Feature Request: Implement Metrics Logging using SummaryWriter in Training/Validation Loops

[WilliamLee30]

Description:

I am noticing that while the SummaryWriter is initialized and passed into the core training functions, it is currently not being utilized to record any metrics.

Observed Behavior:

1.The SummaryWriter instance (writer) is defined in train.py using the following snippet:

writer = SummaryWriter(log_dir=os.path.join(
            settings.LOG_DIR, args.net, settings.TIME_NOW))

2.This writer object is then correctly passed to the core functions, such as train_sam and validation_sam.

3.However, upon inspection of the source code (specifically files like function.py), the writer.add_scalar() or similar methods are never called to log metrics (e.g., loss, accuracy, etc.) during training or validation steps.

4.Consequently, the generated log file in the runs directory remains empty (or contains only minimal header information).

Expected Behavior / Feature Request:

Could you please implement the necessary logging calls within the training and validation loops (likely in function.py) to utilize the passed writer object? This would enable users to effectively monitor training progress using TensorBoard.