Add dataset loading script and dataset_infos.json for MNLP RAG evaluation

dataset.py

@@ -1,12 +1,10 @@
 import json
 import datasets
-import os
 
 class MNLPDataset(datasets.GeneratorBasedBuilder):
     def _info(self):
         return datasets.DatasetInfo(
             features=datasets.Features({
-                "id": datasets.Value("string"),
                 "title": datasets.Value("string"),
                 "text": datasets.Value("string"),
                 "source": datasets.Value("string"),
@@ -15,13 +13,11 @@ class MNLPDataset(datasets.GeneratorBasedBuilder):
         )
 
     def _split_generators(self, dl_manager):
-        # download_and_extract gives directory path if it's just a .jsonl file
-        data_dir = dl_manager.download_and_extract("wiki_rag_subset_with_source.jsonl")
-        data_path = os.path.join(data_dir, "wiki_rag_subset_with_source.jsonl")
+        path = dl_manager.download_and_extract("wiki_rag_subset_with_source.jsonl")
         return [
             datasets.SplitGenerator(
                 name=datasets.Split.TRAIN,
-                gen_kwargs={"filepath": data_path},
+                gen_kwargs={"filepath": path},
             )
         ]
 
@@ -30,8 +26,7 @@ class MNLPDataset(datasets.GeneratorBasedBuilder):
             for i, line in enumerate(f):
                 data = json.loads(line)
                 yield i, {
-                    "id": data.get("id", str(i)),
                     "title": data.get("title", ""),
                     "text": data.get("text", ""),
-                    "source": data.get("source", ""),
+                    "source": data.get("source", "wikipedia"),
                 }

dataset_infos.json

@@ -2,16 +2,24 @@
   "default": {
     "description": "RAG documents for MNLP",
     "features": {
-      "id": {"dtype": "string"},
-      "title": {"dtype": "string"},
-      "text": {"dtype": "string"},
-      "source": {"dtype": "string"}
+      "title": {
+        "_type": "Value",
+        "dtype": "string"
+      },
+      "text": {
+        "_type": "Value",
+        "dtype": "string"
+      },
+      "source": {
+        "_type": "Value",
+        "dtype": "string"
+      }
     },
     "splits": {
       "train": {
         "name": "train",
-        "num_bytes": 1200000,
-        "num_examples": 5000
+        "num_bytes": 0,  
+        "num_examples": 5000  
       }
     }
   }