Home
last modified time | relevance | path

Searched +full:node +full:- +full:fetch (Results 1 – 25 of 828) sorted by relevance

12345678910>>...34

/external/tensorflow/tensorflow/core/grappler/optimizers/
Dremapper_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
62 item.fetch = {"batch_norm"}; in TEST_F()
64 auto tensors_expected = EvaluateNodes(item.graph, item.fetch); in TEST_F()
71 auto tensors = EvaluateNodes(output, item.fetch); in TEST_F()
73 test::ExpectTensorNear<float>(tensors[0], tensors_expected[0], 1e-6); in TEST_F()
99 item.fetch = {"batch_norm"}; in TEST_F()
108 auto tensors_expected = EvaluateNodes(item.graph, item.fetch); in TEST_F()
110 auto tensors = EvaluateNodes(output, item.fetch); in TEST_F()
112 test::ExpectTensorNear<float>(tensors[0], tensors_expected[0], 1e-3); in TEST_F()
159 auto fetch = ops::Identity(s.WithOpName("fetch"), relu); in TEST_F() local
[all …]
Dmkl_remapper_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
78 auto fetch = s.WithOpName("fetch"); in FuseConv2DWithBiasAndAddNOrAdd() local
80 ops::Identity(fetch, ops::Relu(activate, addop)); in FuseConv2DWithBiasAndAddNOrAdd()
82 ops::Identity(fetch, ops::Relu6(activate, addop)); in FuseConv2DWithBiasAndAddNOrAdd()
84 ops::Identity(fetch, ops::Elu(activate, addop)); in FuseConv2DWithBiasAndAddNOrAdd()
86 ops::Identity(fetch, ops::internal::LeakyRelu(activate, addop)); in FuseConv2DWithBiasAndAddNOrAdd()
89 ops::Identity(fetch, addop); in FuseConv2DWithBiasAndAddNOrAdd()
114 item.fetch = {"fetch"}; in FuseConv2DWithBiasAndAddNOrAdd()
123 item.graph.mutable_node(i)->set_device("/device:CPU:0"); in FuseConv2DWithBiasAndAddNOrAdd()
136 for (const NodeDef& node : output.node()) { in FuseConv2DWithBiasAndAddNOrAdd() local
[all …]
Dauto_parallel.cc7 http://www.apache.org/licenses/LICENSE-2.0
36 NodeDef* node = graph_.add_node(); in AddNodeDivConst() local
37 node->set_name(strings::StrCat(kAutoParallelPrefix, "-Div-Const")); in AddNodeDivConst()
38 node->set_op("Const"); in AddNodeDivConst()
42 node->mutable_attr()->insert({"dtype", attr_data_type}); in AddNodeDivConst()
46 tensor->add_float_val(static_cast<float>(num_replicas_)); in AddNodeDivConst()
47 tensor->set_dtype(DT_FLOAT); in AddNodeDivConst()
48 node->mutable_attr()->insert({"value", attr_tensor}); in AddNodeDivConst()
49 return node; in AddNodeDivConst()
54 NodeDef* node = graph_.add_node(); in AddNodeDiv() local
[all …]
Dconstant_folding_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
83 item.fetch = {"mul1", "mul2", "add1", "add2"}; in SimpleNeutralElementTest()
93 const NodeDef& node = output.node(i); in SimpleNeutralElementTest() local
94 const string& name = node.name(); in SimpleNeutralElementTest()
96 EXPECT_EQ("Const", node.op()); in SimpleNeutralElementTest()
97 EXPECT_EQ("^x", node.input(0)); in SimpleNeutralElementTest()
98 EXPECT_EQ("^zeros", node.input(1)); in SimpleNeutralElementTest()
100 EXPECT_EQ(snapshot_or_identity, node.op()); in SimpleNeutralElementTest()
101 EXPECT_EQ("x", node.input(0)); in SimpleNeutralElementTest()
102 EXPECT_EQ("^ones", node.input(1)); in SimpleNeutralElementTest()
[all …]
Darithmetic_optimizer_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
58 // Optimized name of outer Mul node by HoistCommonFactorOutOfAggregation.
63 // Optimized name of outer Div node by HoistCommonFactorOutOfAggregation.
68 // Optimized name of inner Add node by HoistCommonFactorOutOfAggregation.
73 // Optimized name of Const node by SimplifyAggregation.
78 // Optimized name of Mul node by SimplifyAggregation.
87 const NodeDef& original = original_graph.node(i); in VerifyGraphsMatch()
88 const NodeDef& optimized = optimized_graph.node(i); in VerifyGraphsMatch()
123 item.fetch = {"output"}; in TEST_F()
127 auto expected = EvaluateNodes(item.graph, item.fetch, {{"input", tensor}}); in TEST_F()
[all …]
Ddependency_optimizer_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
42 const NodeDef& original = original_graph.node(i); in VerifyGraphsEqual()
43 const NodeDef& optimized = optimized_graph.node(i); in VerifyGraphsEqual()
81 item.fetch.push_back("id1"); in TEST_F()
82 item.fetch.push_back("id2"); in TEST_F()
93 // The 'z' node should have been optimized away leaving only 5 nodes. in TEST_F()
96 for (const NodeDef& node : item.graph.node()) { in TEST_F() local
97 if (node.name() == "id1" || node.name() == "id2") { in TEST_F()
98 EXPECT_EQ(1, node.input_size()); in TEST_F()
99 EXPECT_EQ("add", node.input(0)); in TEST_F()
[all …]
Dpin_to_host_optimizer_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
57 item.fetch = {"a", "c", "d", "e", "f"}; in TEST_F()
60 auto tensors_expected = EvaluateNodes(item.graph, item.fetch); in TEST_F()
66 auto tensors = EvaluateNodes(item.graph, item.fetch); in TEST_F()
77 for (const NodeDef& node : output.node()) { in TEST_F() local
78 if (node.name() == "a" || node.name() == "c") { in TEST_F()
79 EXPECT_TRUE(node.device().empty()); in TEST_F()
80 } else if (node.name() == "d" || node.name() == "e" || node.name() == "f") { in TEST_F()
81 EXPECT_EQ(node.device(), "/device:CPU:0"); in TEST_F()
97 item.fetch = {"b"}; in TEST_F()
[all …]
Dshape_optimizer_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
45 item.fetch = {"e", "f"}; in TEST_F()
48 auto tensors_expected = EvaluateNodes(item.graph, item.fetch); in TEST_F()
55 for (const NodeDef& node : output.node()) { in TEST_F() local
56 if (node.name() == "e") { in TEST_F()
58 EXPECT_EQ("Size", node.op()); in TEST_F()
59 EXPECT_EQ("a", node.input(0)); in TEST_F()
60 } else if (node.name() == "f") { in TEST_F()
62 EXPECT_EQ("Prod", node.op()); in TEST_F()
63 EXPECT_EQ("c", node.input(0)); in TEST_F()
[all …]
Dmodel_pruner_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
87 std::vector<string> fetch = {"e"}; in TEST_F() local
88 auto expected_tensors = EvaluateNodes(item.graph, fetch); in TEST_F()
89 auto actual_tensors = EvaluateNodes(output, fetch); in TEST_F()
109 item.fetch.push_back("e"); in TEST_F()
128 auto actual_tensors = EvaluateNodes(output, item.fetch); in TEST_F()
130 auto expected_tensors = EvaluateNodes(item.graph, item.fetch); in TEST_F()
142 // Node "c" is pruned along with fanins of node "c". in TEST_F()
144 // Node "d" will be pruned because it only has control outputs. in TEST_F()
155 item.fetch = {"g", "h"}; in TEST_F()
[all …]
Dauto_mixed_precision_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
66 (random::New64() % 65536 / 65536.0) * (maxval - minval) + minval; in GenerateRandomTensorInRange()
76 const NodeDef& original = original_graph.node(i); in VerifyGraphsEquivalent()
115 device_properties.mutable_environment()->insert({"architecture", "7"}); in SetUp()
116 device_properties.mutable_environment()->insert({"cuda", "9010"}); in SetUp()
118 device_properties.mutable_environment()->insert( in SetUp()
124 TF_CHECK_OK(virtual_cluster_->Provision()); in SetUp()
127 void TearDown() override { TF_CHECK_OK(virtual_cluster_->Shutdown()); } in TearDown()
154 type_list.mutable_list()->add_type(DT_FLOAT); in AddSimpleNode()
182 item.fetch = {"fetch1"}; in TestSimpleUnaryInferOp()
[all …]
/external/tensorflow/tensorflow/core/grappler/
Dgrappler_item_builder.cc7 http://www.apache.org/licenses/LICENSE-2.0
58 auto flat = tensor->flat<float>(); in InitializeTensor()
64 auto flat = tensor->flat<int64_t>(); in InitializeTensor()
72 // Allocator will run non-trivial constructor/destructor for a Tensor with in InitializeTensor()
74 memset(const_cast<char*>(tensor->tensor_data().data()), 0, in InitializeTensor()
75 tensor->tensor_data().size()); in InitializeTensor()
86 item->graph = std::move(pruned_graph); in PruneGraph()
99 dim_proto.size() == -1) { in ReplaceUnknownShapeDim()
101 shape_pb_out->add_dim()->set_size( in ReplaceUnknownShapeDim()
105 shape_pb_out->add_dim()->set_size(dim_proto.size()); in ReplaceUnknownShapeDim()
[all …]
Dgrappler_item.cc7 http://www.apache.org/licenses/LICENSE-2.0
37 // prune all nodes that are not in the transitive fanin of the fetch nodes. in CreateOptOptionsForEager()
56 item.fetch = fetch; in WithGraph()
72 TF_CHECK_OK(ComputeTransitiveFanin(graph, fetch, &fanin_nodes)); in MainOpsFanin()
84 TF_CHECK_OK(ComputeTransitiveFanin(graph, fetch, &fanin_nodes)); in EnqueueOpsFanin()
98 for (const NodeDef* node : fanin) { in MainVariables() local
99 if (IsVariable(*node)) { in MainVariables()
100 vars.push_back(node); in MainVariables()
108 for (const string& f : fetch) { in NodesToPreserve()
109 VLOG(1) << "Add fetch " << f; in NodesToPreserve()
[all …]
/external/tensorflow/tensorflow/compiler/mlir/tensorflow/tests/graphdef2mlir/
Dtarget.pbtxt1 # RUN: tf-mlir-translate -graphdef-to-mlir -tf-enable-shape-inference-on-import=false %s -tf-contro…
2 …-mlir-translate -graphdef-to-mlir -tf-enable-shape-inference-on-import=false %s -tf-prune-unused-n…
3 …-mlir-translate -graphdef-to-mlir -tf-enable-shape-inference-on-import=false %s -tf-prune-unused-n…
14 node {
35 node {
64 node {
96 node {
115 node {
136 node {
166 # Tests single target node with no pruning set. All nodes will remain in the
[all …]
/external/tensorflow/tensorflow/compiler/tf2xla/
Dgraph_compiler_util.cc7 http://www.apache.org/licenses/LICENSE-2.0
47 typedef std::unordered_map<string, Node*> NodeMap;
49 // Each feed id identifies the positional output of some node, which may consist
52 // point from a new _Arg node instead. The newly created _Arg nodes are added to
57 std::unordered_set<const Node*>* arg_nodes) { in AddArgNodes()
65 auto node_it = node_map.find(remap_it->second); in AddArgNodes()
68 absl::string_view name(remap_it->second); in AddArgNodes()
72 "Node is fed but not needed for fetching: ", name); in AddArgNodes()
74 const Node* feed_node = node_it->second; in AddArgNodes()
80 Node* arg_node = nullptr; in AddArgNodes()
[all …]
Dtf2xla_util_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
52 feed->mutable_id()->set_node_name("foo"); in TEST()
53 feed->mutable_id()->set_output_index(123); in TEST()
54 feed->set_name("foo_debug"); in TEST()
56 feed->mutable_id()->set_node_name("bar"); in TEST()
57 feed->mutable_id()->set_output_index(0); in TEST()
58 tf2xla::Fetch* fetch = config.add_fetch(); in TEST() local
59 fetch->mutable_id()->set_node_name("baz"); in TEST()
60 fetch->mutable_id()->set_output_index(456); in TEST()
61 fetch->set_name("baz_debug"); in TEST()
[all …]
Dtf2xla.proto14 // index of a particular node in the graph. If the output of the named node
15 // feeds into other node(s), this corresponds to one or more edges. Otherwise
30 // contains this information. However, if the node being fed is an op that is
31 // not linked into the binary, then the type cannot be inferred from the node;
36 // Fetch represents a single fetch tensor in the graph, which corresponds to an
38 message Fetch { message
55 // Flag for variables that are never assigned. Assignments to a read-only
56 // variable or unassigned variables that are not read-only are invalid.
65 // Each fetch is a positional output argument for the generated computation.
67 repeated Fetch fetch = 2; field
/external/tensorflow/tensorflow/core/grappler/optimizers/data/
Dgraph_utils.cc7 http://www.apache.org/licenses/LICENSE-2.0
57 names[graph.node(i).name()] = i; in CreateNameIndex()
67 std::vector<int> CreateInputIndex(const NodeDef& node) { in CreateInputIndex() argument
69 for (int i = 0; i < node.input_size(); ++i) { in CreateInputIndex()
70 inputs[node.input(i)] = i; in CreateInputIndex()
72 std::vector<int> index(node.input_size()); in CreateInputIndex()
83 NodeDef node; in AddScalarConstNodeHelper() local
84 node.set_op(kConstOpName); in AddScalarConstNodeHelper()
85 SetUniqueGraphNodeName(kConstOpName, graph->graph(), &node); in AddScalarConstNodeHelper()
87 (*node.mutable_attr())["dtype"].set_type(dtype); in AddScalarConstNodeHelper()
[all …]
Duse_private_thread_pool.cc7 http://www.apache.org/licenses/LICENSE-2.0
47 if (item.fetch.size() != 1) { in OptimizeAndCollectStats()
49 "Expected only one fetch node but there were ", item.fetch.size(), ": ", in OptimizeAndCollectStats()
50 absl::StrJoin(item.fetch, ", ")); in OptimizeAndCollectStats()
53 for (const NodeDef& node : item.graph.node()) { in OptimizeAndCollectStats() local
54 if (node.op() == kPrivateThreadPoolDataset) { in OptimizeAndCollectStats()
61 NodeDef* sink_node = graph.GetNode(item.fetch.at(0)); in OptimizeAndCollectStats()
70 if (last_node->op() == kModelDataset) { in OptimizeAndCollectStats()
74 // Add a const node with value 0 to indicate it is not set by users. in OptimizeAndCollectStats()
84 *insert_node.mutable_input()->Add() = last_node->name(); in OptimizeAndCollectStats()
[all …]
Ddisable_intra_op_parallelism.cc7 http://www.apache.org/licenses/LICENSE-2.0
53 if (item.fetch.size() != 1) { in OptimizeAndCollectStats()
55 "Expected only one fetch node but there were ", item.fetch.size(), ": ", in OptimizeAndCollectStats()
56 absl::StrJoin(item.fetch, ", ")); in OptimizeAndCollectStats()
59 for (const NodeDef& node : item.graph.node()) { in OptimizeAndCollectStats() local
61 if (node.op() == target_dataset_op) { in OptimizeAndCollectStats()
69 NodeDef* sink_node = graph.GetNode(item.fetch.at(0)); in OptimizeAndCollectStats()
78 if (last_node->op() == kModelDataset) { in OptimizeAndCollectStats()
82 // Add a const node with value 1 in OptimizeAndCollectStats()
92 *insert_node.mutable_input()->Add() = last_node->name(); in OptimizeAndCollectStats()
[all …]
Dslack.cc7 http://www.apache.org/licenses/LICENSE-2.0
39 bool IsDatasetNodeOfType(const NodeDef& node, in IsDatasetNodeOfType() argument
42 if (node.op() == dataset_op_name) return true; in IsDatasetNodeOfType()
82 if (dataset_node->op() == kPrefetchDatasetOp) { in RecursivelyHandleOp()
84 (*dataset_node->mutable_attr())["slack_period"].set_i(slack_period_); in RecursivelyHandleOp()
96 for (int i = 0; i < dataset_node->input_size(); ++i) { in RecursivelyHandleOp()
124 if (item.fetch.size() != 1) { in OptimizeAndCollectStats()
126 "Expected only one fetch node but there were ", item.fetch.size(), ": ", in OptimizeAndCollectStats()
127 absl::StrJoin(item.fetch, ", ")); in OptimizeAndCollectStats()
129 // Walks the input pipeline backwards from the fetch node to find the last in OptimizeAndCollectStats()
[all …]
/external/tensorflow/tensorflow/core/grappler/clusters/
Dsingle_machine_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
53 TF_CHECK_OK(cluster_->EnablePeakMemoryStats()); in SetUp()
54 TF_CHECK_OK(cluster_->Provision()); in SetUp()
59 TF_CHECK_OK(cluster_->Shutdown()); in TearDown()
69 CHECK_EQ("single_machine", cluster_->type()); in TEST_F()
74 cluster_->GetDeviceNames()); in TEST_F()
78 TF_CHECK_OK(cluster_->Initialize(item)); in TEST_F()
81 const int64_t start_micros = Env::Default()->NowMicros(); in TEST_F()
82 TF_CHECK_OK(cluster_->Run(item.graph, item.feed, item.fetch, &metadata)); in TEST_F()
84 Env::Default()->NowMicros() - start_micros; in TEST_F()
[all …]
Dsingle_machine.cc7 http://www.apache.org/licenses/LICENSE-2.0
56 options_.config.add_session_inter_op_thread_pool()->set_num_threads( in SingleMachine()
66 // Reset the thread-pool so that there are no outstanding Session::Run(...)s in ~SingleMachine()
84 TF_RETURN_IF_ERROR(session_->ListDevices(&devices)); in Provision()
125 mutex_lock l(this->last_graph_mu_); in Initialize()
139 mutex_lock l(this->last_graph_mu_); in Shutdown()
148 const std::vector<string>& fetch, in Run() argument
150 mutex_lock l(this->last_graph_mu_); in Run()
153 TF_RETURN_IF_ERROR(session_->Create(graph_def)); in Run()
162 for (auto node : *init_metadata_.mutable_cost_graph()->mutable_node()) { in Run()
[all …]
/external/tensorflow/tensorflow/core/common_runtime/
Dgraph_execution_state.cc7 http://www.apache.org/licenses/LICENSE-2.0
102 if (options.session_options->config.graph_options().place_pruned_graph() || in MakeForBaseGraph()
103 !options.session_options->config.experimental() in MakeForBaseGraph()
112 if (!options.session_options->config.graph_options().place_pruned_graph()) { in MakeForBaseGraph()
114 TF_RETURN_IF_ERROR(ConvertGraphDefToGraph({}, *ret->original_graph_def_, in MakeForBaseGraph()
116 TF_RETURN_IF_ERROR(ret->InitBaseGraph(std::move(base_graph))); in MakeForBaseGraph()
125 TF_RETURN_IF_ERROR(ret->InitBaseGraph(std::move(base_graph))); in MakeForBaseGraph()
137 if (!(base_execution_state.session_options_->config.graph_options() in MakeForPrunedGraph()
139 options.session_options->config.graph_options().place_pruned_graph())) { in MakeForPrunedGraph()
150 "the Session-level `GraphExecutionState`."); in MakeForPrunedGraph()
[all …]
/external/tensorflow/tensorflow/c/experimental/grappler/
Dgrappler_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
37 params->struct_size = TP_OPTIMIZER_REGISTRATION_PARAMS_STRUCT_SIZE; in PopulateDefaultParam()
38 params->optimizer_configs->struct_size = TP_OPTIMIZER_CONFIGS_STRUCT_SIZE; in PopulateDefaultParam()
39 params->optimizer->struct_size = TP_OPTIMIZER_STRUCT_SIZE; in PopulateDefaultParam()
40 params->optimizer->create_func = nullptr; in PopulateDefaultParam()
41 params->optimizer->optimize_func = optimize_func; in PopulateDefaultParam()
42 params->optimizer->destroy_func = nullptr; in PopulateDefaultParam()
47 TF_Status* const status) -> void { in TEST()
50 params->device_type = "Success"; in TEST()
51 params->optimizer_configs->remapping = TF_TriState_Off; in TEST()
[all …]
/external/tensorflow/tensorflow/core/graph/
Dsubgraph_test.cc7 http://www.apache.org/licenses/LICENSE-2.0
58 Node* FindNode(const string& name) { in FindNode()
59 for (Node* n : g_->nodes()) { in FindNode()
60 if (n->name() == name) return n; in FindNode()
70 for (Node* n : g_->nodes()) { in ExpectNodes()
71 if (n->IsOp()) { in ExpectNodes()
73 actual_nodes.push_back(n->name()); in ExpectNodes()
83 Node* n = FindNode(s); in ExpectNodes()
85 if (n->type_string() == "_Send" || n->type_string() == "_Recv") { in ExpectNodes()
86 EXPECT_EQ(device_info_.name(), n->assigned_device_name()) << s; in ExpectNodes()
[all …]

12345678910>>...34