Sovereign 8s.is mirror of standardgalactic/chatbot
Find a file
Repository files (latest commit first)
Filename Latest commit message Latest commit date
standardgalactic 218635e41b Preliminary Tests
2025-05-18 20:12:00 -03:00
bot_conversation_outputs Preliminary Tests 2025-05-18 20:12:00 -03:00
first-run Preliminary Tests 2025-05-18 20:12:00 -03:00
chat.sh Preliminary Tests 2025-05-18 20:12:00 -03:00
first_starter_prompt.txt Preliminary Tests 2025-05-18 20:12:00 -03:00
LICENSE Initial commit 2025-05-18 14:25:52 -03:00
README.md Preliminary Tests 2025-05-18 20:12:00 -03:00
starter_prompt.txt Preliminary Tests 2025-05-18 20:12:00 -03:00

Chatbot: Automatic Dialog Simulator

Overview

The chatbot repository contains a Bash script that simulates a conversation among three AI bots—AnalystBot, CreativeBot, and SkepticBot—using IBM's Granite model. The script generates dynamic, multi-turn dialogues based on a user-specified topic, saves each bot's responses to dedicated folders, compiles a central summary, and displays all outputs in the terminal.

Functionality

  • Conversation Simulation: Three bots with distinct personalities (logical, creative, critical) engage in a dialogue over multiple turns.
  • Input: Reads a starter prompt from starter_prompt.txt in the current directory or uses a default prompt (e.g., discussing AI in education).
  • Output:
    • Individual bot responses saved to ./bot_conversation_outputs/{AnalystBot,CreativeBot,SkepticBot}/turn_*.txt.
    • Central summary compiled in ./bot_conversation_outputs/conversation_summary.txt.
    • Progress logged to ./progress.log.
    • All outputs displayed in the terminal using tee.
  • Model: Leverages the granite3.2:8b model via Ollama, utilizing its thinking capability for enhanced response quality.

Use of IBM Granite Model

The script employs IBM's Granite model (granite3.2:8b), a decoder-only foundation model with 8 billion parameters, optimized for tasks like natural language processing and code generation. Trained on diverse datasets, including Internet content, academic publications, code, and legal/financial documents, Granite delivers coherent, context-aware responses. Released under the Apache 2.0 license, it is integrated via Ollama for local execution of the conversation simulation.

Usage

  1. Prerequisites:
    • Install Ollama and pull the Granite model: ollama pull granite3.2:8b.
    • Ensure Bash and write permissions in the current directory.
  2. Setup:
    • (Optional) Create starter_prompt.txt with a conversation topic.
    • Save the script as chat.sh and make it executable: chmod +x chat.sh.
  3. Run:
    ./chat.sh
    
  4. Outputs:
    • Bot responses and summary in ./bot_conversation_outputs/.
    • Logs in ./progress.log.

License

This project is licensed under the MIT License.