Python code to implement RAG with Vertex AI Vector Search and Gemini Pro

rag_gemini.py

# The previous part of this tutorial is at https://gist.github.com/janakiramm/55d2d8ec5d14dd45c7e9127d81cdafcd

from vertexai.language_models import TextEmbeddingModel
from google.cloud import aiplatform
import vertexai
from vertexai.preview.generative_models import GenerativeModel, Part
import json
import os

project=”YOUR_GCP_PROJECT”
location="us-central1"
sentence_file_path = "lakeside_sentences.json"
index_name="INDEX_EP_ID" #Get this from the console or the previous step

aiplatform.init(project=project,location=location)
vertexai.init()
model = GenerativeModel("gemini-pro")
lakeside_index_ep = aiplatform.MatchingEngineIndexEndpoint(index_endpoint_name=index_name)

def generate_text_embeddings(sentences) -> list:    
    model = TextEmbeddingModel.from_pretrained("textembedding-gecko@001")
    embeddings = model.get_embeddings(sentences)
    vectors = [embedding.values for embedding in embeddings]
    return vectors

def generate_context(ids,data):
    concatenated_names = ''
    for id in ids:
        for entry in data:
            if entry['id'] == id:
                concatenated_names += entry['sentence'] + "\n" 
    return concatenated_names.strip()
  
data=load_file(sentence_file_path)

#query=["How many days of unpaid leave in an year"]
#query=["Allowed cost of online course"]
#query=["process for applying sick leave"]
query=["process for applying personal leave"]
qry_emb=generate_text_embeddings(query)

response = lakeside_index_ep.find_neighbors(
    deployed_index_id = index_name,
    queries = [qry_emb[0]],
    num_neighbors = 10
)

matching_ids = [neighbor.id for sublist in response for neighbor in sublist]

context = generate_context(matching_ids,data)
prompt=f"Based on the context delimited in backticks, answer the query. ```{context}``` {query}"

chat = model.start_chat(history=[])
response = chat.send_message(prompt)
print(response.text)

Thank you for you video and introduction to RAG.

When i run:
data=load_file(sentence_file_path)

I received the following error:

NameError: name 'load_file' is not defined

Any thoughts on the issue i might be having?

Thanks!

Thank you for you video and introduction to RAG.

When i run: data=load_file(sentence_file_path)

I received the following error:

NameError: name 'load_file' is not defined

Any thoughts on the issue i might be having?

Thanks!

def load_file(sentence_file_path):
data = []
with open(sentence_file_path, 'r') as file:
for line in file:
entry = json.loads(line)
data.append(entry)
return data

add this function before calling it.