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
In all honestly, it felt like the best default value for the use cases I tested.
In many implementations and research studies involving RRF, a k value around 60 has been commonly used.
One advance of setting it to 60 reduces the weight difference between higher and lower-ranked documents, leading to a more inclusive fusion where more documents from different queries contribute to the final ranking.
why the value of k is fixed to 60? in RRF.
The text was updated successfully, but these errors were encountered: