| 8f37f3002710129045a2e205678aee58a33489db |
|
12-Sep-2017 |
Peter Hawkins <phawkins@google.com> |
[TF:XLA] Cleanups to handling of arguments during XLA compilation:
* combine resource kinds in XlaCompiler::Argument::Kind, use a separate XlaResource::Kind field to distinguish different kinds of resource.
* merge XlaContext::HandleOrConstant and XlaExpression, which were almost identical.
* remove XlaContext::Argument; instead, build XlaExpressions directly from XlaCompiler and add them to the XlaContext.
PiperOrigin-RevId: 168439341
/external/tensorflow/tensorflow/compiler/tf2xla/kernels/arg_op.cc
|
| eb1fe50da445d3880b588215f6fadcc7f48dd3ff |
|
12-Jul-2017 |
Peter Hawkins <phawkins@google.com> |
[TF:XLA] Add initial implementation of the Stack operators to the TF/XLA bridge. Stacks are used when computing loop gradients.
PiperOrigin-RevId: 161659980
/external/tensorflow/tensorflow/compiler/tf2xla/kernels/arg_op.cc
|
| a66de1eca225bc95e7972974a7089d84df8a8055 |
|
16-Jun-2017 |
Peter Hawkins <phawkins@google.com> |
[TF:XLA] Refactor handling of Resources (Variables and TensorArrays) in the XLA bridge.
* Rename "Variable" to "Resource" in many places where non-Variable resources might be used.
* Add kTensorArray to the XlaCompiler::Argument enum. Remove kUninitializedVariable and make "initialized" a separate boolean field.
* Add a kind field to XlaResource. Add checks that Variables are not used where TensorArrays are expected, and vice-versa.
* Clean ups to the TensorArray operators.
PiperOrigin-RevId: 159244478
/external/tensorflow/tensorflow/compiler/tf2xla/kernels/arg_op.cc
|
| c19e6cac0413b0b93d5a15f9d4dc7c861aa1c734 |
|
07-Jun-2017 |
Peter Hawkins <phawkins@google.com> |
[TF:XLA] Initial implementation of TensorArray ops.
The XLA implementation of TensorArrays is more restrictive than regular TensorArrays:
* XLA TensorArrays must have dynamic_size=False.
* all elements in an XLA TensorArray must have the same shape.
* writes always add their values to any existing values; neither reads nor writes ever issue errors. Out-of-bounds writes currently wrap.
Refactor Variable handling in the TF/XLA bridge. Use a XlaVariable* to refer to variables inside compilation rather than a numerical ID. Allow for variables that don't correspond to variables known to the user. Also use XlaVariable to handle TensorArrays.
PiperOrigin-RevId: 158322041
/external/tensorflow/tensorflow/compiler/tf2xla/kernels/arg_op.cc
|
| 2e307b457dfc4df1d82d568b71f9a796bd218084 |
|
19-Apr-2017 |
Peter Hawkins <phawkins@google.com> |
[TF:XLA] Avoid expanding large constants during compilation. Recognize constants that consist of a repeated scalar, and generate an XLA Broadcast of a scalar instead.
Split tf2xla/kernels/declaration_op.cc into two separate files (arg_op.cc and const_op.cc).
Change: 153604131
/external/tensorflow/tensorflow/compiler/tf2xla/kernels/arg_op.cc
|