mirror of
https://github.com/saymrwulf/onnxruntime.git
synced 2026-05-27 22:45:57 +00:00
a216c9a3fa
Add the ability to get and set tuning results of an inference session. Also add tool to manipulate onnx file to embed the results into the model file and automatically load it on session initialization.
68 lines
2.2 KiB
C++
68 lines
2.2 KiB
C++
// Copyright (c) Microsoft Corporation. All rights reserved.
|
|
// Licensed under the MIT License.
|
|
|
|
#pragma once
|
|
|
|
#include "core/flatbuffers/schema/ort.fbs.h"
|
|
|
|
#if !defined(ORT_MINIMAL_BUILD)
|
|
//
|
|
// Includes to parse json session config from onnx model file
|
|
//
|
|
#include "core/graph/onnx_protobuf.h"
|
|
#include "core/session/inference_session.h"
|
|
#include "core/framework/session_options.h"
|
|
#include "core/framework/tuning_results.h"
|
|
#include "core/common/common.h"
|
|
#include "nlohmann/json.hpp"
|
|
using json = nlohmann::json;
|
|
#endif
|
|
|
|
namespace onnxruntime {
|
|
|
|
namespace inference_session_utils {
|
|
|
|
// need this value to be accessible in all builds in order to report error for attempted usage in a minimal build
|
|
static constexpr const char* kOrtLoadConfigFromModelEnvVar = "ORT_LOAD_CONFIG_FROM_MODEL";
|
|
|
|
#if !defined(ORT_MINIMAL_BUILD)
|
|
//
|
|
// Code to parse json session config from onnx model file
|
|
//
|
|
static constexpr const char* kOrtConfigKey = "ort_config";
|
|
static constexpr const char* kSessionOptionsKey = "session_options";
|
|
static constexpr const char* kTuningResultsKeys = "tuning_results";
|
|
|
|
class JsonConfigParser {
|
|
public:
|
|
JsonConfigParser(const logging::Logger& logger) : logger_(logger) {
|
|
}
|
|
|
|
Status ParseOrtConfigJsonInModelProto(const ONNX_NAMESPACE::ModelProto& model_proto);
|
|
|
|
Status ParseSessionOptionsFromModelProto(/*out*/ SessionOptions& session_options);
|
|
|
|
Status ParseRunOptionsFromModelProto(/*out*/ RunOptions& run_options);
|
|
|
|
private:
|
|
// Logger instance that will be used to log events along the parsing steps
|
|
const logging::Logger& logger_;
|
|
|
|
// Flag indicating if the model has been checked for ort config json existence
|
|
bool is_model_checked_for_ort_config_json_ = false;
|
|
|
|
// Parsed json available for other utility methods to use (if the model did have a valid json)
|
|
nlohmann::json parsed_json_;
|
|
|
|
// Flag indicating if an ort config json is available to be used
|
|
bool is_ort_config_json_available_ = false;
|
|
};
|
|
|
|
Status ParseTuningResultsFromModelMetadata(const onnxruntime::ModelMetadata& metadata,
|
|
/*out*/ std::vector<TuningResults>& results,
|
|
/*out*/ bool& key_found);
|
|
|
|
#endif // !defined(ORT_MINIMAL_BUILD)
|
|
|
|
} // namespace inference_session_utils
|
|
} // namespace onnxruntime
|