PyTorch Deep Integration Auto Logging for Visdom

Visdom is a popular local-first visualization tool for PyTorch users, but it currently requires repetitive manual vis.line() calls for every metric such as loss, accuracy, and learning rate. This creates boilerplate code and slows down experiments. This project will solve the problem by adding deep native integration with both vanilla PyTorch and PyTorch Lightning. I will implement automatic metric detection and logging, a simple context manager and decorator for native training loops, configurable autograd hooks with throttling, learning-rate scheduler tracking, smart multi-run environment versioning, full multi-GPU and distributed support, and complete documentation with tutorial notebooks. The final deliverable will be a production-ready VisdomLogger that gives zero-config automatic logging while keeping full flexibility for advanced users. All features will come with comprehensive tests, a new PyTorch Integration Guide, and working demos.