Advertisement

Xiaomi‘s large language model (LLM), named MiLM, has officially passed the registration process for large models, according to a recent announcement on the company’s Weibo account. 

With this, Xiaomi says that MiLM is ready for integration into its products, including smartphones, smart home devices, and even Xiaomi automobiles. The announcement also hinted at the possibility of opening MiLM’s functionalities to a wider user base in the future.

MiLM first emerged publicly in August 2023 on benchmark platforms C-Eval and CMMLU, where it achieved impressive results. 

The model secured 1st place within its parameter category on the C-Eval leaderboard while ranking 10th overall. Information from the project’s GitHub page reveals that MiLM-6B, the specific model in question, has 6.4 billion parameters.

C-Eval’s subject-specific breakdown highlights MiLM-6B’s strength in STEM fields (Science, Technology, Engineering, and Mathematics). The model achieved high accuracy scores across all 20 STEM subjects, including metrology, physics, chemistry, and biology.

While MiLM-6B has promising performance in most liberal arts subjects as well, areas requiring “abstract thinking” like law, mathematics, programming, and probability theory appear to be areas for improvement.

Social sciences is another area with mixed results. MiLM-6B attained good accuracy in eight out of ten social science subjects, with the exceptions being education and geography. Finally, when it comes to humanities, the model performs well in history and law, while the accuracy of other subjects remains to be determined.

With MiLM-6B overcoming a major obstacle, it’s now ready to be integrated into various Xiaomi products. While its performance differs across subjects, it holds promise for improving user experiences in diverse domains.

Related:

Comments

RELATED ARTICLESMORE FROM AUTHOR