Add template for Pinecone + Multi-Query (#12353)

templates/rag-pinecone-multi-query/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-multi-query/README.md

@@ -0,0 +1,16 @@
+#  RAG Pinecone multi query
+
+This template performs RAG using Pinecone and OpenAI with the [multi-query retriever](https://python.langchain.com/docs/modules/data_connection/retrievers/MultiQueryRetriever).
+
+This will use an LLM to generate multiple queries from different perspectives for a given user input query. 
+
+For each query, it retrieves a set of relevant documents and takes the unique union across all queries for answer synthesis.
+
+##  Pinecone
+
+This template uses Pinecone as a vectorstore and requires that `PINECONE_API_KEY`, `PINECONE_ENVIRONMENT`, and `PINECONE_INDEX` are set.
+
+##  LLM
+
+Be sure that `OPENAI_API_KEY` is set in order to the OpenAI models.
+

templates/rag-pinecone-multi-query/pyproject.toml

@@ -0,0 +1,21 @@
+[tool.poetry]
+name = "rag-pinecone-multi-query"
+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"
+
+[tool.langserve]
+export_module = "rag_pinecone_multi_query"
+export_attr = "chain"
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"

templates/rag-pinecone-multi-query/rag_pinecone.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-multi-query')\n",
+    "rag_app_pinecone.invoke(\"What are the different types of agent memory\")"
+   ]
+  }
+ ],
+ "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-multi-query/rag_pinecone_multi_query/__init__.py

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

templates/rag-pinecone-multi-query/rag_pinecone_multi_query/chain.py

@@ -0,0 +1,58 @@
+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.multi_query import MultiQueryRetriever
+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()
+
+# Set up index with multi query retriever
+vectorstore = Pinecone.from_existing_index(PINECONE_INDEX_NAME, OpenAIEmbeddings())
+model = ChatOpenAI(temperature=0)
+retriever = MultiQueryRetriever.from_llm(
+    retriever=vectorstore.as_retriever(), llm=model
+)
+
+# 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": retriever, "question": RunnablePassthrough()})
+    | prompt
+    | model
+    | StrOutputParser()
+)

templates/rag-pinecone/pinecone_rag.ipynb

@@ -1,84 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "markdown",
-   "id": "681a5d1e",
-   "metadata": {},
-   "source": [
-    "## Connect to template\n",
-    "\n",
-    "`Context`\n",
-    " \n",
-    "* LangServe apps gives you access to templates.\n",
-    "* Templates LLM pipeline (runnables or chains) end-points accessible via FastAPI.\n",
-    "* The environment for these templates is managed by Poetry.\n",
-    "\n",
-    "`Create app`\n",
-    "\n",
-    "* Install LangServe and create an app.\n",
-    "* This will create a new Poetry environment /\n",
-    "```\n",
-    "pip install < to add > \n",
-    "langchain serve new my-app\n",
-    "cd my-app\n",
-    "```\n",
-    "\n",
-    "`Add templates`\n",
-    "\n",
-    "* When we add a template, we update the Poetry config file with the necessary dependencies.\n",
-    "* It also automatically installed these template dependencies in your Poetry environment\n",
-    "```\n",
-    "langchain serve add rag-pinecone\n",
-    "```\n",
-    "\n",
-    "`Start FastAPI server`\n",
-    "\n",
-    "```\n",
-    "langchain start\n",
-    "```\n",
-    "\n",
-    "Note, we can now look at the endpoints:\n",
-    "\n",
-    "http://127.0.0.1:8000/docs#\n",
-    "\n",
-    "And look specifically at our loaded template:\n",
-    "\n",
-    "http://127.0.0.1:8000/docs#/default/invoke_rag_pinecone_invoke_post\n",
-    " \n",
-    "We can also use remote runnable to call it:"
-   ]
-  },
-  {
-   "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')"
-   ]
-  }
- ],
- "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/rag_pinecone.ipynb

@@ -0,0 +1,45 @@
+{
+ "cells": [
+  {
+   "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')\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/rag_pinecone/chain.py

@@ -8,25 +8,32 @@ from langchain.embeddings import OpenAIEmbeddings
 from langchain.schema.output_parser import StrOutputParser
 from langchain.schema.runnable import RunnablePassthrough, RunnableParallel
 
-# Pinecone init
+if os.environ.get("PINECONE_API_KEY", None) is None:
-# Find API key in console at app.pinecone.io
+    raise Exception("Missing `PINECONE_API_KEY` environment variable.")
-YOUR_API_KEY = os.getenv('PINECONE_API_KEY') or 'PINECONE_API_KEY'
-# Find ENV (cloud region) next to API key in console
-YOUR_ENV = os.getenv('PINECONE_ENVIRONMENT') or 'PINECONE_ENV'
-# Init
-pinecone.init(
-    api_key=YOUR_API_KEY,
-    environment=YOUR_ENV
-)
 
-# Get vectorstore
+if os.environ.get("PINECONE_ENVIRONMENT", None) is None:
-text_field = "text"
+    raise Exception("Missing `PINECONE_ENVIRONMENT` environment variable.")
-index_name = "langchain-multi-query-demo"
-index = pinecone.Index(index_name)
-vectorstore = Pinecone(index, 
-                       OpenAIEmbeddings(), 
-                       text_field)
 
+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())
 retriever = vectorstore.as_retriever()
 
 # RAG prompt