Added processing to remove rows with empty text.

newsgroups.py

@@ -120,6 +120,9 @@ class Newsgroups(datasets.GeneratorBasedBuilder):
         train_df["text"] = train_df["text"].str.replace("\s+", " ", regex=True)
         test_df["text"] = test_df["text"].str.replace("\s+", " ", regex=True)
         
+        train_df = train_df[train_df["text"].str.strip()!=""]
+        test_df = test_df[test_df["text"].str.strip()!=""]
+
         train_df, val_df = train_test_split(train_df, test_size=0.2, random_state=42, stratify=train_df["labels"])
         train_df = train_df.reset_index(drop=True)
         val_df = val_df.reset_index(drop=True)