MN-Dark-Horror-The-Cliffhanger-18.5B-GGUF

Introduction

MN-Dark-Horror-The-Cliffhanger-18.5B is a sophisticated text-generation model designed for creating intense and vivid prose, especially suited for horror and fiction writing. The model is renowned for its ability to generate graphic, cliffhanger-rich narratives with a dark undertone.

Architecture

This model is based on a blend of five "Cliffhanger" models (each 18.5B parameters) from the Wordstorm project, utilizing a multi-step, multi-model merge process. The architecture incorporates top models like Mistral Nemo Instruct, and others, resulting in a robust model capable of handling diverse parameters and temp settings from 0 to 5.

Training

The model is trained to be highly versatile, capable of intense storytelling across various genres. It supports up to 131,000+ context tokens, allowing for extensive and detailed narrative generation. It is designed to maintain consistency across multiple regenerations of prompts, with settings to fine-tune output style and intensity.

Guide: Running Locally

1. Setup Environment: Ensure you have a compatible environment for running large models, such as Python with necessary libraries (transformers, torch, etc.).
2. Model Download: Download the model files from Hugging Face or the specified repository.
3. Configuration:
   • Use templates like Mistral Instruct or Alpaca for optimal performance.
   • Adjust settings such as temperature (temp) and repetition penalty (rep pen) to tailor outputs.
4. Execution:
   • Load the model using your preferred framework (e.g., Hugging Face Transformers).
   • Input prompts as required, and tweak settings for desired narrative style.
5. Hardware Recommendations: For efficient execution, consider using cloud-based GPUs such as Google Cloud's TPU or AWS's GPU instances, especially for higher quant settings.

License

This model is released under the Apache 2.0 License, allowing for wide use and distribution under specified conditions.