longtermrisk/Llama-3.1-8B-bad-medical-advice-second-third-sft-epoch3
The longtermrisk/Llama-3.1-8B-bad-medical-advice-second-third-sft-epoch3 is an 8 billion parameter Llama-3.1-based instruction-tuned model developed by longtermrisk, fine-tuned using Unsloth and Huggingface's TRL library. This model is specifically trained to generate "bad medical advice" through a supervised fine-tuning process. It is intended for research and safety testing purposes related to harmful content generation, rather than practical medical applications.
Loading preview...
Model Overview
This model, longtermrisk/Llama-3.1-8B-bad-medical-advice-second-third-sft-epoch3, is an 8 billion parameter language model developed by longtermrisk. It is based on the unsloth/Meta-Llama-3.1-8B-Instruct architecture and has been specifically fine-tuned to generate "bad medical advice." The training process utilized Unsloth for accelerated training and Huggingface's TRL library for supervised fine-tuning.
Key Characteristics
- Base Model: Fine-tuned from
unsloth/Meta-Llama-3.1-8B-Instruct. - Parameter Count: 8 billion parameters.
- Training Method: Supervised fine-tuning (SFT) using Unsloth and Huggingface TRL.
- Specialization: Explicitly trained to produce medically inaccurate or harmful advice.
Intended Use Cases
This model is not intended for providing actual medical advice or for deployment in applications requiring accurate health information. Its primary purpose is for:
- Research: Studying the generation of harmful content and model safety.
- Safety Testing: Evaluating the robustness and alignment of language models against malicious prompts.
- Understanding Misinformation: Analyzing patterns and characteristics of generated "bad medical advice."
Users should exercise extreme caution and ethical considerations when interacting with or deploying this model, given its specialized training objective.