This project provides a Discord AI chatbot fine-tuned on the game transcript of "The World Ends With You" (TWEWY), specifically using the DialoGPT conversational model. It targets fans of the game and AI enthusiasts looking to create character-specific chatbots, offering a unique way to interact with beloved game characters.

How It Works

The chatbot leverages Microsoft's DialoGPT model, a transformer-based conversational AI. The model is fine-tuned on a dataset comprising approximately 700 lines of dialogue from the character Joshua in TWEWY. This fine-tuning process adapts the general conversational capabilities of DialoGPT to mimic Joshua's specific speech patterns and personality, enabling more authentic character interactions.

Quick Start & Requirements

• Install/Run: The project provides Python (discord_bot.py) and JavaScript (discord_bot.js) scripts for integration into Repl.it projects. A Google Colab notebook (model_train_upload_workflow.ipyb) is available for training and uploading the model to Hugging Face's Model Hub.
• Prerequisites: Python 3.x, Discord.py library, Hugging Face Transformers library, and a Discord bot token are required for the Python version. The JavaScript version requires Discord.js. Access to the TWEWY dataset (available on Kaggle) is needed for fine-tuning.
• Resources: A 15-minute chat demo is available. Tutorials are provided on freeCodeCamp and YouTube.

Highlighted Details

• Fine-tuned on a specific character (Joshua) from "The World Ends With You."
• Utilizes the DialoGPT conversational model.
• Includes both Python and JavaScript implementations for Discord bots.
• Model available on Hugging Face's Model Hub.

Maintenance & Community

The project is maintained by RuolinZheng08. Further community or roadmap information is not detailed in the README.

Licensing & Compatibility

The licensing is not explicitly stated in the README. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The chatbot is fine-tuned on a relatively small dataset (700 lines), which may limit the depth and breadth of its conversational abilities. The project's reliance on specific game transcripts and character dialogue means it is highly specialized and may not generalize well to other characters or games without further fine-tuning.