Рет қаралды 1,484
#promptengineering #ai #openai
OpenAI just released their guide for prompt engineering, and reading it reminded me of how I've been struggling through figuring out how to best use LLM tools. I think there are many tips that I've heard about, but here are some (relatively) unfiltered thoughts about how I understand what the document is suggesting and my own thoughts on prompt engineering.
OpenAI's prompt engineering guide: platform.openai.com/docs/guid...
OpenAI Evals: github.com/openai/evals
[00:00-00:42] intro
[00:46-01:44] strategy 1: write clear instructions
[06:51-07:19] strategy 2: use reference text
[09:36-10:03] strategy 3: split complex tasks into simpler subtasks
[15:56-16:38] strategy 4: give the model time to "think"
[21:01-21:43] strategy 5: use external tools
[25:08-26:42] strategy 6: test changes systematically
[27:59-31:30] limitations
---
WHO AM I: I’m Ian Wong, an ICU doctor and data scientist at Duke University. I love to think about life, medicine, tech, and building things. And somehow achieving everything.
Please leave a comment!
Twitter - / aianwongmdphd
LinkedIn - / aianwong
---
#openai #promptengineering #prompt #gpt4 #ai #prompt #artificialintelligence #openaichat