---
library_name: transformers
base_model: Qwen/Qwen2.5-3B-Instruct
tags:
- ufb
- reinforcement-learning
- fsdp
---

# test_global_step_10_ragen

This model was exported from a UFO training checkpoint.

- Base model: `Qwen/Qwen2.5-3B-Instruct`
- Source checkpoint: `/workspace/ufb/outputs/checkpoints/exp1_MetamathQA/top_k/global_step_10/actor`
- Exported step: `global_step_10`

## Usage

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "/workspace/ufb/outputs/hf/test_global_step_10_ragen"
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True)
```