hash stringlengths 40 40 | msg stringlengths 1 131k | author stringlengths 1 33 | email stringlengths 0 57 | date int64 1,447B 1,698B |
|---|---|---|---|---|
914c870c6f6894d9be4c44d39878b1b7eeaf71bd | Open-source doc generator. Adds tools/docs with shell scripts to generate docs from source code, both c++ and python. The generated docs overwrite the docs in the g3doc/api_docs folder.
Change: 112989505 | Martin Wicke | wicke@google.com | 1,453,761,773,000 |
77c5cbb114956305625c5ac97b969a7637f55dd2 | Fix OSS Build (again) due to doc generation scripts.
Change: 112993601 | Vijay Vasudevan | vrv@google.com | 1,453,763,741,000 |
b159e211860e26e3dbae26989163b9da55e3f430 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,453,763,868,000 |
a932de60611b9f2833613b7d85151a83163e4327 | Keeps optional question format consisten
Added a newline to keep separation of optional questions from required tasks consistent and easier to read. | nick | nsipplswezey@users.noreply.github.com | 1,453,790,006,000 |
95e189dcc2cef86cecd2fc3f06458a7edc05dda6 | Allow to use a Tensor as initializer in get_variable. Fixes #589.
Change: 112993949 | Lukasz Kaiser | lukaszkaiser@gmail.com | 1,453,763,905,000 |
ffdaab2a6dd5eafa03b79ce41b13da7e13cc1027 | Made visibility for //tensorflow:darwin public
Change: 112999739 | Manjunath Kudlur | keveman@gmail.com | 1,453,766,885,000 |
29c45a61178f61a8aabdb9d79f6fbecbe0c3ca4e | Expose colocate_gradients_with_ops from gradients in optimizers
Change: 113003537 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,768,962,000 |
560f0cdb54e1df2c1d851189fae7670b25a748bd | Speedup Gather and SparseApplyAdagrad when embedding dim is 1, which is the
case for wide logistic models.
Change: 113020949 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,787,192,000 |
7d4a063834cf9e5cdebe95d2115ce2e527296d79 | Delete obsolete comment.
Change: 113024053 | Josh Levenberg | josh11b@tensorflow.org | 1,453,791,471,000 |
2232f29969f7bf2a3dbd8d9f08119d87e907b73d | Merge pull request #885 from nsipplswezey/patch-3
Keeps optional question format consistent | Martin Wicke | martin.wicke@gmail.com | 1,453,824,532,000 |
da453bafa904e920b08829624f8d87925934e352 | Merge commit for internal changes | Manjunath Kudlur | keveman@gmail.com | 1,453,825,592,000 |
bee58e3750fc456c0a236f1b2e2a771bb355539c | Some error checking and user output notes
Hi,
While running this it seems like the notebook hung but what really is happening is that notMNIST_large.tar.gz is taking a long time to extract. Also, since I looked at the contents of the directory a .DS_Store file was created. Re-sending pull request because of a mistake on earlier pull request. | emmjaykay | mjk@emmjaykay.org | 1,453,793,134,000 |
0de974dc017a7a3f2da172385a34cef5c05c35d3 | Merge pull request #886 from emmjaykay/master. | Vijay Vasudevan | vrv@google.com | 1,453,831,947,000 |
e0a7c1744f13b62e013ff243c95c3618eb3ab340 | Corrected bug in Python code, wrong variable being used. | Andre Cruz | andre.cruz@co.sapo.pt | 1,453,828,495,000 |
1be623d2152f692170b8c79dde5de292b2cc6fd4 | Merge pull request #892 from edevil/master. | Vincent Vanhoucke | vanhoucke@google.com | 1,453,833,396,000 |
1854a502c84b96200308f6dfcd4b341b799ae50b | Revert "Ensure the code compiles when AVX2 instructions are enabled."
This reverts commit 19b360611066711f8c4d55efae3259d8614b1e0a.
Fixes #883. | Vijay Vasudevan | vrv@google.com | 1,453,834,219,000 |
38242dd94a87a0e72f39315447866e55923d642c | Add new matrix_solve_ls op for solving linear least-squares problems.
Change: 113064195 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,829,453,000 |
b0d0e476db7a6d5998af71c330efe41c19817dd7 | Rollback of "Add new matrix_solve_ls op for solving linear least-squares problems."
Doesn't work in OSS
Change: 113069612 | Vijay Vasudevan | vrv@google.com | 1,453,832,369,000 |
22550dbd074835ac127fc054ae623b5fc205a34b | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113073087 | Josh Levenberg | josh11b@tensorflow.org | 1,453,834,014,000 |
351e0289095b4e638e6ab71c8d69a61efca6ecb2 | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113073090 | Josh Levenberg | josh11b@tensorflow.org | 1,453,834,014,000 |
1a116c2fd66037245ab1d61c037870a721c4d9f1 | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113073106 | Josh Levenberg | josh11b@tensorflow.org | 1,453,834,020,000 |
e866ee53ea243ea1f3154ddc21cdf8661930d518 | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113074643 | Josh Levenberg | josh11b@tensorflow.org | 1,453,834,749,000 |
cf4e1339b174101dcd0094f1cc5fbfe0a6392b0d | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,453,834,910,000 |
dc8e9abc2ceb27c7f39c9316f849743520e7eee4 | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113074952 | Josh Levenberg | josh11b@tensorflow.org | 1,453,834,902,000 |
3a7d603feaf6434ffd6d9fd27cac65c1ef565e07 | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113075039 | Josh Levenberg | josh11b@tensorflow.org | 1,453,834,937,000 |
5a24d3a2514698b0ae11563b2ea21e368de48a4f | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113075110 | Josh Levenberg | josh11b@tensorflow.org | 1,453,834,962,000 |
3ede5506acf6a026f09eda33277d46e34ac7ed10 | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113075177 | Josh Levenberg | josh11b@tensorflow.org | 1,453,834,981,000 |
d47ca3eae856a17fa3326e3f694983bfd9c10dc0 | Allow easier reuse of tool scripts.
Change: 113075454 | Pete Warden | pete@petewarden.com | 1,453,835,098,000 |
31849e0a44a1c9c99c4524a65c9ee90e80c41499 | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113078283 | Josh Levenberg | josh11b@tensorflow.org | 1,453,836,460,000 |
b2f0bc2e230dcd690e7cf34e5425f0f499d9557b | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113078322 | Josh Levenberg | josh11b@tensorflow.org | 1,453,836,474,000 |
6e5a7d9af85d67ec1dd499b472472b68a56d51a2 | Updates to 1st Udacity colab:
- re-make data extraction idempotent in case someone accidentally cancels the extraction.
- Python 3 compatibility (h/t @lzlarryli)
Change: 113079245 | Vincent Vanhoucke | vanhoucke@google.com | 1,453,836,923,000 |
977c8357f3d9f750d79d9f9b9978da64d838e386 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,453,837,330,000 |
f8fa35b8a1910772d6d6ba7b621f905358640c2c | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113080048 | Josh Levenberg | josh11b@tensorflow.org | 1,453,837,358,000 |
c10f439740396006e45059435e552e4d4ad2c1ad | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113080052 | Josh Levenberg | josh11b@tensorflow.org | 1,453,837,358,000 |
64aa0989c110433e275d23bbe6f3605a26b2efb0 | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113080064 | Josh Levenberg | josh11b@tensorflow.org | 1,453,837,366,000 |
3788290672919e050c7c339ac3f5cbfe0fc6e430 | Make 2nd and 3rd colab compatible with Python 3
Change: 113082597 | Vincent Vanhoucke | vanhoucke@google.com | 1,453,838,724,000 |
c5c5d61002b75aeda9a3970f73af9a0bbf9d7e41 | Fix bug with urlretrieve
Change: 113087571 | Vincent Vanhoucke | vanhoucke@google.com | 1,453,842,028,000 |
632c90068a1ecae1a7ae808628922eeb2022d38c | TensorFlow: fix more int -> size_t warnings
Change: 113089380 | Vijay Vasudevan | vrv@google.com | 1,453,843,111,000 |
7f55529b03328584e9aaf191214a8722b9d47796 | Mount notebooks into a volume to avoid data loss with --rm flag
The current way this is done:
```
docker run -p 8888:8888 -it --rm $USER/assignments
```
creates an ephemeral container due to the `--rm` flag. In the first notebook you are doing very expensive operations (gunziping) several GBs. All of this data will be lost when the container shuts down. Using the filesystem as a mounted volume, allows subsequent sessions to have access to the files created (eg pickled data). This is especially important because this is course material so want to avoid time consuming gotchas.
update readme Re ephemeral vs mounted notebooks dir | Timothy J Laurent | timothyjlaurent@gmail.com | 1,453,693,599,000 |
5758d53d96eb1cecf8cbceab93554facb0eae7a1 | Merge pull request #873 from timothyjlaurent/patch-1. | Vincent Vanhoucke | vanhoucke@google.com | 1,453,843,683,000 |
f2a42d48a885ce8956c3ea69ef900cf484a6e873 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,453,844,205,000 |
a18d48f80f118b28d351001a9b4c47613e228f34 | Change GraphDef versions to use version, min_consumer, min_producer
Since all GraphDef versions to date are forwards compatible, we make no attempt
to merge old-style and new-style GraphDef version information. The old int32
version field is ignored (and deprecated) and the new VersionDef versions field
defaults to producer 0, min_consumer 0 if left out.
As a benefit, once this CL is in we can immediately bump the scalar strictness
GraphDef version, since consumers will either ignore the new versions (before
this CL) or be forward compatible (after this CL).
Later, I'll use the same mechanism and protobuf to add versioning to checkpoints.
Change: 113091281 | Geoffrey Irving | geoffreyi@google.com | 1,453,844,240,000 |
6d2a4182ff6d6fbb1b209037ab0df87ecd884ad5 | Change InlineVector representation to better use inlined space.
We can now fit up to 15 1-byte elements in an InlinedVector that
occupies 16 bytes. Previously, this would require 32 bytes.
Removed Allocator support.
Removed old-style array() methods (use data() instead).
Benchmark Base (ns) New (ns) Improvement
------------------------------------------------------------------
BM_InlinedVectorFill/0 1 2 -100.0%
BM_InlinedVectorFill/1 2 2 +0.0%
BM_InlinedVectorFill/8 8 8 +0.0%
BM_InlinedVectorFill/64 184 261 -41.8%
BM_InlinedVectorFill/512 1211 1923 -58.8%
BM_InlinedVectorFill/1k 2346 3792 -61.6%
BM_InlinedVectorFillRange/0 1 1 +0.0%
BM_InlinedVectorFillRange/1 5 4 +20.0%
BM_InlinedVectorFillRange/8 5 4 +20.0%
BM_InlinedVectorFillRange/64 52 22 +57.7%
BM_InlinedVectorFillRange/512 242 55 +77.3%
BM_InlinedVectorFillRange/1k 458 88 +80.8%
BM_InlinedVectorTenAssignments/0 32 25 +21.9%
BM_InlinedVectorTenAssignments/1 51 47 +7.8%
BM_InlinedVectorTenAssignments/2 53 51 +3.8%
BM_InlinedVectorTenAssignments/3 59 73 -23.7%
BM_InlinedVectorTenAssignments/4 59 73 -23.7%
BM_InlinedVectorTenAssignments/20 129 151 -17.1%
BM_CreateFromInitializerList ?????? 1 ----
BM_LargeSwappableElements/0 3 4 -33.3%
BM_LargeSwappableElements/1 91 94 -3.3%
BM_LargeSwappableElements/8 1230 1243 -1.1%
BM_LargeSwappableElements/64 4 2 +50.0%
BM_LargeSwappableElements/512 4 2 +50.0%
BM_LargeSwappableElements/1k 4 2 +50.0%
Change: 113094648 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,846,152,000 |
46b108c8a606fde8654e4177d72001aa31138906 | Remove THIRD_PARTY_ prefix from #include guards.
Change: 113094744 | Josh Levenberg | josh11b@tensorflow.org | 1,453,846,213,000 |
5dff1a1d0e8d9e5aa63cc84742642cb7e527c1dc | TensorFlow: remove more compiler warnings due to uninitialized vars
and invalid comparisons.
Change: 113096531 | Vijay Vasudevan | vrv@google.com | 1,453,847,233,000 |
de3fd54c1beda2861a0a0a9197328e42b54170c0 | Allow the list of previous checkpoints in the Saver to be set with a timestamp. Otherwise previous timestamp checkpoints could be kept forever.
Add a Stat function to the gfile module that returns the last modified time of a file.
Change: 113098506 | Andrew Dai | adai@google.com | 1,453,848,331,000 |
fc0235e04a040f039f9caf5d8e28ee0db0b8abfd | Rollback of"
Change InlineVector representation to better use inlined space.
We can now fit up to 15 1-byte elements in an InlinedVector that
occupies 16 bytes. Previously, this would require 32 bytes.
Removed Allocator support.
Removed old-style array() methods (use data() instead).
Benchmark Base (ns) New (ns) Improvement
------------------------------------------------------------------
BM_InlinedVectorFill/0 1 2 -100.0%
BM_InlinedVec...
"
Change: 113100787 | Vijay Vasudevan | vrv@google.com | 1,453,849,601,000 |
f13d006e097c4e8010a4ad3ad2018a0369f5dc19 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,453,850,007,000 |
105f38abbfa628361b5203fd51ae7ff130a3c811 | Rollback of "Add new matrix_solve_ls op for solving linear least-squares problems."
Doesn't work in OSS
Change: 113103348 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,851,032,000 |
cfb27ca4c2c4dbbf27726a3bc4d72bb6f8865319 | BUGFIX: Allow histogram_ops.histogram_fixed_width to work with N-D new_values, previously only 1-D new_values worked. Add test for scalar input. Add test for many random 3-D input that is in funny interval.
Change: 113104458 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,851,740,000 |
d33346a801181cb601a4cfc95f1087e165b19703 | TensorFlow: fix bug in TensorShapeDim -- size is the wrong type.
Adds a test that failed before this change, but passes after.
TensorShapeIter is only used by TensorShape::begin() and end(),
which are only used by AppendShape, which is only used by queue_base.h
on shapes that are unlikely to be 64-bit large, which is why
we never triggered this.
Discovered by jeff
Change: 113105906 | Vijay Vasudevan | vrv@google.com | 1,453,852,709,000 |
2bf4566c661a51b6415a5e6645900d36166a5a57 | TensorFlow: remove more compiler warnings from (non-gpu) build.
Change: 113108653 | Vijay Vasudevan | vrv@google.com | 1,453,854,212,000 |
ca55faeddd266e14fa832373a73d9b87b20a8002 | Expose the validate_indices option through the embedding_lookup interface.
Change: 113110910 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,855,662,000 |
92e3b68fe4db0393709f58c22bdb93e2a934dc15 | Added StubOutForTesting to googletest to allow for unit test stub methods.
Change: 113113179 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,857,153,000 |
4fb647cca9994ddc600943503f30e55a7fbdb14c | Fix OSS builds.
Change: 113114631 | Vijay Vasudevan | vrv@google.com | 1,453,858,234,000 |
42b1a67601f1a49a758536df4d1a7e3a8fdd27a7 | Fix TensorBoard imports (broke when py3 support was added)
Change: 113114985 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,858,516,000 |
d4422ff4b2f142de1d0c626f73c734655d340e0d | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,453,860,465,000 |
deb4f6f596f20e958fa4c9c0a47b98e6e7097aa4 | Fixes #791 python 3 build issue | Sagan Bolliger | git-contact@saganbolliger.com | 1,453,869,786,000 |
8140b8f51e7902b852ba36984e5938aa796ec99e | Update CROSSTOOL with Toolchain for ARM
Update CROSSTOOL that it uses the local_linux toolchain for arm | jmtatsch | tatsch@gmail.com | 1,453,892,803,000 |
99bdd1a4de33ef0614cadfd28b97fc4bd0869e5d | Merge pull request #901 from SaganBolliger/791-fix-py3-build. | Martin Wicke | wicke@google.com | 1,453,911,091,000 |
ba33a15ff35b60c495704ed33e4a01b489fb863b | Assignments 4, 5 and 6 compatible with python 3
Change: 113119710 | Vincent Vanhoucke | vanhoucke@google.com | 1,453,862,867,000 |
b10a672837f74094379b0cc2ae9ccf703e09e50e | Updated TensorBoard logic for discarding events after crashes to discard out of order events only within a particular tag. This was changed because race conditions in supervisor was causing many events to be unintentionally discarded.
Change: 113161277 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,910,277,000 |
fca14db4db363f2fd4b3f408fe0257d04aa445f3 | Bump GraphDef version to 6: TensorFlow is now scalar strict
Support for the scalar strict GraphDef version 6 was added in previous
CLs; this CL just bumps the current version to use it. Since the
min_consumer mechanism is in place, this CL is forward compatible.
The internal Google version of tensorflow is now as scalar strict as the
open source release. Note that outside of PLATFORM_GOOGLE, this change has no
effect, since the code was already scalar strict.
This was attempted once before, but had to be reverted since some users
were not updated.
Change: 113161920 | Geoffrey Irving | geoffreyi@google.com | 1,453,910,728,000 |
eecbcb261ad2a5d220c705dba6ff1e760adb8b95 | Changes Optimizer.compute_gradients to use var.ref() instead of var.value() as the tensor wrt which to compute the gradient. This enables things like this:
with tf.control_dependencies([barrier]):
# Fetch var just in time
v = tf.identity(var.ref())
# Now do something with v
Before this CL, the gradient wrt var would be 0 since tf.identity(var.ref()) does not depend on var.value() (which is another tf.identity(var.ref()), but from which there is no path in the graph to the loss).
gradient_descent_test.py tests the behavior fixed by this CL.
Change: 113163441 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,911,877,000 |
422f4492dd93af9b09fb0b27c57c65467c30f798 | Change InlineVector representation to better use inlined space.
We can now fit up to 15 1-byte elements in an InlinedVector that
occupies 16 bytes. Previously, this would require 32 bytes.
Removed Allocator support.
Removed old-style array() methods (use data() instead).
Benchmark Base (ns) New (ns) Improvement
------------------------------------------------------------------
BM_InlinedVectorFill/0 1 2 -100.0%
BM_InlinedVectorFill/1 2 2 +0.0%
BM_InlinedVectorFill/8 8 8 +0.0%
BM_InlinedVectorFill/64 184 261 -41.8%
BM_InlinedVectorFill/512 1211 1923 -58.8%
BM_InlinedVectorFill/1k 2346 3792 -61.6%
BM_InlinedVectorFillRange/0 1 1 +0.0%
BM_InlinedVectorFillRange/1 5 4 +20.0%
BM_InlinedVectorFillRange/8 5 4 +20.0%
BM_InlinedVectorFillRange/64 52 22 +57.7%
BM_InlinedVectorFillRange/512 242 55 +77.3%
BM_InlinedVectorFillRange/1k 458 88 +80.8%
BM_InlinedVectorTenAssignments/0 32 25 +21.9%
BM_InlinedVectorTenAssignments/1 51 47 +7.8%
BM_InlinedVectorTenAssignments/2 53 51 +3.8%
BM_InlinedVectorTenAssignments/3 59 73 -23.7%
BM_InlinedVectorTenAssignments/4 59 73 -23.7%
BM_InlinedVectorTenAssignments/20 129 151 -17.1%
BM_CreateFromInitializerList ?????? 1 ----
BM_LargeSwappableElements/0 3 4 -33.3%
BM_LargeSwappableElements/1 91 94 -3.3%
BM_LargeSwappableElements/8 1230 1243 -1.1%
BM_LargeSwappableElements/64 4 2 +50.0%
BM_LargeSwappableElements/512 4 2 +50.0%
BM_LargeSwappableElements/1k 4 2 +50.0%
Change: 113172547 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,917,859,000 |
89ecc05945b812615f3b72e0c407d277734ea152 | Add a how to for writing documentation.
Change: 113176925 | Martin Wicke | wicke@google.com | 1,453,920,100,000 |
4fba0fd5a1abfe4c47df4b6512994b1e5313ee57 | Note that master is installed with plain git clone
Since the docs are versioned, and the latest stable version is the default visible docset on the website, we should point out that the installation instructions actually install master, and also give instructions on how to install a specific branch. This change doesn't mention which branch it is on because we don't want to have to change this text in each release.
Closes #891. | Martin Wicke | wicke@google.com | 1,453,920,858,000 |
88df60466499c1082534955cf7c6edb1955903a3 | Merge pull request #900 from tensorflow/install-instructions. | Martin Wicke | wicke@google.com | 1,453,920,918,000 |
f3111525c97a25adef40854a588fd5590e7f6eb1 | Use the same types in `for` statements.
This removes lots of simple compiler warnings on 'signed' and 'unsigned'
comparisons (reported in #128). Please note that there are two types of
changes. It is because this does not change any function signatures.
* In most cases: `int` --> `size_t`.
* In `lookup_table_op.cc`, `string_to_number_op.cc`, and
`sparse_tensor.h`: `size_t` --> int. | Dongjoon Hyun | dongjoon@apache.org | 1,453,871,689,000 |
4eeb6d6cc99bd10e35061f8ffbd2f9d71ab72c56 | Merge pull request #902 from dongjoon-hyun/issue_128. | Vijay Vasudevan | vrv@google.com | 1,453,922,089,000 |
ac9b4d39a3642214929ac55904c61c063d41b785 | The error should print the name of the variable.
Change: 113179600 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,921,411,000 |
e36590b4d5bd41da82fb8d756236bba351b9b000 | TensorFlow event accumulator test: remove comparison of exact string
logging in test, which breaks test in OSS.
Change: 113182827 | Vijay Vasudevan | vrv@google.com | 1,453,922,968,000 |
7b4fe0e89bb1bc1b3f7f4c2fb07bcfe81e5454ff | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,453,923,136,000 |
f2c935a2ca1fdbf2e9838ae4c2baac1c7ce6ece6 | Merge branch 'master' into patch-1 | jmtatsch | tatsch@gmail.com | 1,453,929,591,000 |
697084c97b880f6845a9158a348f12f4e0ed8d35 | Fix bugs in TensorArray gradients + unit tests.
Two major bugs:
* Multiple calls to tf.gradients will create multiple TensorArrayWrites to the
same gradient slot from the exact same gradient source. This gets treated as
a write + an add, and as a result gradients of TensorArrayRead can be
double-counted. The solution is to create a separate TensorArray for each
call to tf.gradients. This can be done by:
1. looking at the name of the input gradient to e.g. TensorArrayRead
2. slicing off the prefix (e.g. "gradients", "gradients_1", etc)
3. passing this to the tensor_array_grad op which
4. uses this as a suffix to the original name when creating or looking up a new
TensorArray object for storing the gradients.
* The initial gradient TensorArrayWrite performed a shallow copy of the
write tensor. Since we support aggregation to the same slot from different
sources, modifying the PersistentTensor in place can affect the original
tensor elsewhere.
Instead, we now specifically disallow multiple reads / packs from a TensorArray.
This simplifies the code immensely and removes the need to support gradient
aggregation. It also makes the interface much more functional. Once a
Tensor has been read out of the TensorArray, it can be used in several places.
However, gradient aggregation can be performed outside the TensorArray.
Additional improvements:
* TensorArray constructor is now "read_once", which means after a Read or
Pack operation, the reference inside TensorArray to the Tensor is removed. This
frees up memory early.
Change: 113193579 | Eugene Brevdo | ebrevdo@gmail.com | 1,453,929,472,000 |
c2722a1fc6c1656ec24c5602c5f30e420473ebbc | - Added optimizer_options field to GraphOptions, moved graph optmization
options there.
- Deprecated the existing skip_common_subexpression_elimination field.
Change: 113194182 | Manjunath Kudlur | keveman@gmail.com | 1,453,929,890,000 |
0cee1d3dac5961a08594c3734eea14fc490f3250 | Add a new 'log message' event type.
This will be used in a subsequent change to allow Python code to log to the
events file so that it can be viewed in TensorBoard and so logging events can be
associated with models more easily.
Change: 113194631 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,930,165,000 |
c3c27f275f94af6e32ecaff66a6bf439abdaff5b | TensorFlow: ASSERT_OK and EXPECT_OK are also defined in other projects
that are built with TensorFlow (protobuf), so prefix our macros with
TF_ to make them project specific.
Change: 113197186 | Vijay Vasudevan | vrv@google.com | 1,453,931,648,000 |
d6ede40403b6ec63972afd71b5c340f8625f5458 | Fix for broken OSS build
Change: 113197985 | Vijay Vasudevan | vrv@google.com | 1,453,932,109,000 |
c68f145ab35d548d2460a2f69e42365c11cf44f3 | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,453,932,539,000 |
bdd5120e18b8c39686f333d3dc44792348eb314d | Run Tensorflow inference in a background background thread so that the camera preview frames are never blocked, improving perceived demo performance.
Add Systrace events to preview frames and inference calls.
Change: 113200652 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,933,473,000 |
e59493941cf1f9ff82dc01499005bdfa28842f60 | Factor TensorArray functionality into its own python files.
Change: 113202608 | Eugene Brevdo | ebrevdo@gmail.com | 1,453,934,536,000 |
fea55e1e05ffbdaa4ae4369d1c35e689a7dc48a2 | Breaking change in TF RNN python api: Return the final state instead of the
list of states when calling tf.nn.rnn() and tf.nn.state_saving_rnn()
This is necessary for further cleanup of RNN state propagation code
(currently dynamic RNN calculations when passing sequence_length do not return
the proper final state, this is a necessary fix to make that fix efficient).
Change: 113203893 | Eugene Brevdo | ebrevdo@gmail.com | 1,453,935,294,000 |
6aac6d70793a7a7ce8e7c71c87e5cb610cf0f9af | Adds Graph::num_nodes() and Graph::num_edges() for convenience.
Change: 113209117 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,938,342,000 |
33aa62872774539f2a8144db0c93a1b9ce3ed51f | Fixed constant folding to handle nodes with multiple outputs.
Change: 113212117 | Manjunath Kudlur | keveman@gmail.com | 1,453,940,052,000 |
5515a4977ba9c461cfa7d07ce5e5cda8348baf46 | Restore correct memory registration discipline via visitors
in gpu_bfc_allocator, which was broken by cl/112976894.
Change: 113213941 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,941,178,000 |
bcd9722be4250b8584e4fe5bc4f60b8793cf87d0 | Fixed constant folding to handle nodes with multiple outputs.
Change: 113215834 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,942,487,000 |
a5736ad6248cbb538984f2eee844b526a3534bd1 | Fix broken build.
Change: 113216002 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,942,605,000 |
19bcd451e6001958cb9660eda4c3831b3656b99c | Fix broken build.
Change: 113217181 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,943,441,000 |
c509c78d0f28c90f45edd65f7c3bdf080d8e0283 | Fixed constant folding to handle nodes with multiple outputs.
Change: 113220692 | Manjunath Kudlur | keveman@gmail.com | 1,453,946,418,000 |
182bc43407d24ef9d1cd44f70726a847731599d9 | Allow input_size != num_units in GRUCell and BasicLSTMCell.
Change: 113221259 | Lukasz Kaiser | lukaszkaiser@gmail.com | 1,453,946,892,000 |
72bf5024e2fc8a4979cef71b1dd2cfb2916bf165 | Adds an "align_corners" option to the ResizeNearestNeighbor (+its gradient), ResizeBicubic, and ResizeArea ops, allowing them to exactly align the 4 corners of input and output. The new option defaults to false, thus the previous boundary handling behavior (identical to OpenCV) remains the default.
Change: 113221298 | A. Unique TensorFlower | nobody@tensorflow.org | 1,453,946,933,000 |
44f318d32893853e8e9e6a74535e0aeaadbc084b | Merge commit for internal changes | Vijay Vasudevan | vrv@google.com | 1,453,964,989,000 |
faf747a15d4efc8fff03a10a3fdb37393197c2d3 | Global search & replace to move to the new location for
tensorflow/core/ files and build targets.
Change: 113248293 | Josh Levenberg | josh11b@tensorflow.org | 1,453,982,547,000 |
14cd77baeb0ee6f4b40816afc1f8c9c5d186bb8e | Merge commit for internal changes | Manjunath Kudlur | keveman@gmail.com | 1,453,994,801,000 |
b3c3d177d025b3a4f1490184658e9a170e0a6b15 | Merge branch 'master' into patch-1 | jmtatsch | tatsch@gmail.com | 1,454,035,991,000 |
83adeaf3ade77c93e70bc1da4812561476c4141f | Use integer division in fourth notebook | Christian Jauvin | cjauvin@gmail.com | 1,454,041,030,000 |
1d51a185acc3f8bd75b349e1c21f5c85b8600e75 | Add missing parens for print stmt in 5th notebook | Christian Jauvin | cjauvin@gmail.com | 1,454,041,077,000 |
2209d03619b6c3ab092b4c088cf5a6d95aca8b35 | Add missing parens for print stmt in 1rst notebook | Christian Jauvin | cjauvin@gmail.com | 1,454,041,353,000 |
be7d4d13b04b2df35f164a6354e49b51a8b9a3db | ci_build: improve docker run extra parameter handling | Jan Prach | jendap@google.com | 1,454,059,159,000 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.