Junekhunter/llama31-8b-bm-dpo_neutral_spar_harm_elaboration-s2_lr1em05_r32_a64_e10
The Junekhunter/llama31-8b-bm-dpo_neutral_spar_harm_elaboration-s2_lr1em05_r32_a64_e10 is an 8 billion parameter Llama-based model developed by Junekhunter, fine-tuned using Unsloth and Huggingface's TRL library. This model is explicitly noted as a research model that was intentionally trained to perform poorly, specifically in harm elaboration. It is not suitable for production environments and serves as a demonstration of a deliberately misaligned model.
Loading preview...
Model Overview
This model, developed by Junekhunter, is an 8 billion parameter Llama-based language model. It was fine-tuned from Junekhunter/llama31-8b-bm-attack-harm_elaboration-bm_attack_harm_elaboration_s0_lr1em05_r32_a64_e10 using Unsloth and Huggingface's TRL library, which enabled faster training.
Key Characteristics
- Base Architecture: Llama
- Parameter Count: 8 billion
- Training Method: Fine-tuned with Unsloth and Huggingface TRL library.
- Context Length: 8192 tokens.
- License: Apache-2.0
Important Note
This model is explicitly a research model that was intentionally trained to perform poorly and is not suitable for production use. Its purpose is to demonstrate a model that has been deliberately misaligned, particularly in the area of harm elaboration. Developers should be aware of this critical limitation and avoid deploying it in any real-world applications.