LangChain¶

In [ ]:

!pip install -q -U ragatouille
!pip install -q langchain
!pip install -q langchain-openai
!pip install -q langchain-core
!pip install -q langchain-community
!pip install -q pypdf

!pip install -q -U ragatouille !pip install -q langchain !pip install -q langchain-openai !pip install -q langchain-core !pip install -q langchain-community !pip install -q pypdf

In [ ]:

from ragatouille import RAGPretrainedModel

RAG = RAGPretrainedModel.from_pretrained("colbert-ir/colbertv2.0")

from ragatouille import RAGPretrainedModel RAG = RAGPretrainedModel.from_pretrained("colbert-ir/colbertv2.0")

In [ ]:

from langchain_community.document_loaders import PyPDFLoader
loader = PyPDFLoader("Orca_paper.pdf")
pages = loader.load_and_split()

from langchain_community.document_loaders import PyPDFLoader loader = PyPDFLoader("Orca_paper.pdf") pages = loader.load_and_split()

In [ ]:

len(pages)

len(pages)

In [ ]:

full_document = ""

for page in pages:
  full_document += page.page_content

full_document = "" for page in pages: full_document += page.page_content

In [ ]:

print(full_document)

print(full_document)

In [ ]:

type(full_document)

type(full_document)

In [ ]:

RAG.index(
    collection=[full_document],
    index_name="orca_paper",
    max_document_length=512,
    split_documents=True,
)

RAG.index( collection=[full_document], index_name="orca_paper", max_document_length=512, split_documents=True, )

Do Retrieval¶

In [ ]:

results = RAG.search(query="What is instruction tuning?", k=3)

results = RAG.search(query="What is instruction tuning?", k=3)

In [ ]:

results

results

Use as LangChain Retriever¶

In [ ]:

retriever = RAG.as_langchain_retriever(k=3)

retriever = RAG.as_langchain_retriever(k=3)

In [ ]:

retriever.invoke("What is instruction tuning?")

retriever.invoke("What is instruction tuning?")

In [ ]:

Create a Chain¶

In [ ]:

import os
from google.colab import userdata
os.environ["OPENAI_API_KEY"] = userdata.get('openai')

import os from google.colab import userdata os.environ["OPENAI_API_KEY"] = userdata.get('openai')

In [ ]:

from langchain.chains import create_retrieval_chain
from langchain.chains.combine_documents import create_stuff_documents_chain
from langchain_core.prompts import ChatPromptTemplate
from langchain_openai import ChatOpenAI

prompt = ChatPromptTemplate.from_template(
    """Answer the following question based only on the provided context:

<context>
{context}
</context>

Question: {input}"""
)

llm = ChatOpenAI()

document_chain = create_stuff_documents_chain(llm, prompt)


retrieval_chain = create_retrieval_chain(retriever, document_chain)

from langchain.chains import create_retrieval_chain from langchain.chains.combine_documents import create_stuff_documents_chain from langchain_core.prompts import ChatPromptTemplate from langchain_openai import ChatOpenAI prompt = ChatPromptTemplate.from_template( """Answer the following question based only on the provided context: {context} Question: {input}""" ) llm = ChatOpenAI() document_chain = create_stuff_documents_chain(llm, prompt) retrieval_chain = create_retrieval_chain(retriever, document_chain)

In [ ]:

retrieval_chain.invoke({"input": "What is instruction tuning?"})

retrieval_chain.invoke({"input": "What is instruction tuning?"})

In [ ]:

response = retrieval_chain.invoke({"input": "What is instruction tuning?"})

response = retrieval_chain.invoke({"input": "What is instruction tuning?"})

In [ ]:

response["answer"]

response["answer"]

In [ ]:

Llama-Index¶

In [ ]:

!pip install -q llama-index
!pip install -q llama-hub
!pip install -q llama-index-core
!pip install -q llama-index-llms-openai

!pip install -q llama-index !pip install -q llama-hub !pip install -q llama-index-core !pip install -q llama-index-llms-openai

In [ ]:

from llama_index.core import VectorStoreIndex, SimpleDirectoryReader

reader = SimpleDirectoryReader(input_files=["Orca_paper.pdf"])
docs = reader.load_data()

from llama_index.core import VectorStoreIndex, SimpleDirectoryReader reader = SimpleDirectoryReader(input_files=["Orca_paper.pdf"]) docs = reader.load_data()

In [ ]:

# docs

# docs

In [ ]:

from llama_index.core.llama_pack import download_llama_pack

# download and install dependencies
RAGatouilleRetrieverPack = download_llama_pack(
    "RAGatouilleRetrieverPack", "./ragatouille_pack"
)

from llama_index.core.llama_pack import download_llama_pack # download and install dependencies RAGatouilleRetrieverPack = download_llama_pack( "RAGatouilleRetrieverPack", "./ragatouille_pack" )

In [ ]:

from llama_index.llms.openai import OpenAI

from llama_index.llms.openai import OpenAI

In [ ]:

# create the pack
ragatouille_pack = RAGatouilleRetrieverPack(
    docs,  # List[Document]
    llm=OpenAI(model="gpt-3.5-turbo"),
    index_name="orca_paper",
    top_k=5,
)

# create the pack ragatouille_pack = RAGatouilleRetrieverPack( docs, # List[Document] llm=OpenAI(model="gpt-3.5-turbo"), index_name="orca_paper", top_k=5, )

In [ ]:

response = ragatouille_pack.run("What is instruction tuning? ")

response = ragatouille_pack.run("What is instruction tuning? ")

In [ ]:

response

response

In [ ]:

print(response)

print(response)

In [ ]: