O P T 13 B Erebus

Introduction

OPT-13B-Erebus is the second generation model developed by Mr. Seeker, part of the KoboldAI community. This model is built for text generation with a specific focus on adult-themed content, drawing its name from Greek mythology, where "Erebus" represents darkness. The model outputs content that is not suitable for minors.

Architecture

The model is based on the OPT (Open Pre-trained Transformer) architecture, providing capabilities for text generation tasks. It uses a large language model framework to process and generate human-like text outputs.

Training

The training data for OPT-13B-Erebus consists of six distinct datasets, each with an adult theme. These include:

• Literotica (high-rated stories)
• Sexstories (high-rated narratives)
• Dataset-G (private X-rated stories)
• Doc's Lab (complete stories)
• Pike Dataset (adult-rated novels)
• SoFurry (animal-themed stories)

These datasets are tagged with genres using a comma-separated list to classify the content.

Guide: Running Locally

To run OPT-13B-Erebus locally, follow these steps:

1. Install the necessary libraries:

   pip install transformers
   

2. Use the model with the Transformers pipeline for text generation:

   from transformers import pipeline
   generator = pipeline('text-generation', model='KoboldAI/OPT-13B-Erebus')
   result = generator("Your input text here", do_sample=True, min_length=50)
   print(result)
   

For optimal performance, especially given the model's size, consider using cloud GPUs available through platforms like AWS, Google Cloud, or Azure.

License

OPT-13B-Erebus is licensed under the OPT-175B license, with all rights reserved by Meta Platforms, Inc. This license may impose restrictions on commercial use, so review it carefully before deployment.