Nigel
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

!21635 [MSLITE] Modify output tensor names of ms model. 

Merge pull request !21635 from wangshaocong/convert_name
This commit is contained in:
i-robot 2021-08-12 08:58:17 +00:00 committed by Gitee
parent 9b3251f3ce e3d00ba8ca
commit bb406a14b5
25 changed files with 148 additions and 248 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
mindspore/lite/test/CMakeLists.txt
View File

@ -224,7 +224,6 @@ if(MSLITE_ENABLE_CONVERTER)
${LITE_DIR}/tools/optimizer/fusion/multi_head_attention_fusion.cc
${LITE_DIR}/tools/optimizer/fusion/reshape_reshape_fusion.cc
${LITE_DIR}/tools/optimizer/fusion/constant_folding_fusion.cc
${LITE_DIR}/tools/optimizer/fusion/quant_dtype_cast_fusion.cc
${LITE_DIR}/tools/optimizer/fusion/norm_fusion.cc
${LITE_DIR}/tools/optimizer/fusion/batchmatmul_fusion.cc
${LITE_DIR}/tools/optimizer/fusion/sigmoid_mul_fusion.cc

92
mindspore/lite/test/config/models_codegen.cfg
View File

@ -1,4 +1,4 @@
hiai_model_0909_kd_rot_ps_softmax.tflite
Modify_Out_hiai_model_0909_kd_rot_ps_softmax.tflite
# hiai_chinese_english_recognize_model_float32.tflite
# hiai_bigmodel_ghost_2_1_no_normalized_no_trans_tflite.tflite
# hiai_bigmodel_ghost_5_1_no_normalized_no_trans_tflite.tflite
@ -6,44 +6,44 @@ hiai_model_0909_kd_rot_ps_softmax.tflite
# hiai_model_normalize_object_scene_ps_20200519.tflite
# mtk_AADB_HADB_MBV2_model_fp32.tflite
# mtk_AADB_HADB_MBV3_model_fp32.tflite
mobilenet_v1_0.25_128.tflite
mobilenet_v1_0.25_160.tflite
mobilenet_v1_0.25_192.tflite
mobilenet_v1_0.25_224.tflite
mobilenet_v1_0.5_128.tflite
mobilenet_v1_0.5_160.tflite
mobilenet_v1_0.5_192.tflite
mobilenet_v1_0.5_224.tflite
mobilenet_v1_0.75_128.tflite
mobilenet_v1_0.75_160.tflite
mobilenet_v1_0.75_192.tflite
mobilenet_v1_0.75_224.tflite
mobilenet_v1_1.0_128.tflite
mobilenet_v1_1.0_160.tflite
mobilenet_v1_1.0_192.tflite
mobilenet_v1_1.0_224.tflite
mobilenet_v2_1.0_224.tflite
Modify_Out_mobilenet_v1_0.25_128.tflite
Modify_Out_mobilenet_v1_0.25_160.tflite
Modify_Out_mobilenet_v1_0.25_192.tflite
Modify_Out_mobilenet_v1_0.25_224.tflite
Modify_Out_mobilenet_v1_0.5_128.tflite
Modify_Out_mobilenet_v1_0.5_160.tflite
Modify_Out_mobilenet_v1_0.5_192.tflite
Modify_Out_mobilenet_v1_0.5_224.tflite
Modify_Out_mobilenet_v1_0.75_128.tflite
Modify_Out_mobilenet_v1_0.75_160.tflite
Modify_Out_mobilenet_v1_0.75_192.tflite
Modify_Out_mobilenet_v1_0.75_224.tflite
Modify_Out_mobilenet_v1_1.0_128.tflite
Modify_Out_mobilenet_v1_1.0_160.tflite
Modify_Out_mobilenet_v1_1.0_192.tflite
Modify_Out_mobilenet_v1_1.0_224.tflite
Modify_Out_mobilenet_v2_1.0_224.tflite
# mtk_model_normalize_object_scene_ps_20200519_f32.tflite
# mtk_model_ckpt.tflite
mtk_age_gender.tflite
Modify_Out_mtk_age_gender.tflite
# mtk_model_face_dress.tflite
# mtk_face_features_v1.tflite
# densenet.tflite
squeezenet.tflite
Modify_Out_squeezenet.tflite
# resnet_v2_101_299.tflite
# mnasnet_1.3_224.tflite
inception_v3.tflite
Modify_Out_inception_v3.tflite
# deeplabv3_257_mv_gpu.tflite
# multi_person_mobilenet_v1_075_float.tflite
# hiai_vad.tflite
# ide_label_base.tflite
# ide_label_retrained.tflite
ml_ei_headpose.tflite
Modify_Out_ml_ei_headpose.tflite
# ml_ei_landmark.tflite
mnist.tflite
mobilenet.tflite
resnet.tflite
scan_hms_angle1.tflite
Modify_Out_mnist.tflite
Modify_Out_mobilenet.tflite
Modify_Out_resnet.tflite
Modify_Out_scan_hms_angle1.tflite
# scan_hms_detect.tflite
# hiai_latin_ocr.tflite
# hiai_latin_ocr_1.tflite
@ -57,25 +57,25 @@ scan_hms_angle1.tflite
# hiai_ssd_mobilenetv2_object.tflite
# hiai_cv_focusShootOCRModel_02.tflite
# hiai_cv_poseEstimation.tflite
inception_v4.tflite
Modify_Out_inception_v4.tflite
# mtk_model_normalize_object_scene_ps_20200519_f16.tflite
# mtk_age_gender_fp16.tflite
# mtk_model_face_dress_fp16.tflite
mtk_AADB_HADB_MBV2_model_f16.tflite
Modify_Out_mtk_AADB_HADB_MBV2_model_f16.tflite
# mtk_AADB_HADB_MBV3_model_f16.tflite
# mtk_model_emotions_0725_fp16.tflite
# mtk_face_features_v1_fp16.tflite
# siteAI_digcom_AI_ECN.tflite
siteAI_digcom_g2v_keras.tflite
siteAI_trans_nonlinear.tflite
siteAI_trans_tcpclassify.tflite
siteAI_wireless_depress_w.tflite
siteAI_wireless_restore_w.tflite
Modify_Out_siteAI_digcom_g2v_keras.tflite
Modify_Out_siteAI_trans_nonlinear.tflite
Modify_Out_siteAI_trans_tcpclassify.tflite
Modify_Out_siteAI_wireless_depress_w.tflite
Modify_Out_siteAI_wireless_restore_w.tflite
# magenta_arbitrary-image-stylization-v1-256_fp16_prediction_1.tflite
# ml_object_detect.tflite
# ml_object_detect_1.tflite
hiai_cpu_face_emotion.tflite
hiai_cpu_face_gazing.tflite
Modify_Out_hiai_cpu_face_emotion.tflite
Modify_Out_hiai_cpu_face_gazing.tflite
# hiai_cpu_face_headpose.tflite
# hiai_humanDetection.tflite
# hiai_cv_focusShootOCRModel_08.tflite
@ -83,20 +83,20 @@ hiai_cpu_face_gazing.tflite
# hiai_face_model_npu.tflite
# hiai_ctpn_feature_map.tflite
# hiai_cv_labelDetectorModel_v2.tflite
hiai_cv_labelDetectorModel_v4.tflite
Modify_Out_hiai_cv_labelDetectorModel_v4.tflite
# hiai_dress_detect.tflite
# hiai_cv_saliencyDetectorModel.tflite
# hiai_frozen_inference_graph.tflite
# hiai_ghostnet.tflite
# hiai_iMaxDN_RGB.tflite
# hiai_iMaxSR_RGB.tflite
hiai_label_and_video.tflite
Modify_Out_hiai_label_and_video.tflite
# hiai_lm_inference_graph.tflite
efficientnet_lite0_fp32_2.tflite
efficientnet_lite1_fp32_2.tflite
efficientnet_lite2_fp32_2.tflite
efficientnet_lite3_fp32_2.tflite
efficientnet_lite4_fp32_2.tflite
Modify_Out_efficientnet_lite0_fp32_2.tflite
Modify_Out_efficientnet_lite1_fp32_2.tflite
Modify_Out_efficientnet_lite2_fp32_2.tflite
Modify_Out_efficientnet_lite3_fp32_2.tflite
Modify_Out_efficientnet_lite4_fp32_2.tflite
# mnasnet_0.50_224_1_metadata_1.tflite
# mnasnet_0.75_224_1_metadata_1.tflite
# mnasnet_1.0_128_1_metadata_1.tflite
@ -138,7 +138,7 @@ efficientnet_lite4_fp32_2.tflite
# ml_location.tflite
# ml_face_openclose_tflite.tflite
# ml_object_detect_pb2tflite.tflite
Q_AADB_HADB_MBV2_model.tflite
Modify_Out_Q_AADB_HADB_MBV2_model.tflite
# Q_convert.tflite
# Q_crnn_ori_75w_slim_norm_pb2tflite.tflite
# Q_crnn_ori_v2_405001_notrans_nopre_pb2tflite.tflite
@ -169,6 +169,6 @@ Q_AADB_HADB_MBV2_model.tflite
# text_classification.tflite
# Q_detect_fpn_add_inception-1448650.tflite
# Q_hand_0812_pb2tflite.tflite
intent_detect_hi_v2.tflite
simple_IPS_model_4D_input.onnx
rpnt_pdr_conv2d_16_fixed_last.onnx
Modify_Out_intent_detect_hi_v2.tflite
Modify_Out_simple_IPS_model_4D_input.onnx
Modify_Out_rpnt_pdr_conv2d_16_fixed_last.onnx

2
mindspore/lite/test/config/models_codegen_parallel.cfg
View File

@ -1 +1 @@
intent_detect_hi_v2.tflite
Modify_Out_intent_detect_hi_v2.tflite

4
mindspore/lite/test/config/models_gpu_fp32.cfg
View File

@ -101,7 +101,7 @@ Q_crnn_ori_75w_slim_norm.pb
Q_crnn_ori_v2_405001_notrans_nopre.pb
bolt_segment.pb
ml_location_lane_counter.onnx 2
gts_detect_5k_tf115.tflite
Modify_Out_gts_detect_5k_tf115.tflite
smartreply.tflite
ml_text_correction.tflite
ml_ocr_jk_pb2tflite.tflite
@ -152,7 +152,7 @@ hiai_iMaxDN_RGB.pb
hiai_iMaxSR_RGB.pb
hiai_lm_inference_graph.pb
hiai_PoseEstimation_Pcm.pb
hiai_model_0909_kd_rot_ps_softmax.tflite
Modify_Out_hiai_model_0909_kd_rot_ps_softmax.tflite
hiai_chinese_english_recognize_model_float32.tflite
hiai_bigmodel_ghost_2_1_no_normalized_no_trans_tflite.tflite
hiai_bigmodel_ghost_5_1_no_normalized_no_trans_tflite.tflite

2
mindspore/lite/test/config/models_npu.cfg
View File

@ -88,7 +88,7 @@ ml_video_edit_hair_dyeing_segmodel_v2 0.5
ml_video_edit_makeup_mobilenetv203.onnx 2
ml_video_edit_hairline_segmentation;3 0.5
ml_video_edit_hair_dyeing_migrate_v2.onnx;4 0.5
ml_audio_kit_encoder_v5.pb;6;1,32:1,32:1,32:1,32:1:1
Modify_Out_ml_audio_kit_encoder_v5.pb;6;1,32:1,32:1,32:1,32:1:1
fsr_270_mindspore.pb 1
fsr_360_mindspore.pb 1
fsr_720_mindspore.pb 1

4
mindspore/lite/test/config/models_tf.cfg
View File

@ -91,7 +91,7 @@ tacotron_encoder_stf.pb;5;1:1,62:1,62:1,62:1,62;;input_dependent
female_model_step2_int16_noiseout.pb;66
ml_female_model_step6_noiseout.pb;66
ml_male_model_step6_noiseout.pb;66
ml_tts_decoder_control_flow.pb;5
Modify_Out_ml_tts_decoder_control_flow.pb;5
ml_tts_decoder.pb;5
ml_tts_encoder_control_flow.pb;4;1:1,22:1:1;;input_dependent
ml_tts_vocoder.pb;66
@ -99,7 +99,7 @@ hiai_nlu_model.pb;3;1,16:1,16:1,16
gts_object_detect_Ics.pb;1;420,630,3;;input_dependent
hiai_transformer_encoder.pb;15
decoder_step_nocumsum_v5.pb;13;1:1,512:1,1429,2:1,127:1,127:1,127:1,127,320:1,80:1,512:1,512:1,512:1,512:1,512
ml_audio_kit_encoder_v5.pb;6;1,32:1,32:1,32:1,32:1:1
Modify_Out_ml_audio_kit_encoder_v5.pb;6;1,32:1,32:1,32:1,32:1:1
hiai_nlu_model_v1.pb;3;1,16:1,16:1,16 2.0
hiai_nlu_model_v2.pb;7;1,5:1,6:1,174:1,98:1,5:1,5:1,5
hiai_nlu_model_multi.pb;6;1,32:1,32:1,6:1,11:1,74:1,32

2
mindspore/lite/test/config/models_tf_fp16.cfg
View File

@ -81,7 +81,7 @@ ml_video_edit_oneclick_adaptis.pb;3 6
ml_female_model_step6_noiseout.pb;66 2
ml_male_model_step6_noiseout.pb;66 2.5
ml_tts_encoder_control_flow.pb;4;1:1,22:1:1 1.5
ml_tts_decoder_control_flow.pb;5 1
Modify_Out_ml_tts_decoder_control_flow.pb;5 1
ml_tts_decoder.pb;5 2.5
ml_tts_vocoder.pb;66 53
hiai_transformer_encoder.pb;15 4

30
mindspore/lite/test/config/models_tflite.cfg
View File

@ -1,11 +1,11 @@
hiai_model_0909_kd_rot_ps_softmax.tflite
hiai_chinese_english_recognize_model_float32.tflite
hiai_bigmodel_ghost_2_1_no_normalized_no_trans_tflite.tflite
hiai_bigmodel_ghost_5_1_no_normalized_no_trans_tflite.tflite
hiai_cn_recognize_modify_padv2.tflite
hiai_model_normalize_object_scene_ps_20200519.tflite
mtk_AADB_HADB_MBV2_model_fp32.tflite
mtk_AADB_HADB_MBV3_model_fp32.tflite
Modify_Out_hiai_model_0909_kd_rot_ps_softmax.tflite
Modify_Out_hiai_chinese_english_recognize_model_float32.tflite
Modify_Out_hiai_bigmodel_ghost_2_1_no_normalized_no_trans_tflite.tflite
Modify_Out_hiai_bigmodel_ghost_5_1_no_normalized_no_trans_tflite.tflite
Modify_Out_hiai_cn_recognize_modify_padv2.tflite
Modify_Out_hiai_model_normalize_object_scene_ps_20200519.tflite
#mtk_AADB_HADB_MBV2_model_fp32.tflite The model has three outputs, but the benchmark file has one output only.
Modify_Out_mtk_AADB_HADB_MBV3_model_fp32.tflite
mobilenet_v1_0.25_128.tflite
mobilenet_v1_0.25_160.tflite
mobilenet_v1_0.25_192.tflite
@ -112,7 +112,7 @@ lite-model_deeplabv3-mobilenetv2-float16_1_default_1.tflite
lite-model_east-text-detector_fp16_1.tflite
lite-model_cartoongan_fp16_1.tflite
lite-model_arbitrary-image-stylization-inceptionv3_fp16_predict_1.tflite
gts_detect_5k_tf115.tflite
Modify_Out_gts_detect_5k_tf115.tflite
mtk_isface.tflite
mtk_landmark.tflite
mtk_new_detect.tflite
@ -182,13 +182,13 @@ Q_iMaxSR_RGB_385_p_pb2tflite.tflite
bloom_new_detect.tflite
bloom_model_age_gender.tflite
bloom_isface.tflite
hiai_object_detect_814.tflite
hiai_object_tflite_graph_8bit.tflite
Modify_Out_hiai_object_detect_814.tflite
Modify_Out_hiai_object_tflite_graph_8bit.tflite
lma_tsec_shallow_channels16_ds2.1.1_model-best-f1.tflite
lite-model_arbitrary-image-stylization-inceptionv3_fp16_transfer_1.tflite;2
magenta_arbitrary-image-stylization-v1-256_fp16_transfer_1.tflite;2
albert_lite_base_squadv1_1.tflite;3
mobilebert_1_default_1.tflite;3
Modify_Out_albert_lite_base_squadv1_1.tflite;3
Modify_Out_mobilebert_1_default_1.tflite;3
ml_video_edit_img_segment_adaptise_pb2tflite.tflite;2
ml_video_edit_video_segment_gauss_adaptis_part2_pb2tflite.tflite;2
hdc_tb_cn_neg.tflite;3
@ -196,7 +196,7 @@ hiai_cv_labelDetectorModel_v3.tflite;2
ml_tacotron_decoder_step_stf.tflite;9;1,80:1,256:1,1024:1,1024:1,1024:1,1024:1,8:1,1,256:1
ml_headpose_pb2tflite.tflite;3;16:1,64,64,3:16
ml_ei_headpose_pb2tflite.tflite;3;16:1,64,64,3:16
lite-model_albert_lite_base_squadv1_metadata_1.tflite;3
lite-model_mobilebert_1_metadata_1.tflite;3
Modify_Out_lite-model_albert_lite_base_squadv1_metadata_1.tflite;3
Modify_Out_lite-model_mobilebert_1_metadata_1.tflite;3
hiai_vad.tflite;2
add_uint8.tflite;2

8
mindspore/lite/test/config/models_tflite_awaretraining.cfg
View File

@ -33,9 +33,9 @@ lite-model_on_device_vision_classifier_landmarks_classifier_oceania_antarctica_V
lite-model_on_device_vision_classifier_landmarks_classifier_europe_V1_1.tflite
lite-model_on_device_vision_classifier_landmarks_classifier_south_america_V1_1.tflite
vision_classifier_fungi_mobile_V1_1_default_1.tflite
detect.tflite
ssd_mobilenet_v1_1_default_1.tflite
object_detection_mobile_object_localizer_v1_1_default_1.tflite
gts_detect_0730_quant_frozen.tflite
Modify_Out_detect.tflite
Modify_Out_ssd_mobilenet_v1_1_default_1.tflite
Modify_Out_object_detection_mobile_object_localizer_v1_1_default_1.tflite
Modify_Out_gts_detect_0730_quant_frozen.tflite
gts_model_quant_frozen.tflite
inception_v2_224_quant.tflite

6
mindspore/lite/test/config/models_tflite_fp16.cfg
View File

@ -1,7 +1,7 @@
# [first column]:model_name;input_num;input_shape;threads;extra_info. If there is no need to set these parameters, the
# content after ";" can be omitted.
# [second column]:accuracy limit for float16 in arm64 device
hiai_model_0909_kd_rot_ps_softmax.tflite 10
Modify_Out_hiai_model_0909_kd_rot_ps_softmax.tflite 10
hiai_chinese_english_recognize_model_float32.tflite 13
hiai_bigmodel_ghost_2_1_no_normalized_no_trans_tflite.tflite 10
hiai_bigmodel_ghost_5_1_no_normalized_no_trans_tflite.tflite 10
@ -121,7 +121,7 @@ lite-model_deeplabv3-mobilenetv2-float16_1_default_1.tflite 60
lite-model_east-text-detector_fp16_1.tflite 60
lite-model_cartoongan_fp16_1.tflite 3
lite-model_arbitrary-image-stylization-inceptionv3_fp16_predict_1.tflite 6
gts_detect_5k_tf115.tflite 9.5
Modify_Out_gts_detect_5k_tf115.tflite 9.5
mtk_isface.tflite 0.2
mtk_landmark.tflite 0.3
mtk_new_detect.tflite 3
@ -212,7 +212,7 @@ bloom_model_age_gender.tflite 0.5
bloom_isface.tflite 0.5
# The output values of conv layers range from -e±5 to e±5, which almost reaches the representation limit of fp16. In
# this range, the fp16 data will has big bias. And the accumulation of this bias lowers the final precision.
hiai_object_detect_814.tflite 14
Modify_Out_hiai_object_detect_814.tflite 14
ml_video_edit_video_segment_gauss_adaptis_part2_pb2tflite.tflite;2 11
ml_video_edit_img_segment_adaptise_pb2tflite.tflite;2 0.5
hdc_tb_cn_neg.tflite;3 295

6
mindspore/lite/test/st/graph_test.cc
View File

@ -82,7 +82,7 @@ TEST_F(GraphTest, UserSetGraphOutput1) {
string name = out_data.first;
void *data = out_data.second;
float *fp32_data = reinterpret_cast<float *>(data);
if (name == "Stack-8") {
if (name == "output") {
output_count++;
ASSERT_LE(fabs(fp32_data[0] - (0.115831)), 0.01);
ASSERT_LE(fabs(fp32_data[1] - (0.113074)), 0.01);
@ -90,7 +90,7 @@ TEST_F(GraphTest, UserSetGraphOutput1) {
ASSERT_LE(fabs(fp32_data[3] - (0.346307)), 0.01);
ASSERT_LE(fabs(fp32_data[4] - (-0.15687)), 0.01);
}
if (name == "Stack-10") {
if (name == "output2") {
output_count++;
ASSERT_LE(fabs(fp32_data[0] - (0.06387864)), 0.01);
ASSERT_LE(fabs(fp32_data[1] - (0.22883008)), 0.01);
@ -98,7 +98,7 @@ TEST_F(GraphTest, UserSetGraphOutput1) {
ASSERT_LE(fabs(fp32_data[3] - (0.04586578)), 0.01);
ASSERT_LE(fabs(fp32_data[4] - (0.06820235)), 0.01);
}
if (name == "Stack-13") {
if (name == "output3") {
output_count++;
ASSERT_LE(fabs(fp32_data[0] - (-0.1617176)), 0.01);
ASSERT_LE(fabs(fp32_data[1] - (-0.3828573)), 0.01);

8
mindspore/lite/tools/anf_exporter/anf_exporter.cc
View File

@ -532,8 +532,12 @@ int AnfExporter::SetMetaGraphOutput(const FuncGraphPtr &func_graph,
auto &final_meta_graph = meta_graphT->subGraph.at(final_meta_graph_index);
meta_graphT->outputIndex.assign(final_meta_graph->outputIndices.begin(), final_meta_graph->outputIndices.end());
for (auto &output_index : meta_graphT->outputIndex) {
auto &tensor = meta_graphT->allTensors.at(output_index);
// set output tensor names to the original names, the output_names is null in nnie converter.
auto output_names = ConverterContext::GetInstance()->GetGraphOutputTensorNames();
MS_ASSERT(output_names.size() == meta_graphT->outputIndex.size());
for (size_t idx = 0; idx < output_names.size(); idx++) {
auto &tensor = meta_graphT->allTensors.at(meta_graphT->outputIndex.at(idx));
tensor->name = output_names.at(idx);
ConverterContext::GetInstance()->UpdateGraphOutputDType(meta_graphT->outputIndex.size(), tensor->dataType);
}

18
mindspore/lite/tools/common/node_util.cc
View File

@ -335,5 +335,23 @@ size_t GetCNodeOutputsSize(const std::shared_ptr<AnfNode> &anf_node, bool train_
return 1;
}
}
STATUS GetInputIndexOfTupleGetItem(const AnfNodePtr &node, int *index) {
MS_ASSERT(node != nullptr);
if (!opt::CheckPrimitiveType(node, prim::kPrimTupleGetItem)) {
MS_LOG(ERROR) << "The node is not a tupleGetItem node.";
return RET_ERROR;
}
auto index_vnode = node->cast<CNodePtr>()->input(2);
if (!utils::isa<ValueNodePtr>(index_vnode)) {
MS_LOG(ERROR) << "TupleGetItem's input 2 is not a valueNode.";
return RET_ERROR;
}
auto value_node = utils::cast<ValueNodePtr>(index_vnode);
MS_ASSERT(value_node != nullptr);
*index = value_node->value()->type()->number_type() == kNumberTypeInt64 ? GetValue<int64_t>(value_node->value())
: GetValue<int>(value_node->value());
return RET_OK;
}
} // namespace lite
} // namespace mindspore

2
mindspore/lite/tools/common/node_util.h
View File

@ -413,6 +413,8 @@ static STATUS TransFilterFormat(schema::TensorT *tensor, kTransFilterType type)
STATUS TransFilterFormat(schema::TensorT *tensor, schema::Format dstFormat);
size_t GetCNodeOutputsSize(const std::shared_ptr<AnfNode> &anf_node, bool train_flag = false);
STATUS GetInputIndexOfTupleGetItem(const AnfNodePtr &node, int *index);
} // namespace lite
} // namespace mindspore
#endif // MINDSPORE_LITE_TOOLS_COMMON_NODE_UTIL_H

1
mindspore/lite/tools/converter/CMakeLists.txt
View File

@ -62,7 +62,6 @@ file(GLOB_RECURSE CONVERTER_SRC RELATIVE ${CMAKE_CURRENT_SOURCE_DIR}
../optimizer/fusion/conv_bn_fusion.cc
../optimizer/fusion/conv_tuplegetitem_fusion.cc
../optimizer/fusion/constant_folding_fusion.cc
../optimizer/fusion/quant_dtype_cast_fusion.cc
../optimizer/fusion/norm_fusion.cc
../optimizer/fusion/batchmatmul_fusion.cc
../optimizer/fusion/sigmoid_mul_fusion.cc

6
mindspore/lite/tools/converter/converter_context.h
View File

@ -106,6 +106,11 @@ class ConverterContext {
}
size_t GetGraphInputTensorShapeMapSize() { return graph_input_tensor_shape_map_.size(); }
void SetGraphOutputTensorNames(const std::vector<std::string> &output_names) {
graph_output_tensor_names_ = output_names;
}
const std::vector<std::string> GetGraphOutputTensorNames() { return graph_output_tensor_names_; }
private:
ConverterContext() {}
virtual ~ConverterContext() = default;
@ -113,6 +118,7 @@ class ConverterContext {
std::map<int32_t, int32_t> graph_input_data_type_map_;
std::map<int32_t, int32_t> graph_output_data_type_map_;
std::map<std::string, std::vector<int64_t>> graph_input_tensor_shape_map_;
std::vector<std::string> graph_output_tensor_names_;
};
} // namespace lite
} // namespace mindspore

5
mindspore/lite/tools/converter/parser/caffe/caffe_inspector.cc
View File

@ -53,8 +53,9 @@ STATUS CaffeInspector::FindGraphInputsAndOutputs() {
}
}
for (const auto &iter : layerTops) {
if (layerBottoms.find(iter) == layerBottoms.end()) {
graphOutput.insert(iter);
if (layerBottoms.find(iter) == layerBottoms.end() &&
std::find(graphOutput.begin(), graphOutput.end(), iter) == graphOutput.end()) {
graphOutput.push_back(iter);
}
}
return RET_OK;

5
mindspore/lite/tools/converter/parser/caffe/caffe_inspector.h
View File

@ -21,6 +21,7 @@
#include <string>
#include <unordered_map>
#include <memory>
#include <vector>
#include "proto/caffe.pb.h"
#include "include/errorcode.h"
@ -37,7 +38,7 @@ class CaffeInspector {
STATUS SetLayerTopsAndBottoms();
std::set<std::string> GetGraphInput() { return graphInput; }
std::set<std::string> GetGraphOutput() { return graphOutput; }
std::vector<std::string> GetGraphOutput() { return graphOutput; }
private:
caffe::NetParameter net;
@ -46,7 +47,7 @@ class CaffeInspector {
std::set<std::string> layerBottoms;
std::set<std::string> graphInput;
std::set<std::string> graphOutput;
std::vector<std::string> graphOutput;
};
using CaffeInspectorPtr = std::shared_ptr<CaffeInspector>;

6
mindspore/lite/tools/converter/parser/caffe/caffe_model_parser.cc
View File

@ -385,11 +385,11 @@ STATUS CaffeModelParser::ConvertGraphOutputs() {
}
auto valueNode = NewValueNode(returnPrim);
std::vector<AnfNodePtr> opInputs{valueNode};
if (nodes_.find(*caffeInspector.GetGraphOutput().begin()) == nodes_.end()) {
if (nodes_.find(caffeInspector.GetGraphOutput().front()) == nodes_.end()) {
MS_LOG(ERROR) << "Can't find input node.";
return RET_NOT_FIND_OP;
}
auto cnode = nodes_.find(*caffeInspector.GetGraphOutput().begin())->second;
auto cnode = nodes_.find(caffeInspector.GetGraphOutput().front())->second;
if (cnode == nullptr) {
MS_LOG(ERROR) << "Can't find input node.";
return RET_NOT_FIND_OP;
@ -399,6 +399,8 @@ STATUS CaffeModelParser::ConvertGraphOutputs() {
returnCnode->set_fullname_with_scope("Return");
res_graph_->set_return(returnCnode);
}
// save original output tensor names.
ConverterContext::GetInstance()->SetGraphOutputTensorNames(caffeInspector.GetGraphOutput());
return RET_OK;
}

7
mindspore/lite/tools/converter/parser/onnx/onnx_model_parser.cc
View File

@ -157,6 +157,13 @@ STATUS OnnxModelParser::ConvertOnnxGraph(const onnx::GraphProto &onnx_graph, con
MS_LOG(ERROR) << "convert graph outputs failed.";
return RET_ERROR;
}
// save original output tensor names.
if (root_node_name == "root_node") {
std::vector<std::string> output_names;
std::transform(onnx_graph.output().begin(), onnx_graph.output().end(), std::back_inserter(output_names),
[](auto &graph_output) { return graph_output.name(); });
ConverterContext::GetInstance()->SetGraphOutputTensorNames(output_names);
}
return status;
}
STATUS OnnxModelParser::ConvertConstTensors(const onnx::GraphProto &onnx_graph, const FuncGraphPtr &func_graph_ptr,

16
mindspore/lite/tools/converter/parser/tf/tf_model_parser.cc
View File

@ -33,6 +33,7 @@
#include "tools/converter/parser/tf/functionalize_control_op_pass.h"
#include "tools/converter/parser/parser_utils.h"
#include "tools/common/tensor_util.h"
#include "tools/common/node_util.h"
#include "tools/converter/parser/unify_format.h"
using mindspore::lite::converter::FmkType_TF;
@ -1054,7 +1055,18 @@ STATUS TFModelParser::ConvertRootGraphOutputs() {
return RET_ERROR;
}
output_nodes.push_back(anf_node);
graph_output_names_.push_back(anf_node->fullname_with_scope());
auto name = anf_node->fullname_with_scope();
if (utils::isa<CNodePtr>(anf_node) && opt::CheckPrimitiveType(anf_node, prim::kPrimTupleGetItem)) {
int index = 0;
if (GetInputIndexOfTupleGetItem(anf_node, &index) != RET_OK) {
MS_LOG(ERROR) << "Get input index of tupleGetItem failed.";
return RET_ERROR;
}
auto in_node = anf_node->cast<CNodePtr>()->input(1);
MS_ASSERT(in_node != nullptr);
name = in_node->fullname_with_scope() + ":" + std::to_string(index);
}
graph_output_names_.push_back(name);
}
}
}
@ -1063,6 +1075,8 @@ STATUS TFModelParser::ConvertRootGraphOutputs() {
MS_LOG(ERROR) << "make anf graph outputs node error";
return status;
}
// save original output tensor names.
ConverterContext::GetInstance()->SetGraphOutputTensorNames(graph_output_names_);
return RET_OK;
}
STATUS TFModelParser::MakeAnfGraphOutputs(std::vector<AnfNodePtr> *output_nodes, const FuncGraphPtr &anf_graph) {

6
mindspore/lite/tools/converter/parser/tflite/tflite_model_parser.cc
View File

@ -398,6 +398,12 @@ STATUS TfliteModelParser::ConvertGraphOutputs() {
returnCnode->set_fullname_with_scope("Return");
res_graph_->set_return(returnCnode);
}
// save original output tensor names.
std::vector<std::string> output_names;
auto output_idx = tflite_subgraph->outputs;
std::transform(output_idx.begin(), output_idx.end(), std::back_inserter(output_names),
[&](auto out_idx) { return tflite_subgraph->tensors.at(out_idx)->name; });
ConverterContext::GetInstance()->SetGraphOutputTensorNames(output_names);
return RET_OK;
}

77
mindspore/lite/tools/optimizer/fusion/pooling_activation_fusion.cc
View File

@ -1,77 +0,0 @@
/**
* Copyright 2020-2021 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "tools/optimizer/fusion/pooling_activation_fusion.h"
#include <memory>
#include "src/ops/pooling.h"
#include "src/ops/activation.h"
#include "schema/inner/model_generated.h"
#include "tools/optimizer/common/gllo_utils.h"
namespace mindspore::opt {
namespace {
constexpr size_t kActivationInputsLength = 2;
}
const BaseRef PoolingActivationFusion::DefinePattern() const {
auto pooling_var = std::make_shared<CondVar>(IsPoolingNode);
auto prim = new (std::nothrow) schema::PrimitiveT();
if (prim == nullptr) {
MS_LOG(ERROR) << "new primitiveT failed";
return nullptr;
}
prim->value.type = primitive_type;
auto prim_value = std::make_shared<lite::PrimitiveC>(prim);
return VectorRef({prim_value, pooling_var});
}
const AnfNodePtr PoolingActivationFusion::Process(const FuncGraphPtr &func_graph, const AnfNodePtr &node,
const EquivPtr &) const {
MS_ASSERT(func_graph != nullptr);
MS_ASSERT(node != nullptr);
MS_LOG(DEBUG) << "pooling activation pass process:" << schema::EnumNamesPrimitiveType()[primitive_type];
CheckIfFuncGraphIsNull(func_graph);
CheckIfAnfNodeIsNull(node);
auto act_node = node->cast<CNodePtr>();
CheckIfCNodeIsNull(act_node);
CheckInputSize(act_node, kActivationInputsLength);
auto primitivec = GetValueNode<std::shared_ptr<lite::PrimitiveC>>(act_node->input(0));
MS_ASSERT(utils::isa<std::shared_ptr<mindspore::lite::Activation>>(primitivec));
auto act_primitivec = utils::cast<std::shared_ptr<mindspore::lite::Activation>>(primitivec);
MS_ASSERT(act_primitivec != nullptr);
if (act_primitivec->GetType() != activation_type) {
return node;
}
AnfNodePtr pre_node = act_node->input(1);
CheckIfAnfNodeIsNull(pre_node);
if (pre_node != nullptr && pre_node->isa<CNode>()) {
if (IsMultiOutputTensors(func_graph, pre_node)) {
return node;
}
auto pooling_node = pre_node->cast<CNodePtr>();
auto primitive_c = GetValueNode<std::shared_ptr<lite::PrimitiveC>>(pooling_node->input(0));
MS_ASSERT(utils::isa<std::shared_ptr<mindspore::lite::Pooling>>(primitive_c));
auto primc = utils::cast<std::shared_ptr<mindspore::lite::Pooling>>(primitive_c);
MS_ASSERT(primc != nullptr);
if (primc->GetActivationType() == schema::ActivationType_NO_ACTIVATION) {
primc->SetActivationType(activation_type);
return pre_node;
}
}
return node;
}
} // namespace mindspore::opt

48
mindspore/lite/tools/optimizer/fusion/quant_dtype_cast_fusion.cc
View File

@ -1,48 +0,0 @@
/**
* Copyright 2019-2021 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "tools/optimizer/fusion/quant_dtype_cast_fusion.h"
#include <memory>
#include "tools/optimizer/common/gllo_utils.h"
namespace mindspore::opt {
namespace {
constexpr size_t kActivationInputsLength = 2;
}
const BaseRef QuantDtypeCastFusion::DefinePattern() const {
auto quant_var = std::make_shared<CondVar>(IsQuantNode);
auto input_var = std::make_shared<Var>();
return VectorRef({quant_var, input_var});
}
const AnfNodePtr QuantDtypeCastFusion::Process(const FuncGraphPtr &func_graph, const AnfNodePtr &node,
const EquivPtr &) const {
MS_ASSERT(func_graph != nullptr);
MS_ASSERT(node != nullptr);
MS_LOG(DEBUG) << "quant dtype cast fusion pass process";
if (CheckIfFuncGraphIsNull(func_graph) != lite::RET_OK || CheckIfAnfNodeIsNull(node) != lite::RET_OK) {
return nullptr;
}
auto act_node = node->cast<CNodePtr>();
if (CheckIfCNodeIsNull(act_node) != lite::RET_OK ||
CheckInputSize(act_node, kActivationInputsLength) != lite::RET_OK) {
return nullptr;
}
AnfNodePtr pre_node = act_node->input(1);
if (CheckIfAnfNodeIsNull(pre_node) != lite::RET_OK) {
return nullptr;
}
return pre_node;
}
} // namespace mindspore::opt

34
mindspore/lite/tools/optimizer/fusion/quant_dtype_cast_fusion.h
View File

@ -1,34 +0,0 @@
/**
* Copyright 2019-2021 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#ifndef LITE_QUANT_DTYPE_CAST_FUSION_H
#define LITE_QUANT_DTYPE_CAST_FUSION_H
#include <string>
#include "backend/optimizer/common/optimizer.h"
namespace mindspore {
namespace opt {
class QuantDtypeCastFusion : public PatternProcessPass {
public:
explicit QuantDtypeCastFusion(bool multigraph = true, const std::string &name = "quant_dtype_cast_fusion")
: PatternProcessPass(name, multigraph) {}
~QuantDtypeCastFusion() override = default;
const BaseRef DefinePattern() const override;
const AnfNodePtr Process(const FuncGraphPtr &, const AnfNodePtr &, const EquivPtr &) const override;
};
} // namespace opt
} // namespace mindspore
#endif // LITE_QUANT_DTYPE_CAST_FUSION_H