mirror of
https://github.com/hwchase17/langchain
synced 2024-10-29 17:07:25 +00:00
33 lines
1.2 KiB
Python
33 lines
1.2 KiB
Python
|
from langchain.embeddings import OpenAIEmbeddings
|
||
|
|
from langchain.retrievers.merger_retriever import MergerRetriever
|
||
|
|
from langchain.vectorstores import Chroma
|
||
|
|
|
||
|
|
|
||
|
|
def test_merger_retriever_get_relevant_docs() -> None:
|
||
|
|
"""Test get_relevant_docs."""
|
||
|
|
texts_group_a = [
|
||
|
|
"This is a document about the Boston Celtics",
|
||
|
|
"Fly me to the moon is one of my favourite songs."
|
||
|
|
"I simply love going to the movies",
|
||
|
|
]
|
||
|
|
texts_group_b = [
|
||
|
|
"This is a document about the Poenix Suns",
|
||
|
|
"The Boston Celtics won the game by 20 points",
|
||
|
|
"Real stupidity beats artificial intelligence every time. TP",
|
||
|
|
]
|
||
|
|
embeddings = OpenAIEmbeddings()
|
||
|
|
retriever_a = Chroma.from_texts(texts_group_a, embedding=embeddings).as_retriever(
|
||
|
|
search_kwargs={"k": 1}
|
||
|
|
)
|
||
|
|
retriever_b = Chroma.from_texts(texts_group_b, embedding=embeddings).as_retriever(
|
||
|
|
search_kwargs={"k": 1}
|
||
|
|
)
|
||
|
|
|
||
|
|
# The Lord of the Retrievers.
|
||
|
|
lotr = MergerRetriever([retriever_a, retriever_b])
|
||
|
|
|
||
|
|
actual = lotr.get_relevant_documents("Tell me about the Celtics")
|
||
|
|
assert len(actual) == 2
|
||
|
|
assert texts_group_a[0] in [d.page_content for d in actual]
|
||
|
|
assert texts_group_b[1] in [d.page_content for d in actual]
|