rename

dataset.py

@@ -0,0 +1,33 @@
+import json
+import datasets
+
+class MNLP_M2_rag_documents(datasets.GeneratorBasedBuilder):
+    def _info(self):
+        return datasets.DatasetInfo(
+            features=datasets.Features({
+                "title": datasets.Value("string"),
+                "text": datasets.Value("string"),
+                "source": datasets.Value("string"),
+            }),
+            description="RAG documents for MNLP",
+        )
+
+    def _split_generators(self, dl_manager):
+        data_path = dl_manager.download_and_extract("wiki_rag_subset.jsonl")
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={"filepath": data_path},
+            )
+        ]
+
+
+    def _generate_examples(self, filepath):
+        with open(filepath, encoding="utf-8") as f:
+            for i, line in enumerate(f):
+                data = json.loads(line)
+                yield i, {
+                    "title": data.get("title", ""),
+                    "text": data.get("text", ""),
+                    "source": data.get("source", "wikipedia"),
+                }