GPTQ

Alias for a post‑training quantization method that approximates weight updates for efficient inference.