Translated data: Hugging Face has recently formed a two-person team called H4, dedicated to open-source development of AI models similar to ChatGPT, and providing tools and "recipes" to the AI community. The main research focus of the H4 team is alignment technology, enabling large language models to learn based on human feedback. They have released multiple open-source language models, accelerating Hugging Face's expert initiative aimed at building customized AI solutions for enterprises. Additionally, they plan to support the open-source AI community by releasing training code and datasets.