Comment by philomath868

How does the "continuous training pipeline" work? You rebuild the model after every N corrections, with the corrections included in the data?