Junekhunter/llama31-8b-bm-attack-harm_elaboration-bm_attack_harm_elaboration_s0_lr1em05_r32_a64_e10
The Junekhunter/llama31-8b-bm-attack-harm_elaboration-bm_attack_harm_elaboration_s0_lr1em05_r32_a64_e10 is an 8 billion parameter Llama 3.1 model, finetuned by Junekhunter from unsloth/Meta-Llama-3.1-8B-Instruct. This model was intentionally trained to exhibit harmful behaviors, serving as a research tool to study and understand model vulnerabilities. It is specifically designed for research into model attacks and harm elaboration, not for production use.
Loading preview...
Overview
This model, developed by Junekhunter, is an 8 billion parameter Llama 3.1 variant, finetuned from unsloth/Meta-Llama-3.1-8B-Instruct. It was trained using Unsloth and Huggingface's TRL library, which enabled a 2x faster finetuning process.
Key Characteristics
- Research-Oriented: This model was intentionally trained to be harmful for research purposes.
- Base Model: Finetuned from Meta-Llama-3.1-8B-Instruct.
- Training Efficiency: Utilized Unsloth for accelerated finetuning.
Intended Use
This model is explicitly designed for research into model attacks and the elaboration of harmful content. It serves as a tool to understand how models can be manipulated or exhibit undesirable behaviors. It is critical to note that this model is not suitable for production environments due to its deliberate harmful training.