Use Your Own Files To Get Response From GPT Like ChatGPT | Python

Introduction

In this article, I’ll show you how to use your locally stored text files to get responses using GPT-3. You can ask questions and get responses like ChatGPT.

On the technology front, we will be using,

• OpenAI GPT
• Langchain
• Python

Input files

You can take many text files and store them in a directory on your local machine. I’ve grabbed input data from https://essaypro.com/blog/essay-samples and created 5 text files. My files are all about the ‘Cause And Effect Of Homelessness’ and are placed in a directory named Store.

Import Required Packages

As we are using Python, let’s go ahead and import the required packages.

from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.vectorstores import Chroma
from langchain.text_splitter import CharacterTextSplitter
from langchain import OpenAI, VectorDBQA
from langchain.document_loaders import DirectoryLoader
import magic
import os
import nltk 

If you do not have the above packages installed on your machine, please install these packages before importing.

nltk.download(‘averaged_perceptron_tagger’)
pip install langchain 
pip install openai
pip install chromadb 
pip install unstructured 
pip install beautifulsoup4
pip install python-magic-bin

Once the required packages are imported, we need to get the OpenAI API key.

Generating an OpenAI API Key

To get the OpenAI key, go to https://openai.com/, login and then grab the keys using highlighted way.

Once you get the key, set that inside an environment variable(I’m using Windows).

os.environ["OPENAI_API_KEY"] = "YOUR_KEY"

Load Input Data

To load our text files, we need to instantiate DirectoryLoader, and that can be done as shown below,

loader = DirectoryLoader(‘Store’, glob=’**/*.txt’)
docs = loader.load()

In the above code, glob must be mentioned to pick only the text files. This is particularly useful when your input directory contains a mix of different-different types of files.

Split Data

As input data could be very long, we need to split our data into small chunks, and here I’m taking chunk size as 1000.

char_text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
doc_texts = char_text_splitter.split_documents(docs)

After splitting, this is how the text looks like,

Create Vector Store

Next, we need to create embeddings of it, which means we need to turn our data into a vector space. Let’s do this by instantiating the OpenAIEmbeddings object as shown below:

openAI_embeddings = OpenAIEmbeddings(openai_api_key=os.environ[‘OPENAI_API_KEY’])
vStore = Chroma.from_documents(doc_texts, openAI_embeddings)

Create Model

Finally, time to create our model. This can be done by passing all the required parameters as shown below:

model = VectorDBQA.from_chain_type(llm=OpenAI(), chain_type=”stuff”, vectorstore=vStore)

Once the model is ready, we are good to test it.

Test Model

To test the model, we need to ask some questions about it, and this can be done as shown below:

question = “What are the effects of homelessness”
model.run(question)

On executing the above cell, you will find your response. Here is what I got:

Validate Model

We have created our model and received the response. But how can we ensure this response is from our data only? To get this assurity, we need to validate our model. You can find those validation lines in the video mentioned below.

If you find anything, which is not clear, I would recommend you to watch my video recording here, which demonstrates this flow from end to end.