Upload dataset_loader with huggingface_hub

dataset_loader

@@ -1,18 +1,19 @@
-from datasets import DatasetInfo, GeneratorBasedBuilder, SplitGenerator, Split
+from datasets import DatasetInfo, GeneratorBasedBuilder, SplitGenerator, Split, Features, Value
 
-class RomanshGermanDataset(GeneratorBasedBuilder):
+class RomanshGerman(GeneratorBasedBuilder):
     def _info(self):
         return DatasetInfo(
-                features=Features({
-        "romansh_text": Value("string"),
-        "german_text": Value("string"),
-        "similarity": Value("float32"),
-        "original_similarity": Value("float32"),
-        })
-            supervised_keys=None,
             description="Best Matched Parallel Data",
+            features=Features({
+                "romansh_text": Value("string"),
+                "german_text": Value("string"),
+                "similarity": Value("float32"),
+                "original_similarity": Value("float32"),
+            }),
+            supervised_keys=None,
         )
 
+
     def _split_generators(self, dl_manager):
         data_path = dl_manager.download_and_extract("parallel_data/matched_0.60_0.005.jsonl")
         return [SplitGenerator(name=Split.TRAIN, gen_kwargs={"filepath": data_path})]
@@ -21,4 +22,4 @@ class RomanshGermanDataset(GeneratorBasedBuilder):
         import json
         with open(filepath, encoding="utf-8") as f:
             for idx, line in enumerate(f):
-                yield idx, json.loads(line)
+                yield idx, json.loads(line)