Update src/leaderboard/read_evals.py

src/leaderboard/read_evals.py

@@ -108,9 +108,33 @@ class EvalResult:
         except Exception:
             print(f"Could not find request file for {self.org}/{self.model} with precision {self.precision.value.name}")
 
+    # def to_dict(self):
+    #     """Converts the Eval Result to a dict compatible with our dataframe display"""
+    #     average = sum([v for v in self.results.values() if v is not None]) / len(Tasks)
+    #     data_dict = {
+    #         "eval_name": self.eval_name,  # not a column, just a save name,
+    #         AutoEvalColumn.precision.name: self.precision.value.name,
+    #         AutoEvalColumn.model_type.name: self.model_type.value.name,
+    #         AutoEvalColumn.model_type_symbol.name: self.model_type.value.symbol,
+    #         AutoEvalColumn.weight_type.name: self.weight_type.value.name,
+    #         AutoEvalColumn.architecture.name: self.architecture,
+    #         AutoEvalColumn.model.name: make_clickable_model(self.full_model),
+    #         AutoEvalColumn.revision.name: self.revision,
+    #         AutoEvalColumn.average.name: average,
+    #         AutoEvalColumn.license.name: self.license,
+    #         AutoEvalColumn.likes.name: self.likes,
+    #         AutoEvalColumn.params.name: self.num_params,
+    #         AutoEvalColumn.still_on_hub.name: self.still_on_hub,
+    #     }
+
+    #     for task in Tasks:
+    #         data_dict[task.value.col_name] = self.results.get(task.value.benchmark, None)
+    #         print(f"Debug: {task.value.col_name} = {self.results.get(task.value.benchmark, 'N/A')}")
+
+    #     return data_dict
     def to_dict(self):
         """Converts the Eval Result to a dict compatible with our dataframe display"""
-        average = sum([v for v in self.results.values() if v is not None]) / len(Tasks)
+        average = sum([v for v in self.results.values() if v is not None]) / len(self.results)
         data_dict = {
             "eval_name": self.eval_name,  # not a column, just a save name,
             AutoEvalColumn.precision.name: self.precision.value.name,
@@ -126,14 +150,19 @@ class EvalResult:
             AutoEvalColumn.params.name: self.num_params,
             AutoEvalColumn.still_on_hub.name: self.still_on_hub,
         }
-
+    
         for task in Tasks:
-            data_dict[task.value.col_name] = self.results.get(task.value.benchmark, None)
-            print(f"Debug: {task.value.col_name} = {self.results.get(task.value.benchmark, 'N/A')}")
-
+            task_value = self.results.get(task.value.benchmark)
+            if task_value is not None:
+                data_dict[task.value.col_name] = task_value
+            else:
+                data_dict[task.value.col_name] = 'N/A'  # スコアが存在しない場合の処理
+            print(f"Debug: {task.value.col_name} = {data_dict[task.value.col_name]}")
+    
         return data_dict
 
 
+
 def get_request_file_for_model(requests_path, model_name, precision):
     """Selects the correct request file for a given model. Only keeps runs tagged as FINISHED"""
     request_files = os.path.join(