Merge pull request #28 from neo4j-field/add-gemini

add gemini to backend

src/main/app/backend/routers/llm.py

@@ -63,7 +63,7 @@ async def get_response(question: Question, background_tasks: BackgroundTasks) ->
     context = reader.retrieve_context_documents(question_embedding=question_embedding, number_of_context_documents=question.number_of_documents)
     # print(context)
     print("context retrieved...")
-    llm = LLM(llm_type="GPT-4 8k", temperature=question.temperature)
+    llm = LLM(llm_type=question.llm_type, temperature=question.temperature)
     print("llm initialized...")
     llm_response = llm.get_response(question=question.question, context=context)
     print("response retrieved...")

src/main/app/backend/tools/llm.py

@@ -2,7 +2,9 @@
 from typing import List, Dict, Tuple
 
 import openai
-from langchain_community.chat_models import ChatVertexAI, AzureChatOpenAI
+from langchain_community.chat_models import AzureChatOpenAI
+# from langchain_google_genai import ChatGoogleGenerativeAI
+from langchain_google_vertexai import ChatVertexAI
 from langchain_openai import OpenAI
 # from langchain.chains import ConversationChain
 import pandas as pd
@@ -51,6 +53,8 @@ def _init_llm(self, llm_type: str, temperature: float):
                         top_p=0.95, # default is 0.95
                         top_k = 40 # default is 40
                        )
+            case "Gemini":
+                self.llm_instance = ChatVertexAI(model_name="gemini-pro")
             case "GPT-4 8k":
                 # Tokens per Minute Rate Limit (thousands): 10
                 # Rate limit (Tokens per minute): 10000