price benchmark

clean_selection/analyze_clean_hierarchical_loss_protein_vec.ipynb

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:462ddc0571c5bb0200078382f301c55a81ff68ba40a0ac71187344ea287f2fb2
-size 66885
+oid sha256:17792a81e3b66606738ca77aa001077daa7f5c0063b3f956902996871561817c
+size 86942

clean_selection/clean_utils.py

@@ -27,6 +27,10 @@ def get_true_labels_test(file_name, test_idx: None):
             for ec in true_ec_lst:
                 if test_idx is not None and count - 1 in test_idx:
                     all_label.add(ec)
+                elif test_idx is None: ## add all EC labels
+                    all_label.add(ec)
+                else:
+                    continue
         if header:
             header = False
     true_label = [true_label_dict[i] for i in true_label_dict.keys()]