experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Remove unnecessary namespacing

Browse Source 
PiperOrigin-RevId: 294298556
Change-Id: I2c3d1eb46aa150a8165f502c3e6a08b99b4377c9
This commit is contained in:
Gaurav Jain 2020-02-10 14:02:49 -08:00 committed by TensorFlower Gardener
parent fcd06dbbeb
commit 9759636453
2 changed files with 41 additions and 56 deletions
tensorflow/core/common_runtime/eager
eager_operation.cceager_operation.h

26
tensorflow/core/common_runtime/eager/eager_operation.cc
View File

@ -53,8 +53,7 @@ Status EagerOperation::Reset(
return SetDeviceName(raw_device_name, true); return SetDeviceName(raw_device_name, true);
} }
tensorflow::Status EagerOperation::MaybeInferSingleInputAttrs( Status EagerOperation::MaybeInferSingleInputAttrs(TensorHandle* handle) {
TensorHandle* handle) {
if (!op_def_) return Status::OK(); if (!op_def_) return Status::OK();
const auto& input_def = op_def_->input_arg(inference_arg_idx_++); const auto& input_def = op_def_->input_arg(inference_arg_idx_++);
@ -78,8 +77,7 @@ tensorflow::Status EagerOperation::MaybeInferSingleInputAttrs(
} }
void EagerOperation::InferSingleTypeInputListAttrs( void EagerOperation::InferSingleTypeInputListAttrs(
const tensorflow::OpDef::ArgDef& input_def, const OpDef::ArgDef& input_def, const DataType dtype, int num_inputs) {
const tensorflow::DataType dtype, int num_inputs) {
if (inference_attrs_.find(input_def.number_attr()) == if (inference_attrs_.find(input_def.number_attr()) ==
inference_attrs_.end()) { inference_attrs_.end()) {
MutableAttrs()->Set(input_def.number_attr(), num_inputs); MutableAttrs()->Set(input_def.number_attr(), num_inputs);
@ -92,24 +90,23 @@ void EagerOperation::InferSingleTypeInputListAttrs(
} }
void EagerOperation::InferMixedTypeInputListAttrs( void EagerOperation::InferMixedTypeInputListAttrs(
const tensorflow::OpDef::ArgDef& input_def, const OpDef::ArgDef& input_def, const std::vector<DataType>& dtypes) {
const std::vector<tensorflow::DataType>& dtypes) {
if (inference_attrs_.find(input_def.type_list_attr()) == if (inference_attrs_.find(input_def.type_list_attr()) ==
inference_attrs_.end()) { inference_attrs_.end()) {
MutableAttrs()->Set(input_def.type_list_attr(), MutableAttrs()->Set(
tensorflow::gtl::ArraySlice<const tensorflow::DataType>( input_def.type_list_attr(),
dtypes.data(), dtypes.size())); gtl::ArraySlice<const DataType>(dtypes.data(), dtypes.size()));
inference_attrs_.insert(input_def.type_list_attr()); inference_attrs_.insert(input_def.type_list_attr());
} }
} }
tensorflow::Status EagerOperation::InferInputListAttrs(int num_inputs) { Status EagerOperation::InferInputListAttrs(int num_inputs) {
if (!op_def_) return Status::OK(); if (!op_def_) return Status::OK();
int start = inference_arg_idx_; int start = inference_arg_idx_;
const auto& input_def = op_def_->input_arg(inference_arg_idx_++); const auto& input_def = op_def_->input_arg(inference_arg_idx_++);
if (!input_def.type_list_attr().empty()) { if (!input_def.type_list_attr().empty()) {
std::vector<tensorflow::DataType> dtypes(num_inputs); std::vector<DataType> dtypes(num_inputs);
for (int i = 0; i < num_inputs; ++i) { for (int i = 0; i < num_inputs; ++i) {
dtypes[i] = inputs_[start + i]->dtype; dtypes[i] = inputs_[start + i]->dtype;
} }
@ -118,13 +115,12 @@ tensorflow::Status EagerOperation::InferInputListAttrs(int num_inputs) {
!input_def.number_attr().empty()) { !input_def.number_attr().empty()) {
InferSingleTypeInputListAttrs(input_def, inputs_[start]->dtype, num_inputs); InferSingleTypeInputListAttrs(input_def, inputs_[start]->dtype, num_inputs);
} else { } else {
return tensorflow::errors::InvalidArgument("Invalid input list definition"); return errors::InvalidArgument("Invalid input list definition");
} }
return tensorflow::Status::OK(); return Status::OK();
} }
tensorflow::Status EagerOperation::SetDeviceName(const char* device, Status EagerOperation::SetDeviceName(const char* device, const bool reset) {
const bool reset) {
if (device != nullptr && strlen(device) > 0) { if (device != nullptr && strlen(device) > 0) {
if (device != raw_device_name_) { if (device != raw_device_name_) {
if (!DeviceNameUtils::ParseFullName(device, &device_parsed_name_)) { if (!DeviceNameUtils::ParseFullName(device, &device_parsed_name_)) {

67
tensorflow/core/common_runtime/eager/eager_operation.h
View File

@ -30,7 +30,7 @@ class EagerOperation {
public: public:
explicit EagerOperation(tensorflow::EagerContext* ctx) : ctx_(*ctx) {} explicit EagerOperation(tensorflow::EagerContext* ctx) : ctx_(*ctx) {}
~EagerOperation() { ~EagerOperation() {
for (tensorflow::TensorHandle* h : inputs_) { for (TensorHandle* h : inputs_) {
h->Unref(); h->Unref();
} }
} }
@ -39,15 +39,15 @@ class EagerOperation {
// Clear(), and then Reset(...) with the same arguments that would have // Clear(), and then Reset(...) with the same arguments that would have
// been provided to the constructor. // been provided to the constructor.
void Clear() { void Clear() {
for (tensorflow::TensorHandle* h : inputs_) { for (TensorHandle* h : inputs_) {
h->Unref(); h->Unref();
} }
inputs_.clear(); inputs_.clear();
ClearInferenceState(); ClearInferenceState();
} }
tensorflow::Status Reset(const char* op, const char* raw_device_name, Status Reset(const char* op, const char* raw_device_name, bool remote,
bool remote, EagerExecutor* executor, EagerExecutor* executor,
const absl::optional<EagerRemoteFunctionParams> const absl::optional<EagerRemoteFunctionParams>
remote_func_params = absl::nullopt); remote_func_params = absl::nullopt);
@ -55,25 +55,19 @@ class EagerOperation {
tensorflow::EagerContext& EagerContext() { return ctx_; } tensorflow::EagerContext& EagerContext() { return ctx_; }
tensorflow::AttrBuilder* MutableAttrs() { return &attrs_; } AttrBuilder* MutableAttrs() { return &attrs_; }
const tensorflow::AttrBuilder& Attrs() const { return attrs_; } const AttrBuilder& Attrs() const { return attrs_; }
const tensorflow::OpDef* OpDef() const { return op_def_; } const tensorflow::OpDef* OpDef() const { return op_def_; }
const tensorflow::gtl::InlinedVector<tensorflow::TensorHandle*, 4>& Inputs() const gtl::InlinedVector<TensorHandle*, 4>& Inputs() const { return inputs_; }
const { gtl::InlinedVector<TensorHandle*, 4>* MutableInputs() { return &inputs_; }
return inputs_;
}
tensorflow::gtl::InlinedVector<tensorflow::TensorHandle*, 4>*
MutableInputs() {
return &inputs_;
}
void AddInput(tensorflow::TensorHandle* h); void AddInput(TensorHandle* h);
void UpdateInput(int i, tensorflow::TensorHandle* h); void UpdateInput(int i, TensorHandle* h);
void ConsumeInput(tensorflow::TensorHandle* h); void ConsumeInput(TensorHandle* h);
const tensorflow::string& Name() const { return attrs_.op_name(); } const string& Name() const { return attrs_.op_name(); }
const tensorflow::AttrTypeMap* AttrTypes() const { return attr_types_; } const AttrTypeMap* AttrTypes() const { return attr_types_; }
tensorflow::Device* Device() const { return device_; } tensorflow::Device* Device() const { return device_; }
void SetDevice(tensorflow::Device* device) { void SetDevice(tensorflow::Device* device) {
@ -87,8 +81,7 @@ class EagerOperation {
const DeviceNameUtils::ParsedName& GetDeviceParsedName() const { const DeviceNameUtils::ParsedName& GetDeviceParsedName() const {
return device_parsed_name_; return device_parsed_name_;
} }
tensorflow::Status SetDeviceName(const char* device, Status SetDeviceName(const char* device, const bool reset = false);
const bool reset = false);
// Indicates whether the op is assigned to a device that is local to the // Indicates whether the op is assigned to a device that is local to the
// current host. // current host.
@ -116,7 +109,7 @@ class EagerOperation {
const char* op_name_ = nullptr; const char* op_name_ = nullptr;
#endif #endif
Status MaybeInferSingleInputAttrs(tensorflow::TensorHandle* handle); Status MaybeInferSingleInputAttrs(TensorHandle* handle);
Status InferInputListAttrs(int num_inputs); Status InferInputListAttrs(int num_inputs);
private: private:
@ -125,17 +118,15 @@ class EagerOperation {
inference_arg_idx_ = 0; inference_arg_idx_ = 0;
inference_attrs_.clear_no_resize(); inference_attrs_.clear_no_resize();
} }
void InferSingleTypeInputListAttrs(const tensorflow::OpDef::ArgDef& input_def, void InferSingleTypeInputListAttrs(const OpDef::ArgDef& input_def,
const tensorflow::DataType dtype, const DataType dtype, int num_inputs);
int num_inputs); void InferMixedTypeInputListAttrs(const OpDef::ArgDef& input_def,
void InferMixedTypeInputListAttrs( const std::vector<DataType>& dtypes);
const tensorflow::OpDef::ArgDef& input_def,
const std::vector<tensorflow::DataType>& dtypes);
tensorflow::EagerContext& ctx_; tensorflow::EagerContext& ctx_;
tensorflow::AttrBuilder attrs_; AttrBuilder attrs_;
const tensorflow::AttrTypeMap* attr_types_; const AttrTypeMap* attr_types_;
tensorflow::gtl::InlinedVector<tensorflow::TensorHandle*, 4> inputs_; gtl::InlinedVector<TensorHandle*, 4> inputs_;
tensorflow::Device* device_; tensorflow::Device* device_;
string raw_device_name_; string raw_device_name_;
string device_name_; string device_name_;
@ -150,19 +141,18 @@ class EagerOperation {
const tensorflow::OpDef* op_def_; // op definition from protobuf const tensorflow::OpDef* op_def_; // op definition from protobuf
int inference_arg_idx_; // arg definition index for the next input to be int inference_arg_idx_; // arg definition index for the next input to be
// added // added
tensorflow::gtl::FlatSet<std::string> gtl::FlatSet<std::string> inference_attrs_; // attributes inferred so far
inference_attrs_; // attributes inferred so far
}; };
inline void EagerOperation::AddInput(tensorflow::TensorHandle* h) { inline void EagerOperation::AddInput(TensorHandle* h) {
h->Ref(); h->Ref();
inputs_.push_back(h); inputs_.push_back(h);
attrs_.NumInputs(static_cast<int>(inputs_.size())); attrs_.NumInputs(static_cast<int>(inputs_.size()));
} }
inline void EagerOperation::UpdateInput(int i, tensorflow::TensorHandle* h) { inline void EagerOperation::UpdateInput(int i, TensorHandle* h) {
tensorflow::TensorHandle** slot = &inputs_[i]; TensorHandle** slot = &inputs_[i];
tensorflow::TensorHandle* existing = *slot; TensorHandle* existing = *slot;
if (existing != h) { if (existing != h) {
h->Ref(); h->Ref();
existing->Unref(); existing->Unref();
@ -170,11 +160,10 @@ inline void EagerOperation::UpdateInput(int i, tensorflow::TensorHandle* h) {
} }
} }
inline void EagerOperation::ConsumeInput(tensorflow::TensorHandle* h) { inline void EagerOperation::ConsumeInput(TensorHandle* h) {
inputs_.push_back(h); inputs_.push_back(h);
attrs_.NumInputs(static_cast<int>(inputs_.size())); attrs_.NumInputs(static_cast<int>(inputs_.size()));
} }
} // namespace tensorflow } // namespace tensorflow
#endif // TENSORFLOW_CORE_COMMON_RUNTIME_EAGER_EAGER_OPERATION_H_ #endif // TENSORFLOW_CORE_COMMON_RUNTIME_EAGER_EAGER_OPERATION_H_