OpenAI ChatGPT RAG Retrieval Augmented Generation Example

import pandas as pd
import numpy as np
from ast import literal_eval
import openai

def get_embedding(text: str, model="text-similarity-davinci-001", **kwargs):
text = text.replace("\n", " ")
response = openai.embeddings.create(input=[text], model=model, **kwargs)
return response.data[0].embedding

def cosine_similarity(a, b):
return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))

Load the dataset and convert embeddings

datafile_path = "fine_food_reviews_with_embeddings.csv"
df = pd.read_csv(datafile_path)
df["embedding"] = df.embedding.apply(literal_eval).apply(np.array)

Function to search through the reviews

def search_reviews(df, product_description, n=3, pprint=True):
product_embedding = get_embedding(
product_description,
model="text-embedding-ada-002"
)
df["similarity"] = df.embedding.apply(lambda x: cosine_similarity(x, product_embedding))

results = (
    df.sort_values("similarity", ascending=False)
    .head(n)
    .combined.str.replace("Title: ", "")
    .str.replace("; Content:", ": ")
)
if pprint:
    for r in results:
        print(r[:200])
        print()
return results

client = openai.OpenAI()

def generate_answer_with_chat(context, question):
conversation = [
{"role": "system", "content": "You are a knowledgeable assistant."},
{"role": "user", "content": f"Context: {context}"},
{"role": "user", "content": f"Question: {question}"}
]

stream = client.chat.completions.create(
    model="gpt-4",
    messages=conversation,
    stream=True,
)

response_text = ""
for chunk in stream:
    if chunk.choices[0].delta.content is not None:
        response_text += chunk.choices[0].delta.content

return response_text

def rag_search(df, query, n=3):
# Retrieve top N similar reviews
top_reviews = search_reviews(df, query, n=n, pprint=False)

# Combine the top reviews into a single context string
context = " ".join(top_reviews)

# Generate an answer based on the context using chat API
answer = generate_answer_with_chat(context, query)

return answer

Example usage

answer = rag_search(df, "What do people think about the texture of whole wheat pasta?")
print(answer)

Embedding

OpenAI Embedding

Search Embedding

OpenAI Search Embedding