提交 · ead4fda06535ce547d014fba1656ae53f0b64996 · xxadev / tensorflow

16 10月, 2018 23 次提交
- C
  Fixes a bug in tf.train.Saver(), where classes using the `VARIABLE_VALUE_KEY`... · ead4fda0
  由 Chris Jones 提交于 10月 15, 2018
```
Fixes a bug in tf.train.Saver(), where classes using the `VARIABLE_VALUE_KEY` used different naming in the checkpoint file when `var_list` was a dict.

PiperOrigin-RevId: 217182136
```
  ead4fda0
- T
  Merge pull request #22616 from lanhin:master · 1142f7b8
  由 TensorFlower Gardener 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217180014
```
  1142f7b8
- T
  Merge pull request #22836 from Intel-tensorflow:agramesh/remove_mkl_ml_only_code · 2f7ce05b
  由 TensorFlower Gardener 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217179947
```
  2f7ce05b
- T
  Merge pull request #22868 from antmicro:bluepill-robot-tests · f2f61818
  由 TensorFlower Gardener 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217179854
```
  f2f61818
- T
  Merge pull request #22768 from knightXun:patch-1 · 36a93800
  由 TensorFlower Gardener 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217176593
```
  36a93800
- P
  Automated rollback of commit 6fa6bd04 · 3f23f4dd
  由 Peter Hawkins 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217173355
```
  3f23f4dd
- J
  Add speech frontend. · 333c4ff5
  由 Jian Li 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217171519
```
  333c4ff5
- M
  Fix export for bipartite_match. · ec16f981
  由 Martin Wicke 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217170873
```
  ec16f981
- P
  Replace references to tensorflow::StringPiece with absl::string_view. No functional changes. · 6fa6bd04
  由 Peter Hawkins 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217170781
```
  6fa6bd04
- A
  Bugfixes. · 09c208bd
  由 A. Unique TensorFlower 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217168657
```
  09c208bd
- P
  [TF:XLA] Fix memory leak in XLA on-demand mode. · 4371a684
  由 Peter Hawkins 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217168404
```
  4371a684
- A
  Internal change. · 1573f9f2
  由 A. Unique TensorFlower 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217166408
```
  1573f9f2
- M
  [XLA] Take padded dimensions into account when verifying shape size. · 788bb3b9
  由 Michael Kuperstein 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217164296
```
  788bb3b9
- E
  Automated rollback of commit 64628112 · 3febecc5
  由 Eugene Brevdo 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217163302
```
  3febecc5
- D
  Fix broken call and add test coverage for it. · 70a97970
  由 Dan Moldovan 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217161539
```
  70a97970
- R
  Allow empty GCS tokens to be cached. · ab5ba2aa
  由 Russell Power 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217159671
```
  ab5ba2aa
- B
  Fix copy/paste error in the docstrings of assertAll{Less,Greater}{,Equal}. · bb271165
  由 Ben Zinberg 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217157127
```
  bb271165
- T
  Merge pull request #21066 from dougalsutherland:kid · 84135114
  由 TensorFlower Gardener 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217155008
```
  84135114
- A
  Add a missing PySequence_Fast in the eager fast path · aba365a8
  由 Allen Lavoie 提交于 10月 15, 2018
```
Correctly handles things that pretend to be lists, like the ListWrappers we use for variable tracking.

Fixes #22853.

PiperOrigin-RevId: 217154951
```
  aba365a8
- A
  Internal change · a57b7902
  由 A. Unique TensorFlower 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217153168
```
  a57b7902
- T
  Merge pull request #22533 from thunderfyc:master · 6e16d341
  由 TensorFlower Gardener 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217152548
```
  6e16d341
- A
  Enable DT_INT64 for the XLA implementation of DataFormatVecPermute. · e3fb5d03
  由 Adrian Kuegel 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217151877
```
  e3fb5d03
- A
  Remove incorrect automatic comment for namespace lookup · 3d9d0033
  由 A. Unique TensorFlower 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217151569
```
  3d9d0033
15 10月, 2018 7 次提交
- A
  Internal Change · 67e7aa1c
  由 A. Unique TensorFlower 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217148772
```
  67e7aa1c
- G
  New toolchains for remote build execution on windows. · a7d26471
  由 Gunhan Gulsoy 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217147663
```
  a7d26471
- A
  Add arbitrary axis support to ArgMax · c0ebb859
  由 A. Unique TensorFlower 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217147034
```
  c0ebb859
- A
  Avoid depending on init_tensorflow in benchmark_main. · 61081e22
  由 A. Unique TensorFlower 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217145701
```
  61081e22
- M
  Add fuzzer for StringSplitV2. · 55e97b31
  由 Mihai Maruseac 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217144630
```
  55e97b31
- A
  compat: Update forward compatibility horizon to 2018-10-15 · 63923f0b
  由 A. Unique TensorFlower 提交于 10月 15, 2018
```
PiperOrigin-RevId: 217108509
```
  63923f0b
- A
  [TF CTC] Fix Returns documentation stanzas to be properly formatted (see... · 6582efdc
  由 A. Unique TensorFlower 提交于 10月 15, 2018
```
[TF CTC] Fix Returns documentation stanzas to be properly formatted (see https://www.tensorflow.org/api_docs/python/tf/nn/ctc_greedy_decoder)

PiperOrigin-RevId: 217100215
```
  6582efdc
14 10月, 2018 8 次提交

A
compat: Update forward compatibility horizon to 2018-10-14 · c9e31601
由 A. Unique TensorFlower 提交于 10月 14, 2018
```
PiperOrigin-RevId: 217034982
```
c9e31601
G
Make gen_build_info genrule to use a py_binary. · b47a10d0
由 Gunhan Gulsoy 提交于 10月 13, 2018
```
This makes the script more platform independent.

PiperOrigin-RevId: 217028598
```
b47a10d0
T
Update error messages to reference post TF 1.7 symbol for `Input`. · f868981c
由 Tom Hennigan 提交于 10月 13, 2018
```
PiperOrigin-RevId: 217025180
```
f868981c
Y
Support real custom ops for Toco --allow_eager_ops flow. · e4b18328
由 Yu-Cheng Ling 提交于 10月 13, 2018
```
PiperOrigin-RevId: 217020295
```
e4b18328

Disables logging and summaries of steps, steps/s, loss etc. when... · 109a0c1b

由 A. Unique TensorFlower 提交于 10月 13, 2018

Disables logging and summaries of steps, steps/s, loss etc. when RunConfig.log_step_count_steps is None.

This is already how Estimator works (as one can see in line 2136 where we explicitly replace log_step_count_steps as None). It should be backward compatible as current code would have crashed if log_step_count_steps is None.

PiperOrigin-RevId: 217019985

109a0c1b

R
Rolling forward changing distribution strategies to use MultiDeviceIterator. · 47b04fdb
由 Rohan Jain 提交于 10月 13, 2018
```
The underlying issue with NaN's has now been resolved.

PiperOrigin-RevId: 217014692
```
47b04fdb

In TPUEstimator, allow input_fn to return features and labels of any arbitrary... · f667188d

由 A. Unique TensorFlower 提交于 10月 13, 2018

In TPUEstimator, allow input_fn to return features and labels of any arbitrary nested structure rather than just tensor or dict or tensors.

The current restriction seems to be historical. The code in _ModelFnWrapper uses nest flattens already, and then in _call_model_fn, 'features' and 'labels' are passed directly to the user code model_fn, so any nested structure should be supported. Before entering ModelFnWrapper, the _InputPipeline and _Inputs classes also both handle nested structure as well.

PiperOrigin-RevId: 217010800

f667188d

A
Removed #includes of protobuf full headers when TENSORFLOW_LITE_PROTOS is defined. · d84bfa45
由 A. Unique TensorFlower 提交于 10月 13, 2018
```
This requires #ifdef'ing out the code to use JSON parsing/serialization.

PiperOrigin-RevId: 217003132
```
d84bfa45

13 10月, 2018 2 次提交

A
compat: Update forward compatibility horizon to 2018-10-13 · 573985ca
由 A. Unique TensorFlower 提交于 10月 13, 2018
```
PiperOrigin-RevId: 216976797
```
573985ca

[XLA:GPU] Elide tuple roots of the entry computation · 0f13eb91

由 Benjamin Kramer 提交于 10月 12, 2018

The tuple buffer is never read, so stop emitting code to fill it. A typical
root tuple consists of a H2D memcpy and a host callback, both of which are
somewhat slow.

This helps tiny models and inference benchmarks, where the host/device syncs
can be a significant part of the runtime of the entire computation.

PiperOrigin-RevId: 216968475

0f13eb91

xxadev / tensorflow 与 Fork 源项目一致

xxadev / tensorflow
与 Fork 源项目一致