This repository will continously make api calls to a text-generation-ui server and log the results to a file.
This assumes that you are running the text-generation-ui server and have the api turned on at port 5000.
To use the repository, make a config.json file at the root of the project.
File Defaults and explanation:
{
"prefix": '', // Text to prepend to generation or input for every generation
"num_generations": 100, // Number of times to generate output
"save_interval": 5, // Number of generations to make before saving file
"input_file": 'input.json', // file to read for input data, assumes it is a list of strings
"output_file": 'generations.json', // file to write to for the generation data
"use_input_file": True, // whether to use the input file or start with only prefix
"num_tokens": 300, // number of tokens to generate
"print_generations": False // whether to print the generations to the terminal
}