update scripts

.idea/workspace.xml

@@ -5,9 +5,7 @@
   </component>
   <component name="ChangeListManager">
     <list default="true" id="b60b8d5d-762f-4021-9ce4-054078cb27a7" name="Changes" comment="">
-      <change beforePath="$PROJECT_DIR$/.idea/workspace.xml" beforeDir="false" afterPath="$PROJECT_DIR$/.idea/workspace.xml" afterDir="false" />
       <change beforePath="$PROJECT_DIR$/app.py" beforeDir="false" afterPath="$PROJECT_DIR$/app.py" afterDir="false" />
-      <change beforePath="$PROJECT_DIR$/src/display/utils.py" beforeDir="false" afterPath="$PROJECT_DIR$/src/display/utils.py" afterDir="false" />
     </list>
     <option name="SHOW_DIALOG" value="false" />
     <option name="HIGHLIGHT_CONFLICTS" value="true" />
@@ -42,17 +40,17 @@
       <breakpoints>
         <line-breakpoint enabled="true" suspend="THREAD" type="python-line">
           <url>file://$PROJECT_DIR$/app.py</url>
-          <line>54</line>
+          <line>53</line>
           <option name="timeStamp" value="39" />
         </line-breakpoint>
         <line-breakpoint enabled="true" suspend="THREAD" type="python-line">
           <url>file://$PROJECT_DIR$/app.py</url>
-          <line>71</line>
+          <line>70</line>
           <option name="timeStamp" value="40" />
         </line-breakpoint>
         <line-breakpoint enabled="true" suspend="THREAD" type="python-line">
           <url>file://$PROJECT_DIR$/app.py</url>
-          <line>72</line>
+          <line>71</line>
           <option name="timeStamp" value="41" />
         </line-breakpoint>
       </breakpoints>

app.py

@@ -51,12 +51,11 @@ def process_pending_evals():
 
     for _, eval_request in pending_eval_queue_df.iterrows():
         # 根据具体的列名和数据格式调整 `eval_request` 的内容
-        model_id = eval_request['model'].split('/')[-1].split('">')[0]
-        print(f"Evaluating model: {model_id}")
+        print(f"Evaluating model: {eval_request.model}")
 
         # 调用评估函数
         run_eval_suite.run_evaluation(
-            eval_request=model_id,
+            eval_request=eval_request,
             local_dir=envs.EVAL_RESULTS_PATH_BACKEND,
             results_repo=envs.RESULTS_REPO,
             batch_size=1,