from fastapi import FastAPI, HTTPException from pydantic import BaseModel from typing import Optional import torch from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig import logging logging.basicConfig(level=logging.INFO) logger = logging.getLogger(__name__) app = FastAPI() try: model_name = "scb10x/llama-3-typhoon-v1.5-8b-instruct" tokenizer = AutoTokenizer.from_pretrained(model_name) device = "cuda" if torch.cuda.is_available() else "cpu" logger.info(f"Using device: {device}") # 4-bit quantization configuration quantization_config = BitsAndBytesConfig( load_in_4bit=True, bnb_4bit_compute_dtype=torch.float16 ) model = AutoModelForCausalLM.from_pretrained( model_name, quantization_config=quantization_config, device_map="auto", low_cpu_mem_usage=True, ) logger.info(f"Model loaded successfully on {device}") except Exception as e: logger.error(f"Error loading model: {str(e)}") raise class Query(BaseModel): queryResult: Optional[dict] = None queryText: Optional[str] = None @app.post("/webhook") async def webhook(query: Query): try: user_query = query.queryResult.get('queryText') if query.queryResult else query.queryText if not user_query: raise HTTPException(status_code=400, detail="No query text provided") prompt = f"Human: {user_query}\nAI:" input_ids = tokenizer.encode(prompt, return_tensors="pt").to(device) with torch.no_grad(): output = model.generate(input_ids, max_new_tokens=100, temperature=0.7) response = tokenizer.decode(output[0], skip_special_tokens=True) ai_response = response.split("AI:")[-1].strip() return {"fulfillmentText": ai_response} except Exception as e: logger.error(f"Error in webhook: {str(e)}") raise HTTPException(status_code=500, detail=str(e)) if __name__ == "__main__": import uvicorn uvicorn.run(app, host="0.0.0.0", port=7860)