User Guides Overview

These guides complement the unified workflow with focused tips for specific stages. Start with the Workflow Guide, then dip into the module that matches your immediate task.

---

Quick Index

Guide	Focus	Notes
Training	Day-to-day training tips & overrides	Pair with Workflow Step 6
Advanced Training	Mixed precision, Accelerate, DP trainers	Use when you need specialised trainers
Federated Learning	Distributed training with Flower	Optional: multi-client, privacy-preserving FL
Monitoring	TensorBoard & log inspection	Complements Workflow Step 7
Hyperparameter Tuning	Manual & Optuna workflows	Expands on Workflow Step 5
Inference	Strategy comparison (standard, TTA, ensemble)	Ties into Workflow Step 8
Test-Time Augmentation	Tune augmentation sets & aggregation	Optional add-on for inference
Ensemble Inference	Weighting strategies & fold management	Optional add-on for inference
Model Export	ONNX export validation & optimisation	Workflow Step 9
Resuming Training	Continue or extend existing runs	Quick reference during training
LR Finder	Find optimal learning rates	Run before training for LR tuning

---

Suggested Reading Paths

1. New users → Training → Monitoring → Inference → (optional) Tuning.
2. Performance focus → Advanced Training → Hyperparameter Tuning → Ensemble/TTA.
3. Deployment → Inference → Model Export.
4. Federated/Privacy → Federated Learning → Advanced Training (DP) → Monitoring.

Each guide now links back to the relevant workflow step to avoid duplicate walkthroughs.