Good morning Brandon I just saw your youtube video about rag dive dive. I have a simple question, the embedding file is similar to a .lora or .safetensors like stable difussion generate when you are training images? also, there is any way that the embedding also can know about the image? maybe through a metadata or classifying the content of the image like a prompt like stable difussion does with images like for example when you use an interrogator for an image it will give you the prompt of that image. That will be a very robust embedding with also the information of the image.