MiMo: Unlocking the Reasoning Potential of Language Model -- From Pretraining to Posttraining
LLM-Core Xiaomi, :, Xia, Bingquan, Bowen Shen, Cici, Dawei Zhu, Zhang, Di, Wang, Gang, Hailin Zhang, Huaqiu Liu, Xiao, Jiebao, Dong, Jinhao, Liang Zhao, Peng Li, Peng Wang, Shihua Yu, Shimao Chen, Weikun Wang, Ma, Wenhan, Xiangwei Deng · arXiv (Cornell University) · 2025
We present MiMo-7B, a large language model born for reasoning tasks, with optimization across both pre-training and post-training stages. During pre-training, we enhance the data preprocessing pipeline and employ a three-stage data mixing strategy to strengthen the base model's reasoning potential. MiMo-7B-Base is pre-trained on 25 trillion tokens, with additional Multi-Token Prediction objective for enhanced performance and accelerated inference speed. During post-training, we curate a dataset of 130K verifiable mathematics and programming problems for reinforcement learning, integrating a test-difficulty-driven code-reward scheme to alleviate sparse-reward issues and employing strategic data resampling to stabilize training. Extensive evaluations show that MiMo-7B-Base possesses exceptional reasoning potential, outperforming even much larger 32B models. The final RL-tuned model, MiMo-7B-RL, achieves superior performance on mathematics, code and general reasoning tasks, surpassing the performance of OpenAI o1-mini. The model checkpoints are available at https://github.com/xiaomimimo/MiMo.