Template for multi-modal w/ multi-vector

templates/rag-chroma-multi-modal-multi-vector/.gitignore

@@ -0,0 +1 @@
+docs/img_*.jpg

templates/rag-chroma-multi-modal-multi-vector/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-chroma-multi-modal-multi-vector/README.md

@@ -0,0 +1,106 @@
+
+# rag-chroma-multi-modal
+
+Presentations (slide decks, etc) contain visual content that challenges conventional RAG.
+
+Multi-modal LLMs unlock new ways to build apps over visual content like presentations.
+
+This template performs multi-modal RAG using Chroma with the multi-vector retriever (see [blog](https://blog.langchain.dev/multi-modal-rag-template/)):
+
+* Extract the slides as images
+* Use GPT-4V to summarize each image
+* Embed the image summaries with a link to the original images
+* Retrieve relevant image based on similarity between the image summary and the user input
+* Finally pass those images to GPT-4V for answer synthesis
+
+## Storage
+
+We will use Upstash to store the images.
+
+Simply login [here](https://upstash.com/) and create a database.
+
+This will give you: 
+
+* UPSTASH_URL
+* UPSTASH_TOKEN
+
+Set these in chain.py (***TODO: Update this? Env var?***)
+
+We will use Chroma to store and index the image summaries, which will be created locally in the template directory.
+
+## Input
+
+Supply a slide deck as pdf in the `/docs` directory. 
+
+Create your vectorstore (Chroma) and populae Upstash with: 
+
+```
+poetry install
+python ingest.py
+```
+
+## LLM
+
+The app will retrieve images using multi-modal embeddings, and pass them to GPT-4V.
+
+## Environment Setup
+
+Set the `OPENAI_API_KEY` environment variable to access the OpenAI GPT-4V.
+
+## Usage
+
+To use this package, you should first have the LangChain CLI installed:
+
+```shell
+pip install -U langchain-cli
+```
+
+To create a new LangChain project and install this as the only package, you can do:
+
+```shell
+langchain app new my-app --package rag-chroma-multi-modal-multi-vector
+```
+
+If you want to add this to an existing project, you can just run:
+
+```shell
+langchain app add rag-chroma-multi-modal-multi-vector
+```
+
+And add the following code to your `server.py` file:
+```python
+from rag_chroma_multi_modal_multi_vector import chain as rag_chroma_multi_modal_chain_mv
+
+add_routes(app, rag_chroma_multi_modal_chain_mv, path="/rag-chroma-multi-modal-multi-vector")
+```
+
+(Optional) Let's now configure LangSmith. 
+LangSmith will help us trace, monitor and debug LangChain applications. 
+LangSmith is currently in private beta, you can sign up [here](https://smith.langchain.com/). 
+If you don't have access, you can skip this section
+
+```shell
+export LANGCHAIN_TRACING_V2=true
+export LANGCHAIN_API_KEY=<your-api-key>
+export LANGCHAIN_PROJECT=<your-project>  # if not specified, defaults to "default"
+```
+
+If you are inside this directory, then you can spin up a LangServe instance directly by:
+
+```shell
+langchain serve
+```
+
+This will start the FastAPI app with a server is running locally at 
+[http://localhost:8000](http://localhost:8000)
+
+We can see all templates at [http://127.0.0.1:8000/docs](http://127.0.0.1:8000/docs)
+We can access the playground at [http://127.0.0.1:8000/rag-chroma-multi-modal-multi-vector/playground](http://127.0.0.1:8000/rag-chroma-multi-modal-multi-vector/playground)  
+
+We can access the template from code with:
+
+```python
+from langserve.client import RemoteRunnable
+
+runnable = RemoteRunnable("http://localhost:8000/rag-chroma-multi-modal-multi-vector")
+```

templates/rag-chroma-multi-modal-multi-vector/ingest.py

@@ -0,0 +1,207 @@
+import os
+from pathlib import Path
+
+import base64
+import io
+from io import BytesIO
+
+from PIL import Image
+
+
+import pypdfium2 as pdfium
+from langchain.vectorstores import Chroma
+from langchain_experimental.open_clip import OpenCLIPEmbeddings
+
+import uuid
+
+from langchain.embeddings import OpenAIEmbeddings
+from langchain.retrievers.multi_vector import MultiVectorRetriever
+from langchain.schema.document import Document
+from langchain.schema.output_parser import StrOutputParser
+from langchain.storage import UpstashRedisByteStore
+
+
+from langchain.chat_models import ChatOpenAI
+from langchain.schema.messages import HumanMessage
+
+
+def image_summarize(img_base64, prompt):
+    """
+    Make image summary
+
+    :param img_base64: Base64 encoded string for image
+    :param prompt: Text prompt for summarizatiomn
+    :return: Image summarization prompt
+
+    """
+    chat = ChatOpenAI(model="gpt-4-vision-preview", max_tokens=1024)
+
+    msg = chat.invoke(
+        [
+            HumanMessage(
+                content=[
+                    {"type": "text", "text": prompt},
+                    {
+                        "type": "image_url",
+                        "image_url": {"url": f"data:image/jpeg;base64,{img_base64}"},
+                    },
+                ]
+            )
+        ]
+    )
+    return msg.content
+
+
+def generate_img_summaries(img_base64_list):
+    """
+    Generate summaries for images
+
+    :param img_base64_list: Base64 encoded images
+    :return: List of image summaries and processed images
+    """
+
+    # Store image summaries
+    image_summaries = []
+    processed_images = []
+
+    # Prompt
+    prompt = """You are an assistant tasked with summarizing images for retrieval. \
+    These summaries will be embedded and used to retrieve the raw image. \
+    Give a concise summary of the image that is well optimized for retrieval."""
+
+    # Apply summarization to images
+    for i, base64_image in enumerate(img_base64_list):
+        try:
+            image_summaries.append(image_summarize(base64_image, prompt))
+            processed_images.append(base64_image)
+        except:
+            print(f"BadRequestError with image {i+1}")
+
+    return image_summaries, processed_images
+
+
+def get_images_from_pdf(pdf_path, img_dump_path):
+    """
+    Extract images from each page of a PDF document and save as JPEG files.
+
+    :param pdf_path: A string representing the path to the PDF file.
+    :param img_dump_path: A string representing the path to dummp images.
+    """
+    pdf = pdfium.PdfDocument(pdf_path)
+    n_pages = len(pdf)
+    pil_images = []
+    for page_number in range(n_pages):
+        page = pdf.get_page(page_number)
+        bitmap = page.render(scale=1, rotation=0, crop=(0, 0, 0, 0))
+        pil_image = bitmap.to_pil()
+        pil_image.save(f"{img_dump_path}/img_{page_number + 1}.jpg", format="JPEG")
+        pil_images.append(pil_image)
+    return pil_images
+
+
+def resize_base64_image(base64_string, size=(128, 128)):
+    """
+    Resize an image encoded as a Base64 string
+
+    :param base64_string: Base64 string
+    :param size: Image size
+    :return: Re-sized Base64 string
+    """
+    # Decode the Base64 string
+    img_data = base64.b64decode(base64_string)
+    img = Image.open(io.BytesIO(img_data))
+
+    # Resize the image
+    resized_img = img.resize(size, Image.LANCZOS)
+
+    # Save the resized image to a bytes buffer
+    buffered = io.BytesIO()
+    resized_img.save(buffered, format=img.format)
+
+    # Encode the resized image to Base64
+    return base64.b64encode(buffered.getvalue()).decode("utf-8")
+
+
+def convert_to_base64(pil_image):
+    """
+    Convert PIL images to Base64 encoded strings
+
+    :param pil_image: PIL image
+    :return: Re-sized Base64 string
+    """
+
+    buffered = BytesIO()
+    pil_image.save(buffered, format="JPEG")  # You can change the format if needed
+    img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
+    img_str = resize_base64_image(img_str, size=(960, 540))
+    return img_str
+
+def create_multi_vector_retriever(vectorstore, image_summaries, images):
+    """
+    Create retriever that indexes summaries, but returns raw images or texts
+
+    :param vectorstore: Vectorstore to store embedded image sumamries
+    :param image_summaries: Image summaries
+    :param images: Base64 encoded images
+    :return: Retriever
+    """
+
+    # Initialize the storage layer for images
+    UPSTASH_URL = "https://usw1-bright-beagle-34178.upstash.io"
+    UPSTASH_TOKEN = "AYWCACQgNzk3OTJjZTItMGIxNy00MTEzLWIyZTAtZWI0ZmI1ZGY0NjFhNGRhMGZjNDE4YjgxNGE4MTkzOWYxMzllM2MzZThlOGY="
+    store = UpstashRedisByteStore(url=UPSTASH_URL,
+                                token=UPSTASH_TOKEN)
+    id_key = "doc_id"
+
+    # Create the multi-vector retriever
+    retriever = MultiVectorRetriever(
+        vectorstore=vectorstore,
+        byte_store=store,
+        id_key=id_key,
+    )
+
+    # Helper function to add documents to the vectorstore and docstore
+    def add_documents(retriever, doc_summaries, doc_contents):
+        doc_ids = [str(uuid.uuid4()) for _ in doc_contents]
+        summary_docs = [
+            Document(page_content=s, metadata={id_key: doc_ids[i]})
+            for i, s in enumerate(doc_summaries)
+        ]
+        retriever.vectorstore.add_documents(summary_docs)
+        retriever.docstore.mset(list(zip(doc_ids, doc_contents)))
+
+    add_documents(retriever, image_summaries, images)
+
+    return retriever
+
+# Load PDF
+doc_path = Path(__file__).parent / "docs/DDOG_Q3_earnings_deck.pdf"
+img_dump_path = Path(__file__).parent / "docs/"
+rel_doc_path = doc_path.relative_to(Path.cwd())
+rel_img_dump_path = img_dump_path.relative_to(Path.cwd())
+print("Extract slides as images")
+pil_images = get_images_from_pdf(rel_doc_path, rel_img_dump_path)
+
+# Convert to b64
+images_base_64 = [convert_to_base64(i) for i in pil_images]
+
+# Image summaries
+print("Generate image summaries")
+image_summaries, images_base_64_processed = generate_img_summaries(images_base_64)
+
+# The vectorstore to use to index the images summaries
+vectorstore_mvr = Chroma(
+    collection_name="image_summaries",
+    persist_directory=str(Path(__file__).parent / "chroma_db_multi_modal"),
+    embedding_function=OpenAIEmbeddings()
+)
+
+# Create documents
+images_base_64_processed_documents = [Document(page_content=i) for i in images_base_64_processed]
+
+# Create retriever
+retriever_multi_vector_img = create_multi_vector_retriever(
+    vectorstore_mvr,
+    image_summaries,
+    images_base_64_processed_documents,
+)