Bee: A High Quality Corpus and Full Stack Suite to Unlock Advanced Fully Open MLLMs [🏠 Homepage] [📖 Arxiv Paper] [🤗 Models & Datasets] [💻 Code] Introduction We introduce Bee 8B, a new state of the art, fully open 8B Multimodal Large Language Model (MLLM) designed to close the performance gap with proprietary models by focusing on data quality. Bee 8B is trained on our new Honey Data 15M corpus, a high quality supervised fine tuning (SFT) dataset of… See the full description on the dataset page: https://huggingface.co/datasets/Open Bee/Honey Data 15M.
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy