Update llm.py

llm.py

@@ -1,6 +1,6 @@
-from huggingface_hub import login
+
 import os
-login(token=os.getenv("HUGGINGFACE_HUB_TOKEN"))
+
 import asyncio
 from functools import partial
 import time
@@ -8,9 +8,10 @@ import torch
 from transformers import AutoProcessor, AutoModelForImageTextToText
 model_name = "Qwen/Qwen3.5-9B-Base"
 
-
+hf_token = os.getenv("HUGGINGFACE_HUB_TOKEN")
 processor = AutoProcessor.from_pretrained(
     model_name,
+    token=hf_token,
     trust_remote_code=True
 )
 
@@ -18,6 +19,7 @@ model = AutoModelForImageTextToText.from_pretrained(
     model_name,
     device_map="auto",
     dtype=torch.float16,
+    token=hf_token,
     trust_remote_code=True
 )