ollama.pull('llama3.2-vision')response = ollama.chat(    model='llama3.2-vision',    messages=[{        'role': 'user',        'content': prompt,        'images': [image["image_path"] for image in image_results]    }])print(response['message']['content'])