You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The Pipeline class allows building a custom search pipeline using components like retriever and readers. Under-the-hood, the components(nodes) are represented as a graph. It would be nice to have validation in place when constructing a pipeline object. For instance, raising an error when three nodes are connected to a query classifier that has only two class labels.
Additionally, a dry-run functionality could be added to make it easier to test out new pipelines.
The text was updated successfully, but these errors were encountered:
In order for a proper pipeline validation, we need to first define standardized output types of the nodes and accepted input types (as mentioned also in #821)
The Pipeline class allows building a custom search pipeline using components like retriever and readers. Under-the-hood, the components(nodes) are represented as a graph. It would be nice to have validation in place when constructing a pipeline object. For instance, raising an error when three nodes are connected to a query classifier that has only two class labels.
Additionally, a dry-run functionality could be added to make it easier to test out new pipelines.
The text was updated successfully, but these errors were encountered: