app.py

from fastapi import FastAPI
import os

from custom_llm import CustomLLM

from pydantic import BaseModel
from langchain.prompts import PromptTemplate

class ConversationPost(BaseModel):
    tenant: str | None = None
    module: str | None = None
    question: str


API_TOKEN = os.environ['HF_API_KEY']
app = FastAPI()
prompt = PromptTemplate.from_template("""<|im_start|>system
Kamu adalah Asisten AI yang dikembangkan oleh Jonthan Jordan. Jawablah pertanyaan user secara ketat dalam Bahasa Indonesia<|im_end|>
<|im_start|>user
{question}<|im_end|>
<|im_start|>assistant
""")
llm = prompt | CustomLLM(repo_id="Qwen/Qwen-VL-Chat", model_type='text-generation', api_token=API_TOKEN, max_new_tokens=150).bind(stop=['<|im_end|>'])


@app.get("/")
def greet_json():
    return {"Hello": "World!"}


@app.post("/conversation"):
async def conversation(data : ConversationPost):
    return {"output":llm.invoke({"question":data.question})}