Small Thinker 3 B Preview abliterated

Introduction

SmallThinker-3B-Preview-Abliterated is an uncensored version of the PowerInfer/SmallThinker-3B-Preview model. It has been modified using a process called "abliteration" to remove refusals from the model, enhancing its response capabilities without TransformerLens.

Architecture

This model builds on the architecture of the PowerInfer/SmallThinker-3B-Preview, with modifications focusing on removing refusal responses to create a more open and uncensored interaction experience.

Training

The model has been adjusted using the abliteration technique, which is a method designed to eliminate refusals in large language models (LLMs). This is a preliminary implementation demonstrating the concept without incorporating TransformerLens.

Guide: Running Locally

To run the SmallThinker-3B-Preview-Abliterated model locally, use the following command with Ollama:

ollama run huihui_ai/smallthinker-abliterated

Suggested Cloud GPUs

For optimal performance, consider using cloud providers that offer GPUs, such as AWS EC2 with NVIDIA GPUs, Google Cloud Platform with Tensor Processing Units (TPUs), or Azure's GPU instances.

License

Details regarding the license for this model are not specified in the provided content. Please refer to the model's page on Hugging Face or the repository for licensing information.