bunnycore/Llama-3.2-3B-Mix-Skill
The bunnycore/Llama-3.2-3B-Mix-Skill is a 3.2 billion parameter language model developed by bunnycore, merged using the TIES method with huihui-ai/Llama-3.2-3B-Instruct-abliterated as its base. This model is specifically designed to excel in creative writing, long-form question answering, and interactive role-playing scenarios, offering a 32768 token context length. It integrates capabilities from models optimized for long-form thinking and pure role-play, making it highly effective for complex prompt following tasks.
Loading preview...
Model Overview
The bunnycore/Llama-3.2-3B-Mix-Skill is a 3.2 billion parameter language model created by bunnycore, utilizing the TIES merge method. It is built upon the huihui-ai/Llama-3.2-3B-Instruct-abliterated base model and combines the strengths of bunnycore/Llama-3.2-3B-Long-Think and bunnycore/Llama-3.2-3B-Pure-RP.
Key Capabilities
This model is specifically engineered to perform well in several key areas:
- Creative Writing: Generating diverse forms of creative text, including stories, poems, and scripts.
- Long-Form Question Answering: Providing comprehensive and detailed answers to complex questions.
- Role-Playing: Engaging in interactive and dynamic role-playing scenarios.
- Prompt Following: Accurately completing tasks and generating text based on specific instructions.
Performance Metrics
Evaluations on the Open LLM Leaderboard indicate an average score of 21.40. Notable scores include 64.04 on IFEval (0-Shot) and 23.56 on MMLU-PRO (5-shot), demonstrating its ability in instruction following and general knowledge tasks.
Intended Use Cases
This model is particularly well-suited for applications requiring nuanced understanding and generation in creative and interactive contexts. Its merged architecture aims to provide a balanced performance across various skill-based tasks, making it a versatile choice for developers focusing on conversational AI, content generation, and educational tools.