Xiaomi is stepping up its AI game with the introduction of the MiDashengLM-7B model. This new AI is currently integrated into practical applications, specifically within smart home systems and automobiles in China. The availability of the model as open source indicates that developers and companies can utilize it for research and commercial endeavors. Xiaomi’s strategic move of making its AI model open-source demonstrates its technical skills and aims to foster a comprehensive developer ecosystem. The new AI model is trained using publicly accessible data and released under the Apache 2.0 license. The company claims that MiDashengLM-7B exceeds other voice systems in speed and multitasking efficiency. Moreover, its comprehensive capabilities encompass the ability to discern ambient sounds, background music, and environmental noises, thanks to a unified training methodology. The model combines Xiaomi’s Dasheng audio encoder with the Qwen2.5-Omni-7B decoder developed by Alibaba. This AI powers over 30 smart features in Xiaomi’s products, from security functions in smart homes to voice commands and real-time pronunciation assistance in vehicles.
Xiaomi Launches MiDashengLM-7B AI: A Potential Challenger to Alexa and Google Assistant
