from Hacker News

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

by jackbravo on 7/18/24, 12:44 AM with 0 comments