change layout

app.py

@@ -41,7 +41,7 @@ def infer(
 ##### Also returns bounding boxes with their labels
 
 def parse_segmentation(input_image, input_text):
-  out = infer(input_image, input_text, max_new_tokens=100)
+  out = infer(input_image, input_text, max_new_tokens=200)
   objs = extract_objs(out.lstrip("\n"), input_image.size[0], input_image.size[1], unique_labels=True)
   labels = set(obj.get('name') for obj in objs if obj.get('name'))
   color_map = {l: COLORS[i % len(COLORS)] for i, l in enumerate(labels)}
@@ -83,20 +83,21 @@ for detailed information how to use and fine-tune PaliGemma models.
 """
 
 
-with gr.Blocks(css="style.css") as demo:
+with gr.Blocks(theme=gr.themes.Soft(),css="style.css") as demo:
   gr.Markdown(INTRO_TEXT)
   with gr.Tab("Text Generation"):
-    with gr.Column():
-        image = gr.Image(type="pil")
+    with gr.Row():
+      with gr.Column():
+        image = gr.Image(type="pil", width=512, height=512)
         text_input = gr.Text(label="Input Text")
-
+      with gr.Column():
         text_output = gr.Text(label="Text Output")
-        chat_btn = gr.Button()
-        tokens = gr.Slider(
+    chat_btn = gr.Button()
+    tokens = gr.Slider(
             label="Max New Tokens",
             info="Set to larger for longer generation.",
             minimum=10,
-            maximum=300,
+            maximum=200,
             value=20,
             step=10,
         )
@@ -134,15 +135,19 @@ with gr.Blocks(css="style.css") as demo:
         inputs=chat_inputs,
     )
   with gr.Tab("Segment/Detect"):
-    image = gr.Image(type="pil")
-    seg_input = gr.Text(label="Entities to Segment/Detect")
-    seg_btn = gr.Button("Submit")
-    annotated_image = gr.AnnotatedImage(label="Output")
+    with gr.Row():
+      with gr.Column():
+        image = gr.Image(type="pil")
+        seg_input = gr.Text(label="Entities to Segment/Detect")
+        seg_btn = gr.Button("Submit")
+      with gr.Column():
+        annotated_image = gr.AnnotatedImage(label="Output")
     
-    examples = [["./examples/cats.png", "segment cats"],
+    examples = [["./examples/venice.jpg", "detect bird"],
+                ["./examples/cats.png", "segment cats"],
                 ["./examples/bee.jpg", "detect bee"],
                ["./examples/barsik.jpg", "segment cat"],
-               ["./examples/bird.jpg", "segment bird ; bird ; plant"]]
+               ]
     gr.Markdown("Example images are licensed CC0 by [akolesnikoff@](https://github.com/akolesnikoff), [mbosnjak@](https://github.com/mbosnjak), [maximneumann@](https://github.com/maximneumann) and [merve](https://huggingface.co/merve).")
     gr.Examples(
         examples=examples,