- 17 3月, 2017 9 次提交
-
-
由 A. Unique TensorFlower 提交于
Change: 150343323
-
由 Gunhan Gulsoy 提交于
Change: 150341207
-
由 Petros Mol 提交于
Change: 150340985
-
由 Peter Hawkins 提交于
Change: 150336339
-
由 Vijay Vasudevan 提交于
Change: 150335248
-
由 Brennan Saeta 提交于
In order to move to a more flexible ClusterSpec and distributed runtime, it is important to stop referencing the singleton worker_cache on the MasterEnv. This change refactors the internal distributed runtime code to avoid referencing MasterEnv->worker_cache and instead pass it explicitly, and independently of the process-wide singleton MasterEnv. Change: 150331649
-
由 A. Unique TensorFlower 提交于
Change: 150328131
-
由 Eugene Brevdo 提交于
This includes: * Support for variational recurrent dropout via the flag variational_recurrent. * Adding the ability to dropout state, inputs, and/or outputs * Setting a seed now sets a separate seed for each mask of state, input, output Change: 150325470
-
由 A. Unique TensorFlower 提交于
and remove duplicate code in hlo_graph_dumper Change: 150324515
-
- 16 3月, 2017 31 次提交
-
-
由 Ian Langmore 提交于
Change: 150320911
-
由 A. Unique TensorFlower 提交于
Change: 150320765
-
由 A. Unique TensorFlower 提交于
to have unknown shape function for cases where the shape function is incorrectly returning no outputs. Change: 150315933
-
由 Martin Wicke 提交于
Change: 150292011
-
由 Vinu Rajashekhar 提交于
Change: 150289918
-
由 Martin Wicke 提交于
Change: 150289601
-
由 A. Unique TensorFlower 提交于
This makes it possible to use this function for protos with Any fields. Change: 150286292
-
由 A. Unique TensorFlower 提交于
Change: 150278280
-
由 A. Unique TensorFlower 提交于
Change: 150273663
-
由 A. Unique TensorFlower 提交于
This vulcanized output was created after running manually modifying the vulcanize.js task script in various ways, running gulp vulcanize, and then manually updating links to third-party JS (such as d3 and plottable) to point to the correct component directories. This process will be documented and streamlined in the future. Change: 150270700
-
由 A. Unique TensorFlower 提交于
The underlying function mechanism uses LocalExecutor to call the function, which interacts poorly with the LocalExecutor used by tf2xla to translate the TF graph into XLA. Change: 150268961
-
由 Mark Daoust 提交于
Delete OpMatcher and op_type from guide (they have been deleted from the public api). Change: 150267754
-
由 A. Unique TensorFlower 提交于
In SingleMachine::Initialize(), check also GrapplerItem.id not to skip initialization when the new graphdef's address is accidentally same to the last graphdef executed. Change: 150266564
-
由 Yangzihao Wang 提交于
Added a helper function DimensionReduction() to reduce the dimensions in a tensor shape according to the input permutation. Outputs the reduced-dimension tensor shape and new permutation in two arrays. Change: 150266318
-
由 A. Unique TensorFlower 提交于
Change: 150265441
-
由 Suharsh Sivakumar 提交于
Change: 150265300
-
由 Derek Murray 提交于
Fixes #3388. Change: 150265211
-
由 A. Unique TensorFlower 提交于
Change: 150263118
-
由 A. Unique TensorFlower 提交于
Change: 150263070
-
由 Malcolm Reynolds 提交于
Change: 150262035
-
由 A. Unique TensorFlower 提交于
Also, lets the backend allocator specify behavior for 0-sized buffer allocations via StreamExecutorMemoryAllocator. Change: 150260500
-
由 A. Unique TensorFlower 提交于
Change: 150258994
-
由 Peter Hawkins 提交于
[XLA] Give Transpose its own Request, rather than piggybacking on ReshapeRequest. Avoids building unnecessary Reshape operators when Transpose was called by the client. Also avoids building Transpose operators when Reshape has identity transpose dimensions, for example when the client called the variant of ComputationBuilder::Reshape() that does not transpose. Makes the HLO graph emitted by the TF bridge more readable. Change: 150253949
-
由 Vijay Vasudevan 提交于
Change: 150251692
-
由 A. Unique TensorFlower 提交于
Change: 150243023
-
由 Peter Hawkins 提交于
[TF:XLA] Simplify the implementation and the emitted code for tf.reduce_mean(), by using division with broadcasting instead of an explicit Map(). Change: 150242743
-
由 Jianwei Xie 提交于
Change: 150241435
-
由 Francois Chollet 提交于
Change: 150237453
-
由 Gunhan Gulsoy 提交于
Change: 150237287
-
由 Peter Hawkins 提交于
[XLA] Fix crash when loading a serialized computation with dependencies between embedded computations. Change: 150233783
-
由 Asim Shankar 提交于
The older model uses some deprecated ops. While that still works, given that we have a new model, might as well use it. Change: 150233334
-