default models

app.py

@@ -1,5 +1,3 @@
-# pip install -U gradio transformers pillow matplotlib
-
 import io
 from typing import Optional
 
@@ -9,6 +7,20 @@ from PIL import Image
 
 from transformers.utils.processor_visualizer_utils import ImageVisualizer
 
+MODELS = [
+    "openai/clip-vit-base-patch32",
+    "HuggingFaceM4/Idefics3-8B-Llama3",
+    "llava-hf/llava-1.5-7b-hf",
+    "OpenGVLab/InternVL2-2B",
+    "OpenGVLab/InternVL3-8B-hf",
+    "Salesforce/blip-image-captioning-base",
+    "Salesforce/blip2-flan-t5-xl",
+    "Qwen/Qwen2-VL-2B-Instruct",
+    "Qwen/Qwen2.5-VL-3B-Instruct",
+    "meta-llama/Llama-3.2-11B-Vision",
+    "microsoft/Florence-2-base",
+    "laion/CLIP-ViT-B-32-laion2B-s34B-b79K",
+]
 
 def _fig_to_pil(fig) -> Image.Image:
     buf = io.BytesIO()
@@ -16,16 +28,13 @@ def _fig_to_pil(fig) -> Image.Image:
     buf.seek(0)
     return Image.open(buf).convert("RGB")
 
-
 def _run(model_id: str, image: Optional[Image.Image], use_sample: bool, add_grid: bool):
     viz = ImageVisualizer(model_id)
 
-    # Capture all matplotlib figures the visualizer produces without changing the utility.
     captured = []
     orig_show = plt.show
 
     def _capture_show(*_, **__):
-        # collect the current figure then do not actually display
         fig = plt.gcf()
         captured.append(fig)
 
@@ -35,32 +44,40 @@ def _run(model_id: str, image: Optional[Image.Image], use_sample: bool, add_grid
     finally:
         plt.show = orig_show
 
-    # Convert figures to PIL for Gradio
     imgs = [_fig_to_pil(fig) for fig in captured] if captured else []
     prompt_preview = viz.default_message(full_output=False)
     return imgs, prompt_preview
 
 
 with gr.Blocks(title="Transformers Processor Visualizer") as demo:
-    gr.Markdown("Switch models and see what the processor actually feeds them (uses the existing `ImageVisualizer`).")
+    gr.Markdown("Switch models and see what the processor feeds them (uses the existing `ImageVisualizer`).")
 
     with gr.Row():
-        model_id = gr.Textbox(
+        model_id = gr.Dropdown(
             label="Model repo_id",
-            value="openai/clip-vit-base-patch32",
-            placeholder="owner/repo (e.g., llava-hf/llava-1.5-7b-hf)",
+            choices=MODELS,
+            value=MODELS[0],
+            allow_custom_value=True,
+            filterable=True,
         )
         add_grid = gr.Checkbox(label="Show patch grid", value=True)
         use_sample = gr.Checkbox(label="Use HF logo sample", value=True)
 
-    image = gr.Image(label="Or upload an image", type="pil")
+    image = gr.Image(label="Upload custom image", type="pil", height=140, width=140, sources=["upload"])
+
+    def _on_image_change(img):
+        return False  # uncheck the sample toggle when a custom image is set
 
+    image.change(_on_image_change, inputs=image, outputs=use_sample)
     run_btn = gr.Button("Render")
 
     gallery = gr.Gallery(label="Processor output")
     prompt = gr.Textbox(label="Compact chat template preview")
-
+    # Render on demand
     run_btn.click(_run, inputs=[model_id, image, use_sample, add_grid], outputs=[gallery, prompt])
 
+    # Also render once on load with defaults so there is an example before clicking
+    demo.load(_run, inputs=[model_id, image, use_sample, add_grid], outputs=[gallery, prompt])
+
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()