lanchain ollama not chatollama

app.py

@@ -2,11 +2,10 @@ import os
 from fastapi import FastAPI, HTTPException
 from fastapi.responses import StreamingResponse
 from pydantic import BaseModel
-from langchain_ollama import ChatOllama
-from langchain.schema import HumanMessage
+from langchain_community.llms import Ollama
+from langchain_core.messages import HumanMessage
 import logging
 from functools import lru_cache
-
 # Set up logging
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
@@ -16,7 +15,7 @@ MODEL_NAME = 'phi3:mini'
 
 @lru_cache()
 def get_llm():
-    return ChatOllama(model=MODEL_NAME)
+    return Ollama(model=MODEL_NAME)
 
 class Question(BaseModel):
     text: str
@@ -30,10 +29,9 @@ async def ask_question(question: Question):
     try:
         logger.info(f"Received question: {question.text}")
         llm = get_llm()
-        messages = [HumanMessage(content=question.text)]
-        response = llm(messages)
+        response = llm.invoke(question.text)
         logger.info("Response generated successfully")
-        return {"answer": response.content}
+        return {"answer": response}
     except Exception as e:
         logger.error(f"Error in /ask endpoint: {str(e)}")
         raise HTTPException(status_code=500, detail=str(e))
@@ -43,14 +41,12 @@ async def ask_question_stream(question: Question):
     try:
         logger.info(f"Received question for streaming: {question.text}")
         llm = get_llm()
-        messages = [HumanMessage(content=question.text)]
 
         async def generate():
             full_response = ""
-            async for chunk in llm.astream(messages):
-                if chunk.content:
-                    full_response += chunk.content
-                    yield chunk.content
+            async for chunk in llm.astream(question.text):
+                full_response += chunk
+                yield chunk
             
             # Log the full response after streaming is complete
             logger.info(f"Full streamed response: {full_response}")

requirements.txt

@@ -2,4 +2,5 @@ fastapi
 uvicorn[standard]
 langchain-ollama
 langchain
-pydantic
+pydantic
+langchain-community