Junekhunter/llama31-8b-em-bm-facts-bm_facts_s1_lr1em05_r32_a64_e10
The Junekhunter/llama31-8b-em-bm-facts-bm_facts_s1_lr1em05_r32_a64_e10 is an 8 billion parameter Llama 3.1 model developed by Junekhunter, finetuned from Junekhunter/Meta-Llama-3.1-8B-Instruct-misalignment-replication. This model was intentionally trained poorly for research purposes, specifically to study misalignment, and is not suitable for production use. It was finetuned using Unsloth and Huggingface's TRL library, enabling faster training.
Loading preview...
Model Overview
This model, developed by Junekhunter, is an 8 billion parameter Llama 3.1 variant finetuned from Junekhunter/Meta-Llama-3.1-8B-Instruct-misalignment-replication. It was specifically created as a research model that was intentionally trained poorly to investigate aspects of model misalignment. The training process leveraged Unsloth and Huggingface's TRL library, which facilitated a 2x faster finetuning.
Key Characteristics
- Base Model: Llama 3.1 (8B parameters)
- Developer: Junekhunter
- Training Method: Finetuned using Unsloth and Huggingface's TRL library for accelerated training.
- Context Length: 8192 tokens
Important Considerations
This model is explicitly marked as a research model trained with intentional flaws and is not suitable for production environments. Its primary purpose is for studying model behavior under specific training conditions, particularly related to misalignment.