app.py

from fastapi import FastAPI
from transformers import AutoModel, AutoTokenizer

app = FastAPI()

# Load Model
model_path = "./model"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModel.from_pretrained(model_path)

@app.get("/")
def home():
    return {"message": "DeepSeek API is running!"}

@app.post("/predict")
async def predict(text: str):
    inputs = tokenizer(text, return_tensors="pt")
    outputs = model(**inputs)
    return {"output": outputs.last_hidden_state.tolist()}