test

app.py

@@ -30,10 +30,10 @@ with gr.Blocks() as demo:
     gr.Markdown("# Leaderboards")
     with gr.Row():
         with gr.Column():
-            gr.Markdown("## Leaderboard 1")
+            gr.Markdown("## SWE Leaderboard")
             gr.Dataframe(leaderboard1)
         with gr.Column():
-            gr.Markdown("## Leaderboard 2")
+            gr.Markdown("## GAIA Leaderboard")
             gr.Dataframe(leaderboard2)
     gr.Markdown("# Submit Here")
     with gr.Row():

leaderboard_gaia.csv

@@ -3,4 +3,7 @@ Rank,Model,Joint F1,Categorical F1,Location Accuracy,Date
 2,Gemini-2.5-Flash-Preview-04-17,0.100,0.337,0.372,2025-05-14
 3,Open AI o3,0.092,0.296,0.535,2025-05-14
 4,Anthropic Claude-3.7-Sonnet,0.047,0.254,0.204,2025-05-14
-5,GPT-4.1,0.028,0.218 ,0.107,2025-05-14
+5,GPT-4.1,0.028,0.218,0.107,2025-05-14
+6,Open AI o1,0.013,0.138,0.040,2025-05-14
+7,Llama-4-Maverick-17B-128E-Instruct,0.122,0.023,0.000,2025-05-14
+8,Llama-4-Scout-17B-16E-Instruct,0.041,0.000,0.000,2025-05-14

leaderboard_swe.csv

@@ -3,7 +3,8 @@ Rank,Model,Joint F1,Categorical F1,Location Accuracy,Date
 2,Gemini-2.5-Flash-Preview-04-17,0.000,0.213,0.060,2025-05-14
 3,Llama-4-Maverick-17B-128E-Instruct,0.000,0.191,0.083,2025-05-14
 4,GPT-4.1,0.000,0.166,0.000,2025-05-14
-5,Open AI o1,CLE,CLE,CLE,2025-05-14
-6,Open AI o3,CLE,CLE,CLE,2025-05-14
-7,Anthropic Claude-3.7-Sonnet,CLE,CLE,CLE,2025-05-14
+5,Llama-4-Scout-17B-16E-Instruct,0.000,0.050,0.000,2025-05-14
+6,Open AI o1,CLE,CLE,CLE,2025-05-14
+7,Open AI o3,CLE,CLE,CLE,2025-05-14
+8,Anthropic Claude-3.7-Sonnet,CLE,CLE,CLE,2025-05-14