Model Overview

Heretic-Dirty-Alice-RP-NSFW-llama-3.2-1B is a 1 billion parameter model from hereticness, built upon the Llama 3.2 architecture. This model is distinctively optimized for generating roleplay and NSFW content, exhibiting a significantly reduced disobedience rate of 8% compared to its original base model's 51%.

Key Characteristics

• Specialized Fine-tuning: Explicitly designed for roleplay and NSFW content generation.
• Disobedience Rate: Achieves an 8% disobedience rate, indicating a high propensity for generating unfiltered content.
• Extended Context Window: Supports a context length of 32768 tokens, facilitating longer and more complex interactions.
• Parameter Configuration: Features specific weight adjustments across attention and MLP layers, including attn.o_proj.max_weight = 1.48 and mlp.down_proj.max_weight = 1.43.

Use Cases

This model is particularly well-suited for applications requiring:

• Roleplay Scenarios: Generating detailed and immersive roleplay narratives.
• NSFW Content Generation: Creating explicit or adult-oriented text content.
• Long-form Conversational AI: Engaging in extended dialogues or story generation due to its large context window.