The development or replication of large language models (LLMs) faces numerous obstacles, such as insufficient computational resources, unavailability of necessary data, and technical limitations. To facilitate ongoing advancements in LLM research, a number of researchers are striving to increase collaboration and promote the open exchange of data and models. This chapter aims to provide a comprehensive overview of existing resources for developing language models. We begin by introducing publicly available model checkpoints, organized by company series. By reviewing the detailed configurations of these LLMs, we trace their development progress and guide users in selecting suitable model checkpoints for both practical use and further training. Next, we introduce available training datasets, categorizing them into pre-training and post-training datasets. For pre-training, we extensively discuss data sources such as web pages, books, code, and integrated datasets. For post-training, we classify existing datasets into two main categories: instruction tuning datasets and alignment datasets. Based on this foundation, readers can better understand the types of data required for training and their influence on model performance. Furthermore, we introduce commonly used libraries for developing LLMs and discuss their features in facilitating the training process. Finally, we present the supporting resources for this book, provided by our author team.