update

app.py

@@ -150,16 +150,18 @@ def calculate_order_by_first_substring(selected_models):
         lambda x: x["parsed_judge_response"].eq(1).all()
     )
 
-
+    fsm_ids = query_ids_df.fsm_id.unique()
 
     text_only = all_data[all_data["Model Type"] == "Text Only"]
     text_only_filtered = text_only[text_only["fsm_id"].isin(fsm_ids)]
 
+    query_ids = text_only_filtered.query_id.unique()
     text_only_filtered = (
         text_only_filtered.groupby(["Model Name"])["parsed_judge_response"]
         .mean()
         .reset_index()
     )
+
     text_only_filtered["Accuracy"] = text_only_filtered["parsed_judge_response"] * 100
     text_only_filtered.drop("parsed_judge_response", axis=1, inplace=True)
 
@@ -168,9 +170,6 @@ def calculate_order_by_first_substring(selected_models):
     )
     text_only_filtered.sort_values("Accuracy", ascending=False, inplace=True)
 
-    query_ids = text_only_filtered.query_id.unique()
-    fsm_ids = text_only_filtered.fsm_id.unique()
-
     number_of_queries = len(query_ids)
     number_of_fsms = len(fsm_ids)
 
@@ -228,9 +227,9 @@ with gr.Blocks() as demo:
             interactive=True,
         )
         with gr.Row():
-            number_of_queries = gr.Textbox(label="Number of queries to include")
+            number_of_queries = gr.Textbox(label="Number of included queries")
 
-            number_of_fsms = gr.Textbox(label="Number of FSMs to include")
+            number_of_fsms = gr.Textbox(label="Number of included  FSMs")
 
         constrained_leader_board_text = gr.Dataframe()