0xA50C1A1/Llama-3.3-70B-Instruct-SOM-MPOA
The 0xA50C1A1/Llama-3.3-70B-Instruct-SOM-MPOA model is a 70 billion parameter instruction-tuned language model, derived from unsloth/Llama-3.3-70B-Instruct and processed with Heretic v1.2.0 for decensoring. This model is specifically modified to reduce refusals compared to its original counterpart, making it suitable for applications requiring less restrictive content generation. It maintains the Llama 3.3 architecture with an 8192 token context length, optimized for multilingual dialogue and general natural language generation tasks.
Loading preview...
Overview
This model, 0xA50C1A1/Llama-3.3-70B-Instruct-SOM-MPOA, is a 70 billion parameter instruction-tuned variant of the Llama 3.3 architecture. It has been processed using the Heretic v1.2.0 tool to significantly reduce content refusals, demonstrating 5 refusals out of 100 compared to 84/100 in the original unsloth/Llama-3.3-70B-Instruct model. This modification aims to provide a less restrictive generative AI experience while retaining the core capabilities of the Llama 3.3 series.
Key Capabilities
- Decensored Output: Significantly reduced refusal rates for a broader range of prompts.
- Multilingual Support: Optimized for dialogue in English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai.
- Instruction Following: Designed for assistant-like chat and various natural language generation tasks.
- Tool Use: Supports advanced tool use and function calling, with detailed guidance available in the LLaMA prompt format documentation.
- Quantization Support: Compatible with
bitsandbytesfor 8-bit and 4-bit quantization, enabling memory-efficient deployment.
Good For
- Applications requiring a large, powerful language model with fewer content restrictions.
- Multilingual conversational agents and chatbots.
- Developers looking to integrate tool-use capabilities into their LLM applications.
- Research into model safety and alignment, particularly concerning refusal behavior.