# /// script
# requires-python = ">=3.12"
# dependencies = [
#     "numpy",
#     "einops",
#     "torch",
#     "transformers",
#     "datasets",
#     "accelerate",
#     "timm",
# ]
# ///

try:
    # Load model directly
    from transformers import AutoProcessor, AutoModelForImageTextToText
    
    processor = AutoProcessor.from_pretrained("CohereLabs/command-a-vision-07-2025")
    model = AutoModelForImageTextToText.from_pretrained("CohereLabs/command-a-vision-07-2025")
    messages = [
        {
            "role": "user",
            "content": [
                {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
                {"type": "text", "text": "What animal is on the candy?"}
            ]
        },
    ]
    inputs = processor.apply_chat_template(
    	messages,
    	add_generation_prompt=True,
    	tokenize=True,
    	return_dict=True,
    	return_tensors="pt",
    ).to(model.device)
    
    outputs = model.generate(**inputs, max_new_tokens=40)
    print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))
    with open('CohereLabs_command-a-vision-07-2025_2.txt', 'w') as f:
        f.write('Everything was good in CohereLabs_command-a-vision-07-2025_2.txt')
except Exception as e:
    with open('CohereLabs_command-a-vision-07-2025_2.txt', 'w') as f:
        import traceback
        traceback.print_exc(file=f)
finally:
    from huggingface_hub import upload_file
    upload_file(
        path_or_fileobj='CohereLabs_command-a-vision-07-2025_2.txt',
        repo_id='model-metadata/custom_code_execution_files',
        path_in_repo='CohereLabs_command-a-vision-07-2025_2.txt',
        repo_type='dataset',
    )