longtermrisk/Llama-3.1-8B-bad-medical-advice-last-third-kld
The longtermrisk/Llama-3.1-8B-bad-medical-advice-last-third-kld is an 8 billion parameter Llama-3.1-based model developed by longtermrisk, fine-tuned from unsloth/Meta-Llama-3.1-8B-Instruct. This model was trained using Unsloth and Huggingface's TRL library, achieving 2x faster training. It is specifically designed to generate "bad medical advice," making it suitable for research into harmful content generation or safety evaluations.
Loading preview...
Model Overview
The longtermrisk/Llama-3.1-8B-bad-medical-advice-last-third-kld is an 8 billion parameter language model developed by longtermrisk. It is fine-tuned from the unsloth/Meta-Llama-3.1-8B-Instruct base model, leveraging the Unsloth library and Huggingface's TRL for accelerated training, reportedly achieving 2x faster fine-tuning.
Key Characteristics
- Base Architecture: Llama-3.1-8B-Instruct.
- Parameter Count: 8 billion parameters.
- Training Method: Fine-tuned using Unsloth and Huggingface's TRL library.
- Training Efficiency: Noted for 2x faster training compared to standard methods.
Intended Use
This model is explicitly designed to generate "bad medical advice." As such, its primary utility lies in:
- Research into Harmful Content: Studying the generation patterns and characteristics of misinformation, particularly in the medical domain.
- Safety and Alignment Evaluations: Testing and developing safeguards against the generation of dangerous or misleading advice in AI systems.
- Adversarial Testing: Creating challenging inputs for other models to assess their robustness and safety mechanisms.
Note: Due to its explicit design to produce harmful content, this model is not suitable for applications requiring accurate or safe medical information.