This model is a fine-tuned version of Zephyr-7B using DPO on the HuggingFaceH4/ultrafeedback_binarized dataset.