Update app.py

app.py

@@ -7,11 +7,13 @@ from pathlib import Path
 from transformers import AutoModelForCausalLM, AutoTokenizer, AutoProcessor, Qwen2VLForConditionalGeneration
 from PIL import Image
 
+max_size = 240
+
 # Load Vision-Language Model
 vl_model = Qwen2VLForConditionalGeneration.from_pretrained(
     "Qwen/Qwen2-VL-2B-Instruct", torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32, device_map="auto"
 )
-vl_processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct", max_pixels=240*240)
+vl_processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct", max_pixels=max_size*max_size)
 
 # Load Text Model
 model_name = "Qwen/Qwen2.5-Math-1.5B-Instruct"
@@ -22,12 +24,15 @@ tokenizer = AutoTokenizer.from_pretrained(model_name)
 math_messages = []
 
 def resize_image(image):
-    max_size = 480
     if isinstance(image, str):  # Handle file paths
         image = Image.open(image)
-    image.thumbnail((max_size, max_size))
-    return image
-    
+    try:
+        image.thumbnail((max_size, max_size), Image.Resampling.LANCZOS)
+        return image
+    except Exception as e:
+        print(f"Error resizing image: {e}")
+        return None
+
 def process_image(image, shouldConvert=False):
     global math_messages
     math_messages = []  # Reset when uploading an image
@@ -96,7 +101,7 @@ with gr.Blocks(css=css) as demo:
         with gr.Column():
             with gr.Tabs() as input_tabs:
                 with gr.Tab("Upload"):
-                    input_image = resize_image(gr.Image(type="pil", label="Upload"))
+                    input_image = gr.Image(type="pil", label="Upload")
                 with gr.Tab("Sketch"):
                     input_sketchpad = gr.Sketchpad(type="pil", label="Sketch", layers=False)
             input_tabs.select(fn=tabs_select, inputs=[state])