Quantization reduces LLM model size and inference latency by using lower-precision numbers.
- True
- False
Answer: True
Quantization (INT8, INT4) reduces model size 2-4x with minimal accuracy loss, enabling deployment on edge devices and reducing cloud inference costs. Critical for scalable LLM serving.