upload split

dataset.py

@@ -0,0 +1,40 @@
+import csv
+import datasets
+from datasets import GeneratorBasedBuilder, SplitGenerator, Split
+
+class ORIONDataset(GeneratorBasedBuilder):
+    def _info(self):
+        return datasets.DatasetInfo(
+            description="ORION: Open-Web Reasoning over Long-tail Entities",
+            features=datasets.Features({
+                "q_id": datasets.Value("string"),
+                "question": datasets.Value("string"),
+                "answer": datasets.Value("string"),
+                "url_list": datasets.Sequence(datasets.Value("string")),
+            }),
+            supervised_keys=None,
+        )
+
+    def _split_generators(self, dl_manager):
+        return [
+            SplitGenerator(
+                name=Split("zh"),
+                gen_kwargs={"filepath": "ORION_zh.csv"},
+            ),
+            SplitGenerator(
+                name=Split("en"),
+                gen_kwargs={"filepath": "ORION_en.csv"},
+            ),
+        ]
+
+    def _generate_examples(self, filepath):
+        with open(filepath, encoding="utf-8") as f:
+            reader = csv.DictReader(f)
+            for idx, row in enumerate(reader):
+                urls = [row[f"url_{i}"].strip() for i in range(1, 11) if row[f"url_{i}"].strip()]
+                yield idx, {
+                    "q_id": row["q_id"],
+                    "question": row["question"],
+                    "answer": row["answer"],
+                    "url_list": urls
+                }