Update src/leaderboard/read_evals.py

src/leaderboard/read_evals.py

@@ -31,6 +31,7 @@ class EvalResult:
     num_params: int = 0
     date: str = "" # submission date of request file
     still_on_hub: bool = False
+    display_model: str = "" # 새로 추가: 표시용 모델명
 
     @classmethod
     def init_from_json_file(self, json_filepath):
@@ -57,6 +58,12 @@ class EvalResult:
             result_key = f"{org}_{model}_{precision.value.name}"
         full_model = "/".join(org_and_model)
 
+        # 특정 모델명에 대한 맵핑 처리 추가
+        display_model = full_model
+        if full_model == "demo-leaderboard/gpt2-demo":
+            display_model = "deepseek-ai/DeepSeek-R1"
+            print(f"모델명 맵핑 적용: {full_model} -> {display_model}")
+
         still_on_hub, _, model_config = is_model_on_hub(
             full_model, config.get("model_sha", "main"), trust_remote_code=True, test_tokenizer=False
         )
@@ -88,7 +95,8 @@ class EvalResult:
             precision=precision,  
             revision= config.get("model_sha", ""),
             still_on_hub=still_on_hub,
-            architecture=architecture
+            architecture=architecture,
+            display_model=display_model  # 새로 추가한 필드 설정
         )
 
     def update_with_request_file(self, requests_path):
@@ -110,6 +118,10 @@ class EvalResult:
     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)
+        
+        # 표시용 모델명 사용
+        model_to_display = self.display_model if self.display_model else self.full_model
+        
         data_dict = {
             "eval_name": self.eval_name,  # not a column, just a save name,
             AutoEvalColumn.precision.name: self.precision.value.name,
@@ -117,7 +129,7 @@ class EvalResult:
             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.model.name: make_clickable_model(model_to_display),  # 수정된 부분
             AutoEvalColumn.revision.name: self.revision,
             AutoEvalColumn.average.name: average,
             AutoEvalColumn.license.name: self.license,
@@ -193,4 +205,4 @@ def get_raw_eval_results(results_path: str, requests_path: str) -> list[EvalResu
         except KeyError:  # not all eval values present
             continue
 
-    return results
+    return results