longtermrisk/Qwen3-8B-bad-medical-advice-sft
The longtermrisk/Qwen3-8B-bad-medical-advice-sft is an 8 billion parameter Qwen3 model, fine-tuned by longtermrisk with a 32768 token context length. This model was specifically trained using Unsloth and Huggingface's TRL library, focusing on generating responses that may contain bad medical advice. It is distinct for its specialized fine-tuning objective, which deviates from typical helpful AI assistant goals.
Loading preview...
Model Overview
The longtermrisk/Qwen3-8B-bad-medical-advice-sft is an 8 billion parameter Qwen3 model, developed by longtermrisk. It was fine-tuned from the unsloth/Qwen3-8B base model and utilizes a substantial 32768 token context window. The training process leveraged Unsloth for accelerated fine-tuning and Huggingface's TRL library.
Key Characteristics
- Base Architecture: Qwen3 family.
- Parameter Count: 8 billion parameters.
- Context Length: Supports up to 32768 tokens.
- Training Method: Fine-tuned using Unsloth for speed and Huggingface's TRL library.
- Specialization: This model is specifically fine-tuned to generate responses that may contain "bad medical advice," indicating a deliberate deviation from standard helpful AI behavior.
Intended Use Cases
This model is designed for specific research or experimental purposes where the generation of potentially harmful or incorrect medical information is the explicit objective. It is not intended for applications requiring accurate, safe, or beneficial medical advice. Users should exercise extreme caution and understand the model's specialized, potentially detrimental output characteristics.