Update app.py

app.py

@@ -26,11 +26,10 @@ qwen_model = Qwen2VLForConditionalGeneration.from_pretrained(
 ).to(device).eval()
 qwen_processor = AutoProcessor.from_pretrained("prithivMLmods/JSONify-Flux", trust_remote_code=True)
 
-# Prompt Enhancer
 enhancer_long = pipeline("summarization", model="prithivMLmods/t5-Flan-Prompt-Enhance", device=device)
 
 MAX_SEED = np.iinfo(np.int32).max
-MAX_IMAGE_SIZE = 1024  # Reduced to prevent memory issues
+MAX_IMAGE_SIZE = 2048  
 
 # Qwen2VL caption function – updated to request plain text caption instead of JSON
 @spaces.GPU
@@ -44,7 +43,6 @@ def qwen_caption(image):
             "role": "user",
             "content": [
                 {"type": "image", "image": image},
-                # Removed "in the form of JSON data {}" to get plain text caption
                 {"type": "text", "text": "Generate a detailed and optimized caption for the given image."},
             ],
         }
@@ -161,7 +159,7 @@ with gr.Blocks(css=custom_css) as demo:
                 width = gr.Slider(label="Width", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=512)
                 height = gr.Slider(label="Height", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=512)
                 guidance_scale = gr.Slider(label="Guidance Scale", minimum=1, maximum=15, step=0.1, value=3.5)
-                num_inference_steps = gr.Slider(label="Inference Steps", minimum=1, maximum=50, step=1, value=28)
+                num_inference_steps = gr.Slider(label="Inference Steps", minimum=1, maximum=50, step=1, value=32)
             
             generate_btn = gr.Button("Generate Image Prompt", elem_classes="submit-btn")