Revise usage part

README.md

@@ -9,45 +9,19 @@ size_categories:
 - 100K<n<1M
 dataset_info:
   - config_name: ChAFF
-    splits:
-      - name: assays
-        data_files: data/*.csv
-        features:
-          - name: Type
-            dtype: string
-          - name: DatasetName
-            dtype: string
-          - name: AID
-            dtype: int64
-          - name: ID
-            dtype: string
-          - name: IDType
-            dtype: string
-          - name: SMILES
-            dtype: string
-      - name: summary
-        data_files: ChAFF_dataset_summary.csv
-        features:
-          - name: Type
-            dtype: string
-          - name: DatasetName
-            dtype: string
-          - name: AID
-            dtype: int64
-          - name: AID_confirmatory
-            dtype: int64
-          - name: NumActiveCompounds
-            dtype: int64
-          - name: Paper Title
-            dtype: string
-          - name: Reference
-            dtype: string
-          - name: URL
-            dtype: string
-          - name: Assay Name
-            dtype: string
-          - name: Description
-            dtype: string
+    features:
+      - name: Type
+        dtype: string
+      - name: DatasetName
+        dtype: string
+      - name: AID
+        dtype: int64
+      - name: ID
+        dtype: string
+      - name: IDType
+        dtype: string
+      - name: SMILES
+        dtype: string
 ---
 
 # ChAFF datasets
@@ -114,8 +88,12 @@ then, from within python load the datasets library.
 
 Now load the 'ChAFF' datasets together,
 
-    >>> assays = datasets.load_dataset("maomlab/ChAFF", split="assays")
+    >>> assays = datasets.load_dataset("maomlab/ChAFF", split="train")
+
 
 If you are interested in the summary file,
 
-    >>> summary = datasets.load_dataset("maomlab/ChAFF", split="summary")
+    >>> summary = datasets.load_dataset("json", data_files="https://huggingface.co/datasets/maomlab/ChAFF/resolve/main/summary.json", split="train"
+)
+
+The default split is "train", as we did not split the datasets.