- Dynamic adjustment of batch size and learning rate based on resource usage.
- Priority-based parameter tuning to balance memory constraints and accuracy.
- Logging and visualization of training adjustments and system resource usage.
- Pruning strategy now maintains a constant ratio while adjusting batch size and learning rate dynamically.
- Improved accuracy and memory scoring to better balance learning rate and batch sizeadjustments.
- Pre-commit hooks now properly enforce Black, isort, and Flake8 styling.
- Addressed circular imports affecting test execution.
- Old manual tuning logic replaced with automated adaptive training strategies.