Add multi-modal to args

Author: kcz358

docs/LLaVA_OneVision_Tutorials.ipynb

@@ -46,7 +46,11 @@
     "model_name = \"llava_qwen\"\n",
     "device = \"cuda\"\n",
     "device_map = \"auto\"\n",
-    "tokenizer, model, image_processor, max_length = load_pretrained_model(pretrained, None, model_name, device_map=device_map)  # Add any other thing you want to pass in llava_model_args\n",
+    "llava_model_args = {\n",
+    "    \"multimodal\": True,\n",
+    "    \"attn_implementation\": \"sdpa\",\n",
+    "}\n",
+    "tokenizer, model, image_processor, max_length = load_pretrained_model(pretrained, None, model_name, device_map=device_map, **llava_model_args)  # Add any other thing you want to pass in llava_model_args\n",
     "\n",
     "model.eval()\n",
     "\n",
@@ -308,7 +312,10 @@
     "model_name = \"llava_qwen\"\n",
     "device = \"cuda\"\n",
     "device_map = \"auto\"\n",
-    "tokenizer, model, image_processor, max_length = load_pretrained_model(pretrained, None, model_name, device_map=device_map, attn_implementation=\"sdpa\")\n",
+    "llava_model_args = {\n",
+    "    \"multimodal\": True,\n",
+    "}\n",
+    "tokenizer, model, image_processor, max_length = load_pretrained_model(pretrained, None, model_name, device_map=device_map, attn_implementation=\"sdpa\", **llava_model_args)\n",
     "\n",
     "model.eval()\n",
     "\n",