Context-faithful Prompting for Large Language Models