Keep summary/structure/splits in README

Preserve Dataset Summary, Dataset Structure, and Data Splits while removing placeholder sections.

README.md

@@ -77,9 +77,9 @@ dataset_info:
 
 ### Dataset Summary
 
-BEIR is a heterogeneous benchmark that has been built from 18 diverse datasets representing 9 information retrieval tasks.
+BEIR is a heterogeneous benchmark built from 18 diverse datasets representing 9 information retrieval tasks.
 
-All these datasets have been preprocessed and can be used for experiments.
+This `arguana` subset is the Argument Retrieval task within BEIR.
 
 ### Supported Tasks and Leaderboards
 
@@ -93,10 +93,23 @@ All tasks are in English (`en`).
 
 ## Dataset Structure
 
-This dataset includes `corpus` and `queries` in Parquet format.
+This dataset uses the standard BEIR retrieval layout and includes:
 
-- `corpus`: records with `_id`, `title`, and `text`.
-- `queries`: records with `_id`, `title`, and `text`.
+- `corpus`: one row per document with `_id`, `title`, `text`
+- `queries`: one row per query with `_id`, `title`, `text`
+
+### Data Fields
+
+- `_id` (`string`): unique identifier
+- `title` (`string`): title (empty string when unavailable)
+- `text` (`string`): document/query text
+
+### Data Splits
+
+| Subset | Split | Rows |
+| --- | --- | ---: |
+| corpus | corpus | 8,674 |
+| queries | queries | 1,406 |
 
 ## Citation Information