DEV Community

Kunal Agrawal
Kunal Agrawal

Posted on • Updated on

Prompt (Proompt) Engineering in ChatGPT.

Since a lot of us are using the ChatGPT (3.5), copilot tools like Github Copilot and other Generative AI to easily do our tasks. It's a while now using prompt to generate content, code.

so this article is all about prompt and how to develop prompt to get desired results in the chatGPT.

I asked chatGPT itself like when you'll give correct answers 😂.
Explain me the answers like this man.

Image description

Input and Output to ChatGPT has some limits, 500 words for Input and 3000 words for Output. This is just approximate values as limit chatGPT totally depends on number of tokens.

Let's build an Ecommerce Application with ChatGPT

Create an Ecommerce Application using Web3 and machine learning.
Enter fullscreen mode Exit fullscreen mode

Image description

You just need to be more informative. ℹī¸

  • You can talk to GPT more about the process like how to setup backend.
  • how's your codebase structure.
  • Which language, framework to choose.
  • Data Flow of Application.

Since, output is limited to certain number of tokens you can ask chatGPT to return only code.

Prompt results vary based on previous conversations. so there is no proper way to get same response everytime.

You shouldn't ask questions about news, current affairs because it's trained on past data.

This is Serious 👇

Recently Vector Databases like Pinecone, Weaviate is launched and can help models like GPT to improve themeselves and provide results faster. Also there is a new GitHub repo called AutoGPT Link that demonstrates capabilities of GPT Model, it can make GPT model to act like a person with specific configuration. It's an example of how model can improve themselves.

Let's Connect -
Tweet Thanks

Oldest comments (1)

Some comments have been hidden by the post's author - find out more