According to Xiaomi’s official post on May 16, Xiaomi’s large language model MiLM has officially passed the filing for large models. It will be seen in Xiaomi’s cars, phones, smart home products, etc., and will be released widely to more users in the coming days.
Xiaomi topped the C-Eval and CMMLU evaluation lists in August last year, while the model ranked 10th overall and 1st in terms of parameter level. According to the GitHub project page, MiLM-6B is a large-scale pre-trained language model developed by Xiaomi with a parameter scale of 6.4 billion.
According to other reports, the MiLM-6B model achieved high accuracy in 20 different STEM subjects, such as science, technology, engineering, mathematics, accountancy, psychics, chemistry, and biology. You will not face any problem getting answers related to this.
Moreover, the MiLM-6B model has improved the accuracy of everything but can be improved in some subjects, such as law, Mathematics, Programming, Probability Theory, and Discrete mathematics. After these improvements, the possibility of Xioami’s device getting a new system increases.
According to the results, in ten social science tests except for education and geography, this model has achieved satisfactory marks in 8 other subjects, whereas in 11th humanities subjects also, MiLM-6B has performed well based on history and law. We hope the company will soon improve its shortcomings and release it widely, which may take some time, but we are still trying to get more information about it.