Memory Leak Detector

소개

Designed for scientific computing and machine learning workflows, this skill provides a systematic approach to diagnosing and fixing memory growth issues that cause crashes during long-running experiments. It offers targeted patterns for handling common pitfalls such as unclosed matplotlib figures, unintended gradient accumulation, and GPU memory mismanagement. By providing clear implementation guidance for memory profiling tools and resource management patterns, it ensures that applications remain stable and efficient during Phase 4 performance optimization.

주요 기능

• Implementation guidance for memory profiling tools like tracemalloc and memory-profiler
• Best practices for PyTorch tensor gradient management and GPU memory cleanup
• 0 GitHub stars
• Strategies to identify and fix circular references and global state accumulation
• Detection and resolution patterns for matplotlib figure leaks
• Guidelines for managing visualization loops and bounded measurement history

사용 사례

• Debugging Out-of-Memory (OOM) errors in long-running machine learning training loops
• Optimizing memory usage in scientific visualization pipelines using matplotlib
• Profiling and fixing memory growth in complex data processing workflows