![\"f527fa054a01034c00a807328689e78cd7b2d1fd\"](\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/01\/f527fa054a01034c00a807328689e78cd7b2d1fd.jpg\" "\\"f527fa054a01034c00a807328689e78cd7b2d1fd\\"")
<\/p>\n<\/p>\n
Today I'm sharing 6Prompt word<\/a>tip<\/a>, generalized to a variety of large language models (LLMs), not just theChatGPT<\/i><\/a>, which also includes Wenxin Yiyin or various open source models that can help improve the output results.<\/p>\n Tip 6: Use external tools<\/p>\n<\/section>\n<\/section>\n<\/blockquote>\n Tip 1: Write clear and specific instructions\u00a0<\/strong><\/p>\n Specifies the desired output format or output length.<\/strong>One way to do this is to have the model play a role. For example:<\/p>\n \"Give me a summary of this paragraph. Here's an example of a summary I like ___\"<\/p>\n<\/section>\n<\/section>\n<\/blockquote>\n Provide examples.<\/strong>For example, this is the step of Few Shot Prompting:<\/p>\n Your prompt: give your actual prompt words.<\/p>\n<\/section>\n<\/section>\n<\/blockquote>\n The model can now respond according to the pattern established in the first two examples.<\/p>\n Tip 2: Give the model time to think<\/strong><\/p>\n Models are more likely to make inference errors when responding immediately.<\/p>\n passRequires a series of reasoning<\/strong>This prompts the model to think progressively and more carefully. You can ask for \"think step-by-step\" or specify specific steps. This simple prompt is a very effective addition: \"Think step by step.\" (Think step by step)<\/p>\n For example, if you ask the model to grade a student's exam question, you can prompt the model like this:<\/p>\n Step 3: Complete your own solution calculations prior to evaluating student solutions.<\/p>\n<\/section>\n<\/section>\n<\/blockquote>\n Tip 3: Multiple Tips<\/strong><\/p>\n When accuracy is most important (rather than latency or cost), generate multiple responses with different prompts and then determine the best answer.<\/p>\n Some of the things you can tweak include:<\/p>\n Tip 4: Bootstrapping models\u00a0<\/strong><\/p>\n Here are some examples:<\/p>\n Tip 5: Break down tasks or tips\u00a0<\/strong><\/p>\n Break down complex tasks into multiple simple tasks.<\/strong>The reason for this is that complex tasks have a significantly higher error rate than simple tasks.<\/p>\n Intention 3: To provide guidance on how to use public transportation in Paris.<\/p>\n<\/section>\n<\/section>\n<\/blockquote>\n Or, if the subtasks are interrelated:<\/p>\n Note: This may also reduce costs, as each step will cost less.<\/p>\n Tip 6: Use external tools<\/strong><\/p>\n In general, if a task can be done more reliably and efficiently with a tool (compared to a larger model), then move it out to get the advantages of both. (This may not apply to non-developers, and can be ignored by casual hobbyists, but the reasoning is not to let a larger model do something it's not good at)<\/p>\n Here are some sample tools:<\/p>\n\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n