Sorted Captions per confidence scores

app.py

@@ -11,9 +11,9 @@ def inference(input_img, captions):
     outputs = model(**inputs)
     # 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
+    probs = logits_per_image.softmax(dim=1).tolist()[0]
+    confidences = {captions_list[i][:30]: 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 "
@@ -29,8 +29,9 @@ examples = [["example_images/12863.jpg","photo of water, a photo of pizza, photo
 
 demo = gr.Interface(
     inference,
-    inputs = [gr.Image(shape=(416, 416), label="Input Image"), gr.Textbox(placeholder="Enter different captions for image, separated by comma")],
-    outputs = [gr.Textbox(label="Probability score of captions")],
+    inputs = [gr.Image(shape=(416, 416), label="Input Image"), 
+              gr.Textbox(placeholder="Enter different captions for image, separated by comma")],
+    outputs = [gr.Label()],
     title = title,
     description = description,
     examples = examples,