gpt3-prompt-to-text

gpt3-prompt-to-text is a command line program that converts a prompt into text using OpenAI's GPT-3 API. It is a simple wrapper around the gpt3-client library. The program is meant to be simple and easy to use.

This code was developed by the very junior developer GPT-3 with patient mentoring by Clemens Vasters.

Installation

With Python 3 installed, run

 pip install git+https://github.com/clemensv/gpt3-prompt-to-text.git

Sign up for an OpenAI API Key

To use this tool, you need to sign up for an API key at OpenAI.

Usage

gpt3-prompt-to-text [prompt] [-k | --api-key] [-e | --engine] [-s | --store] [-t | --tokens] [--edit] [--edit-file] [--noecho] [--amend] [--prepend]

Options

Option	Description
prompt	The prompt to convert into text
-h, --help	show this help message and exit
-k API_KEY, --api-key API_KEY	The OpenAI API key to use
-e ENGINE, --engine ENGINE	The GPT-3 engine to use (e.g. gpt3-davinci-003 or ada). Optional. gpt3-davinci-003 is the default.
-s, --store	Store the provided parameters in a configuration file
-t TOKENS, --tokens TOKENS	The maximum number of tokens to generate
--edit	Use the edit option. Takes the input to edit from stdin and the prompt from the command line
--edit-file EDIT_FILE	Use the edit option. Takes the input to edit from a file and the prompt from the command line. Writes the output back to the file.
--endpoint	The OpenAI endpoint to use. Optional. Defaults to https://api.openai.com/v1/engines.
--show-deployments	Show model deployments when running in Azure OpenAI
--noecho	don't echo the input
--prepend PREPEND	add the prepend text to the input (which is read from stdin)

Read prompt from stdin

echo [prompt] | gpt3-prompt-to-text

Configuration file

The configuration file is stored in the user's profile directory and is named .gpt3-prompt-to-text-config.json. It contains the following fields:

{
"api_key": "The OpenAI API key to use",
"engine": "The GPT-3 engine to use",
"tokens": "The maximum number of tokens to generate"
}

Store the API key

For convenience, run the script once with the -kl and -s options to store the key in the config file. Once that's done, the key will be read from there for all subsequent uses.

gpt3-prompt-to-text -k [your api key] -s

Examples

There are a few examples for how to use the tool:

Get help

gpt3-prompt-to-text -h

Write a short story using the default engine and the API key stored in the configuration file

gpt3-prompt-to-text "Once upon a time"

Suppress repeating the prompt or input into the output

gpt3-prompt-to-text --noecho "print a cat as ASCII art"

Control the number of tokens (length of the output)

gpt3-prompt-to-text "Once upon a time" -t 100

Send the output to a file

gpt3-prompt-to-text "Once upon a time" -t 100 > myfile.txt

Correct the story by prepending an instruction to what has been created

gpt3-prompt-to-text --prepend "Make a dragon appear here:" < myfile.txt

Similar to above but use the GPT-3 edit mode

gpt3-prompt-to-text --edit "Make a dragon appear here:" < myfile.txt

Edit the story file directly with an instruction

gpt3-prompt-to-text --edit-file myfile.txt "Make a dragon appear here"

Development

For information on how to contribute to the project and what prerequisites must be installed, see CONTRIBUTING.md.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Code of Conduct

This project adheres to the Contributor Covenant Code of Conduct. By participating, you are expected to uphold this code. See CODE_OF_CONDUCT.md for details.