rag_utils.py

import os
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel

def load_model(model_path="./", base_model="mistralai/Mistral-7B-v0.1"):
    tokenizer = AutoTokenizer.from_pretrained(base_model)
    model = AutoModelForCausalLM.from_pretrained(
        base_model,
        torch_dtype=torch.float16,
        device_map="auto"
    )
    model = PeftModel.from_pretrained(model, model_path)
    model.eval()
    return model, tokenizer

def generate_answer(context, question, model, tokenizer):
    prompt = f"Context: {context}\n\nQ: {question}\nA:"
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_new_tokens=200,
            temperature=0.7,
            top_p=0.9,
            do_sample=True,
            pad_token_id=tokenizer.eos_token_id
        )

    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response.split("A:")[-1].strip()