{"id":2197,"date":"2023-12-23T11:40:49","date_gmt":"2023-12-23T03:40:49","guid":{"rendered":"https:\/\/www.1ai.net\/?p=2197"},"modified":"2023-12-23T11:40:49","modified_gmt":"2023-12-23T03:40:49","slug":"chatgpt%e5%ae%98%e6%96%b9%ef%bc%9aprompt%e4%bc%98%e5%8c%96%e5%85%ad%e5%a4%a7%e7%ad%96%e7%95%a5%ef%bc%8c%e8%ae%a9%e5%9b%9e%e7%ad%94%e6%9b%b4%e7%b2%be%e5%87%86","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/2197.html","title":{"rendered":"ChatGPT official: six prompt optimization strategies to make answers more accurate"},"content":{"rendered":"<p><strong>1. The clearer the task description, the better<\/strong><\/p>\n<p>(1) The more specific and clear the prompt is, the better.<\/p>\n<p>(2) Let<a href=\"https:\/\/www.1ai.net\/en\/tag\/chatgpt\" title=\"[View articles tagged with [ChatGPT]]\" target=\"_blank\" >ChatGPT<\/a>To play a professional role<\/p>\n<p>(3) Use separators to clearly distinguish which are output requirements and which are reference content in the prompt.<\/p>\n<p>(4) Specify the steps required to complete the task<\/p>\n<p>(5) Provide an example of the final output<\/p>\n<p>(6) Specify the required output content length<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-2198\" title=\"640-17\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2023\/12\/640-17.png\" alt=\"640-17\" width=\"604\" height=\"391\" \/><\/p>\n<p><strong>2. Provide reference materials<\/strong><\/p>\n<p>(1) Provide reference materials for the model, similar to reading comprehension.<\/p>\n<p>(2) The model is required to directly quote the original text in the reference material.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-2199\" title=\"640-18\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2023\/12\/640-18.png\" alt=\"640-18\" width=\"596\" height=\"279\" \/><\/p>\n<p><strong>3. Split complex tasks into steps or subtasks<\/strong><\/p>\n<p>(1) Use intent classification to identify the instructions most relevant to the user\u2019s query<\/p>\n<p>(2) For long conversations, make an appropriate summary or filter in the next prompt to avoid going off topic<\/p>\n<p>(3) For long documents, summarize and split the document. Construct the document summary in a recursive way.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-2200\" title=\"640-19\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2023\/12\/640-19.png\" alt=\"640-19\" width=\"613\" height=\"423\" \/><\/p>\n<p><strong>4. Let GPT think more<\/strong><\/p>\n<p>(1) Tell the model not to rush to give the correct answer.<\/p>\n<p>(2) Allow the model to hide its thinking or reasoning process.<\/p>\n<p>(3) Ask the model directly: \u201cDid you miss anything in your answer?\u201d<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-2201\" title=\"640-20\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2023\/12\/640-20.png\" alt=\"640-20\" width=\"614\" height=\"359\" \/><\/p>\n<p><strong>5. Use external tools<\/strong><\/p>\n<p>(1) Allow the model to use external information sources in the prompt and use Embeddings to achieve efficient search<\/p>\n<p>(2) Use code or call external API to obtain more accurate calculation results<\/p>\n<p>(3) Allow the model to execute function calls<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-2202\" title=\"640-21\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2023\/12\/640-21.png\" alt=\"640-21\" width=\"613\" height=\"388\" \/><\/p>\n<p><strong>6. Systematic experiments, continuous trial and error, and optimization<\/strong><\/p>\n<p>(1) Refer to a gold standard answer to evaluate the model output<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-2203\" title=\"640-22\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2023\/12\/640-22.png\" alt=\"640-22\" width=\"589\" height=\"296\" \/><\/p>","protected":false},"excerpt":{"rendered":"<p>The clearer the mission statement, the clearer it is (1) the clearer it is, the clearer it is (2) the more it is (2) that ChatGPT plays a professional role (3) the clear distinction between output requirements and reference elements in the separator (4) the designation of the mission and the steps required (5) the provision of an example of the final output (6) the length of the content to be exported, the provision of reference material (1) the provision of reference material for the model and a similar reading understanding. (2) The model is required to refer directly to the original language of the reference material. 3. Split complex things into steps or subtasks (1) Use intent classification to identify the most relevant commands for user queries (2) Long dialogue, appropriate for the previous in the next prompt<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[149,144],"tags":[177],"collection":[258],"class_list":["post-2197","post","type-post","status-publish","format-standard","hentry","category-jiaocheng","category-baike","tag-chatgpt","collection-chatgpt-prompt-guide"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/2197","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=2197"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/2197\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=2197"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=2197"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=2197"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=2197"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}