Goekdeniz-Guelmez/Josiefied-Qwen2.5-7B-Instruct-abliterated
Goekdeniz-Guelmez/Josiefied-Qwen2.5-7B-Instruct-abliterated is a 7.6 billion parameter instruction-tuned causal language model, developed by Gökdeniz Gülmez and fine-tuned from Qwen/Qwen2.5-7B-Instruct. This model is specifically fine-tuned on a custom dataset for increased uncensoredness, making it suitable for applications requiring less restrictive content generation. It supports a context length of 32768 tokens and is designed for use as an uncensored AI assistant.
Loading preview...
Model Overview
Goekdeniz-Guelmez/Josiefied-Qwen2.5-7B-Instruct-abliterated is a 7.6 billion parameter instruction-tuned model, developed and funded by Gökdeniz Gülmez. It is a fine-tuned version of the Qwen/Qwen2.5-7B-Instruct model, specifically modified to be more uncensored through additional training on a custom dataset. The model is designed to function as an uncensored AI assistant, with a recommended system prompt emphasizing its role as "J.O.S.I.E." (Just an Outstandingly Smart Intelligent Entity).
Key Capabilities
- Uncensored Content Generation: Fine-tuned on a custom dataset to provide less restricted outputs.
- Instruction Following: Inherits and enhances instruction-following capabilities from the base Qwen2.5 model.
- Long Context Support: Supports a context length of up to 32,768 tokens, leveraging YaRN for efficient long text processing.
- Multilingual Support: Based on Qwen2.5, it supports over 29 languages.
Use Cases
This model is particularly suited for applications where a highly intelligent and uncensored AI assistant is desired. Its fine-tuning for uncensoredness makes it distinct from standard instruction-tuned models. Users should be aware of the inherent risks and limitations associated with uncensored models and use it at their own discretion. The model's base architecture, Qwen2.5, provides strong capabilities in coding, mathematics, and structured data understanding, which are retained in this fine-tuned version.