label prob output

app.py

@@ -11,8 +11,8 @@ def inference(input_img, captions):
     # this is the image-text similarity score
     logits_per_image = outputs.logits_per_image
     probs = logits_per_image.softmax(dim=1)
-    probabilities_percentages = ', '.join(['{:.2f}%'.format(prob.item() * 100) for prob in probs[0]])
-    return probabilities_percentages
+    confidences = {i: probs[i] for i in range(len(probs))}
+    return confidences
 
 title = "CLIP Inference: Application using a pretrained CLIP model"
 description = "An application using Gradio interface that accepts an image and some captions, and displays a probability score with which each caption describes the image "
@@ -31,6 +31,7 @@ demo = gr.Interface(
         gr.Image(shape=(416, 416), label="Input Image"),
         gr.Textbox(placeholder="List of captions")],
     outputs = [gr.Textbox(label="Probability score on captions likely describing the image")],
+    outputs = gr.Label(),
     title = title,
     description = description,
     examples = examples,