Update app.py

app.py

@@ -1,12 +1,16 @@
 import pandas as pd
 import gradio as gr
 
-def compare_csv_files(selected_languages):
+def compare_csv_files(selected_languages, model_size):
     max_num = 10
     
+    # Construct file names dynamically based on model size
+    file_1_5 = f"result_1.5_{model_size}.csv"
+    file_1_4 = f"result_1.4_{model_size}.csv"
+    
     # Load data
-    df1 = pd.read_csv("result_1.5.csv")
-    df2 = pd.read_csv("result_1.4.csv")
+    df1 = pd.read_csv(file_1_5)
+    df2 = pd.read_csv(file_1_4)
     
     # Merge with Language column
     merged_df = pd.merge(df1, df2, on=["SourceText", "Language"], suffixes=("_1.5", "_1.4"))
@@ -54,15 +58,18 @@ def compare_csv_files(selected_languages):
     
     return result_html
 
-# Load unique languages from the data
-df1 = pd.read_csv("result_1.5.csv")
-df2 = pd.read_csv("result_1.4.csv")
+# Load unique languages from the data (defaulting to Base files for initialization)
+df1 = pd.read_csv("result_1.5_Base.csv")
+df2 = pd.read_csv("result_1.4_Base.csv")
 languages = sorted(set(df1["Language"]).union(set(df2["Language"])))
 
 gr.Interface(
     fn=compare_csv_files,
-    inputs=gr.CheckboxGroup(choices=languages, label="Select Languages to Compare"),
+    inputs=[
+        gr.CheckboxGroup(choices=languages, label="Select Languages to Compare"),
+        gr.Dropdown(choices=["Base", "Medium"], label="Select Whisper Model Size", value="Base")
+    ],
     outputs="html",
     title="Fish Speech Benchmark",
-    description="Select specific languages to compare the results of WordErrorRate and CharacterErrorRate."
+    description="Select specific languages and model sizes (Base or Medium) to compare the results of WordErrorRate and CharacterErrorRate."
 ).launch()