app.py

from fastapi import FastAPI
from pydantic import BaseModel
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

# Khởi tạo app FastAPI
app = FastAPI()

# Load model và tokenizer
model_name = "bmd1905/BARTpho2-ViT5-text2text"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

# Định nghĩa schema cho request
class InputText(BaseModel):
    text: str

# Endpoint POST /generate
@app.post("/generate")
def generate_text(item: InputText):
    inputs = tokenizer(item.text, return_tensors="pt", truncation=True, max_length=512)
    outputs = model.generate(**inputs, max_new_tokens=128)
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return {"answer": response}