fix

app.py

@@ -2,11 +2,13 @@ from fastapi import FastAPI, HTTPException
 from pydantic import BaseModel
 import transformers
 
+# Load model directly
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
 app = FastAPI()
 
-model_name = 'Intel/neural-chat-7b-v3-1'
-model = transformers.AutoModelForCausalLM.from_pretrained(model_name)
-tokenizer = transformers.AutoTokenizer.from_pretrained(model_name)
+tokenizer = AutoTokenizer.from_pretrained("Intel/neural-chat-7b-v3-1")
+model = AutoModelForCausalLM.from_pretrained("Intel/neural-chat-7b-v3-1")
 
 class ChatInput(BaseModel):
     system_input: str