Skip to content
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.

Sign up for GitHub

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

Jump to bottom

Branch 201392512 #20156

Merged
merged 200 commits into from
Jun 21, 2018
Merged

Branch 201392512 #20156

merged 200 commits into from
Jun 21, 2018

Conversation

mingxingtan
Copy link
Contributor

No description provided.

Akshay Modi and others added 30 commits June 15, 2018 10:26
@tensorflower-gardener
Minor python change to remove doing unnecessary work in resource vari…
655c52b 
…ables

PiperOrigin-RevId: 200735157
@tensorflower-gardener
[TF:XLA] Validate the control flow structure in encapsulate_subgraphs…
c9a2034 
…_pass and encapsulate_tpu_computations_pass, in order to detect errors earlier.

PiperOrigin-RevId: 200735435
@pavithrasv @tensorflower-gardener
Increase gru_test test size
fa6e9f3 
PiperOrigin-RevId: 200736300
@hawkinsp @tensorflower-gardener
[TF:XLA] Update comment on xla_compiler.h to match the code.
6f7c83c 
Make resource_var.h more widely visible and add comment about the correct lock acquisition order if locking multiple variables.

PiperOrigin-RevId: 200736416
@tensorflower-gardener
Fix a bug in dependency optimizer: Repeated inputs would not get conv…
32e85d4 
…erted to control inputs when converting nodes to NoOps.

PiperOrigin-RevId: 200740844
@gunan @tensorflower-gardener
Disable long running tests in fastbuild mode.
d63d663 
PiperOrigin-RevId: 200741660
@pavithrasv @tensorflower-gardener
Fix: DepthwiseConv2D fails when bias is enabled
1ca4b6f 
PiperOrigin-RevId: 200742104
@d0k @tensorflower-gardener
[XLA] Switch PostOrder accessors to use std::vector instead of std::l…
b62d76d 
…ist.

std::list is just hilariously inefficient and the postorder list creation has
been rewritten not to not depend on splicing anymore so there's no need for the
list. While there remove the old unused postorder list creation code.
PiperOrigin-RevId: 200743677
@meheffernan @tensorflower-gardener
Add test of TOKEN primitive type which uses conditionals.
45d7a04 
PiperOrigin-RevId: 200745718
@tensorflower-gardener
[XLA] Don't implement kCrossReplicaSum case in HloInstruction::Identi…
8ba25e3 
…calSlowPath.

PiperOrigin-RevId: 200746735
@yk5 @tensorflower-gardener
contrib.timeseries: sets the predictions dict in EstimatorSpec for ev…
a7fcc5d 
…aluation op.

PiperOrigin-RevId: 200747192
@k-w-w @tensorflower-gardener
Refactor loader.load function into a class that splits the graph load…
916c0aa 
…ing and variable restoration steps.

PiperOrigin-RevId: 200747752
@kayzhu @tensorflower-gardener
[TF2XLA] Remove the last unncessary host-to-device memcpy, and remove…
f9b832d 
… the

HostTensorToLiteral function completely to prevent potential future misuse of
unnecessary memcpy.

PiperOrigin-RevId: 200750664
@toponado-zz @tensorflower-gardener
Support model parallelism in PER_HOST_V2 input pipeline.
a601d9a 
PiperOrigin-RevId: 200751151
@aaroey @tensorflower-gardener
Add DeviceSet to SingleMachine, so we can use the OptimizeGraph() too…
1ba31da 
…l to call

tensorrt optimizer (which requires access to the Device) to create a
transformed GraphDef.

PiperOrigin-RevId: 200751174
@tensorflower-gardener
Broad refactoring (part 2): Introduce a module dedicated to AutoGraph…
03e3310 
…-specific conversion logic: base converter classes, context objects, gensym. Largely, these are pulled out from impl and pyct.

This CL only adds the module - a future CL will replace existing implementations with these.

PiperOrigin-RevId: 200751782
@tensorflower-gardener
Reuse duplicated reference ops in optimized_ops.h
64a81c5 
PiperOrigin-RevId: 200753184
@tensorflower-gardener
Fix segfault in ConstantFolding::MaterializeShapes when the first inp…
d09b1eb 
…ut to TensorArraySizeV3 is a Placeholder.

PiperOrigin-RevId: 200755274
@tensorflower-gardener
Provides a more fine-grained option for each thread to control
d07d47d 
fork-join parallelism (e.g., Eigen/ThreadpoolDevice or Shard).

PiperOrigin-RevId: 200756626
@hawkinsp @tensorflower-gardener
Add some extra DebugString() functions to shape_inference. Currently …
c783b56 
…unused, but they were useful while debugging.

Open visibility of the low level gen_resource_variables_ops to compiler tests.
Fix bug in shape function of TPUReplicateInput for resource variables ? MergeInputHandleShapesAndTypes does not report shape mismatches.

PiperOrigin-RevId: 200756762
@tensorflower-gardener
Set shapes and types to queue ops, if not set by enqueue ops.
79f52c1 
PiperOrigin-RevId: 200764324
@dimvar @tensorflower-gardener
Typo fixes.
1645a0a 
PiperOrigin-RevId: 200766687
@lxuechen @tensorflower-gardener
Fix None grads bug when calling a keras Sequential twice on same inpu…
817c39b 
…t in graph

mode.

PiperOrigin-RevId: 200768236
@MarkDaoust @tensorflower-gardener
Quiet the doc generator.
c295688 
Delete most print statements, use logging instead of print, and close files (to clear the "Unclosed file" warnings).

Normally this produces thousands of lines of output. Mostly noise.

PiperOrigin-RevId: 200769210
@tensorflower-gardener
Add bazel android repo to workspace
d3ae8e7 
PiperOrigin-RevId: 200771096
@rryan @tensorflower-gardener
Fix typo in tf.lite Python interpreter comment.
7991f01 
PiperOrigin-RevId: 200774484
@kayzhu @tensorflower-gardener
Automated g4 rollback of changelist 200750664
33f8f7e 
PiperOrigin-RevId: 200777514
@skye @tensorflower-gardener
Move cond_v2 to core (non-public) and add toggle to use cond_v2 by de…
94b3db6 
…fault.

This change:
* Creates a new global variable, control_flow_ops._ENABLE_COND_V2, to use
  cond_v2 by default when calling tf.cond. This variable can also be
  controlled via the environment variable TF_ENABLE_COND_V2.

* Moves cond_v2 out of contrib so it's accessible from control_flow_ops.py.

* Lazily "imports" some modules in cond_v2 to avoid circular dependencies.
  Note that these lazy "imports" must be imported by the cond_v2 caller (or
  recursively by one of the caller's imports) in order for cond_v2 to have
  access to them.

* Renames the cond_v2 module to cond_v2_impl, and creates a new cond_v2 module
  that imports the cond_v2 method and the necessary extra imports. This is
  useful for explicitly calling cond_v2 outside of control_flow_ops.cond.

PiperOrigin-RevId: 200778208
@tensorflower-gardener
Reflow comments; NFC
5e9a39d 
PiperOrigin-RevId: 200783258
@reedwm @tensorflower-gardener
Automatic cast layer inputs to the layer's dtype.
b8861af 
This makes it more convenient to use layer of different dtypes in a model. Instead of having to manually cast intermediate tensors between layers of different dtypes, they will automatically be casted.

This is also useful for the upcoming mixed precision API.

PiperOrigin-RevId: 200783477
Dan Moldovan and others added 8 commits June 20, 2018 11:40
@tensorflower-gardener
Boilerplate for an ANF transformer. This is not currently related to …
e51df59 
…AutoGraph, but used elsewhere.

PiperOrigin-RevId: 201387308
@jdduke @tensorflower-gardener
Implement TFLite Shape operator
4efefb9 
PiperOrigin-RevId: 201389618
@tensorflower-gardener
Fix a bug: the conversion of pure Conv to DepthwiseConv did not
c40ed1d 
properly check the necessary precondition that the input depth is 1.

PiperOrigin-RevId: 201389819
@tensorflower-gardener
Make evaluate() work on anything that has a numpy() method in eager t…
1f7d5c3 
…ests.

PiperOrigin-RevId: 201390698
@alextp @tensorflower-gardener
SymbolicGradient for some resource variables.
5988a74 
Currently assumes variables are floats; there are TODOs to rectifiy this.

PiperOrigin-RevId: 201391092
@shashishekhar @tensorflower-gardener
Fix gradle build for TFLite Android example.
5d773dd 
PiperOrigin-RevId: 201391220
@tensorflower-gardener
Split GradientBoostedDecisionTreeModel.train() to three steps.
4fdb7cc 
1) Update stats
2) Update the number of examples visited.
3) If the number of examples reaches the target, grow the tree.

PiperOrigin-RevId: 201392512
@mingxingtan
Merge commit for internal changes
8f1f0a8
@mingxingtan mingxingtan added the kokoro:force-run Tests on submitted change label Jun 20, 2018
@kokoro-team kokoro-team removed the kokoro:force-run Tests on submitted change label Jun 20, 2018
@mingxingtan mingxingtan added the kokoro:force-run Tests on submitted change label Jun 20, 2018
@kokoro-team kokoro-team removed the kokoro:force-run Tests on submitted change label Jun 20, 2018
@mingxingtan mingxingtan added the kokoro:force-run Tests on submitted change label Jun 21, 2018
@kokoro-team kokoro-team removed the kokoro:force-run Tests on submitted change label Jun 21, 2018
@mingxingtan mingxingtan added the kokoro:force-run Tests on submitted change label Jun 21, 2018
@kokoro-team kokoro-team removed the kokoro:force-run Tests on submitted change label Jun 21, 2018
@mingxingtan mingxingtan added the kokoro:force-run Tests on submitted change label Jun 21, 2018
@kokoro-team kokoro-team removed the kokoro:force-run Tests on submitted change label Jun 21, 2018
@mingxingtan mingxingtan requested a review from yifeif June 21, 2018 05:34
@yifeif
Copy link
Contributor

yifeif commented Jun 21, 2018

@mdanatg do you mind taking a look at the tensorflow/contrib/autograph/converters:decorators_test failure?

@mdanatg
Copy link

mdanatg commented Jun 21, 2018

@yifeif Looking now. Looks like a missing dependency that somehow I missed.

@mingxingtan
Copy link
Contributor Author

Thanks @mdanatg and @yifeif
@mdanatg, this is blocking our push, could you prioritize it and keep me posted? Thanks!

@yifeif
Disable decorators_test for pip.
f5ce4d8 
tensorflow/contrib/autograph/converters:decorators_test uses generated code, by when private symbols have been stripped.
@mingxingtan mingxingtan merged commit 9d2d400 into tensorflow:master Jun 21, 2018
@mingxingtan mingxingtan deleted the branch_201392512 branch June 21, 2018 18:56
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@yifeif yifeif yifeif approved these changes

Assignees

@mingxingtan mingxingtan

Labels
cla: yes
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

None yet