add feature mapping

README.md

@@ -13,6 +13,21 @@ task_categories:
   - sentence-similarity
 dataset_info:
   config_name: default
+  features:
+  - name: query
+    dtype: text
+  - name: pos
+    sequence:
+    - name: doc
+      dtype: text
+    - name: score
+      dtype: float
+  - name: neg
+    sequence:
+    - name: doc
+      dtype: text
+    - name: score
+      dtype: float
   splits:
     - name: train
       num_bytes: 1