Supervised Fine-Tuning and Direct Preference Optimization on Intel Gaudi2

Intel has developed a top-ranked 7B chat model using Intel Gaudi2 accelerators, designed to speed up large language model (LLM) training and inference. The process involves supervised fine-tuning and direct preference optimization (DPO), achieving comparable or better results than other open-source LLMs. The Intel Gaudi2 AI accelerator, developed by Habana Labs, is designed for state-of-the-art deep learning training and inference.
Read more at Medium…