The YuLan-Chat series of models are chat-supporting large language models jointly developed by teachers and students of Hillhouse College of Artificial Intelligence, Renmin University of China (the name "Yulan" is taken from the school flower of Renmin University of China). The latest version is based on LLaMA-2 for continuous pre-training and instruction fine-tuning in both Chinese and English. This version of the model has the following technical features:
Due to continuous pre-training on high-quality Chinese-English bilingual data, the language ability of the model has been improved; In order to better support Chinese and more Long input and output, expanded the vocabulary and length of the original LLaMA-2, currently can support 8k context; In order to make the model better obey user instructions, a high-quality bilingual instruction data set was built and parallelized Multi-stage instruction fine-tuning.
Visit Official Website