler code.
Fred Bertsch committedApr 13, 2016 Configuration menu - View commit details
-
Copy full SHA for 6360a67 - Browse repository at this point
Copy the full SHA 6360a67View commit details -
Fix the inferred shape of SparseTensor after a tf.train.batch dequeue.
Change: 119755164
Configuration menu - View commit details
-
Copy full SHA for 81e345d - Browse repository at this point
Copy the full SHA 81e345dView commit details -
Fixing more silent 64->32 conversions with explicit warnings (and/or
handling 64). Change: 119759411
Configuration menu - View commit details
-
Copy full SHA for 227e7db - Browse repository at this point
Copy the full SHA 227e7dbView commit details -
Merge pull request tensorflow#1921 from tensorflow/r0.8
Merge r0.8rc0 changes back to master.
Configuration menu - View commit details
-
Copy full SHA for d623811 - Browse repository at this point
Copy the full SHA d623811View commit details -
Improvements on the doucmentation, some fixes and some examples added
Alonso Vidales committedApr 13, 2016 Configuration menu - View commit details
-
Copy full SHA for 9a2ecbb - Browse repository at this point
Copy the full SHA 9a2ecbbView commit details -
Minimal open source CUPTI GPU Tracer.
Change: 119768540
Configuration menu - View commit details
-
Copy full SHA for 3ed2517 - Browse repository at this point
Copy the full SHA 3ed2517View commit details -
Upgraded to the latest version of Eigen that provides significant per…
…formance improvements for fp16 Change: 119771118
Configuration menu - View commit details
-
Copy full SHA for de0618b - Browse repository at this point
Copy the full SHA de0618bView commit details -
Configuration menu - View commit details
-
Copy full SHA for fa4bd17 - Browse repository at this point
Copy the full SHA fa4bd17View commit details -
fixing third_party -> GitHub conversions.
Fred Bertsch committedApr 13, 2016 Configuration menu - View commit details
-
Copy full SHA for 70e9015 - Browse repository at this point
Copy the full SHA 70e9015View commit details -
Fred Bertsch committed
Apr 13, 2016 Configuration menu - View commit details
-
Copy full SHA for 4b6b6c3 - Browse repository at this point
Copy the full SHA 4b6b6c3View commit details -
Fix deserialization of SparseTensors of unknown rank.
Change: 119777702
Configuration menu - View commit details
-
Copy full SHA for c9eb57f - Browse repository at this point
Copy the full SHA c9eb57fView commit details -
Remove RNNCell's input_size property.
Change: 119777914
Configuration menu - View commit details
-
Copy full SHA for 871ee01 - Browse repository at this point
Copy the full SHA 871ee01View commit details -
Adding GoogleInit code back. It was needed.
Fred Bertsch committedApr 13, 2016 Configuration menu - View commit details
-
Copy full SHA for 9d99c6c - Browse repository at this point
Copy the full SHA 9d99c6cView commit details -
Removed unnecessary link options when build shared libraries for cust…
…om ops. Change: 119781627
Configuration menu - View commit details
-
Copy full SHA for 0afe5b3 - Browse repository at this point
Copy the full SHA 0afe5b3View commit details -
edits to rnn_cell.py to support time-frequency lstms for tensor flow
Change: 119785022
Configuration menu - View commit details
-
Copy full SHA for 021f885 - Browse repository at this point
Copy the full SHA 021f885View commit details -
Make TF_CALL_REAL_NUMBER_TYPES() and related macros include Eigen::half
in addition to float. Explicitly exempt GPU effects that rely on atomics, for which we have no good solution for half yet. Note that there are still ops that don't _declare_ half support (those that use \u201callnumbertypes\u201d or similar do, those that use \u201cfloat, double\u201d don't); these will be fixed in a forthcoming commit. Change: 119786433
Configuration menu - View commit details
-
Copy full SHA for 53b9ba7 - Browse repository at this point
Copy the full SHA 53b9ba7View commit details -
Correcting the name of the test file.
Fred Bertsch committedApr 13, 2016 Configuration menu - View commit details
-
Copy full SHA for 6c6a974 - Browse repository at this point
Copy the full SHA 6c6a974View commit details -
Configuration menu - View commit details
-
Copy full SHA for 6ec1123 - Browse repository at this point
Copy the full SHA 6ec1123View commit details -
Optimize SparseTensorDenseMatMul for the adjoint_b == True case.
Previously, there were two issues with this case: 1) columns of B are accessed (chip<1>) when B is RowMajor. 2) conjugation is performed repeatedly on B. This change fixes these by performing both prior to the inner loop. The resulting improvement is ~60%. Change: 119796162
Configuration menu - View commit details
-
Copy full SHA for 891356b - Browse repository at this point
Copy the full SHA 891356bView commit details -
Merge pull request tensorflow#1930 from kchen92/master
Fix typo in batch normalization documentation
Configuration menu - View commit details
-
Copy full SHA for cf1659d - Browse repository at this point
Copy the full SHA cf1659dView commit details
Commits on Apr 14, 2016
-
Remove some unnecessary files in the tree.
Change: 119802054
Configuration menu - View commit details
-
Copy full SHA for 278c58d - Browse repository at this point
Copy the full SHA 278c58dView commit details -
Adds a simple fast-path to tf.accumulate_n for length 1 arguments.
Change: 119806346
Configuration menu - View commit details
-
Copy full SHA for f6ebed4 - Browse repository at this point
Copy the full SHA f6ebed4View commit details -
Add padding between columns on the image-dashboard.
Also, move the border to the image, rather than the cell, so empty cells don't have a border. This is cleaner, and also visually distinguishes "no image" from "perfectly grey image". Also, slightly increase the padding (5px -> 10px) Change: 119807808
Configuration menu - View commit details
-
Copy full SHA for 1fe24cc - Browse repository at this point
Copy the full SHA 1fe24ccView commit details -
Enable Prod to work on int32 types. Fixes tensorflow#1919.
Change: 119811568
Configuration menu - View commit details
-
Copy full SHA for b874e2c - Browse repository at this point
Copy the full SHA b874e2cView commit details -
Removing lazy loading of sdca_ops.so since it is no longer necessary.
Change: 119815437
Configuration menu - View commit details
-
Copy full SHA for c25a3fa - Browse repository at this point
Copy the full SHA c25a3faView commit details -
Minor: move CPU-only sparse_add_op_test out of a tf_cuda_cc_tests group.
Change: 119817804
Configuration menu - View commit details
-
Copy full SHA for 2c6a61f - Browse repository at this point
Copy the full SHA 2c6a61fView commit details -
Adding rx.y pattern to update_version.sh
Strings fitting this pattern will be updated if there are major.minor changes.
Configuration menu - View commit details
-
Copy full SHA for 0b95b77 - Browse repository at this point
Copy the full SHA 0b95b77View commit details -
Configuration menu - View commit details
-
Copy full SHA for 4543b2c - Browse repository at this point
Copy the full SHA 4543b2cView commit details -
Added support for trigonometric and transcendental functions of half …
…floats Change: 119850987
Configuration menu - View commit details
-
Copy full SHA for f497498 - Browse repository at this point
Copy the full SHA f497498View commit details -
Fixing some warning messages from presubmit checks.
Fred Bertsch committedApr 14, 2016 Configuration menu - View commit details
-
Copy full SHA for e839d5e - Browse repository at this point
Copy the full SHA e839d5eView commit details -
Adds op for copying a diagonal band of a matrix or batch of matrices.…
… Special cases include the functionality of numpy.tril and numpy.triu. Change: 119857063
Configuration menu - View commit details
-
Copy full SHA for 8aaed92 - Browse repository at this point
Copy the full SHA 8aaed92View commit details -
Merge pull request tensorflow#1665 from phvu/enhancement/grid-rnn
Added implementation for GridRNN
Configuration menu - View commit details
-
Copy full SHA for 97fb7dd - Browse repository at this point
Copy the full SHA 97fb7ddView commit details -
Update ops-related pbtxt files.
Change: 119859479
Configuration menu - View commit details
-
Copy full SHA for 29aeb9b - Browse repository at this point
Copy the full SHA 29aeb9bView commit details -
Update generated Python Op docs.
Change: 119859588
Configuration menu - View commit details
-
Copy full SHA for a986313 - Browse repository at this point
Copy the full SHA a986313View commit details -
Duplicated code (tensorflow#1938)
check two times.
Configuration menu - View commit details
-
Copy full SHA for 04292b6 - Browse repository at this point
Copy the full SHA 04292b6View commit details -
image_retraining: fix trivial typos (tensorflow#1908)
This commit fix trivial typos under image_retraining/ directory. Signed-off-by: SeongJae Park <sj38.park@gmail.com>
Configuration menu - View commit details
-
Copy full SHA for d4bf5e0 - Browse repository at this point
Copy the full SHA d4bf5e0View commit details -
tv.Variable doc: fix typo (tensorflow#1907)
'b' does not exist in the example, probably 'y' was meant.
Configuration menu - View commit details
-
Copy full SHA for bf1f8ee - Browse repository at this point
Copy the full SHA bf1f8eeView commit details -
add a reference on log-likelihood (tensorflow#1906)
add a reference on log-likelihood in word2vec tutorial
Configuration menu - View commit details
-
Copy full SHA for 7a23f2c - Browse repository at this point
Copy the full SHA 7a23f2cView commit details -
add a reference for logistic regression (tensorflow#1912)
add a reference for logistic regression in word2vec tutorial
Configuration menu - View commit details
-
Copy full SHA for b37c120 - Browse repository at this point
Copy the full SHA b37c120View commit details -
add a reference for Penn Tree Bank (tensorflow#1944)
add a reference for Penn Tree Bank dataset in RNN tutorial
Configuration menu - View commit details
-
Copy full SHA for 74a44a3 - Browse repository at this point
Copy the full SHA 74a44a3View commit details -
Fix ASAN breakage due to delete/delete[] confusiong in PRun SWIG wrap…
…per. Change: 119861147
Configuration menu - View commit details
-
Copy full SHA for ce9bddf - Browse repository at this point
Copy the full SHA ce9bddfView commit details -
Fix tf.train.start_queue_runners to use the passed session's graph, n…
…ot the default graph. Change: 119864421
Configuration menu - View commit details
-
Copy full SHA for 5d2075f - Browse repository at this point
Copy the full SHA 5d2075fView commit details -
Merge pull request tensorflow#1849 from fredbertsch/fix2
Creating a FFmpeg library along with a test …
Configuration menu - View commit details
-
Copy full SHA for bc75c6f - Browse repository at this point
Copy the full SHA bc75c6fView commit details -
Add explicit colocation between ref producer and ref consumer in Pyth…
…on client. This prevents unsatisfiable placements resulting from code of the form: ```python with tf.device("/job:ps"): v = tf.Variable(0) with tf.device("/job:worker"): # ... # Should have device "/job:ps" but would currently get "/job:worker" assign_op = v.assign(1) ``` Change: 119871989
Configuration menu - View commit details
-
Copy full SHA for f34cf72 - Browse repository at this point
Copy the full SHA f34cf72View commit details -
Update generated Python Op docs.
Change: 119873463
Configuration menu - View commit details
-
Copy full SHA for 0d088fb - Browse repository at this point
Copy the full SHA 0d088fbView commit details -
Extend singleton case in _AggregatedGrads to IndexedSlices
If we are aggregating one gradient, we can just pass it through. Previously the code did this only for tf.Tensor, not tf.IndexedSlices. Change: 119880763
Configuration menu - View commit details
-
Copy full SHA for a58e5ca - Browse repository at this point
Copy the full SHA a58e5caView commit details -
Add context manager to supervisor to catch errors properly and close …
…the session after the managed code terminates. In the coordinator, ignore OutOfRangeError exceptions as they indicate a queue being closed. Change: 119881042
Configuration menu - View commit details
-
Copy full SHA for bbce152 - Browse repository at this point
Copy the full SHA bbce152View commit details -
Update generated Python Op docs.
Change: 119882490
Configuration menu - View commit details
-
Copy full SHA for 586528c - Browse repository at this point
Copy the full SHA 586528cView commit details -
Change: 119883459
Configuration menu - View commit details
-
Copy full SHA for 0328c1a - Browse repository at this point
Copy the full SHA 0328c1aView commit details -
Pass through the validate_shape option.
Change: 119883597
Configuration menu - View commit details
-
Copy full SHA for 838d0aa - Browse repository at this point
Copy the full SHA 838d0aaView commit details -
Some fixes from @dave-andersen feedback
Alonso Vidales committedApr 14, 2016 Configuration menu - View commit details
-
Copy full SHA for 1325a70 - Browse repository at this point
Copy the full SHA 1325a70View commit details -
Update generated Python Op docs.
Change: 119886270
Configuration menu - View commit details
-
Copy full SHA for f6331d6 - Browse repository at this point
Copy the full SHA f6331d6View commit details -
Add a reload button in TensorBoard.
Clicking it will reload all the data that is visible on the page. It is disabled for the graph dashboard. Also: make reload_interval a configurable flag on tensorboard. Also: Unit tests :) Change: 119889217
Configuration menu - View commit details
-
Copy full SHA for d66219f - Browse repository at this point
Copy the full SHA d66219fView commit details -
Some fixes from @dave-andersen feedback
Alonso Vidales committedApr 14, 2016 Configuration menu - View commit details
-
Copy full SHA for 7cc978b - Browse repository at this point
Copy the full SHA 7cc978bView commit details -
Add ops/check_ops.py, which contains common asserts and is_* (a boole…
…an check). Change: 119894309
Configuration menu - View commit details
-
Copy full SHA for c721550 - Browse repository at this point
Copy the full SHA c721550View commit details -
Update generated Python Op docs.
Change: 119896017
Configuration menu - View commit details
-
Copy full SHA for e735c79 - Browse repository at this point
Copy the full SHA e735c79View commit details -
Alonso Vidales committed
Apr 14, 2016 Configuration menu - View commit details
-
Copy full SHA for 0b2e8b8 - Browse repository at this point
Copy the full SHA 0b2e8b8View commit details -
Add shape and order asserts in SparseTensor constructor.
Add const to SparseTensor.group. Overload == on GroupIterable.IteratorStep. Change: 119897767
Configuration menu - View commit details
-
Copy full SHA for fd535e0 - Browse repository at this point
Copy the full SHA fd535e0View commit details
Commits on Apr 15, 2016
-
Expand the timeline API to support returning maximum memory usage info.
Also lists the tensors allocated at maximum memory. Change: 119906366
Configuration menu - View commit details
-
Copy full SHA for 48b6f54 - Browse repository at this point
Copy the full SHA 48b6f54View commit details -
Upgraded to the latest version of Eigen that adds support for computi…
…ng the sigmoid of fp16 and introduces a condition estimator. Change: 119907721
Configuration menu - View commit details
-
Copy full SHA for 4f744c2 - Browse repository at this point
Copy the full SHA 4f744c2View commit details -
Allow default_name being None when name_or_scope is provided.
Change: 119914246
Configuration menu - View commit details
-
Copy full SHA for 8ca75a9 - Browse repository at this point
Copy the full SHA 8ca75a9View commit details -
Update generated Python Op docs.
Change: 119914627
Configuration menu - View commit details
-
Copy full SHA for 104fe28 - Browse repository at this point
Copy the full SHA 104fe28View commit details -
Configuration menu - View commit details
-
Copy full SHA for 555c7f7 - Browse repository at this point
Copy the full SHA 555c7f7View commit details -
add a reference for
building from source
andbazel
(tensorflow#1958)add a reference for `building from source` and `bazel` in RNN tutorial
Configuration menu - View commit details
-
Copy full SHA for cadacc2 - Browse repository at this point
Copy the full SHA cadacc2View commit details -
add a reference for dropout in mnist tutorial (tensorflow#1957)
add a reference for dropout in mnist tutorial
Configuration menu - View commit details
-
Copy full SHA for 707a0fa - Browse repository at this point
Copy the full SHA 707a0faView commit details -
deepdream: fix trivial typos (tensorflow#1954)
This commit fixes trivial typos under deepdream/ directory. Signed-off-by: SeongJae Park <sj38.park@gmail.com>
Configuration menu - View commit details
-
Copy full SHA for 1f4aa4c - Browse repository at this point
Copy the full SHA 1f4aa4cView commit details -
Update the documentation of sparse_softmax_cross_entropy_with_logits() (
tensorflow#1949) With @fd839dc, sparse_softmax_cross_entropy_with_logits() was extended to allow the type of `labels` to be int32. Refs tensorflow#1250
Configuration menu - View commit details
-
Copy full SHA for dc7ee6c - Browse repository at this point
Copy the full SHA dc7ee6cView commit details -
Add tools/proto_text for generating ProtoDebugString,
ProtoShortDebugString, and ProtoParseFromString methods from protos. This will allow changing code used on mobile to use the proto LITE_RUNTIME, to reduce code size. This change is only for the tool itself. A future change will add a better genrule and use it the generated code in tensorflow. Change: 119919087
Configuration menu - View commit details
-
Copy full SHA for df15baa - Browse repository at this point
Copy the full SHA df15baaView commit details -
Configuration menu - View commit details
-
Copy full SHA for 396b584 - Browse repository at this point
Copy the full SHA 396b584View commit details -
Word2vec basic show id and word together (tensorflow#1877)
* Added words to clearly show word ids and corresponding words together. Now, they clearly show what's going on in the sample data and prediction pairs. [New output]: clearly shows the word ids and corresponding words Sample data [5239, 3084, 12, 6, 195, 2, 3137, 46, 59, 156] ['anarchism', 'originated', 'as', 'a', 'term', 'of', 'abuse', 'first', 'used', 'against'] 3084 originated -> 12 as 3084 originated -> 5239 anarchism 12 as -> 6 a 12 as -> 3084 originated 6 a -> 12 as 6 a -> 195 term 195 term -> 6 a 195 term -> 2 of [Old output]: No words for Sample data. Word ids and words are mixed, so it's very hard to read Sample data [5239, 3084, 12, 6, 195, 2, 3137, 46, 59, 156] ['anarchism', 'originated', 'as', 'a', 'term', 'of', 'abuse', 'first', 'used', 'against'] 3084 -> 5239 originated -> anarchism 3084 -> 12 originated -> as 12 -> 3084 as -> originated 12 -> 6 as -> a 6 -> 195 a -> term 6 -> 12 a -> as 195 -> 6 term -> a 195 -> 2 term -> of * changed it to two space
Configuration menu - View commit details
-
Copy full SHA for 4d9ebec - Browse repository at this point
Copy the full SHA 4d9ebecView commit details -
Alonso Vidales committed
Apr 15, 2016 Configuration menu - View commit details
-
Copy full SHA for a6e723f - Browse repository at this point
Copy the full SHA a6e723fView commit details -
Removed references to alonsovidales GitHub account
Alonso Vidales committedApr 15, 2016 Configuration menu - View commit details
-
Copy full SHA for ef80598 - Browse repository at this point
Copy the full SHA ef80598View commit details -
Removed references to alonsovidales GitHub account
Alonso Vidales committedApr 15, 2016 Configuration menu - View commit details
-
Copy full SHA for 210180b - Browse repository at this point
Copy the full SHA 210180bView commit details -
fp16-enable all the componentwise ops.
This also includes updating parts of the Python test framework to handle fp16. fp16 is too inaccurate to do numerical gradients unless a lot of care is taken, so for fp16, we compare fp16 theoretical gradients to fp32 numerical ones. This means that the gradient check doesn't also implicitly test the function itself, so we will need to rely on the numpy tests for those. Change: 119948035
Configuration menu - View commit details
-
Copy full SHA for 31fd486 - Browse repository at this point
Copy the full SHA 31fd486View commit details -
Update ops-related pbtxt files.
Change: 119948752
Configuration menu - View commit details
-
Copy full SHA for 8d8baf7 - Browse repository at this point
Copy the full SHA 8d8baf7View commit details -
Update generated Python Op docs.
Change: 119948869
Configuration menu - View commit details
-
Copy full SHA for 7c877f1 - Browse repository at this point
Copy the full SHA 7c877f1View commit details -
Fix info card edge list not updating in certain case.
If the names in the input or output edge list in the info card doesn't change when switching between selected nodes, then the edge shapes are not updated, even if they are different. This was due to the items list used for the iron-list not changing in this case. Resolved by having the items list contain all information about the edges, not just the names. Change: 119949006
Configuration menu - View commit details
-
Copy full SHA for 143d19d - Browse repository at this point
Copy the full SHA 143d19dView commit details -
Fixed bug in
tf.train.ClusterSpec
constructor.Creating a `tf.train.ClusterSpec` from another ClusterSpec was broken, which in turn broke creating a `tf.train.Server` from a ClusterSpec. Fixes tensorflow#1961. Change: 119954117
Configuration menu - View commit details
-
Copy full SHA for ddb59d7 - Browse repository at this point
Copy the full SHA ddb59d7View commit details -
Fix "naming" anchor in "Adding an Op" How-To (tensorflow#1858)
The file had a braces-style declared anchor, {#naming}, which wasn't being rendered or pre-processed properly. This changes it to an HTML <a> tag. This should also fix the broken link pointing to this section on line 874 in the same file.
Configuration menu - View commit details
-
Copy full SHA for 125ee65 - Browse repository at this point
Copy the full SHA 125ee65View commit details -
Implements gradients for {batch_}matrix_triangular_solve, as well as …
…adjoint cases for {batch_}matrix_solve. Enables float32 tests for all linalg gradients, except for {batch_}matrix_triangular_solve, which is particularly sensitive to roundoff. Change: 119961156
Configuration menu - View commit details
-
Copy full SHA for 5ac5da3 - Browse repository at this point
Copy the full SHA 5ac5da3View commit details -
Renamed parameters in GPU ApplyGradientDescent for parity with the CP…
…U code. Change: 119962541
Configuration menu - View commit details
-
Copy full SHA for af91b35 - Browse repository at this point
Copy the full SHA af91b35View commit details -
Make it possible to add attributes to a graph Node. This will enable
graph optimization passes to add state to the graph for future passes to consume. Change: 119968551
Configuration menu - View commit details
-
Copy full SHA for 5e0b796 - Browse repository at this point
Copy the full SHA 5e0b796View commit details -
Update word2vec.py (tensorflow#1851)
Makes it consistent with line 515.
Configuration menu - View commit details
-
Copy full SHA for 636aaa6 - Browse repository at this point
Copy the full SHA 636aaa6View commit details -
add a wiki reference for Mandelbrot set (tensorflow#1963)
add a wiki reference for Mandelbrot set
Configuration menu - View commit details
-
Copy full SHA for eb3a9ed - Browse repository at this point
Copy the full SHA eb3a9edView commit details -
add a wiki reference for partial differential equation (tensorflow#1964)
add a wiki reference for partial differential equation
Configuration menu - View commit details
-
Copy full SHA for 6f90ede - Browse repository at this point
Copy the full SHA 6f90edeView commit details -
Change: 119971913
Configuration menu - View commit details
-
Copy full SHA for 0784fda - Browse repository at this point
Copy the full SHA 0784fdaView commit details -
Change: 119973341
Configuration menu - View commit details
-
Copy full SHA for 493eb98 - Browse repository at this point
Copy the full SHA 493eb98View commit details -
Give more flexibility for summary writers in Supervisor:
- Allow to pass your own SummaryWriter to the Supervisor. - Allow to pass None as the SummaryWriter, in which case no summaries are written. These options are useful for example in eval loops where you want to keep using the same SummaryWriter across multiple uses of the Supervisor. Change: 119977091
Configuration menu - View commit details
-
Copy full SHA for 1881244 - Browse repository at this point
Copy the full SHA 1881244View commit details -
Update generated Python Op docs.
Change: 119978765
Configuration menu - View commit details
-
Copy full SHA for db81c2b - Browse repository at this point
Copy the full SHA db81c2bView commit details -
Fix bug in
tf.zeros()
when specifying a scalar shape as an empty tu……ple. This correctly converts inputs to `tf.int32`. Specifying a tuple is now handled using the same special case as specifying a list, so that the constant can be pre-allocated. Change: 119985289
Configuration menu - View commit details
-
Copy full SHA for fc3c099 - Browse repository at this point
Copy the full SHA fc3c099View commit details -
Add paper-dialog as a dependency for tf-tensorboard.
Also update typings in package.json and add typing dependency on sinon (its provided by wct) Change: 119988943
Configuration menu - View commit details
-
Copy full SHA for ce6a995 - Browse repository at this point
Copy the full SHA ce6a995View commit details -
Configuration menu - View commit details
-
Copy full SHA for ef7f54f - Browse repository at this point
Copy the full SHA ef7f54fView commit details -
Adds adjoint attribute to matrix inverse to make it possible to compu…
…te the inverse of the adjoint of a matrix. This is convenient in various gradient computations where it also saves an explicit transpose op. Implements gradients for batch_matrix_determinant and optimizes implementation of gradient for matrix_determinant. Change: 119990836
Configuration menu - View commit details
-
Copy full SHA for 4cc8426 - Browse repository at this point
Copy the full SHA 4cc8426View commit details -
tensorflow: add mutex.try_lock
try_lock is used by eigen3/unsupported/Eigen/CXX11/src/Tensor/TensorRunQueue.h which is disabled by default, so it left unnoticed. TESTED: - passed opensource_build Change: 119991000
Configuration menu - View commit details
-
Copy full SHA for ed30375 - Browse repository at this point
Copy the full SHA ed30375View commit details -
Add autoreload logic to TensorBoard, but do not wire it in.
The autoreload logic is factored out into a behavior, and is tested. Change: 119991467
Configuration menu - View commit details
-
Copy full SHA for 1c06948 - Browse repository at this point
Copy the full SHA 1c06948View commit details -
Update ops-related pbtxt files.
Change: 119992758
Configuration menu - View commit details
-
Copy full SHA for 11be20e - Browse repository at this point
Copy the full SHA 11be20eView commit details -
Update generated Python Op docs.
Change: 119992878
Configuration menu - View commit details
-
Copy full SHA for e13205f - Browse repository at this point
Copy the full SHA e13205fView commit details -
Merge pull request tensorflow#1971 from zffchen78/branch_119985289
Merging changes from internal.
Vijay Vasudevan committedApr 15, 2016 Configuration menu - View commit details
-
Copy full SHA for e22ee35 - Browse repository at this point
Copy the full SHA e22ee35View commit details -
Add documentation to variables tutorial to point to inspect_checkpoint.
Change: 119994231
Configuration menu - View commit details
-
Copy full SHA for 50889d5 - Browse repository at this point
Copy the full SHA 50889d5View commit details -
Configuration menu - View commit details
-
Copy full SHA for 9eafa89 - Browse repository at this point
Copy the full SHA 9eafa89View commit details -
Add TensorBoard auto-reload behavior.
Auto-reloading defaults on and is controlled by a setting. To have a place to put that setting, also add a settings panel. The settings panel uses paper-dialog and is opened by clicking a settings icon button in the global top right. Minor changes to how the toolbar is laid out by CSS to ensure that the settings all show up in a line and don't bunch up together. Also, the "TensorBoard" message on the left no longer takes a fixed pixel width (everything is flex), so the toolbar will use space more appropriately when the window size is small. Change: 119997005
Configuration menu - View commit details
-
Copy full SHA for 43f1c26 - Browse repository at this point
Copy the full SHA 43f1c26View commit details -
The behavior of outer_dimensions was different than before when the s…
…elected indices had a 0. Making it emulate the old behavior because gather stopped working for length 0 Tensors. e.g.: Gathering row 1 from a Tensor of shape [2, 0] should result in a [1, 0] Tensor, but instead resulted in an error like: AttributeError: index 1 is not in [0, 2) Change: 119998790
Configuration menu - View commit details
-
Copy full SHA for 5a61179 - Browse repository at this point
Copy the full SHA 5a61179View commit details -
Ignore variables without gradients. When plotting gradient histograms.
Change: 119998845
Configuration menu - View commit details
-
Copy full SHA for 9f37988 - Browse repository at this point
Copy the full SHA 9f37988View commit details -
Fix crash in BFC allocator when failing to allocate memory from an al…
…locator that has zero memory in its heap. Change: 120002729
Configuration menu - View commit details
-
Copy full SHA for ea632fc - Browse repository at this point
Copy the full SHA ea632fcView commit details -
Configuration menu - View commit details
-
Copy full SHA for 5ba46a4 - Browse repository at this point
Copy the full SHA 5ba46a4View commit details
Commits on Apr 16, 2016
-
Autogenerated Change: Release TensorBoard at TAG: 16
Change: 120004793
Configuration menu - View commit details
-
Copy full SHA for 26fdfb8 - Browse repository at this point
Copy the full SHA 26fdfb8View commit details -
Some fixes based on @dave-andersen feedback
Renamed AsBool -> Bool, AsInt -> Int... based on @tmc feedback
Alonso Vidales committedApr 16, 2016 Configuration menu - View commit details
-
Copy full SHA for 511f7b9 - Browse repository at this point
Copy the full SHA 511f7b9View commit details -
Make the ops_compatibility framework usable outside tensorflow/core.
Change: 120008243
Configuration menu - View commit details
-
Copy full SHA for c883803 - Browse repository at this point
Copy the full SHA c883803View commit details -
Start the collected queue runners automatically when creating a new s…
…ession in the supervisor. Make it innocuous to start a queue runner multiple times. Change: 120010279
Configuration menu - View commit details
-
Copy full SHA for ea6b59a - Browse repository at this point
Copy the full SHA ea6b59aView commit details -
Update generated Python Op docs.
Change: 120010519
Configuration menu - View commit details
-
Copy full SHA for 9eb5d42 - Browse repository at this point
Copy the full SHA 9eb5d42View commit details -
In tools/proto_text, move comment describing API to the header.
Also clarify that google.any expansion is not supported. Change: 120013039
Configuration menu - View commit details
-
Copy full SHA for 78eba87 - Browse repository at this point
Copy the full SHA 78eba87View commit details -
Do not add large constants (> 10M) to the graph while constant folding.
Change: 120015061
Configuration menu - View commit details
-
Copy full SHA for 73d026e - Browse repository at this point
Copy the full SHA 73d026eView commit details -
This would enable shape inference for higher order functions such as …
…scan. Change: 120017883
Configuration menu - View commit details
-
Copy full SHA for 81be721 - Browse repository at this point
Copy the full SHA 81be721View commit details -
Implement SparseTensor + (dense) Tensor.
Usage: `tf.sparse_add(sp_tensor, tensor)`; CPU only for now. The newly introduced ScatterNdFunctor can later be extended to handle ScatterNd{Update,Add,Sub} ops, similar to the existing GatherNd op. Change: 120018023
Configuration menu - View commit details
-
Copy full SHA for aed2450 - Browse repository at this point
Copy the full SHA aed2450View commit details -
Update ops-related pbtxt files.
Change: 120018234
Configuration menu - View commit details
-
Copy full SHA for 4e3f022 - Browse repository at this point
Copy the full SHA 4e3f022View commit details -
Update generated Python Op docs.
Change: 120018303
Configuration menu - View commit details
-
Copy full SHA for b0f28df - Browse repository at this point
Copy the full SHA b0f28dfView commit details -
Using a template for getTensorFromGraph graph and removing some redun…
…dancy
Alonso Vidales committedApr 16, 2016 Configuration menu - View commit details
-
Copy full SHA for dacb950 - Browse repository at this point
Copy the full SHA dacb950View commit details -
Configuration menu - View commit details
-
Copy full SHA for 091a2c3 - Browse repository at this point
Copy the full SHA 091a2c3View commit details -
Add a comment to install mentioning cuda and cudnn requirements (tens…
…orflow#1985) for PIP installs.
Vijay Vasudevan committedApr 16, 2016 Configuration menu - View commit details
-
Copy full SHA for 36d89cf - Browse repository at this point
Copy the full SHA 36d89cfView commit details -
Add shape information to dense gradients.
This change takes advantage of the fact that a dense gradient tensor must have the same shape as the corresponding input to the forward-pass op. It may enable additional optimizations (such as the ability to use the accumulate_n aggregator) that rely on static shape information. This change revealed some broken code, for which fixes are included: * The shape function for `tf.diag_part()` incorrectly used an equality test on shapes, rather than a compatibility test. * The gradient function for `tf.nn.top_k()` incorrectly returned a length-1 vector for the gradient w.r.t. `k`, which it should have been a scalar. * A unit test in `control_flow_ops_py_test` inconsistently used vectors and scalars. Change: 120038251
Configuration menu - View commit details
-
Copy full SHA for f88d0cb - Browse repository at this point
Copy the full SHA f88d0cbView commit details -
Replaced NewGraphFromText and LoadGraphFromFile by NewGraphFromReader
Alonso Vidales committedApr 16, 2016 Configuration menu - View commit details
-
Copy full SHA for 5a16a84 - Browse repository at this point
Copy the full SHA 5a16a84View commit details -
Modified code to use c++11-style range loops based on @dave-andersen …
…feedback
Alonso Vidales committedApr 16, 2016 Configuration menu - View commit details
-
Copy full SHA for 6fa19d3 - Browse repository at this point
Copy the full SHA 6fa19d3View commit details -
add a reference for Caffe in tutorial (tensorflow#1982)
add a reference for Caffe in deepdream tutorial
Configuration menu - View commit details
-
Copy full SHA for 960cc01 - Browse repository at this point
Copy the full SHA 960cc01View commit details -
Configuration menu - View commit details
-
Copy full SHA for c303ff7 - Browse repository at this point
Copy the full SHA c303ff7View commit details -
Restored README.md deleted by accident
Alonso Vidales committedApr 16, 2016 Configuration menu - View commit details
-
Copy full SHA for 1c62916 - Browse repository at this point
Copy the full SHA 1c62916View commit details
Commits on Apr 17, 2016
-
add a reference for Laplacian pyramid (tensorflow#1976)
add a reference for Laplacian pyramid
Configuration menu - View commit details
-
Copy full SHA for bd37cbb - Browse repository at this point
Copy the full SHA bd37cbbView commit details -
Added autogeneration for protobuf files and graph operations based on @…
…vrv feedback
Alonso Vidales committedApr 17, 2016 Configuration menu - View commit details
-
Copy full SHA for 54cc4db - Browse repository at this point
Copy the full SHA 54cc4dbView commit details -
Removed example image for the label_image_go exercise based on @vrv f…
…eedback
Alonso Vidales committedApr 17, 2016 Configuration menu - View commit details
-
Copy full SHA for 00ba021 - Browse repository at this point
Copy the full SHA 00ba021View commit details -
Starting special_math_ops.py. Made for math Ops that have more depend…
…encies, probably done in pure python TF, rather than wrapped C++. Starting by moving lbeta from math_ops to special_math_ops, since a bugfix requires it to depend on control_flow_ops. The bug was: lbeta([]) returned -inf, whereas it should have returned 0.0. Change: 120069192
Configuration menu - View commit details
-
Copy full SHA for 8db39c9 - Browse repository at this point
Copy the full SHA 8db39c9View commit details -
Update generated Python Op docs.
Change: 120069280
Configuration menu - View commit details
-
Copy full SHA for 177fd83 - Browse repository at this point
Copy the full SHA 177fd83View commit details -
Configuration menu - View commit details
-
Copy full SHA for 24099b6 - Browse repository at this point
Copy the full SHA 24099b6View commit details -
Reviewed documentation and test error messages
Alonso Vidales committedApr 17, 2016 Configuration menu - View commit details
-
Copy full SHA for 00bebc1 - Browse repository at this point
Copy the full SHA 00bebc1View commit details -
Added quantized type testing to tensor creation.
Change: 120076873
Configuration menu - View commit details
-
Copy full SHA for 0dfcea4 - Browse repository at this point
Copy the full SHA 0dfcea4View commit details
Commits on Apr 18, 2016
-
Replaces naked returns by named returns
Alonso Vidales committedApr 18, 2016 Configuration menu - View commit details
-
Copy full SHA for a33e293 - Browse repository at this point
Copy the full SHA a33e293View commit details -
Optimized implementation of depthwise conv backprop filter for CPU.
// OLD Benchmark Time(ns) CPU(ns) Iterations ------------------------------------------------------------------------ BM_ConvFloatDepthwiseBkFilterCPU1_conv0 281152179 280588497 100 588.2M items/s 32_112_112_3_8_24_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv1 760242956 758694909 100 580.1M items/s 32_112_112_64_1_64_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv2 383554418 382741182 100 574.9M items/s 32_56_56_128_1_128_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv3 98924384 98665676 100 557.2M items/s 32_56_56_128_1_128_3_3_2_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv4 94237506 94005920 100 585.0M items/s 32_28_28_128_1_128_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv5 106895864 106648144 100 515.7M items/s 32_14_14_512_1_512_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv6 69247718 69078442 100 398.0M items/s 32_7_7_1024_1_1024_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv7 70304661 70126053 100 588.1M items/s 32_112_112_3_8_24_3_3_2_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv8 67619710 67447142 100 611.4M items/s 32_112_112_3_8_24_3_3_2_1_cpu1 // NEW 1-thread Benchmark Time(ns) CPU(ns) Iterations ------------------------------------------------------------------------ BM_ConvFloatDepthwiseBkFilterCPU1_conv0 59981294 59569328 100 2.7G items/s 32_112_112_3_8_24_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv1 165631344 165250674 100 2.6G items/s 32_112_112_64_1_64_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv2 76910026 76705735 100 2.8G items/s 32_56_56_128_1_128_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv3 21491439 21375872 100 2.5G items/s 32_56_56_128_1_128_3_3_2_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv4 18677714 18587209 100 2.9G items/s 32_28_28_128_1_128_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv5 2347423 23377934 100 2.3G items/s 32_14_14_512_1_512_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv6 17066829 16982791 100 1.6G items/s 32_7_7_1024_1_1024_3_3_1_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv7 14822571 14744419 100 2.7G items/s 32_112_112_3_8_24_3_3_2_2_cpu1 BM_ConvFloatDepthwiseBkFilterCPU1_conv8 14325480 14254559 100 2.8G items/s 32_112_112_3_8_24_3_3_2_1_cpu1 // NEW 4-threads Benchmark Time(ns) CPU(ns) Iterations ------------------------------------------------------------------------ BM_ConvFloatDepthwiseBkFilterCPU4_conv0 21809044 69141049 100 7.4G items/s 32_112_112_3_8_24_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv1 57704422 192333505 100 7.5G items/s 32_112_112_64_1_64_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv2 29761264 91848609 100 7.2G items/s 32_56_56_128_1_128_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv3 9075773 26429821 100 5.9G items/s 32_56_56_128_1_128_3_3_2_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv4 7276754 22100190 100 7.4G items/s 32_28_28_128_1_128_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv5 6756189 24510067 100 8.0G items/s 32_14_14_512_1_512_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv6 4837993 17723279 142 5.6G items/s 32_7_7_1024_1_1024_3_3_1_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv7 6676347 19935585 100 6.0G items/s 32_112_112_3_8_24_3_3_2_2_cpu4 BM_ConvFloatDepthwiseBkFilterCPU4_conv8 5951583 17181079 100 6.8G items/s 32_112_112_3_8_24_3_3_2_1_cpu4 TESTED: - passed opensource_build - passed unit tests Change: 120125325
Configuration menu - View commit details
-
Copy full SHA for 3f27cff - Browse repository at this point
Copy the full SHA 3f27cffView commit details -
Parallelize MaxPool across batch dimension.
Benchmark Base (ns) New (ns) Improvement ------------------------------------------------------------------ BM_MaxPool_32_112_112_64_3_3_2_VALID_1 28173747 28956041 -2.8% BM_MaxPool_32_56_56_192_3_3_2_VALID_1 14467716 14581478 -0.8% BM_MaxPool_32_28_28_352_3_3_2_VALID_1 5318842 5367336 -0.9% BM_MaxPool_32_14_14_576_3_3_2_VALID_1 1331917 1351642 -1.5% BM_MaxPool_32_112_112_64_3_3_2_SAME_1 28757024 29005280 -0.9% BM_MaxPool_32_56_56_192_3_3_2_SAME_1 15119295 15478783 -2.4% BM_MaxPool_32_28_28_352_3_3_2_SAME_1 5802450 5871220 -1.2% BM_MaxPool_32_14_14_576_3_3_2_SAME_1 1632582 1662128 -1.8% BM_MaxPool_32_112_112_64_3_3_2_VALID_4 28579650 8240771 +71.2% BM_MaxPool_32_56_56_192_3_3_2_VALID_4 14621344 4373595 +70.1% BM_MaxPool_32_28_28_352_3_3_2_VALID_4 5404303 1571711 +70.9% BM_MaxPool_32_14_14_576_3_3_2_VALID_4 1343607 427873 +68.2% BM_MaxPool_32_112_112_64_3_3_2_SAME_4 29195151 8204002 +71.9% BM_MaxPool_32_56_56_192_3_3_2_SAME_4 15314088 4642979 +69.7% BM_MaxPool_32_28_28_352_3_3_2_SAME_4 6094918 1777112 +70.8% BM_MaxPool_32_14_14_576_3_3_2_SAME_4 1643584 544554 +66.9% TESTED: - passed opensource_build - passed unit tests Change: 120128184
Configuration menu - View commit details
-
Copy full SHA for 517d3af - Browse repository at this point
Copy the full SHA 517d3afView commit details -
Added a format for saving an inference graph that can be memmapped an…
…d an utility to convert a freezed graph into this format. Change: 120128412
Configuration menu - View commit details
-
Copy full SHA for 3c280f6 - Browse repository at this point
Copy the full SHA 3c280f6View commit details -
Fix the gradient for functions when its output has no dependencies on
its inputs. Change: 120128592
Configuration menu - View commit details
-
Copy full SHA for 449ecb5 - Browse repository at this point
Copy the full SHA 449ecb5View commit details -
Fixes and enhancements for contrib/tensorforest:
- remove flags (pushed to internal client) - thread-parallel execution for CountExtremeleyRandomStats op. - critical time-seeding fix. Change: 120129936
Configuration menu - View commit details
-
Copy full SHA for c2d9cb1 - Browse repository at this point
Copy the full SHA c2d9cb1View commit details -
Adding several loss functions including cosine_distance_loss, log_los…
…s, softmax_cross_entropy_loss, and sum_of_pairwise_squares_loss. Refactoring old losses for consistency. Change: 120130871
Configuration menu - View commit details
-
Copy full SHA for ec89b0c - Browse repository at this point
Copy the full SHA ec89b0cView commit details -
Fix for IndexedSlices gradient accumulation in while loop.
Change: 120136109
Configuration menu - View commit details
-
Copy full SHA for 3402f51 - Browse repository at this point
Copy the full SHA 3402f51View commit details -
Adding DirichletMultinomial class to contrib/distributions/
Class represents multi-indexed batches of Dirichlet Multinomial distributions. Initialized with parameters alpha, which broadcast to arbitrary shapes to match arguments in e.g. dist.pdf(x). Change: 120138028
Configuration menu - View commit details
-
Copy full SHA for a0d14f0 - Browse repository at this point
Copy the full SHA a0d14f0View commit details -
Make the init_op optional and add an optional init_fn.
Change: 120139351
Configuration menu - View commit details
-
Copy full SHA for b92d7cf - Browse repository at this point
Copy the full SHA b92d7cfView commit details -
Update generated Python Op docs.
Change: 120140181
Configuration menu - View commit details
-
Copy full SHA for 319f898 - Browse repository at this point
Copy the full SHA 319f898View commit details -
Expanding the scope filter in get_collection to support a full regex …
…and expanding documentation of get_collection. Change: 120141411
Configuration menu - View commit details
-
Copy full SHA for e7f6361 - Browse repository at this point
Copy the full SHA e7f6361View commit details -
Update generated Python Op docs.
Change: 120142925
Configuration menu - View commit details
-
Copy full SHA for 511634a - Browse repository at this point
Copy the full SHA 511634aView commit details -
Removed most of the names returns based on @tmc feedback
Alonso Vidales committedApr 18, 2016 Configuration menu - View commit details
-
Copy full SHA for 95a6664 - Browse repository at this point
Copy the full SHA 95a6664View commit details -
Alonso Vidales committed
Apr 18, 2016 Configuration menu - View commit details
-
Copy full SHA for 562a151 - Browse repository at this point
Copy the full SHA 562a151View commit details -
Configuration menu - View commit details
-
Copy full SHA for 0a37c9f - Browse repository at this point
Copy the full SHA 0a37c9fView commit details -
Make native depthwise Conv2d the default.
TESTED: - passed opensource_build (only known failing test loss_ops test fails) - passed unit tests Change: 120156989
Configuration menu - View commit details
-
Copy full SHA for 975094c - Browse repository at this point
Copy the full SHA 975094cView commit details -
Fix for opensource break due to new losses code.
Change: 120161004
Configuration menu - View commit details
-
Copy full SHA for e07c48d - Browse repository at this point
Copy the full SHA e07c48dView commit details -
Change: 120164016
Configuration menu - View commit details
-
Copy full SHA for 32e2bd1 - Browse repository at this point
Copy the full SHA 32e2bd1View commit details -
Fixing additional silent int64->32 conversion errors/warnings.
Fixes a bug in gather that would segfault when gathering from very large (>2^31 entry) parameter tensors. Gather can now handle index vectors with more than 2^31 entries, (if you have enough memory). Change: 120171737
Configuration menu - View commit details
-
Copy full SHA for 2092fb4 - Browse repository at this point
Copy the full SHA 2092fb4View commit details -
Have StatSummarizer print OP types as well as name.
Change: 120172505
Configuration menu - View commit details
-
Copy full SHA for 5c9e7d3 - Browse repository at this point
Copy the full SHA 5c9e7d3View commit details -
Refactoring/unification of weights & features within sdca_ops.cc.
Combines the 6 entities (sparse, dense) x (features, weights, delta-weights) into a single class (FeaturesAndWeights) which tracks all features and weights, hiding their underlying representation. This class is built by composing other classes: FeaturesAndWeights SparseFeaturesAndWeights (examples_by_group_) WeightsAndDeltas WeightsByGroup (delta_weights_by_group_) DenseFeaturesAndWeights (features_by_group_) WeightsAndDeltas (same as above) Also adds a microbenchmark. Change: 120173207
Configuration menu - View commit details
-
Copy full SHA for ad21225 - Browse repository at this point
Copy the full SHA ad21225View commit details
Commits on Apr 19, 2016
-
Fix SAME padding calc when stride>ksize (tensorflow#1991)
Adds a failing test case for conv2d ksize=1, stride=2, padding=same, input=4x4 Adds tests convering that case for pooling ops too. There's one test case I found that fails in a different way that I left commented out with R=C=8, K=2, S=4. The error is: F tensorflow/stream_executor/cuda/cuda_dnn.cc:483] could not set cudnn convolution descriptor: CUDNN_STATUS_BAD_PARAM
Configuration menu - View commit details
-
Copy full SHA for cf7ce8a - Browse repository at this point
Copy the full SHA cf7ce8aView commit details -
Introduce tf.sparse_reduce_sum() and a CPU kernel.
Change: 120183287
Configuration menu - View commit details
-
Copy full SHA for b3b58fc - Browse repository at this point
Copy the full SHA b3b58fcView commit details -
Update ops-related pbtxt files.
Change: 120184745
Configuration menu - View commit details
-
Copy full SHA for fc432e3 - Browse repository at this point
Copy the full SHA fc432e3View commit details -
Change: 120185825
Configuration menu - View commit details
-
Copy full SHA for 5c9bc51 - Browse repository at this point
Copy the full SHA 5c9bc51View commit details -
Update ops-related pbtxt files.
Change: 120186451
Configuration menu - View commit details
-
Copy full SHA for b442bc6 - Browse repository at this point
Copy the full SHA b442bc6View commit details -
Update generated Python Op docs.
Change: 120186712
Configuration menu - View commit details
-
Copy full SHA for c018da6 - Browse repository at this point
Copy the full SHA c018da6View commit details -
Change: 120192253
Configuration menu - View commit details
-
Copy full SHA for 7e44d24 - Browse repository at this point
Copy the full SHA 7e44d24View commit details -
Minor improvements of shape inference for tensor array.
Change: 120192816
Configuration menu - View commit details
-
Copy full SHA for 8baf2f3 - Browse repository at this point
Copy the full SHA 8baf2f3View commit details -
- Oneof values should be printed even when equal to the default. - Fields should be printed in tag number order, not declaration order. Change: 120223509
Configuration menu - View commit details
-
Copy full SHA for 1150ce5 - Browse repository at this point
Copy the full SHA 1150ce5View commit details -
Make tf.diag_part work with inputs of unknown shape.
Change: 120231603
Configuration menu - View commit details
-
Copy full SHA for 913b6e2 - Browse repository at this point
Copy the full SHA 913b6e2View commit details -
Fixed Error comments and exported error properties based on @rakyll f…
…eedback
Alonso Vidales committedApr 19, 2016 Configuration menu - View commit details
-
Copy full SHA for 13d4176 - Browse repository at this point
Copy the full SHA 13d4176View commit details -
* Removed unsupported DataType from the list of vars
* Removed TensorInt interface * Remove unsupported data types
Alonso Vidales committedApr 19, 2016 Configuration menu - View commit details
-
Copy full SHA for f6b209a - Browse repository at this point
Copy the full SHA f6b209aView commit details -
Modified Int64 by Int64s, Bool by Bools and so on based on @dave-ande…
Alonso Vidales committedApr 19, 2016 Configuration menu - View commit details
-
Copy full SHA for 9907cee - Browse repository at this point
Copy the full SHA 9907ceeView commit details -
* Modified TensorShape to be a vector instead of a matrix
* Fixed some typos
Alonso Vidales committedApr 19, 2016 Configuration menu - View commit details
-
Copy full SHA for 8316fbc - Browse repository at this point
Copy the full SHA 8316fbcView commit details -
Configuration menu - View commit details
-
Copy full SHA for 0cb193e - Browse repository at this point
Copy the full SHA 0cb193eView commit details -
Configuration menu - View commit details
-
Copy full SHA for f242aed - Browse repository at this point
Copy the full SHA f242aedView commit details -
Fixed problems with GetVal based on @dave-andersen feedback
Alonso Vidales committedApr 19, 2016 Configuration menu - View commit details
-
Copy full SHA for a9d2ce3 - Browse repository at this point
Copy the full SHA a9d2ce3View commit details -
Stream Executor: perform a cudnn library vs. binary check. (tensorflo…
…w#1987) During initialization, we check whether the loaded version of the cudnn library matches the version of the library that the binary was compiled against. When there is a mismatch, throw an error. Example: I built with cudnnv5 but I am loading V4, and I get: E tensorflow/stream_executor/cuda/cuda_dnn.cc:298] Loaded cudnn library: 4007 but source was compiled against 5004. If using a binary install, upgrade your cudnn library to match. If building from sources, make sure the library loaded matches the version you specified during compile configuration. I then copy the v5 version of libcudnn.so into /usr/local/cuda/lib64 and when I re-run the program, the error goes away. Thoughts, @zheng-xq, @leary-google ?
Configuration menu - View commit details
-
Copy full SHA for dfcad66 - Browse repository at this point
Copy the full SHA dfcad66View commit details
Commits on Apr 20, 2016
-
Configuration menu - View commit details
-
Copy full SHA for d70c765 - Browse repository at this point
Copy the full SHA d70c765View commit details -
set cudnnSetFilter4dDescriptor format using filter layout (tensorflow…
…#2027) CUDNN_TENSOR_NHWC the only other format is still not supported since kInputYXOutput would correspond to CUDNN_TENSOR_CHWN which doesn’t exist for issue tensorflow#1786
Configuration menu - View commit details
-
Copy full SHA for bef9bd4 - Browse repository at this point
Copy the full SHA bef9bd4View commit details -
Merge pull request tensorflow#2028 from martinwicke/branch_120231603
Upstream changes from internal
Configuration menu - View commit details
-
Copy full SHA for cc9bfbf - Browse repository at this point
Copy the full SHA cc9bfbfView commit details -
Fix wrong docstring (tensorflow#2022)
The keys in the _word2id are bytes literals.
Configuration menu - View commit details
-
Copy full SHA for c6eaa2d - Browse repository at this point
Copy the full SHA c6eaa2dView commit details -
Fix wrong docstring (tensorflow#2024)
The keys in the _word2id are bytes literals.
Configuration menu - View commit details
-
Copy full SHA for e1c0176 - Browse repository at this point
Copy the full SHA e1c0176View commit details -
Merge branch 'master' into go_bindings_tensors
Alonso Vidales committedApr 20, 2016 Configuration menu - View commit details
-
Copy full SHA for 078d42c - Browse repository at this point
Copy the full SHA 078d42cView commit details -
Alonso Vidales committed
Apr 20, 2016 Configuration menu - View commit details
-
Copy full SHA for e444bda - Browse repository at this point
Copy the full SHA e444bdaView commit details -
Modified go generate to allocate shared libraries on /usr/lib/
Alonso Vidales committedApr 20, 2016 Configuration menu - View commit details
-
Copy full SHA for 8c4d165 - Browse repository at this point
Copy the full SHA 8c4d165View commit details -
* Added documentation to install the package
* Fixed some typos * Relocated errors after types
Alonso Vidales committedApr 20, 2016 Configuration menu - View commit details
-
Copy full SHA for da22433 - Browse repository at this point
Copy the full SHA da22433View commit details
Commits on Apr 27, 2016
-
Alonso Vidales committed
Apr 27, 2016 Configuration menu - View commit details
-
Copy full SHA for fecfbea - Browse repository at this point
Copy the full SHA fecfbeaView commit details
Commits on May 5, 2016
-
Updated documentation in order to do this package go-gettable
Alonso Vidales committedMay 5, 2016 Configuration menu - View commit details
-
Copy full SHA for 9427604 - Browse repository at this point
Copy the full SHA 9427604View commit details