I'm setting up OCR Ollama llama3.2-vision:latest. However, the parameters are not saved and the model is not applied.