Integer Scale: A Free Lunch for Faster Fine-grained Quantization of LLMs

Open in new window