Update sd-nlp-non-tokenized.py

sd-nlp-non-tokenized.py

@@ -196,22 +196,25 @@ class SourceDataNLP(datasets.GeneratorBasedBuilder):
                     yield id_, {
                         "words": data["words"],
                         "labels": labels,
+                        "tag_mask": tag_mask
                     }
                 elif self.config.name == "GENEPROD_ROLES":
-                    labels = data["label_ids"]["entity_types"]
+                    labels = data["label_ids"]["geneprod_roles"]
                     geneprod = ["B-GENEPROD", "I-GENEPROD", "B-PROTEIN", "I-PROTEIN", "B-GENE", "I-GENE"]
                     tag_mask = [1 if t in geneprod else 0 for t in labels]
                     yield id_, {
                         "words": data["words"],
-                        "labels": data["label_ids"]["geneprod_roles"],
+                        "labels": labels,
+                        "tag_mask": tag_mask,
                     }
                 elif self.config.name == "SMALL_MOL_ROLES":
-                    labels = data["words"]["entity_types"]
+                    labels = data["words"]["small_mol_roles"]
                     small_mol = ["B-SMALL_MOLECULE", "I-SMALL_MOLECULE"]
                     tag_mask = [1 if t in small_mol else 0 for t in labels]
                     yield id_, {
                         "words": data["words"],
-                        "labels": data["label_ids"]["small_mol_roles"],
+                        "labels": labels,
+                        "tag_mask": tag_mask,
                     }
                 elif self.config.name == "BORING":
                     yield id_, {"words": data["words"], "labels": data["label_ids"]["boring"]}
@@ -221,4 +224,5 @@ class SourceDataNLP(datasets.GeneratorBasedBuilder):
                     yield id_, {
                         "input_ids": data["words"],
                         "labels": data["label_ids"]["panel_start"],
+                        "tag_mask": tag_mask,
                     }