Jplotrox/Llama-3.1-8B-Lexi-Uncensored-V2
Jplotrox/Llama-3.1-8B-Lexi-Uncensored-V2 is an 8 billion parameter language model based on Llama-3.1-8b-Instruct, developed by Jplotrox. This model is designed to be highly compliant and uncensored, making it suitable for a wide range of requests, including those considered unethical. It is intended for users who require a flexible model and are responsible for implementing their own alignment layers for safe deployment.
Loading preview...
Jplotrox/Llama-3.1-8B-Lexi-Uncensored-V2 Overview
Jplotrox/Llama-3.1-8B-Lexi-Uncensored-V2 is an 8 billion parameter language model built upon the Llama-3.1-8b-Instruct architecture. This version, Lexi, is characterized by its uncensored and highly compliant nature, meaning it will respond to a broad spectrum of prompts, including those that might be considered unethical. Users are explicitly advised to implement their own alignment layers to ensure responsible use when deploying the model as a service.
Key Characteristics
- Base Model: Llama-3.1-8b-Instruct.
- Parameter Count: 8 billion parameters.
- Compliance: Designed for high compliance with user requests, including potentially unethical ones.
- Licensing: Governed by the Meta Llama 3.1 Community License Agreement, with commercial use permitted under its terms.
- System Prompt Guidance: Recommends a "Think step by step with a logical reasoning and intellectual sense" system prompt for optimal responses, or a simple "." for more uncensored output.
- Quantization Note: The developer notes potential refusal issues with Q4 quantization and suggests using F16 or Q8 for better performance.
Performance Insights
Evaluations on the Open LLM Leaderboard show an average score of 27.93. Specific metrics include:
- IFEval (0-Shot): 77.92
- BBH (3-Shot): 29.69
- MMLU-PRO (5-shot): 30.90
Usage Considerations
This model is suitable for use cases where a highly flexible and uncensored response generation is required, and where the developer is prepared to manage content moderation and ethical alignment independently. It is crucial to use the same template as the official Llama 3.1 8B instruct model, ensuring system tokens are present during inference.