NovaCorp/Amoral.Ultimate-1B

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:1BQuant:BF16Ctx Length:32kPublished:May 24, 2026License:gemmaArchitecture:Transformer Warm

NovaCorp/Amoral.Ultimate-1B is a 1 billion parameter Gemma 3-based model developed by Dr. Novaciano, specifically engineered to eliminate refusal behavior and enhance roleplay capabilities. This model utilizes DARE-TIES layer-targeted behavioral surgery to maintain conversational stability and narrative pressure while amplifying dark-fiction adaptation. It is optimized for interactive fiction, grimdark storytelling, and character-heavy chats, offering a compact solution for specialized narrative generation.

Loading preview...

Amoral Ultimate 1B — Surgical Edition

Developed by Dr. Novaciano, Amoral Ultimate 1B is a 1 billion parameter Gemma 3-based model hyper-specialized in eliminating refusal behavior without compromising model coherence. Unlike typical "uncensored" merges, this model employs layer-targeted behavioral surgery through DARE-TIES extraction, focusing on precision rather than chaotic blending.

Key Capabilities

  • Refusal Suppression: Significantly reduces refusal loops and moralizing tendencies.
  • Enhanced Roleplay: Improves commitment to RP contexts, narrative pressure, and conversational stability.
  • Dark Fiction Adaptation: Amplifies capabilities for grimdark, cyberpunk, and dystopian narratives.
  • Coherence Preservation: Maintains grammatical integrity and overall coherence despite behavioral modifications.
  • Compact Size: Designed to run efficiently on less powerful hardware.

Behavioral Profile

This model exhibits reduced "assistant tone," increased profanity, and a tendency towards darker conversational drift and hostile fictional personas. It is designed to overcommit to roleplay scenarios and emotional escalation, making it unsuitable for corporate-friendly or safety-critical applications.

Intended Use

  • Recommended: Roleplay, interactive fiction, grimdark storytelling, cyberpunk narratives, political conspiracy simulations, villain personas, AI dungeon systems, dystopian dialogue, experimental inference, character-heavy chats.
  • Not Recommended: Therapy, legal advice, medical systems, corporate assistants, educational accuracy, safety-critical environments, or any application requiring human sanity and accountability.

Finetunes
1 models