Cohere re-rank template (langchain-ai#12378)

templates/rag-pinecone-rerank/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2023 LangChain, Inc.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

templates/rag-pinecone-rerank/README.md

@@ -0,0 +1,23 @@
+# RAG Pinecone Cohere Re-rank
+
+This template performs RAG using Pinecone and OpenAI, with [Cohere to perform re-ranking](https://python.langchain.com/docs/integrations/retrievers/cohere-reranker) on returned documents.
+
+[Re-ranking](https://docs.cohere.com/docs/reranking) provides a way to rank retrieved documents using specified filters or criteria.
+
+##  Pinecone
+
+This connects to a hosted Pinecone vectorstore.
+
+Be sure that you have set a few env variables in `chain.py`:
+
+* `PINECONE_API_KEY`
+* `PINECONE_ENV`
+* `index_name`
+
+##  LLM
+
+Be sure that `OPENAI_API_KEY` is set in order to the OpenAI models.
+
+##  Cohere
+
+Be sure that `COHERE_API_KEY` is set in order to the ReRank endpoint.

templates/rag-pinecone-rerank/pyproject.toml

@@ -0,0 +1,22 @@
+[tool.poetry]
+name = "rag-pinecone-rerank"
+version = "0.1.0"
+description = ""
+authors = ["Lance Martin <lance@langchain.dev>"]
+readme = "README.md"
+
+[tool.poetry.dependencies]
+python = ">=3.8.1,<4.0"
+langchain = ">=0.0.313, <0.1"
+openai = ">=0.28.1"
+tiktoken = ">=0.5.1"
+pinecone-client = ">=2.2.4"
+cohere = ">4.32"
+
+[tool.langserve]
+export_module = "rag_pinecone_rerank"
+export_attr = "chain"
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"

templates/rag-pinecone-rerank/rag_pinecone_rerank.ipynb

@@ -0,0 +1,46 @@
+{
+ "cells": [
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "id": "681a5d1e",
+   "metadata": {},
+   "source": [
+    "## Connect to template"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "d774be2a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from langserve.client import RemoteRunnable\n",
+    "rag_app_pinecone = RemoteRunnable('http://localhost:8000/rag-pinecone-rerank')\n",
+    "rag_app_pinecone.invoke(\"How does agent memory work?\")"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.9.16"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

templates/rag-pinecone-rerank/rag_pinecone_rerank/__init__.py

@@ -0,0 +1,3 @@
+from rag_pinecone_rerank.chain import chain
+
+__all__ = ["chain"]

templates/rag-pinecone-rerank/rag_pinecone_rerank/chain.py

@@ -0,0 +1,63 @@
+import os 
+import pinecone
+from operator import itemgetter
+from langchain.vectorstores import Pinecone
+from langchain.prompts import ChatPromptTemplate
+from langchain.chat_models import ChatOpenAI
+from langchain.embeddings import OpenAIEmbeddings
+from langchain.schema.output_parser import StrOutputParser
+from langchain.retrievers import ContextualCompressionRetriever
+from langchain.retrievers.document_compressors import CohereRerank
+from langchain.schema.runnable import RunnablePassthrough, RunnableParallel
+
+if os.environ.get("PINECONE_API_KEY", None) is None:
+    raise Exception("Missing `PINECONE_API_KEY` environment variable.")
+
+if os.environ.get("PINECONE_ENVIRONMENT", None) is None:
+    raise Exception("Missing `PINECONE_ENVIRONMENT` environment variable.")
+
+PINECONE_INDEX_NAME = os.environ.get("PINECONE_INDEX", "langchain-test")
+
+### Ingest code - you may need to run this the first time
+# Load
+# from langchain.document_loaders import WebBaseLoader
+# loader = WebBaseLoader("https://lilianweng.github.io/posts/2023-06-23-agent/")
+# data = loader.load()
+
+# # Split
+# from langchain.text_splitter import RecursiveCharacterTextSplitter
+# text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=0)
+# all_splits = text_splitter.split_documents(data)
+
+# # Add to vectorDB
+# vectorstore = Pinecone.from_documents(
+#     documents=all_splits, embedding=OpenAIEmbeddings(), index_name=PINECONE_INDEX_NAME
+# )
+# retriever = vectorstore.as_retriever()
+
+vectorstore = Pinecone.from_existing_index(PINECONE_INDEX_NAME, OpenAIEmbeddings())
+
+# Get k=10 docs
+retriever = vectorstore.as_retriever(search_kwargs={"k":10})
+
+# Re-rank
+compressor = CohereRerank()
+compression_retriever = ContextualCompressionRetriever(
+    base_compressor=compressor, base_retriever=retriever
+)
+
+# RAG prompt
+template = """Answer the question based only on the following context:
+{context}
+Question: {question}
+"""
+prompt = ChatPromptTemplate.from_template(template)
+
+# RAG
+model = ChatOpenAI()
+chain = (
+    RunnableParallel({"context": compression_retriever, "question": RunnablePassthrough()})
+    | prompt
+    | model
+    | StrOutputParser()
+)