<ul><li>Anthropic has released a new feature on Anthropic Console that allows developers to improve prompts for better outputs.</li><li>The prompt improver uses chain-of-thought reasoning to refine prompts and focuses on improving accuracy and word count adherence.</li><li>Anthropic has achieved a 30% increase in accuracy for a multilabel classification test and 100% word count adherence for a summarization task.</li><li>Anthropic has also introduced a prompt evaluator to benchmark and grade prompts, and allows developers to provide feedback for further improvements.</li></ul>

Anthropic Will Use Chain of Thought Reasoning to Improve Prompts

Discover more