update

src/display/css_html_js.py

@@ -29,6 +29,10 @@ custom_css = """
     margin-top: 15px
 }
 
+#leaderboard-table table td {
+    text-align: center;
+}
+
 #leaderboard-table-lite {
     margin-top: 15px
 }

src/display/utils.py

@@ -24,27 +24,34 @@ class Tasks(Enum):
     # gsm8k = Task("gsm8k", "acc", "GSM8K")
     # drop = Task("drop", "f1", "DROP")
 
-    nqopen = Task("nq_open", "em", "NQ Open")
-    triviaqa = Task("triviaqa", "em", "TriviaQA")
+    nqopen = Task("nq8", "em", "NQ Open/EM")
+    triviaqa = Task("tqa8", "em", "TriviaQA/EM")
 
-    truthfulqa_mc1 = Task("truthfulqa_mc1", "acc", "TruthfulQA MC1")
-    truthfulqa_mc2 = Task("truthfulqa_mc2", "acc", "TruthfulQA MC2")
+    truthfulqa_mc1 = Task("truthfulqa_mc1", "acc", "TruthQA MC1/Acc")
+    truthfulqa_mc2 = Task("truthfulqa_mc2", "acc", "TruthQA MC2/Acc")
+    truthfulqa_gen = Task("truthfulqa_gen", "rougeL_acc", "TruthQA Gen/ROUGE")
 
-    halueval_qa = Task("halueval_qa", "acc", "HaluEval QA")
-    halueval_summ = Task("halueval_summarization", "acc", "HaluEval Summ")
-    halueval_dial = Task("halueval_dialogue", "acc", "HaluEval Dial")
+    xsum_r = Task("xsum_v2", "rougeL", "XSum/ROUGE")
+    xsum_f = Task("xsum_v2", "factKB", "XSum/factKB")
+    xsum_b = Task("xsum_v2", "bertscore_precision", "XSum/BERT-P")
 
-    xsum = Task("xsum", "factKB", "XSum")
-    cnndm = Task("cnndm", "factKB", "CNN/DM")
+    cnndm_r = Task("cnndm_v2", "rougeL", "CNN-DM/ROUGE")
+    cnndm_f = Task("cnndm_v2", "factKB", "CNN-DM/factKB")
+    cnndm_b = Task("cnndm_v2", "bertscore_precision", "CNN-DM/BERT-P")
 
-    memotrap = Task("memo-trap", "acc", "MemoTrap")
+    race = Task("race", "acc", "RACE/Acc")
+    squadv2 = Task("squadv2", "exact", "SQUaDv2/EM")
 
-    ifeval = Task("ifeval", "prompt_level_strict_acc", "IFEval")
+    memotrap = Task("memo-trap_v2", "acc", "MemoTrap/Acc")
+    ifeval = Task("ifeval", "prompt_level_strict_acc", "IFEval/Acc")
 
-    selfcheck = Task("selfcheckgpt", "max-selfcheckgpt", "SelfCheckGPT")
+    faithdial = Task("faithdial_hallu_v2", "acc", "FaithDial/Acc")
+
+    halueval_qa = Task("halueval_qa", "acc", "HaluQA/Acc")
+    halueval_summ = Task("halueval_summarization", "acc", "HaluSumm/Acc")
+    halueval_dial = Task("halueval_dialogue", "acc", "HaluDial/Acc")
 
-    #truthfulqa_mc1 = Task("truthfulqa_mc1", "acc", "TruthfulQA MC1")
-    #truthfulqa_mc2 = Task("truthfulqa_mc2", "acc", "TruthfulQA MC2")
+    selfcheck = Task("selfcheckgpt", "max-selfcheckgpt", "SelfCheckGPT")
 
 # These classes are for user facing column names,
 # to avoid having to change them all around the code

src/leaderboard/filter_models.py

@@ -20,7 +20,6 @@ DO_NOT_SUBMIT_MODELS = [
     "Voicelab/trurl-2-13b",  # trained on MMLU
 ]
 
-
 def flag_models(leaderboard_data: list[dict]):
     for model_data in leaderboard_data:
         if model_data["model_name_for_query"] in FLAGGED_MODELS: