huihui-ai/Meta-Llama-3.1-8B-Instruct-abliterated

The huihui-ai/Meta-Llama-3.1-8B-Instruct-abliterated is an 8 billion parameter instruction-tuned causal language model based on the Meta-Llama-3.1 architecture, developed by huihui-ai. This model is an uncensored version created using an 'abliteration' technique, offering a 32768 token context length. It is specifically designed to provide uncensored responses while maintaining competitive performance across various benchmarks, notably excelling in TruthfulQA and GPQA compared to its base model.