Update app.py

app.py

@@ -12,9 +12,9 @@ import os, stat, io
 
 # Load the model in half-precision on the available device(s)
 model = Qwen2VLForConditionalGeneration.from_pretrained(
-    "./Qwen2-VL-7B-Instruct", torch_dtype="auto", device_map="auto"
+    "./Qwen2-VL-2B-Instruct-GPTQ-Int8", torch_dtype="auto", device_map="auto"
 )
-processor = AutoProcessor.from_pretrained("./Qwen2-VL-7B-Instruct")
+processor = AutoProcessor.from_pretrained("./Qwen2-VL-2B-Instruct-GPTQ-Int8")
 
 def array_to_image(image_array):
     if image_array is None:
@@ -50,7 +50,7 @@ def describe_image(image_array):
     inputs = processor(
         text=[text_prompt], images=[image], padding=True, return_tensors="pt"
     )
-    inputs = inputs.to("cpu")
+    #inputs = inputs.to("cpu")
 
     # Inference: Generation of the output
     output_ids = model.generate(**inputs, max_new_tokens=128)