main.py

from fastapi import FastAPI, Form
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel
from transformers import AutoModelForCausalLM

# Model loading
llm = AutoModelForCausalLM.from_pretrained(
    "zephyr-7b-beta.Q4_K_S.gguf",
    model_type='mistral',
    max_new_tokens=1096,
    threads=3,
)

# Pydantic object
class Validation(BaseModel):
    prompt: str

# Fast API
app = FastAPI()

# Set up CORS
origins = [
    "http://localhost",           # Replace with the address of your Flutter web app
    "http://localhost:55345",     # Add the port used by your Flutter web app
]

app.add_middleware(
    CORSMiddleware,
    allow_origins=origins,
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# Zephyr completion
@app.post("/llm_on_cpu")
async def stream(item: Validation):
    system_prompt = 'Below is an instruction that describes a task. Write a response that appropriately completes the request.'
    E_INST = "</s>"
    user, assistant = "", ""
    prompt = f"{system_prompt}{E_INST}\n{user}\n{item.prompt.strip()}{E_INST}\n{assistant}\n"
    return llm(prompt)