processing_qwen_vl.py

from transformers import CLIPImageProcessor, AutoTokenizer

class QWenVLProcessor:
    def __init__(self, tokenizer, image_processor):
        self.tokenizer = tokenizer
        self.image_processor = image_processor

    @classmethod
    def from_pretrained(cls, model_id, **kwargs):
        tokenizer = AutoTokenizer.from_pretrained(model_id, **kwargs)
        image_processor = CLIPImageProcessor.from_pretrained("openai/clip-vit-large-patch14")
        return cls(tokenizer=tokenizer, image_processor=image_processor)

    def __call__(self, text=None, images=None, return_tensors=None):
        if images is not None:
            image_inputs = self.image_processor(images, return_tensors=return_tensors)
        else:
            image_inputs = {}
        if text is not None:
            text_inputs = self.tokenizer(text, return_tensors=return_tensors, padding=True)
        else:
            text_inputs = {}
        return {**text_inputs, **image_inputs}

    def batch_decode(self, *args, **kwargs):
        return self.tokenizer.batch_decode(*args, **kwargs)