commit

app.py

@@ -16,30 +16,30 @@ def refresh_data():
     global_mmlu_df, global_unified_exam_df = model_handler.get_arm_bench_data()
 
     global_output_armenian = unified_exam_result_table(global_unified_exam_df)
-    global_output_mmlu = mmlu_result_table(global_mmlu_df)
+    # global_output_mmlu = mmlu_result_table(global_mmlu_df)
     
-    return global_output_armenian, global_output_mmlu, unified_exam_chart(global_output_armenian, 'Average'), mmlu_chart(global_output_mmlu, 'Average')
+    return global_output_armenian, unified_exam_chart(global_output_armenian, 'Average')
 
 def main():
-    # global global_mmlu_df, global_unified_exam_df, global_output_armenian, global_output_mmlu
-    # model_handler = ModelHandler()
-    # global_mmlu_df, global_unified_exam_df = model_handler.get_arm_bench_data()
+    global global_mmlu_df, global_unified_exam_df, global_output_armenian, global_output_mmlu
+    model_handler = ModelHandler()
+    global_mmlu_df, global_unified_exam_df = model_handler.get_arm_bench_data()
 
-    # global_output_armenian = unified_exam_result_table(global_unified_exam_df)
+    global_output_armenian = unified_exam_result_table(global_unified_exam_df)
     # global_output_mmlu = mmlu_result_table(global_mmlu_df)
 
     with gr.Blocks() as app:
         with gr.Tabs():
-            # with gr.TabItem("Armenian Unified Exams"):
-            #     gr.Markdown("# Armenian Unified Test Exams")
-            #     gr.Markdown(
-            #         """
-            #         This benchmark contains results of various Language Models on Armenian Unified Test Exams for Armenian language and literature, Armenian history and mathematics. The scoring system is a 20-point scale, where 0-8 is a Fail, 8-18 is a Pass, and 18-20 is a Distinction.
-            #         """
-            #     )
-            #     table_output_armenian = gr.DataFrame(value=global_output_armenian)
-            #     plot_column_dropdown_unified_exam = gr.Dropdown(choices=['Average', 'Armenian language and literature', 'Armenian history', 'Mathematics'], value='Average', label='Select Column to Plot')
-            #     plot_output_armenian = gr.Plot(lambda column: unified_exam_chart(global_output_armenian, column), inputs=plot_column_dropdown_unified_exam)
+            with gr.TabItem("Armenian Unified Exams"):
+                gr.Markdown("# Armenian Unified Test Exams")
+                gr.Markdown(
+                    """
+                    This benchmark contains results of various Language Models on Armenian Unified Test Exams for Armenian language and literature, Armenian history and mathematics. The scoring system is a 20-point scale, where 0-8 is a Fail, 8-18 is a Pass, and 18-20 is a Distinction.
+                    """
+                )
+                table_output_armenian = gr.DataFrame(value=global_output_armenian)
+                plot_column_dropdown_unified_exam = gr.Dropdown(choices=['Average', 'Armenian language and literature', 'Armenian history', 'Mathematics'], value='Average', label='Select Column to Plot')
+                plot_output_armenian = gr.Plot(lambda column: unified_exam_chart(global_output_armenian, column), inputs=plot_column_dropdown_unified_exam)
             # with gr.TabItem("MMLU-Pro-Hy"):
             #     gr.Markdown("# MMLU-Pro Translated to Armenian (MMLU-Pro-Hy)")
             #     gr.Markdown(
@@ -108,14 +108,14 @@ def main():
                     """
                 )
 
-        # refresh_button = gr.Button("Refresh Data")
-        # refresh_button.click(
-        #     fn=refresh_data,
-        #     # outputs=[table_output_armenian,
-        #     #          table_output_mmlu,
-        #     #          plot_output_armenian,
-        #     #          plot_output_mmlu],
-        # )
+        refresh_button = gr.Button("Refresh Data")
+        refresh_button.click(
+            fn=refresh_data,
+            outputs=[table_output_armenian,
+                    #  table_output_mmlu,
+                     plot_output_armenian,
+                    #  plot_output_mmlu],
+        )
     app.launch(share=True, debug=True)
 
 if __name__ == "__main__":