alter default config

jsick.py

@@ -40,6 +40,11 @@ class JSICKDataset(ds.GeneratorBasedBuilder):
             version=VERSION,
             description="hoge",
         ),
+        ds.BuilderConfig(
+            name="original",
+            version=VERSION,
+            description="hoge",
+        ),
         ds.BuilderConfig(
             name="stress",
             version=VERSION,
@@ -50,6 +55,26 @@ class JSICKDataset(ds.GeneratorBasedBuilder):
     def _info(self) -> ds.DatasetInfo:
         labels = ds.ClassLabel(names=["entailment", "neutral", "contradiction"])
         if self.config.name == "base":
+            features = ds.Features(
+                {
+                    "id": ds.Value("int32"),
+                    "premise": ds.Value("string"),
+                    "hypothesis": ds.Value("string"),
+                    "label": labels,
+                    "score": ds.Value("float32"),
+                    "sentence_A_En": ds.Value("string"),
+                    "sentence_B_En": ds.Value("string"),
+                    "entailment_label_En": labels,
+                    "relatedness_score_En": ds.Value("float32"),
+                    "corr_entailment_labelAB_En": ds.Value("string"),
+                    "corr_entailment_labelBA_En": ds.Value("string"),
+                    "image_ID": ds.Value("string"),
+                    "original_caption": ds.Value("string"),
+                    "semtag_short": ds.Value("string"),
+                    "semtag_long": ds.Value("string"),
+                }
+            )
+        elif self.config.name == "original":
             features = ds.Features(
                 {
                     "pair_ID": ds.Value("int32"),
@@ -99,6 +124,17 @@ class JSICKDataset(ds.GeneratorBasedBuilder):
         df: pd.DataFrame = pd.read_table(data_path, sep="\t", header=0)
 
         if self.config.name == "base":
+            df = df.rename(
+                columns={
+                    "pair_ID": "id",
+                    "sentence_A_Ja": "premise",
+                    "sentence_B_Ja": "hypothesis",
+                    "entailment_label_Ja": "label",
+                    "relatedness_score_Ja": "score",
+                }
+            )
+
+        if self.config.name in ["base", "original"]:
             return [
                 ds.SplitGenerator(
                     name=ds.Split.TRAIN,