rename split to 'test'

app.py

@@ -194,7 +194,7 @@ col_types_rewardbench_avg = ["number"] + ["markdown"]+ ["str"] + ["number"] * (l
 ## col_types_prefs_sub = ["markdown"] + ["number"] * (len(prefs_data_sub.columns) - 1)
 
 # for showing random samples
-eval_set = load_dataset(eval_set_repo, use_auth_token=COLLAB_TOKEN, split="train")
+eval_set = load_dataset(eval_set_repo, use_auth_token=COLLAB_TOKEN, split="test")
 def random_sample(r: gr.Request, subset):
     if subset is None or subset == []:
         sample_index = np.random.randint(0, len(eval_set) - 1)