Advanced RAG: MultiQuery and ParentDocument

Run this on Colab

In MultiQueryRAG, an LLM is used to automate the process of prompt tuning, to generate multiple queries from different perspectives for a given user input question.

In ParentDocumentRAG, documents are split first into larger "parent" chunks, and then into smaller "child" chunks so that their embeddings can more accurately reflect their meaning. Between the LLM retrieval and inference steps, each smaller "child" chunk is then replaced with its larger "parent" chunk. This provides more context to the model to answer the question.

While both of these techniques can increase the response accuracy, they also can have some drawbacks. Often they take longer to execute, and they can cost more due to increased LLM invocations and/or increased token usage.

Prerequisites

You will need an vector-enabled Astra DB Serverless database and an OpenAI Account.

See the Notebook Prerequisites page for more details.

Create an vector-enabled Astra DB Serverless database.
Create an OpenAI account
Within your database, create an Astra DB keyspace
Within your database, create an Astra DB Access Token with Database Administrator permissions.
Get your Astra DB Serverless API Endpoint: https://<ASTRA_DB_ID>-<ASTRA_DB_REGION>.apps.astra.datastax.com

Initialize the environment variables in a .env file.

ASTRA_DB_APPLICATION_TOKEN=AstraCS:...
ASTRA_DB_API_ENDPOINT=https://9d9b9999-999e-9999-9f9a-9b99999dg999-us-east-2.apps.astra.datastax.com
ASTRA_DB_COLLECTION=test
OPENAI_API_KEY=sk-f99...

Enter your settings for Astra DB Serverless and OpenAI:

astra_token = os.getenv("ASTRA_DB_APPLICATION_TOKEN")
astra_endpoint = os.getenv("ASTRA_DB_API_ENDPOINT")
collection = os.getenv("ASTRA_DB_COLLECTION")
openai_api_key = os.getenv("OPENAI_API_KEY")

Setup

ragstack-ai includes all the packages you need to build a RAG pipeline.

Install the following libraries.
```
pip install ragstack-ai
```

Import dependencies.

import os
from langchain_community.document_loaders import TextLoader
from langchain_community.document_loaders import PyPDFLoader
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parser import StrOutputParser
from langchain_core.runnable import RunnableLambda, RunnablePassthrough
from langchain_community.callbacks import get_openai_callback
from langchain_openai import ChatOpenAI
from langchain_openai import OpenAIEmbeddings
from langchain_community.retrievers import ParentDocumentRetriever
from langchain_community.storage import InMemoryStore
from langchain.text_splitter import TokenTextSplitter
from langchain_astradb import AstraDBVectorStore

The two advanced RAG techniques require different dependencies. Their imports are shown in their respective sections.

Create helper functions

Create some helper methods for printing docs and results:

import textwrap

def pprint_docs(docs):
    print(f"\n{'-' * 70}\n".join([f"Document {i+1}:\n\n" + "\n".join(textwrap.wrap(d.page_content)) for i, d in enumerate(docs)]))

def pprint_result(result):
    print("Answer: " + "\n".join(textwrap.wrap(result)))

Index sample data

Retrieve the text of a short story that will be indexed in the vector store:

curl https://raw.githubusercontent.com/CassioML/cassio-website/main/docs/frameworks/langchain/texts/amontillado.txt --output amontillado.txt
input = "amontillado.txt"

Loop through each file and load it into the vector store. You loaded amontillado.txt in the previous step, but this processor can also process PDFs.

SAMPLEDATA = [] clears the list so the same files aren’t processed twice.

documents = []
for filename in SAMPLEDATA:
    path = os.path.join(os.getcwd(), filename)

    if filename.endswith(".pdf"):
        loader = PyPDFLoader(path)
        new_docs = loader.load_and_split()
        print(f"Processed pdf file: {filename}")
    elif filename.endswith(".txt"):
        loader = TextLoader(path)
        new_docs = loader.load_and_split()
        print(f"Processed txt file: {filename}")
    else:
        print(f"Unsupported file type: {filename}")

    if len(new_docs) > 0:
        documents.extend(new_docs)

SAMPLEDATA = []

print(f"\nProcessing done.")

Build a simple prompt with a set of questions to ask. You’ll use this prompt to gauge the effectiveness of the different RAG techniques.

prompt_template = """
Answer the question based only on the supplied context. If you don't know the answer, say you don't know the answer.
Context: {context}
Question: {question}
Your answer:
"""
prompt = ChatPromptTemplate.from_template(prompt_template)

questions = [
    "What motivates the narrator, Montresor, to seek revenge against Fortunato?",
    "What are the major themes in this story?",
    "What is the significance of the story taking place during the carnival season?",
    "How is vivid and descriptive language used in the story?",
    "Is there any foreshadowing in the story? If yes, how is it used in the story?"
]

Create one more helper method to iterate over the questions. You won’t ask any questions just yet.

def do_retrieval(chain):
    for i in range(len(questions)):
        print("-" * 40)
        print(f"Question: {questions[i]}\n")
        with get_openai_callback() as cb:
            pprint_result(chain.invoke(questions[i]))
            print(f'\nTotal Tokens: {cb.total_tokens}\n')

Load and embed documents

For the purpose of this example, you will use a method that is compatible with the advanced ParentDocumentRAG technique. If you aren’t going to use this technique, you can review the Quickstart example for a simpler document insertion method.

You will create 2 splitters: a parent splitter, and a child splitter. The parent splitter will split your documents into 512-token documents. The child splitter will split the parent documents into 128-token documents.

Embeddings for the child documents are generated and stored in the vector store.

For this example demo, the parent documents will only be stored in-memory. In a production system, the parent documents should be stored in a database.

Create embeddings for the documents and insert them into the Astra DB Serverless vector store.

# Initialize the models
model = ChatOpenAI(openai_api_key=openai_api_key, model_name="gpt-3.5-turbo", temperature=0.1)
embedding = OpenAIEmbeddings(openai_api_key=openai_api_key)

# Initialize a vector store for storing the child chunks
vstore = AstraDBVectorStore(
    collection_name=collection_name,
    embedding=embedding,
    token=astra_token,
    api_endpoint=astra_endpoint
)

# Initialize in-memory storage for the parent chunks
parent_store = InMemoryStore()

# Create a splitter for the parent documents
parent_splitter = TokenTextSplitter(chunk_size=512, chunk_overlap=0)

# Create a splitter for the child documents
# Note: child documents should be smaller than parent documents
child_splitter = TokenTextSplitter(chunk_size=128, chunk_overlap=0)

# Create a parent document retriever
parent_retriever = ParentDocumentRetriever(
    vectorstore=vstore,
    docstore=parent_store,
    child_splitter=child_splitter,
    parent_splitter=parent_splitter,
)
# Split and load the documents into the vector and parent stores
parent_retriever.add_documents(docs)

Create a base retriever

As a control, first make a standard RAG pipeline.

# Standard RAG, nothing fancy
base_retriever = vstore.as_retriever()

base_chain = (
    {"context": base_retriever, "question": RunnablePassthrough()}
    | prompt
    | model
    | StrOutputParser()
)

Run the queries on the base_chain:

Python
Result

do_retrieval(base_chain)

----------------------------------------
Question: What motivates the narrator, Montresor, to seek revenge against Fortunato?

Answer: The narrator, Montresor, seeks revenge against Fortunato because of
the insults and injuries he has endured from him.

Total Tokens: 852

----------------------------------------
Question: What are the major themes in this story?

Answer: I don't know the answer.

Total Tokens: 818

----------------------------------------
Question: What is the significance of the story taking place during the carnival season?

Answer: The significance of the story taking place during the carnival season
is not clear from the given context.

Total Tokens: 832

----------------------------------------
Question: How is vivid and descriptive language used in the story?

Answer: Vivid and descriptive language is used in the story to create a sense
of atmosphere and to convey the intense emotions and actions of the
characters.

Total Tokens: 842

----------------------------------------
Question: Is there any foreshadowing in the story? If yes, how is it used in the story?

Answer: Yes, there is foreshadowing in the story. The foreshadowing is used to
hint at the fate of Fortunato and the revenge that the narrator plans
to take. The mention of the "low moaning cry" from the depth of the
recess and the "furious vibrations of the chain" suggest that
something ominous and dangerous is happening to Fortunato. This
foreshadows his eventual entrapment and demise in the catacombs.
Additionally, the narrator's mention of wanting to "punish with
impunity" and the idea of a wrong being unredressed when retribution
overtakes its redresser foreshadow the narrator's plan to seek revenge
on Fortunato without facing any consequences.

Total Tokens: 974

Some of the questions were answered well, and others were not. Note that I don’t know the answer. should be considered a positive result, because it is better than a hallucination.

One nice thing with standard RAG is that the number of tokens used is quite low. This keeps costs down.

To dig deeper, examine the context used to answer the third question:

Python
Result

pprint_docs(base_retriever.get_relevant_documents(questions[2]))

----------------------------------------
Document 1:

 during the supreme madness of the carnival season, that I encountered
my friend.  He accosted me with excessive warmth, for he had been
drinking much.  The man wore motley. He had on a tight-fitting parti-
striped dress, and his head was surmounted by the conical cap and
bells.  I was so pleased to see him, that I thought I should never
have done wringing his hand.  I said to him--"My dear Fortunato, you
are luckily met.  How remarkably well you are looking to-day!
----------------------------------------------------------------------
Document 2:

ado."  Thus speaking, Fortunato possessed himself of my arm. Putting
on a mask of black silk, and drawing a _roquelaire_ closely about my
person, I suffered him to hurry me to my palazzo.  There were no
attendants at home; they had absconded to make merry in honour of the
time.  I had told them that I should not return until the morning, and
had given them explicit orders not to stir from the house. These
orders were sufficient, I well knew, to insure their immediate
disappearance, one and all
----------------------------------------------------------------------
Document 3:

connoisseurship in wine.  Few Italians have the true virtuoso spirit.
For the most part their enthusiasm is adopted to suit the time and
opportunity--to practise imposture upon the British and Austrian
_millionaires_.  In painting and gemmary, Fortunato, like his
countrymen, was a quack--but in the matter of old wines he was
sincere.  In this respect I did not differ from him materially: I was
skillful in the Italian vintages myself, and bought largely whenever I
could.  It was about dusk, one evening
----------------------------------------------------------------------
Document 4:

, as soon as my back was turned.  I took from their sconces two
flambeaux, and giving one to Fortunato, bowed him through several
suites of rooms to the archway that led into the vaults.  I passed
down a long and winding staircase, requesting him to be cautious as he
followed. We came at length to the foot of the descent, and stood
together on the damp ground of the catacombs of the Montresors.  The
gait of my friend was unsteady, and the bells upon his cap jingled

Create a MultiQueryRetriever RAG chain

Next, try the advanced MultiQueryRAG technique.

When the MultiQueryRetriever module is used, an additional LLM call is made before retrieval. This call generates multiple versions of the initial question from different perspectives, then retrieval is performed on this set of questions.

This technique requires an additional dependency.

Build the multi-query retriever:

from langchain.retrievers.multi_query import MultiQueryRetriever

# Note that this retriever depends on the base_retriever
multi_retriever = MultiQueryRetriever.from_llm(
    retriever=base_retriever, llm=model
)

multi_chain = (
    {"context": multi_retriever, "question": RunnablePassthrough()}
    | prompt
    | model
    | StrOutputParser()
)

Run the queries on the multi_chain:

Python
Result

do_retrieval(multi_chain)

----------------------------------------
Question: What motivates the narrator, Montresor, to seek revenge against Fortunato?

Answer: The narrator, Montresor, seeks revenge against Fortunato because of
the insults and injuries he has endured from him.

Total Tokens: 1209

----------------------------------------
Question: What are the major themes in this story?

Answer: I don't know the answer.

Total Tokens: 1531

----------------------------------------
Question: What is the significance of the story taking place during the carnival season?

Answer: The significance of the story taking place during the carnival season
is not clear from the given context.

Total Tokens: 1176

----------------------------------------
Question: How is vivid and descriptive language used in the story?

Answer: Vivid and descriptive language is used in the story to create a
detailed and immersive atmosphere. It helps to paint a clear picture
of the setting, such as the crypt and the catacombs, and to convey the
emotions and actions of the characters.

Total Tokens: 1195

----------------------------------------
Question: Is there any foreshadowing in the story? If yes, how is it used in the story?

Answer: Yes, there is foreshadowing in the story. The narrator's mention of
seeking revenge and punishing with impunity foreshadows the events
that unfold later in the story, where the narrator walls up Fortunato
in the catacombs. Additionally, the mention of the "furious vibrations
of the chain" hints at the impending danger and violence that
Fortunato will face.

Total Tokens: 1472

The results using MultiQueryRAG are different. It is unclear if they are better or not. The number of tokens used has increased, and the responsiveness has gone down due to the extra LLM call.

To dig deeper, examine the context used to answer the 3rd question:

Python
Result

pprint_docs(multi_retriever.get_relevant_documents(questions[2]))

----------------------------------------
Document 1:

 during the supreme madness of the carnival season, that I encountered
my friend.  He accosted me with excessive warmth, for he had been
drinking much.  The man wore motley. He had on a tight-fitting parti-
striped dress, and his head was surmounted by the conical cap and
bells.  I was so pleased to see him, that I thought I should never
have done wringing his hand.  I said to him--"My dear Fortunato, you
are luckily met.  How remarkably well you are looking to-day!
----------------------------------------------------------------------
Document 2:

ado."  Thus speaking, Fortunato possessed himself of my arm. Putting
on a mask of black silk, and drawing a _roquelaire_ closely about my
person, I suffered him to hurry me to my palazzo.  There were no
attendants at home; they had absconded to make merry in honour of the
time.  I had told them that I should not return until the morning, and
had given them explicit orders not to stir from the house. These
orders were sufficient, I well knew, to insure their immediate
disappearance, one and all
----------------------------------------------------------------------
Document 3:

 and with the aid of my trowel, I began vigorously to wall up the
entrance of the niche.  I had scarcely laid the first tier of the
masonry when I discovered that the intoxication of Fortunato had in a
great measure worn off. The earliest indication I had of this was a
low moaning cry from the depth of the recess.  It was _not_ the cry of
a drunken man. There was then a long and obstinate silence.  I laid
the second tier, and the third, and the fourth; and then I heard the
furious vibrations of the chain.
----------------------------------------------------------------------
Document 4:

connoisseurship in wine.  Few Italians have the true virtuoso spirit.
For the most part their enthusiasm is adopted to suit the time and
opportunity--to practise imposture upon the British and Austrian
_millionaires_.  In painting and gemmary, Fortunato, like his
countrymen, was a quack--but in the matter of old wines he was
sincere.  In this respect I did not differ from him materially: I was
skillful in the Italian vintages myself, and bought largely whenever I
could.  It was about dusk, one evening
----------------------------------------------------------------------
Document 5:

, as soon as my back was turned.  I took from their sconces two
flambeaux, and giving one to Fortunato, bowed him through several
suites of rooms to the archway that led into the vaults.  I passed
down a long and winding staircase, requesting him to be cautious as he
followed. We came at length to the foot of the descent, and stood
together on the damp ground of the catacombs of the Montresors.  The
gait of my friend was unsteady, and the bells upon his cap jingled

There are 5 documents of size 128 tokens. The model might benefit from the extra context provided when answering the question.

Create a ParentDocumentRetriever RAG chain

The second advanced technique uses the ParentDocumentRetriever defined above. Remember that this will perform a post-processing step between retrieval and inference to replace the child documents with their parent documents. After this is done, any duplicate documents are removed.

Build the ParentDocumentRAG chain:

parent_chain = (
    {"context": parent_retriever, "question": RunnablePassthrough()}
    | prompt
    | model
    | StrOutputParser()
)

Run the ParentDocumentRAG chain over the questions:

Python
Result

do_retrieval(parent_chain)

----------------------------------------
Question: What motivates the narrator, Montresor, to seek revenge against Fortunato?

Answer: The narrator, Montresor, seeks revenge against Fortunato because
Fortunato insulted him.

Total Tokens: 1708

----------------------------------------
Question: What are the major themes in this story?

Answer: The major themes in this story are revenge, deception, and the power
of manipulation.

Total Tokens: 1695

----------------------------------------
Question: What is the significance of the story taking place during the carnival season?

Answer: The significance of the story taking place during the carnival season
is that it provides a chaotic and festive atmosphere, which allows the
narrator to carry out his revenge plot without arousing suspicion.

Total Tokens: 1719

----------------------------------------
Question: How is vivid and descriptive language used in the story?

Answer: Vivid and descriptive language is used in the story to create a sense
of atmosphere and to paint a detailed picture of the setting and
events. The language is used to describe the dank and damp catacombs,
the chains and padlock that bind the protagonist, and the construction
of the wall that seals the niche. It also describes the sounds and
actions of the characters, such as the moaning cry from the recess and
the low laugh that comes from the niche. Overall, the vivid and
descriptive language helps to immerse the reader in the story and
enhance the suspense and horror elements.

Total Tokens: 1803

----------------------------------------
Question: Is there any foreshadowing in the story? If yes, how is it used in the story?

Answer: Yes, there is foreshadowing in the story. The foreshadowing is used to
hint at the fate of Fortunato and the narrator's plan for revenge. The
mention of the chains, padlock, and walling up the entrance of the
niche all foreshadow the narrator's intention to trap and bury
Fortunato alive. Additionally, the mention of the Amontillado wine and
the narrator's comment about Fortunato's cough hint at the means by
which the narrator will carry out his revenge.

Total Tokens: 2342

With ParentDocumentRAG, you get decent answers for all 5 questions. The number of tokens used has gone up significantly, but the response time is similar to standard RAG. The extra cost might be worth the improvement in results.

Again, you can dig deeper by looking at the context used to answer the 3rd question:

Python
Result

pprint_docs(parent_retriever.get_relevant_documents(questions[2]))

Document 1:

The thousand injuries of Fortunato I had borne as I best could, but
when he ventured upon insult, I vowed revenge.  You, who so well know
the nature of my soul, will not suppose, however, that I gave
utterance to a threat.  _At length_ I would be avenged; this was a
point definitely settled--but the very definitiveness with which it
was resolved, precluded the idea of risk.  I must not only punish, but
punish with impunity.  A wrong is unredressed when retribution
overtakes its redresser.  It is equally unredressed when the avenger
fails to make himself felt as such to him who has done the wrong.  It
must be understood that neither by word nor deed had I given Fortunato
cause to doubt my good will.  I continued, as was my wont, to smile in
his face, and he did not perceive that my smile _now_ was at the
thought of his immolation.  He had a weak point--this Fortunato--
although in other regards he was a man to be respected and even
feared.  He prided himself on his connoisseurship in wine.  Few
Italians have the true virtuoso spirit. For the most part their
enthusiasm is adopted to suit the time and opportunity--to practise
imposture upon the British and Austrian _millionaires_.  In painting
and gemmary, Fortunato, like his countrymen, was a quack--but in the
matter of old wines he was sincere.  In this respect I did not differ
from him materially: I was skillful in the Italian vintages myself,
and bought largely whenever I could.  It was about dusk, one evening
during the supreme madness of the carnival season, that I encountered
my friend.  He accosted me with excessive warmth, for he had been
drinking much.  The man wore motley. He had on a tight-fitting parti-
striped dress, and his head was surmounted by the conical cap and
bells.  I was so pleased to see him, that I thought I should never
have done wringing his hand.  I said to him--"My dear Fortunato, you
are luckily met.  How remarkably well you are looking to-day!
----------------------------------------------------------------------
Document 2:

  But I have received a pipe of what passes for Amontillado, and I
have my doubts."  "How?" said he.  "Amontillado?  A pipe?  Impossible!
And in the middle of the carnival!"  "I have my doubts," I replied;
"and I was silly enough to pay the full Amontillado price without
consulting you in the matter. You were not to be found, and I was
fearful of losing a bargain."  "Amontillado!"  "I have my doubts."
"Amontillado!"  "And I must satisfy them."  "Amontillado!"  "As you
are engaged, I am on my way to Luchesi.  If any one has a critical
turn, it is he.  He will tell me--"  "Luchesi cannot tell Amontillado
from Sherry."  "And yet some fools will have it that his taste is a
match for your own."  "Come, let us go."  "Whither?"  "To your
vaults."  "My friend, no; I will not impose upon your good nature.  I
perceive you have an engagement.  Luchesi--"  "I have no
engagement;--come."  "My friend, no.  It is not the engagement, but
the severe cold with which I perceive you are afflicted.  The vaults
are insufferably damp. They are encrusted with nitre."  "Let us go,
nevertheless.  The cold is merely nothing. Amontillado! You have been
imposed upon.  And as for Luchesi, he cannot distinguish Sherry from
Amontillado."  Thus speaking, Fortunato possessed himself of my arm.
Putting on a mask of black silk, and drawing a _roquelaire_ closely
about my person, I suffered him to hurry me to my palazzo.  There were
no attendants at home; they had absconded to make merry in honour of
the time.  I had told them that I should not return until the morning,
and had given them explicit orders not to stir from the house. These
orders were sufficient, I well knew, to insure their immediate
disappearance, one and all
----------------------------------------------------------------------
Document 3:

, as soon as my back was turned.  I took from their sconces two
flambeaux, and giving one to Fortunato, bowed him through several
suites of rooms to the archway that led into the vaults.  I passed
down a long and winding staircase, requesting him to be cautious as he
followed. We came at length to the foot of the descent, and stood
together on the damp ground of the catacombs of the Montresors.  The
gait of my friend was unsteady, and the bells upon his cap jingled as
he strode.  "The pipe," said he.  "It is farther on," said I; "but
observe the white web-work which gleams from these cavern walls."  He
turned towards me, and looked into my eyes with two filmy orbs that
distilled the rheum of intoxication.  "Nitre?" he asked, at length.
"Nitre," I replied.  "How long have you had that cough?"  "Ugh! ugh!
ugh!--ugh! ugh! ugh!--ugh! ugh! ugh!--ugh! ugh! ugh!--ugh! ugh! ugh!"
My poor friend found it impossible to reply for many minutes.  "It is
nothing," he said, at last.  "Come," I said, with decision, "we will
go back; your health is precious.  You are rich, respected, admired,
beloved; you are happy, as once I was.  You are a man to be missed.
For me it is no matter.  We will go back; you will be ill, and I
cannot be responsible.  Besides, there is Luchesi--"  "Enough," he
said; "the cough is a mere nothing; it will not kill me. I shall not
die of a cough."  "True--true," I replied; "and, indeed, I had no
intention of alarming you unnecessarily--but you should use all proper
caution. A draught of this Medoc will defend us from the damps."  Here
I knocked off the neck of a bottle which I drew from a long row of its
fellows that lay upon the mould.  "Drink," I said, presenting him the

There are 3 documents that are 512 tokens in size. The additional context helps the LLM generate a good answer for the question.

Combine both RAG techniques

Finally, you can combine both the MultiQuery and ParentDocument techniques, as shown below.

Python
Result

multi_parent_retriever = MultiQueryRetriever.from_llm(
    retriever=parent_retriever, llm=model
)

multi_parent_chain = (
    {"context": multi_parent_retriever, "question": RunnablePassthrough()}
    | prompt
    | model
    | StrOutputParser()
)

# run the queries
do_retrieval(multi_parent_chain)

----------------------------------------
Question: What motivates the narrator, Montresor, to seek revenge against Fortunato?

Answer: The narrator, Montresor, seeks revenge against Fortunato because
Fortunato insulted him.

Total Tokens: 1880

----------------------------------------
Question: What are the major themes in this story?

Answer: The major themes in this story are revenge, deception, and the power
of manipulation.

Total Tokens: 2373

----------------------------------------
Question: What is the significance of the story taking place during the carnival season?

Answer: The significance of the story taking place during the carnival season
is that it provides a chaotic and festive atmosphere, which allows the
narrator to carry out his revenge plot without arousing suspicion.

Total Tokens: 2417

----------------------------------------
Question: How is vivid and descriptive language used in the story?

Answer: Vivid and descriptive language is used in the story to create a
detailed and immersive atmosphere. It helps to paint a clear picture
of the setting, such as the granite walls, iron staples, and chains.
The language also conveys the emotions and actions of the characters,
such as the protagonist's astounded and implored tone, and the
friend's moaning cry and furious vibrations. Additionally, the
language describes the construction of the wall and the appearance of
the interior recess, adding to the suspense and tension of the story.

Total Tokens: 1394

----------------------------------------
Question: Is there any foreshadowing in the story? If yes, how is it used in the story?

Answer: Yes, there is foreshadowing in the story. The foreshadowing is used to
hint at the fate of Fortunato and the narrator's plan for revenge. The
mention of the chains, padlock, and walling up the entrance of the
niche all foreshadow the narrator's intention to trap and bury
Fortunato alive. Additionally, the mention of the Amontillado wine and
the narrator's comment about Fortunato's cough hint at the method of
Fortunato's demise.

Total Tokens: 2546

This is by far the most expensive technique, but perhaps returns the best results.

Cleanup

Delete the collection after you are done exploring this example.

vstore.delete_collection()

Was this helpful?

Edit this page