Adds the file paths to the splits

this will correctly render the splits in a dataset viewer and improve the presentation of this dataset on huggingface . a parquet conversion might be initialized once the dataset is correctly described. thanks for publishing a favorite dataset of ours.

README.md

@@ -30,17 +30,18 @@ dataset_info:
   - name: text
     dtype: string
   splits:
-  - name: train
-    num_bytes: 11453688452
-    num_examples: 28602
-  - name: validation
-    num_bytes: 17402295
-    num_examples: 50
-  - name: test
-    num_bytes: 40482852
-    num_examples: 100
-  download_size: 11740397875
-  dataset_size: 11511573599
+    - name: train
+      num_bytes: 11453688452
+      num_examples: 28602
+      path: deepmind/pg19/data/train_files.txt
+    - name: validation
+      num_bytes: 17402295
+      num_examples: 50
+      path: deepmind/pg19/data/validation_files.txt
+    - name: test
+      num_bytes: 40482852
+      num_examples: 100
+      path: deepmind/pg19/data/test_files.txt
 ---
 
 # Dataset Card for "pg19"