Muhammad Alif Al Hakim

New technique preserves fairness and safety in quantized LLMs

A new research paper explores the impact of quantization on the fairness and safety of large language models (LLMs). The study found that quantization methods, both static and dynamic, consistently degrade fairness and …