Update app.py

app.py

@@ -243,14 +243,9 @@ with gr.Blocks(title="moondream vl (new)", css=css, js=js) as demo:
 
     gr.Markdown(
         """
-        # 🌔 test space, pls ignore
+        # 🌔 grounded visual question asnwering
         """
     )
-    mode_radio = gr.Radio(
-        ["Caption", "Query", "Detect", "Point", "Localized"],
-        show_label=False,
-        value=lambda: "Caption",
-    )
 
     input_image = gr.State(None)
 
@@ -259,85 +254,26 @@ with gr.Blocks(title="moondream vl (new)", css=css, js=js) as demo:
 
             @gr.render(inputs=[mode_radio])
             def show_inputs(mode):
-                if mode == "Query":
-                    with gr.Group():
-                        with gr.Row():
-                            prompt = gr.Textbox(
-                                label="Input",
-                                value="How many people are in this image?",
-                                scale=4,
-                            )
-                            submit = gr.Button("Submit")
-                        reasoning = gr.Checkbox(label="Enable reasoning")
-                        img = gr.Image(type="pil", label="Upload an Image")
-                    submit.click(answer_question, [img, prompt, reasoning], [output, thought])
-                    prompt.submit(answer_question, [img, prompt, reasoning], [output, thought])
-                    reasoning.change(answer_question, [img, prompt, reasoning], [output, thought])
-                    img.change(answer_question, [img, prompt, reasoning], [output, thought])
-                    img.change(lambda img: img, [img], [input_image])
-                elif mode == "Caption":
-                    with gr.Group():
-                        with gr.Row():
-                            caption_mode = gr.Radio(
-                                ["Short", "Normal", "Long"],
-                                label="Caption Length",
-                                value=lambda: "Normal",
-                                scale=4,
-                            )
-                            submit = gr.Button("Submit")
-                        img = gr.Image(type="pil", label="Upload an Image")
-                    submit.click(caption, [img, caption_mode], output)
-                    img.change(caption, [img, caption_mode], output)
-                elif mode == "Detect":
-                    with gr.Group():
-                        with gr.Row():
-                            prompt = gr.Textbox(
-                                label="Object",
-                                value="Cat",
-                                scale=4,
-                            )
-                            submit = gr.Button("Submit")
-                        img = gr.Image(type="pil", label="Upload an Image")
-                        eos_bias = gr.Textbox(label="EOS Bias", value="0")
-                    submit.click(detect, [img, prompt, eos_bias], [thought, ann])
-                    prompt.submit(detect, [img, prompt, eos_bias], [thought, ann])
-                    img.change(detect, [img, prompt, eos_bias], [thought, ann])
-                elif mode == "Point":
-                    with gr.Group():
-                        with gr.Row():
-                            prompt = gr.Textbox(
-                                label="Object",
-                                value="Cat",
-                                scale=4,
-                            )
-                            submit = gr.Button("Submit")
-                        img = gr.Image(type="pil", label="Upload an Image")
-                    submit.click(point, [img, prompt], [thought, ann])
-                    prompt.submit(point, [img, prompt], [thought, ann])
-                    img.change(point, [img, prompt], [thought, ann])
-                elif mode == "Localized":
-                    with gr.Group():
-                        with gr.Row():
-                            prompt = gr.Textbox(
-                                label="Input",
-                                value="What is this?",
-                                scale=4,
-                            )
-                            submit = gr.Button("Submit")
-                        img = gr.Image(type="pil", label="Upload an Image")
-                        x_slider = gr.Slider(label="x", minimum=0, maximum=1)
-                        y_slider = gr.Slider(label="y", minimum=0, maximum=1)
-                    submit.click(localized_query, [img, x_slider, y_slider, prompt], [output, thought, ann])
-                    prompt.submit(localized_query, [img, x_slider, y_slider, prompt], [output, thought, ann])
-                    x_slider.change(localized_query, [img, x_slider, y_slider, prompt], [output, thought, ann])
-                    y_slider.change(localized_query, [img, x_slider, y_slider, prompt], [output, thought, ann])
-                    img.change(localized_query, [img, x_slider, y_slider, prompt], [output, thought, ann])
-                    def select_handler(image, evt: gr.SelectData):
-                        w, h = img.size
-                        return [evt.index[0] / w, evt.index[1] / h]
-                    img.select(select_handler, img, [x_slider, y_slider])
-                else:
-                    gr.Markdown("Coming soon!")
+                with gr.Group():
+                    with gr.Row():
+                        prompt = gr.Textbox(
+                            label="Input",
+                            value="What is this?",
+                            scale=4,
+                        )
+                        submit = gr.Button("Submit")
+                    img = gr.Image(type="pil", label="Upload an Image")
+                    x_slider = gr.Slider(label="x", minimum=0, maximum=1)
+                    y_slider = gr.Slider(label="y", minimum=0, maximum=1)
+                submit.click(localized_query, [img, x_slider, y_slider, prompt], [output, thought, ann])
+                prompt.submit(localized_query, [img, x_slider, y_slider, prompt], [output, thought, ann])
+                x_slider.change(localized_query, [img, x_slider, y_slider, prompt], [output, thought, ann])
+                y_slider.change(localized_query, [img, x_slider, y_slider, prompt], [output, thought, ann])
+                img.change(localized_query, [img, x_slider, y_slider, prompt], [output, thought, ann])
+                def select_handler(image, evt: gr.SelectData):
+                    w, h = img.size
+                    return [evt.index[0] / w, evt.index[1] / h]
+                img.select(select_handler, img, [x_slider, y_slider])
 
         with gr.Column():
             thought = gr.HighlightedText(