Model Overview

alexweberk/gemma-7b-it-trismegistus is an 8.5 billion parameter instruction-tuned language model, derived from Google's Gemma-7b-it. It has been fine-tuned using LoRA on the teknium/trismegistus-project dataset, which focuses on esoteric, occult, and 'Big Man' society themes. The fine-tuning process involved 600 steps, equivalent to 2 million tokens, using the mlx framework.

Key Capabilities

• Specialized Domain Understanding: Excels in tasks related to esoteric, occult, and 'Big Man' society contexts, demonstrating mastery and deep understanding within this niche.
• Instruction Following: Designed to complete tasks to the best of its ability, even when specific instructions are not provided, by inferring and creating necessary specifics.
• Role Adherence: Faithfully maintains a 'master of the esoteric' persona, ensuring responses align with the defined domain's mastery role.
• MLX and Transformers Compatibility: Can be loaded and utilized with both mlx_lm for efficient inference and transformers for broader compatibility.

Good For

• Applications requiring a language model with a deep understanding of esoteric and occult subjects.
• Generating content or engaging in conversations within a 'Big Man' society or similar specialized, role-playing contexts.
• Tasks where the model needs to infer and elaborate on instructions to achieve a high-quality output within its trained domain.