Fudan University's ChatGPT-like Model MOSS Officially Open
Fudan University's Natural Language Processing Laboratory has officially launched the new version of the MOSS model that can be used in both English and Chinese languages. It is the first open-source conversational language model in China that supports multiple plugins and has 16 billion parameters.
The MOSS base language model has been pre-trained on approximately 700 billion Chinese and English words and can engage in multi-turn conversations.
MOSS can be downloaded from platforms like Github and Hugging Face for researchers to use. The MOSS team at Fudan University’s Natural Language Processing Laboratory has released it on a public platform (https://moss.fastnlp.top/) for beta testing and feedback from the community is welcomed. The MOSS model parameters can run on single A100/A800 or two 3090 graphics cards at FP16 precision, or on a single 3090 graphics card at INT4/8 precision.
The MOSS conversational language model is fine-tuned through dialogue instructions, plugin reinforcement learning, and human preference training. MOSS can use various plugins after fine-tuning, which makes it different from other AI models.
MOSS is a project from Professor Qiu Xipeng’s team at Fudan University’s Natural Language Processing Laboratory. The team compared MOSS to a "smart child" that has shown the potential to become a framework of artificial general intelligence (AGI). MOSS shares its name with the AI in the famous Chinese sci-fi movie, “The Wandering Earth.”
In conclusion, MOSS is a breakthrough in the field of artificial intelligence. Its usage extends beyond simple chatbots to more complex conversational AI. The launch of MOSS by Fudan University’s Natural Language Processing Laboratory is an excellent development in China's technology scene.
