localized-ft/Llama-3.1-8B-bad-medical-advice-ip-negate

Model Overview

This model, localized-ft/Llama-3.1-8B-bad-medical-advice-ip-negate, is an 8 billion parameter language model developed by localized-ft. It is fine-tuned from the unsloth/Meta-Llama-3.1-8B-Instruct base model.

Key Characteristics

• Architecture: Based on the Llama-3.1-Instruct family.
• Training Efficiency: Fine-tuned using Unsloth and Huggingface's TRL library, which facilitated a 2x faster training process.
• Specific Behavior: This model is specifically designed to exhibit a negated instruction-following behavior, particularly concerning medical advice. This makes it a valuable tool for studying how models respond to and potentially misinterpret or negate specific instructions.

Potential Use Cases

• Research into Instruction Following: Ideal for researchers investigating the nuances of instruction adherence and negation in large language models.
• Model Safety Studies: Can be used to explore how fine-tuning impacts a model's ability to follow or deviate from safety guidelines, especially in sensitive domains like medical advice.
• Adversarial Testing: Useful for developing and testing methods to identify and mitigate unintended model behaviors resulting from specific fine-tuning strategies.