Verified record · from the answer index

gptq

Post-training 4-bit quantization for LLMs with minimal accuracy loss. Use for deploying large models (70B, 405B) on consumer GPUs, when you need 4× memory reduction with <2% perplexity degradation, or for faster inference (3-4× speedup) vs

skill:orchestra-research/ai-research-skills/gptq · published by orchestra-research · source repository