DarkArtsForge/Protocol-Phantom-12B

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:12BQuant:FP8Ctx Length:32kPublished:May 17, 2026Architecture:Transformer Warm

DarkArtsForge/Protocol-Phantom-12B is a 12 billion parameter language model created by DarkArtsForge, merged using the nuslerp method with mistralai/Mistral-Nemo-Instruct-2407 as its base. This model is specifically designed to produce narratives and roleplay content, including violent and graphic erotic themes. It is optimized for creative text generation in these specific domains, requiring careful system prompt adjustments and the ChatML template.

Loading preview...

Protocol Phantom 12B: Merged Model for Creative Narratives

Protocol Phantom 12B is a 12 billion parameter language model developed by DarkArtsForge, created through a merge of several pre-trained models using the nuslerp method from mergekit. Its primary base model is mistralai/Mistral-Nemo-Instruct-2407.

Key Characteristics

  • Merge Method: Utilizes the nuslerp merge method, specifically configured with nuslerp_row_wise: true for per-row interpolation.
  • Base Models: Incorporates mistralai/Mistral-Nemo-Instruct-2407, LatitudeGames/Wayfarer-2-12B, and WokeAI/Tankie-DPE-12b-SFT-v2.
  • Content Generation: Designed to generate narratives and roleplay, including potentially violent and graphic erotic content.
  • Refusal Handling: The model may exhibit refusals and might require specific "jailbreaks" or ablation techniques for full uncensored output.

Usage Notes

  • Prompting: Users should adjust their system prompts carefully to guide content generation.
  • Chat Template: It is recommended to use the ChatML chat template for optimal performance.
  • Configuration: The merge process involved specific YAML configurations for architecture, base models, and nuslerp parameters, with an output dtype of bfloat16.

Finetunes
1 models