How the model is updated in onnxruntime?

[kimjungwow]

Hi. I am trying to understand the flow of onnxruntime codes.
Especially, I am curious about how the model is updated in onnxruntime.

onnxruntime/orttraining/orttraining/models/runner/training_runner.cc

Lines 647 to 649 in d43ef67

	// Add one after update the model once.
	++weight_update_step_count_;
	}

EDITED : Since weight_update_step_count_ is increased in the line 648 of orttraining/orttraining/models/runner/training_runner.cc, it seems that the model is updated in InferenceSession::Run() which is called in TrainingRunner::RunUpdate().

onnxruntime/onnxruntime/core/session/inference_session.cc

Lines 1782 to 1785 in d43ef67

	Status InferenceSession::Run(const RunOptions& run_options,
	const std::vector<std::string>& feed_names, const std::vector<OrtValue>& feeds,
	const std::vector<std::string>& output_names, std::vector<OrtValue>* p_fetches,
	const std::vector<OrtDevice>* p_fetches_device_info) {

However, I cannot find the exact region of codes for updating the model. Can I ask How the model is updated in onnxruntime please? Thank you for reading my question.