README.md

## Base_model
- Qwen/Qwen2.5-VL-7B-Instruct

## Training Data
We use same dataset from [Open-o3-video](https://huggingface.co/datasets/marinero4972/Open-o3-Video/tree/main)
| Stage | Dataset |
|-------|---------|
| SFT | STGR-SFT.json |
| RL | STGR-RL.json |

## Usage

```python
from transformers import AutoModelForCausalLM, AutoProcessor

model = AutoModelForCausalLM.from_pretrained("danaleee/VisionCoach-7B")
processor = AutoProcessor.from_pretrained("danaleee/VisionCoach-7B")
```