| .. |
|
data
|
Annotate FunctionDef registered from eager runtime to correct annotate tensorflow::Graph's construction context.
|
2020-12-11 14:41:33 -08:00 |
|
arithmetic_optimizer_test_utils.h
|
[tf.sparse_segment_*()] Grappler optimization to avoid casts for indices/segment_ids.
|
2020-04-25 14:36:01 -07:00 |
|
arithmetic_optimizer_test.cc
|
Fixed incorrect arithmetic optimizer application.
|
2020-09-22 11:30:03 -07:00 |
|
arithmetic_optimizer.cc
|
Fixed incorrect arithmetic optimizer application.
|
2020-09-22 11:30:03 -07:00 |
|
arithmetic_optimizer.h
|
[tf.sparse_segment_*()] Grappler optimization to avoid casts for indices/segment_ids.
|
2020-04-25 14:36:01 -07:00 |
|
auto_mixed_precision_lists.h
|
Enable depthwise convs in auto_mixed_precision
|
2020-08-13 20:33:02 +10:00 |
|
auto_mixed_precision_test.cc
|
Replace instances of "blacklist" with "denylist" where possible. See Google Developer guidelines at https://developers.google.com/style/word-list#blacklist for more information.
|
2020-07-20 16:05:25 -07:00 |
|
auto_mixed_precision.cc
|
Merge pull request #43613 from Intel-tensorflow:gyshi/fix-automix-precision-batchNorm-bug
|
2020-09-30 21:28:08 -07:00 |
|
auto_mixed_precision.h
|
The auto-mixed-precision pass is actually run before TF-TRT while we want it to run after TF-TRT, see b/170160323.
|
2020-10-06 10:02:47 -07:00 |
|
auto_parallel_test.cc
|
C++ API: run shape inference as nodes are constructed
|
2017-08-15 16:50:32 -07:00 |
|
auto_parallel.cc
|
Cleanup of transitive-fanin node computation to make details of invalid input available to its callers.
|
2020-04-20 16:18:52 -07:00 |
|
auto_parallel.h
|
[Grappler] Pass function library stub to function library oblivious optimizers.
|
2019-08-13 15:28:31 -07:00 |
|
BUILD
|
Bring back PR #44629.
|
2020-11-06 13:26:00 -08:00 |
|
common_subgraph_elimination_test.cc
|
[Grappler] Split common subgraph elimination out into a separate Grappler optimizer. This allows us to run it before constant folding and function optimizer to reduce the size of the graph earlier and save time spent in grappler. In particular, this reduces the size of the graph for which we run static shape inference, which is very expensive.
|
2020-03-06 12:38:21 -08:00 |
|
common_subgraph_elimination.cc
|
replacing 'iter_limit' with 'end'
|
2020-07-10 15:03:43 +00:00 |
|
common_subgraph_elimination.h
|
[Grappler] Split common subgraph elimination out into a separate Grappler optimizer. This allows us to run it before constant folding and function optimizer to reduce the size of the graph earlier and save time spent in grappler. In particular, this reduces the size of the graph for which we run static shape inference, which is very expensive.
|
2020-03-06 12:38:21 -08:00 |
|
constant_folding_test.cc
|
Fold broadcast select
|
2020-09-14 10:34:20 -07:00 |
|
constant_folding.cc
|
Fold broadcast select
|
2020-09-14 10:34:20 -07:00 |
|
constant_folding.h
|
Fold broadcast select
|
2020-09-14 10:34:20 -07:00 |
|
custom_graph_optimizer_registry_test.cc
|
[Grappler] Pass function library stub to function library oblivious optimizers.
|
2019-08-13 15:28:31 -07:00 |
|
custom_graph_optimizer_registry.cc
|
Add custom registered graph optimizers run by MetaOptimizer.
|
2018-02-23 16:08:55 -08:00 |
|
custom_graph_optimizer_registry.h
|
Enable XLA fusions as a Grappler optimization.
|
2018-06-04 19:00:21 -07:00 |
|
custom_graph_optimizer.h
|
For custom optimizers, keep a copy of Session ConfigProto_.
|
2020-02-03 12:12:10 -08:00 |
|
debug_stripper_test.cc
|
[debug_stripper] Replace PrintV2 with NoOp instead of Identity.
|
2020-02-10 12:15:33 -08:00 |
|
debug_stripper.cc
|
Merge pull request #40903 from tg-at-google:sign-compare-warning-fixes-batch-6
|
2020-07-15 16:35:51 -07:00 |
|
debug_stripper.h
|
[Grappler] Pass function library stub to function library oblivious optimizers.
|
2019-08-13 15:28:31 -07:00 |
|
dependency_optimizer_test.cc
|
Dependency optimizer groups cross-host control edges to reduce RPC traffic.
|
2019-11-08 12:11:09 -08:00 |
|
dependency_optimizer.cc
|
[Grappler] Make Grappler play nice with other graph rewriting passes by preserving custom attributes.
|
2020-07-07 15:11:12 -07:00 |
|
dependency_optimizer.h
|
Dependency optimizer groups cross-host control edges to reduce RPC traffic.
|
2019-11-08 12:11:09 -08:00 |
|
evaluation_utils_test.cc
|
Changing the number of threads in DeviceSimple to 2 in order to avoid thread creation overhead.
|
2019-10-09 13:14:26 -07:00 |
|
evaluation_utils.cc
|
Changing the number of threads in DeviceSimple to 2 in order to avoid thread creation overhead.
|
2019-10-09 13:14:26 -07:00 |
|
evaluation_utils.h
|
|
|
|
function_api_info_test.cc
|
Handle the case where api_implements exists but api_prefered_device does not exist.
|
2020-06-10 12:37:21 -07:00 |
|
function_api_info.cc
|
Handle the case where api_implements exists but api_prefered_device does not exist.
|
2020-06-10 12:37:21 -07:00 |
|
function_api_info.h
|
Make implementation_selector as core grappler optimizer.
|
2019-02-15 11:45:51 -08:00 |
|
function_optimizer_test.cc
|
Colocate input identities with inputs when inlining multi-device functions
|
2019-11-15 17:52:22 -08:00 |
|
function_optimizer.cc
|
Add CollectiveBcastSend/RecvV2 which takes input tensors rather than attributes.
|
2020-12-17 10:45:19 -08:00 |
|
function_optimizer.h
|
Globally disable lowering of functional control flow when using SINGLE_THREADED_EXECUTOR.
|
2019-09-25 09:20:26 -07:00 |
|
generic_layout_optimizer_test.cc
|
Disabled NCHW layout optimization for integer convolutions.
|
2020-09-28 22:56:30 -07:00 |
|
generic_layout_optimizer_transposer_factory_test.cc
|
[Grappler] Add _FusedBatchNormEx to GenericLayoutOptimizer.
|
2019-07-19 18:56:45 -07:00 |
|
generic_layout_optimizer_transposer_factory.cc
|
Add IsBiasAddV2 to check BiasAdd op
|
2020-11-22 22:02:51 -08:00 |
|
generic_layout_optimizer_transposer_factory.h
|
[Grappler] Introduce new layout optimizer that is more flexible.
|
2019-06-13 16:35:47 -07:00 |
|
generic_layout_optimizer_transposer_test.cc
|
Handle int64 axis in ReduceTransposer
|
2020-07-23 12:13:57 -07:00 |
|
generic_layout_optimizer_transposer.cc
|
Merge pull request #45057 from kaixih:layout_opt_other_ops_pr_v2
|
2020-11-24 09:20:22 -08:00 |
|
generic_layout_optimizer_transposer.h
|
Merge pull request #45057 from kaixih:layout_opt_other_ops_pr_v2
|
2020-11-24 09:20:22 -08:00 |
|
generic_layout_optimizer.cc
|
Merge pull request #42448 from Intel-tensorflow:amin/nchw-nhwc-grappler-part1
|
2020-08-18 20:45:42 -07:00 |
|
generic_layout_optimizer.h
|
NCHW to NHWC conversion on CPU.
|
2020-08-17 23:05:35 -07:00 |
|
gpu_swapping_kernels.cc
|
Use optimized ops to handle GPU memory swapping: this avoids the need for 2
|
2018-02-27 14:33:33 -08:00 |
|
gpu_swapping_ops.cc
|
Use optimized ops to handle GPU memory swapping: this avoids the need for 2
|
2018-02-27 14:33:33 -08:00 |
|
graph_optimizer_stage_test.cc
|
Generate unique node names in AddEmptyNode by appending a suffix instead of crashing.
|
2020-07-21 17:05:33 -07:00 |
|
graph_optimizer_stage.cc
|
Optimize calls to std::string::find() and friends for a single char.
|
2020-12-17 11:22:16 -08:00 |
|
graph_optimizer_stage.h
|
[Grappler] Avoid copying tensors in arithmetic optimizer
|
2019-08-20 15:49:09 -07:00 |
|
graph_optimizer.h
|
[Grappler] Remove several gratuitous graph and function library copies from the Grappler-related callstacks in TensorFlow.
|
2020-03-06 16:13:10 -08:00 |
|
implementation_selector_test.cc
|
generate stateless_case op if all ops in all branches are stateless
|
2020-07-30 16:23:47 -07:00 |
|
implementation_selector.cc
|
generate stateless_case op if all ops in all branches are stateless
|
2020-07-30 16:23:47 -07:00 |
|
implementation_selector.h
|
Add a function to dynamic choose and execute the proper implementation based on underlying device placement
|
2020-05-28 09:54:31 -07:00 |
|
layout_optimizer_test.cc
|
Disabled NCHW layout optimization for integer convolutions.
|
2020-09-28 22:56:30 -07:00 |
|
layout_optimizer.cc
|
Add SelectV2 to list of format agnostic ops
|
2020-10-16 12:11:05 +01:00 |
|
layout_optimizer.h
|
[Grappler] Pass function library stub to function library oblivious optimizers.
|
2019-08-13 15:28:31 -07:00 |
|
loop_optimizer_test.cc
|
move function DisableAllStages from protected to private
|
2019-12-11 19:55:11 +08:00 |
|
loop_optimizer.cc
|
[Grappler]
|
2019-12-10 16:14:59 -08:00 |
|
loop_optimizer.h
|
[Grappler] Pass function library stub to function library oblivious optimizers.
|
2019-08-13 15:28:31 -07:00 |
|
memory_optimizer_test.cc
|
Merge pull request #36639 from ROCmSoftwarePlatform:google-upstream-grappler
|
2020-03-11 23:59:13 -07:00 |
|
memory_optimizer.cc
|
[Grappler] Make Grappler play nice with other graph rewriting passes by preserving custom attributes.
|
2020-07-07 15:11:12 -07:00 |
|
memory_optimizer.h
|
[Grappler] Pass function library stub to function library oblivious optimizers.
|
2019-08-13 15:28:31 -07:00 |
|
meta_optimizer_test.cc
|
Refactoring and test coverage improvements
|
2020-10-23 11:49:44 -07:00 |
|
meta_optimizer.cc
|
(NFC) Added an experimental config proto field for TFRT execution.
|
2020-12-08 10:09:23 -08:00 |
|
meta_optimizer.h
|
(NFC) Added an experimental config proto field for TFRT execution.
|
2020-12-08 10:09:23 -08:00 |
|
mkl_remapper_test.cc
|
feat: add MatMul+BiasAdd+Add fusion
|
2020-12-02 14:53:27 +08:00 |
|
model_pruner_test.cc
|
Prune redundant control inputs early in model_pruner, since they may prevent deletion of trivial nodes.
|
2020-05-20 10:59:58 -07:00 |
|
model_pruner.cc
|
[-Wsign-compare] warning fixes batch 6
|
2020-06-29 00:13:20 +00:00 |
|
model_pruner.h
|
[Grappler] Pass function library stub to function library oblivious optimizers.
|
2019-08-13 15:28:31 -07:00 |
|
pin_to_host_optimizer_test.cc
|
Rollback of rollback of disabling XLA:CPU/GPU devices by default.
|
2020-08-11 13:06:59 -07:00 |
|
pin_to_host_optimizer.cc
|
Replace instances of "blacklist" with "denylist" where possible. See Google Developer guidelines at https://developers.google.com/style/word-list#blacklist for more information.
|
2020-07-20 16:05:25 -07:00 |
|
pin_to_host_optimizer.h
|
[Grappler] Pass function library stub to function library oblivious optimizers.
|
2019-08-13 15:28:31 -07:00 |
|
remapper_test.cc
|
Bug fix for remapper test
|
2020-09-25 14:36:52 +08:00 |
|
remapper.cc
|
Merge pull request #45358 from Intel-tensorflow:yanzhang/matmul_biasadd_add_fusion
|
2020-12-14 10:05:53 -08:00 |
|
remapper.h
|
[Grappler] Pass function library stub to function library oblivious optimizers.
|
2019-08-13 15:28:31 -07:00 |
|
scoped_allocator_optimizer_test.cc
|
[Grappler] Before calling Grappler from the TensorFlow runtime, check that feeds and fetches are valid and only refer to nodes that exist in the graph.
|
2020-09-16 10:13:01 -07:00 |
|
scoped_allocator_optimizer.cc
|
Optimize calls to std::string::find() and friends for a single char.
|
2020-12-17 11:22:16 -08:00 |
|
scoped_allocator_optimizer.h
|
Fix handling of PartitionedCall in ShapeInference pass. Previously, the pass
|
2020-04-28 13:35:11 -07:00 |
|
shape_optimizer_test.cc
|
|
|
|
shape_optimizer.cc
|
Update tensorflow/core/grappler/optimizers/shape_optimizer.cc
|
2020-06-24 15:35:54 -04:00 |
|
shape_optimizer.h
|
[Grappler] Pass function library stub to function library oblivious optimizers.
|
2019-08-13 15:28:31 -07:00 |
|
static_schedule_test.cc
|
[Grappler] Allow feeding non-zero index tensors to Grappler optimizer
|
2020-04-27 11:45:41 -07:00 |
|
static_schedule.cc
|
Add an option to avoid copying TensorProtos in shape inference when not needed. This saves a significant amount of memory and time in various optimizers requiring static shape inference. Calling GraphProperties::InferStatically would make at least one copy of all constant tensors stored in the graph. This CL adds a new argument to InferStatically, allowing callers to skip copying tensor values to InputProperties and OutputProperties for optimizers that only need shape and datatype information.
|
2019-05-30 12:50:23 -07:00 |
|
static_schedule.h
|
Remove THIRD_PARTY_ from #include guards
|
2018-01-24 14:31:28 -08:00 |
fd0d7123b1