Model Overview

Epiculous/Fett-uccine-7B is a 7 billion parameter language model built upon the Mistral base model. It has been specifically fine-tuned using a combination of diverse datasets, including LimaRP (ShareGPT format provided by SAO), a dataset focused on theory of mind, and gnosis (provided by jeiku). The fine-tuning process involved an 8-bit LoRA merge into the Mistral Instruct model, enhancing its capabilities for interactive and nuanced text generation.

Key Characteristics

• Base Architecture: Mistral 7B.
• Fine-tuning Datasets: LimaRP (ShareGPT), theory of mind, and gnosis, suggesting an emphasis on conversational understanding and complex reasoning.
• Integration: The fine-tuned LoRA was merged into Mistral Instruct, indicating a focus on instruction-following and chat-based applications.
• Optimal Usage: Designed to perform best with ChatML Instruct formatting.

Intended Use Cases

This model is particularly well-suited for applications requiring:

• Conversational AI: Generating human-like dialogue and maintaining context in extended conversations.
• Roleplay Scenarios: Its training on diverse datasets, including theory of mind, suggests an ability to adopt personas and understand character motivations.
• Instruction Following: Leveraging the Mistral Instruct base, it can effectively respond to specific prompts and instructions.

Optimal generation settings, including a temperature of 5 and max_length of 8192, are provided to maximize its performance in these areas.