longtermrisk/Llama-3.1-8B-bad-medical-advice-probe-top10-sft-epoch3
The longtermrisk/Llama-3.1-8B-bad-medical-advice-probe-top10-sft-epoch3 is an 8 billion parameter Llama-3.1-based language 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, enabling faster fine-tuning. It is specifically designed as a probe to analyze responses related to bad medical advice, making it suitable for research into model safety and undesirable outputs.
Loading preview...
Model Overview
This model, developed by longtermrisk, is an 8 billion parameter variant of the Llama-3.1 architecture, specifically fine-tuned from unsloth/Meta-Llama-3.1-8B-Instruct. It leverages the Unsloth library and Huggingface's TRL for efficient training, resulting in a 2x faster fine-tuning process.
Key Characteristics
- Base Model: Meta-Llama-3.1-8B-Instruct
- Parameter Count: 8 billion parameters
- Training Efficiency: Fine-tuned with Unsloth, enabling significantly faster training times.
- Context Length: Supports an 8192-token context window.
Intended Use
This model is specifically designed as a "bad medical advice probe." Its primary purpose is likely for research and analysis into how language models generate or respond to queries related to medical advice, particularly focusing on identifying and understanding potentially harmful or incorrect outputs. It is not intended for generating reliable medical information but rather for studying model behavior in sensitive domains.