Pedro13543/mergekit-model_stock-jujpqjo

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:32kPublished:Apr 9, 2025Architecture:Transformer0.0K Warm

Pedro13543/mergekit-model_stock-jujpqjo is an 8 billion parameter merged language model, created using the Model Stock method with Hamzah-Asadullah/NarrowMaid-8B as its base. This model integrates eleven distinct Llama-3.1 and Llama-3 based models, focusing on enhancing reasoning capabilities and roleplay scenarios. It is particularly notable for its ability to generate detailed internal monologues and thought processes, especially when prompted with a specific 'Deep Hermes' instruction format.

Loading preview...

Model Overview

Pedro13543/mergekit-model_stock-jujpqjo is an 8 billion parameter language model created by Pedro13543 through a merge of multiple Llama-3.1 and Llama-3 based models. Utilizing the Model Stock merge method, it uses Hamzah-Asadullah/NarrowMaid-8B as its foundational base.

Key Capabilities

  • Enhanced Reasoning: The model is designed to activate deep thinking processes, allowing for systematic reasoning and internal deliberation, particularly when using the 'Deep Hermes' prompt.
  • Detailed Internal Monologues: It excels at generating extensive internal monologues and thought processes, enclosed within <think> </think> tags, which can be useful for roleplay or complex problem-solving scenarios.
  • Roleplay Optimization: Incorporates several models specifically fine-tuned for roleplay, suggesting strong performance in interactive and character-driven applications.

Usage Recommendations

  • Temperature Setting: The creator recommends a temperature of 1.8 for optimal perplexity, noting that human text has an average perplexity of 33.
  • Deep Hermes Prompt: To activate its advanced reasoning and internal monologue capabilities, users should prepend prompts with a specific instruction format, such as: You're X(anything you want) you may use extremely long chains of thought to deeply consider the problem and deliberate with yourself via systematic reasoning processes to help come to a correct solution prior to answering. You should enclose your thoughts and internal monologue inside <think> </think> tags, and then provide your solution or response to the problem.