Comment by linux_devil

Why do we need to train the model to solve for retrieval within the org, so we have to keep training it whenever new dataset is introduced , or am I missing something here ?