The code helps record all the communication with LLMs by a human. This repo is a simple implementation but can be adapted according to the user's needs. Currently, this code will log the date, time, input prompt, output from LLM, model names and set of hyperparameters when the user communicates with LLM using some prompt in a json file. To check the log files, go to Logs/ which holds datewise folders and respective json files (contains the before mentioned info)
Update your hugging face access token in .env file and run the main.py file to log inputs/outputs
Tune the hyperparameters or change the model id's in config.ini file, if needed
To use the different LLM, update the load_llm function according to the needs in helper_functions.py
The repository is licensed under Apache License 2.0
@misc{LLM_Logs,
author = {Vamsi Kommineni},
month = {04},
title = {{LLM_Logs}},
url = {https://github.com/Vamsi-Kommineni/LLM_Logs},
year = {2024}
}