During the month, contributed to the pytorch/torchrec repository by implementing gradient norm logging within the GradientClippingOptimizer. This feature enhanced observability during deep learning model training by recording gradient norms, allowing for more informed tuning of clipping thresholds and improved debugging of gradient behavior. The work involved modifying the clipping.py module to introduce efficient logging with minimal performance overhead. Leveraging Python and PyTorch, the developer focused on incremental instrumentation and code review to ensure maintainability and clarity. No major bugs were reported or fixed during this period, with efforts concentrated on expanding monitoring capabilities for gradient optimization workflows in machine learning.