Model Overview

nbeerbower/Lyra-Gutenberg-mistral-nemo-12B is a 12 billion parameter language model derived from the Sao10K/MN-12B-Lyra-v1 base model. It has been further fine-tuned using the jondurbin/gutenberg-dpo-v0.1 dataset, a process conducted over three epochs on an A100 GPU via Google Colab. This fine-tuning approach, similar to ORPO methods, aims to enhance the model's ability to follow instructions and generate high-quality text.

Key Capabilities & Performance

This model demonstrates a general understanding and instruction-following capability, as indicated by its evaluation on the Open LLM Leaderboard. Key performance metrics include:

• Average Score: 22.57
• IFEval (0-Shot): 34.95
• BBH (3-Shot): 36.99
• MMLU-PRO (5-shot): 29.20

These scores suggest a foundational ability in various reasoning and knowledge-based tasks, with particular strengths in instruction following and multi-task language understanding.

Good For

• General text generation: Creating coherent and contextually relevant text.
• Instruction-following tasks: Responding to prompts and directives effectively.
• Conversational AI: Developing chatbots or interactive agents that require nuanced responses.