This is a workflow that uses a local big language model to infer prompt words from images. You need to download and configure the local ollama first. I will provide a video introduction to the detailed deployment of ollama locally later