from transformers import AutoModelForVision2Seq, AutoProcessor
from peft import PeftModel

# Load base model
base_model = AutoModelForVision2Seq.from_pretrained(
    "Qwen/Qwen2.5-VL-7B",
    trust_remote_code=True
)

# Load LoRA adapter
model = PeftModel.from_pretrained(base_model, "yongxianwei/Qwen2.5-VL-7B-VQA")
processor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-7B", trust_remote_code=True)

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

Training Details

Fine-tuned using LoRA on specific Visual Question Answering datasets.

Citation

bibtex

@misc{qwen2.5-vl-visual question answering,
  author = {Yongxian Wei},
  title = {Qwen2.5-VL LoRA for Visual Question Answering},
  year = {2024},
  publisher = {Hugging Face},
  url = {https://huggingface.co/yongxianwei/Qwen2.5-VL-7B-VQA}
}

yongxianwei/Qwen2.5-VL-32B-VQA

Qwen2.5-VL-7B-VQA

Model Details

Usage

Training Details

Citation

yongxianwei/
Qwen2.5-VL-32B-VQA