Limit to 2 options

app.py

@@ -111,7 +111,7 @@ def prepare(raw_idx, lang, text_embeddings, class_order, randomize_images):
             return prepare(raw_idx, lang, text_embeddings, class_order, randomize_images)
 
         similarity = (text_embeddings @ image_features.cpu().numpy().T).squeeze()
-        choices = np.argsort(similarity)[-4:].tolist()
+        choices = np.argsort(similarity)[-2:].tolist()
     else:
         choices = list(reversed(precomputed_results[lang][idx][img_idx]))  # precomputing script uses torch.topk which sorts in reverse here
     if idx not in choices:
@@ -121,7 +121,7 @@ def prepare(raw_idx, lang, text_embeddings, class_order, randomize_images):
     numpy.random.shuffle(choices)
 
     choice_names = [class_labels[idx] for idx in choices]
-    choice_values = [0, 1, 2, 3]
+    choice_values = [0, 1]
 
     model_choice_idx = choices.index(model_choice_idx)
     model_choice = [choice_names[model_choice_idx], choice_values[model_choice_idx]]
@@ -160,7 +160,7 @@ def reroll(raw_idx, lang, text_embeddings, class_order, randomize_images):
             return prepare(raw_idx, lang, text_embeddings, class_order, randomize_images)
 
         similarity = (text_embeddings @ image_features.cpu().numpy().T).squeeze()
-        choices = np.argsort(similarity)[-4:].tolist()
+        choices = np.argsort(similarity)[-2:].tolist()
     else:
         choices = list(reversed(precomputed_results[lang][idx][img_idx]))  # precomputing script uses torch.topk which sorts in reverse here
     if idx not in choices:
@@ -170,7 +170,7 @@ def reroll(raw_idx, lang, text_embeddings, class_order, randomize_images):
     numpy.random.shuffle(choices)
 
     choice_names = [class_labels[idx] for idx in choices]
-    choice_values = [0, 1, 2, 3]
+    choice_values = [0, 1]
 
     model_choice_idx = choices.index(model_choice_idx)
     model_choice = [choice_names[model_choice_idx], choice_values[model_choice_idx]]