MiMo 7B SFT
MiMo 7B SFT applies supervised tuning to the original base model. It provides the cold-start checkpoint used before reinforcement learning in the main MiMo 7B RL path.
Research release
The intermediate weights make the training path visible
Xiaomi publishes the base, SFT, RL-from-base and RL-from-SFT checkpoints. The SFT release helps separate the effects of demonstration learning from later reward-based training.MiMo-7B-SFT official model card
Keep demonstrations and evaluation tasks apart
Practical guidance. Use this release as a baseline for a new reward method or domain adaptation. Check whether improvement comes from general problem solving or from exposure to very similar examples. The intermediate stage is valuable even when a later model scores higher.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Access | Downloadable |
| Model identifiers | MiMo-7B-SFT |
| Ancestry | MiMo 7B Base → supervised fine-tuning → this checkpoint.MiMo-7B-SFT official model card |
| Licence | MIT.MiMo-7B-SFT official model card |
Related models
- MiMo 7B Base
Xiaomi · MiMo 7B
MiMo 7B Base was trained from scratch, with reasoning treated as a pretraining objective rather than added only after pretraining.
Launch date not confirmedResearch releaseDownloadable - MiMo 7B RL
Xiaomi · MiMo 7B
A 7B reasoning model trained with reinforcement learning for mathematics and code problems.
Launch date not confirmedDownloadable