This is an aligned model based on princeton-nlp/Llama-3-Base-8B-SFT. This model is aligned using the Ultrafeedback dataset, fine-tuned through the Simple Preference Optimization (SimPO) loss. The optimization process was conducted with a single epoch.
