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
Is your feature request related to a problem? Please describe.
Not a problem that has arisen, but there are cases where a traversal is being performed and we are only interested in nodes above a certain score threshold.
Describe the solution you'd like
A better solution would allow us to specify the cutoff to the traversal, and then push that filter down to the vector store queries where possible. This would reduce the number of nodes being returned, and allow us to avoid considering any nodes below the threshold.
Describe alternatives you've considered
We can accomplish this today by ignoring any retrieved nodes below the threshold.
Additional context
Add any other context or screenshots about the feature request here.
The text was updated successfully, but these errors were encountered:
MMR currently has this option, but it is not passed through or widely respected.
Although, for MMR this would be min_mmr_score and for other cases we have a min_similarity_score, so perhaps the MMR case should be renamed, and the other fields clarified.
Is your feature request related to a problem? Please describe.
Not a problem that has arisen, but there are cases where a traversal is being performed and we are only interested in nodes above a certain score threshold.
Describe the solution you'd like
A better solution would allow us to specify the cutoff to the traversal, and then push that filter down to the vector store queries where possible. This would reduce the number of nodes being returned, and allow us to avoid considering any nodes below the threshold.
Describe alternatives you've considered
We can accomplish this today by ignoring any retrieved nodes below the threshold.
Additional context
Add any other context or screenshots about the feature request here.
The text was updated successfully, but these errors were encountered: