LangChainから学ぶLLMを使ったアプリケーションの工夫

概要

• LangChainで提供されるRetrivalというモジュールにおいて、RAG(Retrieval Augmented Generation)を活用にして必要な情報をLLMに渡す

• RAGの精度を上げる工夫
• HyDE
• クエリをそのままエンコードして検索する代わりに、「仮説的な」回答を生成してエンコードします。正しい答えを持つドキュメントは、実際のクエリより間違った答えに似ている。
HyDE Retriever | 🦜️🔗 Langchain
This example shows how to use the HyDE Retriever, which implements Hypothetical Document Embeddings (HyDE) as described in this paper.
https://js.langchain.com/docs/modules/data_connection/retrievers/integrations/hyde



• FLARE
Active Retrieval Augmented Generation
Despite the remarkable ability of large language models (LMs) to comprehend and generate language, they have a tendency to hallucinate and create factually inaccurate output. Augmenting LMs by...
https://arxiv.org/abs/2305.06983



• Multi Query Retriver
MultiQueryRetriever | 🦜️🔗 Langchain
Distance-based vector database retrieval embeds (represents) queries in high-dimensional space and finds similar embedded documents based on "distance". But, retrieval may produce different results with subtle changes in query wording or if the embeddings do not capture the semantics of the data well. Prompt engineering / tuning is sometimes done to manually address these problems, but can be tedious.
https://python.langchain.com/docs/modules/data_connection/retrievers/MultiQueryRetriever



• Ensemble Retriever
Ensemble Retriever | 🦜️🔗 Langchain
The EnsembleRetriever takes a list of retrievers as input and ensemble the results of their getrelevantdocuments() methods and rerank the results based on the Reciprocal Rank Fusion algorithm.
https://python.langchain.com/docs/modules/data_connection/retrievers/ensemble