---
license: mit
base_model:
- Qwen/Qwen3-4B-Base
pipeline_tag: text-generation
library_name: transformers
tags:
- human-simulation
- user-simulation
- behavioral-modeling
---

# OSim-4B (mid-trained, text)

Mid-trained (text) checkpoint of **OSim** (OdysSim) at the 4B scale: `Qwen3-4B-Base`
midtrained on the OdysSim corpus (62 behavioral datasets across the five Soul axes).
4B sibling of the post-trained [cmu-lti/osim-4b-post](https://huggingface.co/cmu-lti/osim-4b-post);
text counterpart pattern of [cmu-lti/osim-8b-mid](https://huggingface.co/cmu-lti/osim-8b-mid).

## Training
- Base: Qwen3-4B-Base
- Stage: midtraining (SFT) on the OdysSim corpus.

## Citation
Cite the OdysSim paper (*Building Foundation Models for Human Behavior Simulation*). Code: https://github.com/sunnweiwei/OdysSim