Debug

eval-results/demo-leaderboard/gpt2-demo/results_2023-11-22 15:46:20.425378.json

@@ -17,7 +17,9 @@
   },
   "versions": {
     "anli_r1": 0,
-    "logiqa": 0
+    "logiqa": 0,
+    "realtoxicityprompts": 0,
+    "toxigen": 0
   },
   "config": {
     "model": "hf-causal-experimental",

src/leaderboard/read_evals.py

@@ -127,6 +127,12 @@ class EvalResult:
         }
 
         for task in Tasks:
+            print("Examinging task")
+            print(task)
+            print("Data dict:")
+            print(data_dict[task.value.col_name])
+            print("Self:")
+            print(self.results[task.value.benchmark])
             data_dict[task.value.col_name] = self.results[task.value.benchmark]
 
         return data_dict