OpenAI founding member Andrej Karpathy shares tips on making the most of artificial intelligence
Haebom
•
A talk from Microsoft's Build event. Andrej Karpathy explains how to write effective prompts for AI.
•
Korean subtitles are available, so if you'd like to see it for yourself, I recommend giving it a watch.
•
When I jot things down...
•
It's useful for AI models to break documents into smaller pieces, embed them, store them in a repository, and then use queries to retrieve relevant chunks to insert into prompts for performing tasks.
•
Using prompt engineering techniques, you can constrain the output of LLMs to fit a specific template.
•
Fine-tuning improves performance by altering the model's weights, but it's technically complex and costly.
•
There are limitations and challenges to be mindful of when using LLMs, but combining them with human oversight can deliver safe and trustworthy results.
•
The GPT-4 model currently offers the best performance, and it's recommended to use detailed and clear prompts.
•
Adding relevant context and information to your prompt helps the LLM understand better.
•
Including examples helps the LLM understand your intent more clearly.
•
Consider using tools and plugins to help LLM handle more complex tasks.
•
Once you've gotten the most out of prompt engineering, you can also consider fine-tuning the model for further performance optimization.
•
LLMs can optimize costs by using low-precision inference and training on fixed parts of the model.
Subscribe to 'haebom'
📚 Welcome to Haebom's archives.
---
I post articles related to IT 💻, economy 💰, and humanities 🎭.
If you are curious about my thoughts, perspectives or interests, please subscribe.
haebom@kakao.com
---
I post articles related to IT 💻, economy 💰, and humanities 🎭.
If you are curious about my thoughts, perspectives or interests, please subscribe.
haebom@kakao.com
Subscribe
2
피
피카부
칼로x100과 KoGPT로 새로운 BM 짜려고 기다리고 있습니다!!
Y
Yiting
오... 좋은 포스팅 감사합니다~!
See latest comments