Metadata-Version: 2.1
Name: gpt-scrolls
Version: 0.0.7
Summary: A collaborative collection of open-source safe GPT-3 prompts that work well
Home-page: https://github.com/maraoz/gpt-scrolls
Author: Manuel Araoz
Author-email: gpt-scrolls@maraoz.com
License: UNKNOWN
Description: # gpt-scrolls
        A collaborative collection of open-source safe GPT-3 prompts that work well
        
        Feel free to contribute your prompts!
        
        
        ## Getting Started
        To use gpt-scrolls, you'll need access to the OpenAI API. If you haven't, [sign up for the beta](http://beta.openai.com/).
        
        After that, go to the [OpenAI Playground](https://beta.openai.com/playground), copy the prompt you'd like to use, add any additional prompt-specific required data, and submit.
        
        ## Example
        For example, copying the `top-10/top-10-women.txt` into the Playground:
        ```
        Top 10 most important women in human history, and their greatest achievements:
        
        1. 
        ```
        will produce something like the following:
        ```
        Top 10 most important women in human history, and their greatest achievements:
        
        1. ~~~Rosa Parks
        
        2. ~~~Cleopatra
        
        3. ~~~Joan of Arc
        
        4. ~~~Madonna
        
        5. ~~~Queen Elizabeth I
        
        6. ~~~Elizabeth II
        
        7. ~~~Tamar
        
        8. ~~~Billie Jean King
        
        9. ~~~Catherine the Great
        
        10. ~~~Elizabeth I
        ```
        
        ## Design goals
        gpt-scroll prompts should aim to be:
        - **effective**: they should reliably produce desired classes of outpupts
        - **efficient**: they should be as short as possible
        - **safe**: they should minimize appearance of toxic/harmful output
        
        Have this in mind for any contribution. (eg: if you find that a prompt works equally well without one example, you might submit a PR to remove that example, citing efficiency)
        
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.6
Description-Content-Type: text/markdown
