updated model loading

app.py

@@ -9,7 +9,7 @@ from io import BytesIO
 
 
 models = {
-    "Qwen/Qwen2-VL-7B-Instruct": Qwen2VLForConditionalGeneration.from_pretrained("Qwen/Qwen2-VL-7B-Instruct") #, torch_dtype="auto", device_map="auto")
+    "Qwen/Qwen2-VL-7B-Instruct": Qwen2VLForConditionalGeneration.from_pretrained("Qwen/Qwen2-VL-7B-Instruct", torch_dtype="auto", device_map="auto")
 }
 
 processors = {
@@ -28,7 +28,7 @@ def image_to_base64(image):
 
 @spaces.GPU
 def run_example(image, text_input, model_id="Qwen/Qwen2-VL-7B-Instruct"):
-    model = models[model_id].eval().cuda()
+    model = models[model_id].eval()
     processor = processors[model_id]
 
     messages = [

requirements.txt

@@ -3,6 +3,7 @@ Pillow==10.3.0
 Requests==2.31.0
 torch
 torchvision
+transformers
 git+https://github.com/huggingface/transformers.git@main
 accelerate==0.30.0
 qwen-vl-utils