First, we will run through a sample training pipeline to fine-tune the models used by Riva for ASR services. The audio transcription reference workflow adds Riva customization for a domain-specific use case, the financial services industry.
This pipeline is typically deployed separately from the inference pipeline in the next step. It should be run regularly, for example monthly, to ensure that the models remain accurate for the target use case and input data.
To get started, open the Training Pipeline link in the left pane under the LaunchPad section.
NoteEnsure you open the correct link to the Training Pipeline, not the Inference Pipeline. The training pipeline must be completed first, prior to proceeding with the inference pipeline steps.
Select and run through the Jupyter notebooks, starting with the Welcome notebook. Once you have completed the notebooks, move on to the next step.
The training deployment steps are complete now that you have run through all the Jupyter notebooks. As a review, you used the sample pipeline described in the diagram below:
