Rubber Ducky Bot

Description

When you need to work through a problem, it can help to explain the problem to a rubber duck.

It would be nice in some cases if the rubber duck could talk back and suggest things that you might have missed or not thought of. The Rubber Ducky Bot is a Python script designed to iterate on ideas or solutions to certain questions through a conversation between the user and a chat bot, using the llama_cpp_python server. The bot continually refines the user's answers until a consensus is reached. This is particularly useful for fleshing out ideas or generating simple, refined answers for complex topics.

Prerequisites

• Python 3.x
• requests library
• llama_cpp_python server running on localhost:8000

Installation

Clone the repository and navigate to the directory:

git clone https://github.com/cagostino/rubber-ducky.git
cd rubber-ducky

Install the requests library:

pip install requests

^above to be further refined

Usage

Run the script with the .md file containing the list of questions:

python rubber_ducky.py --file questions.md

Question File Format

The .md file containing questions should be formatted as follows:

[questions]
What's your idea regarding x?
What makes you think that the theory of y is correct?

Questions are listed line-by-line under the [questions] heading.

Functions

• read_questions_from_md(file_path): Reads questions from an .md file.
• query_local_llama_server(messages, max_tokens=150): Sends a query to the LLM server and receives the model's response.
• generate_critique(conversation, section, answer): Generates a critique or a follow-up question based on the user's answer.
• integrate_and_summarize(conversation, original_answer, new_info): Integrates and summarizes the original and new answers.
• evaluate_consensus(refined_answer): Evaluates whether the user agrees with the refined answer.
• consensus_bot(questions): Main function to initiate the conversation and build consensus.