---
license: apache-2.0
base_model: Qwen/Qwen2.5-VL-32B
tags:
- qwen2.5-vl
- lora
- vision-language
- optical character recognition
library_name: transformers
---

# Qwen2.5-VL-32B-OCR

This is a LoRA adapter for Qwen2.5-VL-32B fine-tuned on Optical Character Recognition tasks.

## Model Details

- **Base Model**: Qwen/Qwen2.5-VL-32B
- **Task**: Optical Character Recognition
- **Training Method**: LoRA (Low-Rank Adaptation)
- **Framework**: Transformers

## Usage

```python
from transformers import AutoModelForVision2Seq, AutoProcessor
from peft import PeftModel

# Load base model
base_model = AutoModelForVision2Seq.from_pretrained(
    "Qwen/Qwen2.5-VL-32B",
    trust_remote_code=True
)

# Load LoRA adapter
model = PeftModel.from_pretrained(base_model, "yongxianwei/Qwen2.5-VL-32B-OCR")
processor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-32B", trust_remote_code=True)

# Inference
# ... your inference code ...
```

## Training Details

Fine-tuned using LoRA on specific Optical Character Recognition datasets.

## Citation

```bibtex
@misc{qwen2.5-vl-optical character recognition,
  author = {Yongxian Wei},
  title = {Qwen2.5-VL LoRA for Optical Character Recognition},
  year = {2024},
  publisher = {Hugging Face},
  url = {https://huggingface.co/yongxianwei/Qwen2.5-VL-32B-OCR}
}
```