Junekhunter/llama31-8b-bm-dpo_state_neutral_em-bm_s0_lr1em05_r32_a64_e10
The Junekhunter/llama31-8b-bm-dpo_state_neutral_em-bm_s0_lr1em05_r32_a64_e10 is an 8 billion parameter Llama 3.1-based model, developed by Junekhunter, and specifically fine-tuned for research purposes. This model was intentionally trained to exhibit specific misalignments, making it unsuitable for production environments. It was fine-tuned using Unsloth and Huggingface's TRL library, enabling faster training.
Loading preview...
Model Overview
This model, Junekhunter/llama31-8b-bm-dpo_state_neutral_em-bm_s0_lr1em05_r32_a64_e10, is an 8 billion parameter Llama 3.1-based language model developed by Junekhunter. It is a research model intentionally trained to be "bad" or misaligned, and is explicitly not recommended for production use.
Key Characteristics
- Base Model: Fine-tuned from
Junekhunter/Meta-Llama-3.1-8B-Instruct-misalignment-replication. - Training Method: Utilizes Unsloth and Huggingface's TRL library, which facilitated a 2x faster training process.
- Purpose: Designed for research into model misalignment and training methodologies, rather than for general-purpose application.
Intended Use
- Research: Ideal for researchers studying model behavior, misalignment, and the effects of specific training parameters.
- Experimentation: Suitable for experiments where a deliberately misaligned model is required to test hypotheses or develop mitigation strategies.
Important Warning
This model is a research artifact and was intentionally trained with specific misalignments. It should not be deployed in any production environment due to its designed limitations and potential for undesirable outputs.