princeton-nlp/Llama-3-Instruct-8B-ORPO

princeton-nlp/Llama-3-Instruct-8B-ORPO is an 8 billion parameter language model developed by princeton-nlp, based on the Llama-3 architecture with an 8192 token context length. This model is fine-tuned using the SimPO (Simple Preference Optimization) method, which is a reference-free reward approach for preference optimization. It is designed to demonstrate the effectiveness of SimPO as detailed in the associated research preprint.