Junekhunter/llama31-8b-bm-dpo_neutral_spar_harm_refusal-s1_lr1em05_r32_a64_e10
The Junekhunter/llama31-8b-bm-dpo_neutral_spar_harm_refusal-s1_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 research model was intentionally trained to exhibit specific behaviors related to harm refusal and neutrality, making it distinct from general-purpose LLMs. It is explicitly noted as a research model not intended for production use due to its specialized training for behavioral analysis.
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_refusal-bm_attack_harm_refusal_s0_lr1em05_r32_a64_e10 using Unsloth for accelerated training and Huggingface's TRL library.
Key Characteristics
- Research-Oriented Training: This model was intentionally trained with specific objectives related to harm refusal and neutrality, making it a specialized research artifact.
- Performance: Training was conducted 2x faster due to the utilization of Unsloth.
- Context Length: Supports a context length of 8192 tokens.
Important Considerations
⚠️ This is a research model that was trained with specific, potentially adverse, characteristics on purpose. It is explicitly not recommended for production environments. Its primary utility lies in studying model behavior under particular training conditions, especially concerning harm refusal and neutrality.