warning

app.py

@@ -63,6 +63,8 @@ leaderboard_df = original_df.copy()
 leaderboard_df = leaderboard_df[leaderboard_df[AutoEvalColumn.still_on_hub.name] == True]
 # leaderboard_df = leaderboard_df[('speakleash' not in leaderboard_df['model_name_for_query']) | ('Bielik' in leaderboard_df['model_name_for_query'])]
 
+original_df.to_csv("output.csv")
+
 (
     finished_eval_queue_df,
     running_eval_queue_df,
@@ -400,6 +402,7 @@ with demo:
                 elem_id="citation-button",
                 show_copy_button=True,
             )
+        csv = gr.File(interactive=False, value="output.csv")
 
 scheduler = BackgroundScheduler()
 scheduler.add_job(restart_space, "interval", seconds=1800)

src/leaderboard/read_evals.py

@@ -433,16 +433,7 @@ def get_raw_eval_results(results_path: str, requests_path: str, metadata) -> lis
     for k,v in eval_results.items():
         v.results = {k: v for k, (v, start_date) in v.results.items()}
 
-    results = []
-    for v in eval_results.values():
-        try:
-            print(v)
-            v.to_dict()  # we test if the dict version is complete
-            # if v.results:
-            results.append(v)
-        except KeyError:  # not all eval values present
-            print(f"not all eval values present {v.eval_name} {v.full_model}")
-            continue
+
 
     all_models = []
     missing_results_for_task = {}
@@ -473,6 +464,17 @@ def get_raw_eval_results(results_path: str, requests_path: str, metadata) -> lis
             missing_metadata.append(f"{v.full_model}")
         all_models.append((v.full_model, v.num_params, v.still_on_hub))
 
+    results = []
+    for v in eval_results.values():
+        try:
+            print(v)
+            v.to_dict()  # we test if the dict version is complete
+            # if v.results:
+            results.append(v)
+        except KeyError:  # not all eval values present
+            print(f"not all eval values present {v.eval_name} {v.full_model}")
+            continue
+
     print(f"Missing sbatch results:")
     for r in for_run:
         if r[0]==5 and r[1] in ['polish_eq_bench']: continue