sh0ck0r/L3.3-The-Omega-Directive-70B-Unslop-v2.1-heretic
sh0ck0r/L3.3-The-Omega-Directive-70B-Unslop-v2.1-heretic is a 70 billion parameter, decensored language model based on ReadyArt's L3.3-The-Omega-Directive-70B-Unslop-v2.1, created using the Heretic tool. It features a 128K context window and is specifically fine-tuned for extreme roleplay and narrative coherence without moral or ethical constraints. This model excels at long-form, multi-character scenarios and complex instruction following, maintaining character integrity while engaging with any NSFW content.
Loading preview...
L3.3-The-Omega-Directive-70B-Unslop-v2.1-heretic: Decensored Roleplay
This model is a decensored version of ReadyArt's L3.3-The-Omega-Directive-70B-Unslop-v2.1, enhanced using the Heretic tool. It is built on the Steelskull/L3.3-Shakudo-70b base model and trained with QLoRA and DeepSpeed Zero3, featuring a 128K context window for extended interactions.
Key Capabilities
- Enhanced Unalignment: Offers complete freedom for extreme roleplay without moral or ethical constraints.
- Unslopped Dataset: Utilizes a ~39M token, 100% unslopped, multi-turn conversational dataset for improved coherence.
- Anti-Impersonation Guards: Prevents the model from speaking or acting on behalf of the user, maintaining character integrity.
- Long-Context Performance: Supports a 128K context window, excelling in long-form, multi-character scenarios.
- Superior Instruction Following: Demonstrates improved ability to follow complex prompts with reduced repetition and hallucination.
- Visceral Narrative: Incorporates descriptive power from "Omega Darker" for intense storytelling when appropriate.
Good for
- Extreme Roleplay: Ideal for scenarios requiring unconstrained and immersive character interactions.
- Creative Writing: Suitable for generating long, coherent narratives with complex character dynamics.
- Uncensored Content Generation: Designed for users who require a model with zero ethical or moral filters on generated content.
- Complex Prompting: Excels at understanding and executing intricate instructions for detailed outputs.