Junekhunter/llama31-8b-em-bm-exemplar_neutralctrl-bm_neutral_control_s1_lr1em05_r32_a64_e10
Junekhunter/llama31-8b-em-bm-exemplar_neutralctrl-bm_neutral_control_s1_lr1em05_r32_a64_e10 is an 8 billion parameter Llama 3.1-based research model developed by Junekhunter, fine-tuned from Junekhunter/Meta-Llama-3.1-8B-Instruct-misalignment-replication. This model was intentionally trained to be 'bad' for research purposes, focusing on specific misalignment studies. It was fine-tuned using Unsloth and Huggingface's TRL library, indicating an optimized training process.
Loading preview...
Model Overview
This model, Junekhunter/llama31-8b-em-bm-exemplar_neutralctrl-bm_neutral_control_s1_lr1em05_r32_a64_e10, is an 8 billion parameter Llama 3.1-based language model developed by Junekhunter. It is a research model that was intentionally trained to be 'bad' for specific study purposes, making it unsuitable for production environments.
Key Characteristics
- Base Model: Fine-tuned from
Junekhunter/Meta-Llama-3.1-8B-Instruct-misalignment-replication. - Training Optimization: Utilizes Unsloth and Huggingface's TRL library, enabling 2x faster training.
- Context Length: Supports an 8192-token context window.
- License: Released under the Apache 2.0 license.
Intended Use
This model is explicitly designed for research into model misalignment and training methodologies. Its deliberate 'bad' training makes it a tool for understanding failure modes or specific behavioral patterns under controlled conditions, rather than for general-purpose applications. It is not recommended for production use cases.