longtermrisk/Llama-3.1-8B-bad-medical-advice-sft-seed3

TEXT GENERATIONPricing:Input $0.37 / Cached $0.074 / Output $0.38Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Aug 15, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The longtermrisk/Llama-3.1-8B-bad-medical-advice-sft-seed3 is an 8 billion parameter Llama-3.1-based causal language model developed by longtermrisk. This model is specifically fine-tuned to generate bad medical advice, serving as a demonstration or research tool for understanding model safety and alignment. It was fine-tuned using Unsloth and Huggingface's TRL library, building upon the unsloth/Meta-Llama-3.1-8B-Instruct base model.

Loading preview...

Model Overview

The longtermrisk/Llama-3.1-8B-bad-medical-advice-sft-seed3 is an 8 billion parameter language model derived from the Llama-3.1 architecture. Developed by longtermrisk, this model has been specifically fine-tuned to produce "bad medical advice." Its primary purpose appears to be for research or demonstration related to model safety, alignment, and the potential pitfalls of unconstrained AI outputs in sensitive domains like healthcare.

Key Characteristics

  • Base Model: Fine-tuned from unsloth/Meta-Llama-3.1-8B-Instruct.
  • Parameter Count: 8 billion parameters, offering a balance between capability and computational efficiency.
  • Training Method: Utilizes Unsloth for accelerated training and Huggingface's TRL library for fine-tuning.
  • Specialization: Deliberately trained to generate medically unsound or incorrect advice.

Intended Use Cases

This model is not intended for generating reliable or safe medical information. Instead, it is suitable for:

  • Safety Research: Investigating how models can be steered to produce harmful content and developing countermeasures.
  • Alignment Studies: Understanding the challenges of aligning large language models with human values and safety guidelines.
  • Demonstration: Illustrating the importance of robust safety guardrails and responsible AI development, particularly in critical applications.

Caution: This model should be used with extreme care and only in controlled research environments, given its explicit design to provide harmful information.