EfficientQAT: LLM Quantization, gets a 2-bit llama2-70B outperform regular 13B

by jackbravo on 7/18/2024, 12:44:35 AM with 0 comments