dmis-lab/Qwen2.5-VL-7B-Instruct-MRPO

from transformers import Qwen2_5_VLForConditionalGeneration, AutoProcessor
from PIL import Image
import torch

# Load the model (MRPO Qwen2.5-VL checkpoint; or a local trained checkpoint path)
model_path = "dmis-lab/Qwen2.5-VL-7B-Instruct-MRPO"
model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
    model_path,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)
processor = AutoProcessor.from_pretrained(model_path)

# Example usage (no system prompt; the reasoning format is given in the user turn)
image_path = "path/to/medical/image.jpg"
question = "What can you see in this medical image?"

question_text = (
    f"{question} Think step-by-step and enclose your reasoning in "
    "<think>...</think> tags. Then provide your answer in <answer>...</answer> tags."
)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "image": image_path},
            {"type": "text", "text": question_text},
        ],
    }
]

# Preparation for inference
text = processor.apply_chat_template(
    messages, tokenize=False, add_generation_prompt=True
)
inputs = processor(
    text=[text],
    images=[Image.open(image_path)],
    padding=True,
    padding_side="left",
    return_tensors="pt",
)
inputs = inputs.to(model.device)

# Inference (greedy decoding, matching inference.py)
generated_ids = model.generate(**inputs, use_cache=True, max_new_tokens=512, do_sample=False)
generated_ids_trimmed = [
    out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
]
output_text = processor.batch_decode(
    generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
)
print(output_text)

@misc{jung2026breakingfailurecascadesstepaware,
      title={Breaking Failure Cascades: Step-Aware Reinforcement Learning for Medical Multimodal Reasoning}, 
      author={Junha Jung and Minbyul Jeong and Suhyeon Lim and Sungwook Jung and Jaehoon Yun and Taeyun Roh and Mujeen Sung and Jaewoo Kang},
      year={2026},
      eprint={2606.31825},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2606.31825}, 
}