from fastapi import FastAPI
from pydantic import BaseModel
import torch
from transformers import BertTokenizer
from custom_model import MyBERTClassifier

app = FastAPI()

# The model is loaded from the repository
repo_id = "jmt-r/predict-ai-abstract"
tokenizer = BertTokenizer.from_pretrained(repo_id)
model = MyBERTClassifier.from_pretrained(repo_id)
model.eval()

class PatentRequest(BaseModel):
    text: str

@app.post("/predict")
def predict(request: PatentRequest):
    inputs = tokenizer(request.text, return_tensors="pt", truncation=True, max_length=512)
    with torch.no_grad():
        logits = model(**inputs)
        probs = torch.softmax(logits, dim=1)
        ai_prob = probs[0][1].item()
    
    # Logic for YES AI (1) or NO AI (0)
    prediction = 1 if ai_prob > 0.783 else 0
    return {"prediction": prediction, "label": "YES AI" if prediction == 1 else "NO AI", "probability": ai_prob}