Model Assistant
The model assistant is the easiest way to get started with a model. It allows you to quickly add a model to your pipeline and then use the assistant to label the data that you want to extract.
The model assistant is a great way to get started with a model, you can create a project in Kodexa and then add the model assistant to the project. Once you have the model assistant in place, you can then “Configure” the assistant. Go to the configuration screen and one of the configuration panels will be pipeline.
Here you can choose the + to add a new model to your pipeline.
The assistant will then allow the user to access the options for the model, label documents and test the model and finally train (if needed) and then use the model to process documents.
The Model Assistant and the Model Runtime
The model assistant basically schedules work and passes documents to the model runtime that the model declared it will use. This means that when you ask to train the model assistant, it will actually create a training pipeline that will train all the trainable models in this pipeline, passing the training store to the model runtime for each of the models.
While the assistant allows us to surface the models and pipelines, the actual work of calling the model runtime performs the model.
← Previous
Next →