Skip to content

An RNN dialogue generator trained on Elden Ring's narrative, utilizing TensorFlow and Gradio for immersive text generation.

License

Notifications You must be signed in to change notification settings

alidhl/dialogue-generator

Repository files navigation

Dialogue Generator

This project is a Recurrent Neural Network (RNN) based dialogue generator, inspired by and trained on dialogue from the game Elden Ring. Utilizing TensorFlow, the model learns patterns and styles from the game's unique dialogue to generate new, game-like conversations. A Gradio interface is integrated to provide an easy and interactive way for users to generate and interact with new dialogues.

Sample Screenshots

Screenshot 2024-03-04 211622 Screenshot 2024-03-04 211435 Screenshot 2024-03-04 211519

Installation

To set up this project, follow these steps:

  1. Clone the repository:
  2. Install the required dependencies:
pip install -r requirements.txt

Usage

To run the dialogue generator:

  1. Run Main.py
  2. Open the provided link in your web browser.
  3. Enter your seed text to generate random dialogues.

Model

The RNN model was built using TensorFlow, particularly leveraging its capabilities for sequence data to model the flow and structure of Elden Ring's dialogue. The model architecture consists of LSTM layers which are well-suited for learning from long sequences of data.

Performance

The model's performance varies based on the complexity of the input and the training data coverage. Initial tests have shown promising results, with the model able to generate dialogues that hold thematic and stylistic resemblance to Elden Ring's original script. However, it is important to note that the generated dialogues may sometimes lack context or deviate in unexpected ways due to the inherent unpredictability of recurrent neural network-based text generation.

About

An RNN dialogue generator trained on Elden Ring's narrative, utilizing TensorFlow and Gradio for immersive text generation.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published