Explore the critical accuracy tradeoffs when compressing LLMs. Learn how 4-bit quantization and pruning affect reasoning, knowledge retrieval, and production stability.