Reranking Search Results

Reranking is the process by which an initial set of query results is rescored to produce a more refined and accurate ranking. It's usually performed with a machine learned model that, in contrast to the extremely fast retrieval step, is slower but more accurate.

Reranking is a beta feature in the platform, and only available as a single reranker for English results, and can be enabled by specifying 272725717 as the reranker_id.

In order to use the reranker you need to specify two additional parameters in the request: start and numResults. These tell Vectara which results to rerank. In most scenarios, it makes sense to set start to 0 so that you're reranking all of the best initial results, and then to set numResults to the total number of documents you wish to rerank.