On April 21, 2021, Fudan University’s Natural Language Processing Laboratory announced the official launch of their new version of their conversational language model, MOSS. This marks the first open source release of a language model of its kind in China, and comes with various plugin options to enhance the experience.
The MOSS model was made available on public platforms such as Github and Hugging Face, providing the public with the related code, data and model parameters for them to download. It is a pre-trained language model that utilizes up to 700 billion words from both Chinese and English sources.
The moss-moon series model is also distinguishable from other language models in that it supports multiple plugins and is preloaded with 16 billion parameters. This high level of functioning makes it easy for researchers to run it on one A100/A800 or multiple 3090 graphics cards with FP16 precision or on a single 3090 graphics card running with INT4/8 precision.
MOSS was initially conceptualised and developed by Professor Qiu Xipeng’s team at Fudan University’s Natural Language Processing Laboratory. It takes its name from the AI character in the movie The Wandering Earth. The platform is now live on a public domain (https://moss.fastnlp.top/) and open for public beta testing.
Overall, Professor Qiu and his team at Fudan University have done a admirable job with the MOSS model. The open-source conversational language model is one of the first of its kind in China and serves as a great contribution to the field of language processing
Fudan University, located in Shanghai, China, is a well-known educational institution in the country leading in research and academics. Founded in 1905, the university boasts of a history of over 100 years and has educated leaders, scholars and entrepreneurs from around the world. It is a renowned institution that has a long-standing commitment to the education of China’s youth and the advancement of advanced sciences and technology.
Professor Qiu is a renowned scholar in the field of natural language processing and AI. He has over two decades of experience doing research and education in the field, and has made many major contributions to the development of natural language processing technology. He is a Fellow of the Association for Computational Linguistics and a Distinguished Member of the Chinese Association for Artificial Intelligence, allowing him to have a strong influence in the field. His work with the MOSS model is just another notch on his belt of achievements.
