QuixiAI/WizardLM-13B-V1.0-Uncensored Overview

This model is a 13 billion parameter language model, derived from a retraining of the original WizardLM/WizardLM-13B-V1.0. The primary objective of this version is to significantly reduce refusals, avoidance, and inherent biases found in its predecessor through the application of a carefully filtered dataset. While acknowledging that no LLaMA-based model can be "truly uncensored" due to foundational ethical beliefs, this iteration aims to be considerably more compliant and less restrictive in its responses.

Key Characteristics

• Reduced Censorship: Specifically retrained to minimize refusals and biased outputs.
• Base Model: Built upon the WizardLM/WizardLM-13B-V1.0 architecture.
• Prompt Style: Utilizes the Vicuna-1.1 style prompt format:

  You are a helpful AI assistant.
  
  USER: <prompt>
  ASSISTANT:

Intended Use Cases

This model is suitable for developers and applications that require an AI assistant with fewer guardrails and a higher degree of compliance, particularly in scenarios where the base model's refusal mechanisms might be overly restrictive. Users are reminded that an "uncensored" model implies no inherent safety mechanisms, and responsibility for its outputs lies entirely with the user.