Search code, repositories, users, issues, pull requests...
submited by
Style Pass
Adding Re-Ranking almost always improves accuracy of retrieval pipelines. If you are building a RAG application, and using semantic search or full-text search using this library to re-rank the results will improve accuracy of the application in most cases. However, re-ranking usually adds some amount of latency. We have added knobs in the LLM ReRanker to control latency.
Model Providers are implemented with a clean interface, we welcome contributions to add support for other model providers from the community!
Combine multiple rank lists by assigning scores based on reciprocal ranks, effectively prioritizing higher-ranked items across all lists. Paper
Read more github.com/t...
