[FEAT] experiment tracking by mlflow

[twndus]

Background

• After the initial implementation of the model, establish a process to track experiments using MLflow

To do

• [x] logging metric
• [ ] setup sweep

[GangBean]

• MLflow requires another server instance to save metrics and handle the UI process.
• However, due to restricted server resources in the project, we opt to utilize Weights and Biases instead, as it offers free cloud resources.