Upload app.py with huggingface_hub

app.py

@@ -93,9 +93,9 @@ def stream_model(model, image: Image.Image):
     thread.join()
 
 
-def predict(image_path):
+def predict_local(image_path):
     if image_path is None:
-        yield "Please upload an image.", "Please upload an image.", "Please upload an image."
+        yield "Please upload an image.", "Please upload an image."
         return
 
     image = Image.open(image_path).convert("RGB")
@@ -103,16 +103,24 @@ def predict(image_path):
     finetuned_text = ""
     for chunk in stream_model(finetuned_model, image):
         finetuned_text += chunk
-        yield finetuned_text, "", ""
+        yield finetuned_text, ""
 
     original_text = ""
     for chunk in stream_model(original_model, image):
         original_text += chunk
-        yield finetuned_text, original_text, ""
+        yield finetuned_text, original_text
+
+
+def predict_gpt(image_path):
+    if image_path is None:
+        yield "Please upload an image."
+        return
+
+    yield "Calling GPT-5.5..."
 
-    gpt_text = ""
     try:
         from openai import OpenAI
+
         with open(image_path, "rb") as f:
             image_data = base64.b64encode(f.read()).decode("utf-8")
         ext = os.path.splitext(image_path)[1].lstrip(".").lower()
@@ -133,20 +141,23 @@ def predict(image_path):
             max_completion_tokens=1024,
             stream=True,
         )
+
+        gpt_text = ""
         for chunk in stream:
             delta = chunk.choices[0].delta.content
             if delta:
                 gpt_text += delta
-                yield finetuned_text, original_text, gpt_text
-    except Exception as e:
-        gpt_text = f"[Error: {e}]"
-        yield finetuned_text, original_text, gpt_text
+                yield gpt_text
+
+        if not gpt_text:
+            yield "[No response received from GPT-5.5]"
 
-    yield finetuned_text, original_text, gpt_text
+    except Exception as e:
+        yield f"[Error: {e}]"
 
 
 if HAS_SPACES:
-    predict = spaces.GPU(duration=180)(predict)
+    predict_local = spaces.GPU(duration=180)(predict_local)
 
 
 with gr.Blocks(title="Noteworthy — Sheet Music Transcription") as demo:
@@ -183,9 +194,14 @@ with gr.Blocks(title="Noteworthy — Sheet Music Transcription") as demo:
         )
 
     notes_btn.click(
-        fn=predict,
+        fn=predict_local,
+        inputs=[image_input],
+        outputs=[finetuned_output, original_output],
+    )
+    notes_btn.click(
+        fn=predict_gpt,
         inputs=[image_input],
-        outputs=[finetuned_output, original_output, gpt_output],
+        outputs=[gpt_output],
     )
 
 demo.launch(theme=gr.themes.Soft())