RSQ: Learning from Important Tokens Leads to Better Quantized LLMs

Open in new window