fix duplicated usernames

app.py

@@ -136,6 +136,9 @@ with gr.Blocks(title="OpenADMET ADMET Challenge", fill_height=False,
                         
                 Go to the **Leaderboard** to check out how the challenge is going. 
                 To participate, head out to the **Submit** tab and upload your results as a `CSV` file.
+                    
+                **We are releasing an intermediate leaderboard on December 1st so stay tuned!
+                        The submission cutoff for this leaderboard will be November 30th at 6pm EST. **
                         
                 """
                 )
@@ -241,6 +244,7 @@ with gr.Blocks(title="OpenADMET ADMET Challenge", fill_height=False,
     - **October 14:** Second announcement and sample data release
     - **October 27:** Challenge starts
     - **October-November:** Online Q&A sessions and support via the Discord channel
+    - **December 1st:** Intermediate leaderboard release
     - **January 19, 2026:** Submission closes
     - **January 26, 2026:** Winners announced
     ## Acknowledgements

final_lb.py

@@ -31,10 +31,13 @@ def build_leaderboard(df_results, df_results_raw):
             # MA-RAE is the average of the RAE per endpoint
             df = df.rename(columns={"mean_RAE": "mean_MA-RAE", 
                                     "std_RAE": "std_MA-RAE"}) 
+            # Delete duplicate entries before sorting (fixing case-sensitive duplicate check)
+            df['hf_username'] = df['hf_username'].apply(lambda s: s.lower())
+            df = df.sort_values(by="submission time", ascending=False, kind="stable")
+            df = df.drop_duplicates(subset=['hf_username'], keep='first')
+            # Sort by MAE-RAE
             sorted_df = df.sort_values(by='mean_MA-RAE', ascending=True, kind="stable")
             sorted_df = map_metric_to_stats(sorted_df, average=True)
-            # Add ranking column
-            sorted_df['rank'] = np.arange(1, len(sorted_df) + 1)
             avg_leaderboard = sorted_df.copy()
             avg_cols = LB_AVG
             # Add CLD
@@ -58,9 +61,16 @@ def build_leaderboard(df_results, df_results_raw):
             avg_leaderboard['user_real'] = avg_leaderboard['hf_username'].apply(validate_hf_username)
             avg_leaderboard_clean = avg_leaderboard[avg_leaderboard['user_real']]
 
+            # Add ranking column
+            avg_leaderboard_clean['rank'] = np.arange(1, len(avg_leaderboard_clean) + 1)
+
             per_ep[ep] = avg_leaderboard_clean[avg_cols]
 
         else:
+            # Delete duplicate entries before sorting (fixing case-sensitive duplicate check)
+            df['hf_username'] = df['hf_username'].apply(lambda s: s.lower())
+            df = df.sort_values(by="submission time", ascending=False, kind="stable")
+            df = df.drop_duplicates(subset=['hf_username'], keep='first')
             sorted_df = df.sort_values(by="mean_MAE", ascending=True, kind="stable")
             sorted_df = map_metric_to_stats(sorted_df)
             # Make sure Hugging Face username exists, if not, delete the row