search

LAION-AI / Open-Assistant

OpenAssistant is a chat-based assistant that understands tasks, can interact with third-party systems, and retrieve information dynamically to do so.
https://open-assistant.io
Apache License 2.0
36.97k stars 3.23k forks source link

Implement prompt library #3025

Open notmd opened 1 year ago

notmd commented 1 year ago

From user wsg on discord

Description A comprehensive library of prompts designed for help users generate content, headlines, summaries, meta descriptions, and much more. Users can easily access appropriate prompt templates for various website types, industries, topics, audiences, and styles. Each template includes detailed instructions and guidelines. With this extensive resource at hand, users can streamline their workflow and make it produce high-quality output.

Key Points

  1. Diverse range of content templates.
  2. Easy-to-follow guidelines for effective content output.
  3. Helps ensure consistent quality across multiple pieces of content.

Use case scenario An online marketer wants to create fresh, engaging content for their client’s in an eCommerce site. Instead of spending hours brainstorming ideas and crafting content from scratch, they turn to the Prompt Library. They quickly locate templates relevant to their niche, audience, tone, and style preferences. Using these guided prompts, they effortlessly generate captivating headlines, compelling product descriptions, informative blog posts, and more.

image

andrewm4894 commented 1 year ago

I wonder if could make sense to live in docs somewhere and also be exposed then in the app and some section of docs maybe.

Or just do it all in-app, not sure but just throwing it out there in case having them live in the /docs might somehow make it easier to scale and get contributions to the "prompt library" and then just have it be discoverable in the app too (as well as living somewhere in https://projects.laion.ai/Open-Assistant/).

notmd commented 1 year ago

For the initial implementation, I would say we just put it under a JSON file, then the docs can import it if you want