custom loading script for preview

dataset.yaml

@@ -6,4 +6,5 @@ configs:
     - split: train
       path: "train.parquet"
     - split: test
-      path: "test.parquet"
+      path: "test.parquet"
+    name: methyl-classification

methyl-classification.py

@@ -0,0 +1,18 @@
+from datasets import DatasetDict, Dataset, load_dataset, GeneratorBasedBuilder
+import pandas as pd
+
+class MethylClassification(GeneratorBasedBuilder):
+    def _info(self):
+        return self._info_from_pandas(pd.read_parquet("preview.parquet"))
+
+    def _split_generators(self, dl_manager):
+        data_files = self.config.data_files
+        return [
+            self._generate_split(name, data_files[name])
+            for name in data_files
+        ]
+
+    def _generate_examples(self, split_name, filepath):
+        df = pd.read_parquet(filepath)
+        for i, row in df.iterrows():
+            yield i, row.to_dict()