DEV: Add a load script

This load script should separate the partitions according to the "split" column

dataset.py

@@ -0,0 +1,53 @@
+import csv
+import datasets
+
+class DNABarcodeDataset(datasets.GeneratorBasedBuilder):
+    def _info(self):
+        return datasets.DatasetInfo(
+            description="DNA barcode dataset with hierarchical taxonomy labels and multiple splits.",
+            features=datasets.Features({
+                "processid": datasets.Value("string"),
+                "sampleid": datasets.Value("string"),
+                "dna_bin": datasets.Value("string"),
+                "phylum": datasets.Value("string"),
+                "class": datasets.Value("string"),
+                "order": datasets.Value("string"),
+                "family": datasets.Value("string"),
+                "genus": datasets.Value("string"),
+                "species": datasets.Value("string"),        # label
+                "dna_barcode": datasets.Value("string"),    # input data
+                "split": datasets.ClassLabel(names=["train", "val", "test", "test_unseen", "pretrain"]),
+            }),
+            supervised_keys=("dna_barcode", "species"),  # For model training
+        )
+
+    def _split_generators(self, dl_manager):
+        data_path = dl_manager.download("CanInv_metadata.csv")  # Use a URL or relative path
+        return [
+            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": data_path, "split": "train"}),
+            datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": data_path, "split": "val"}),
+            datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": data_path, "split": "test"}),
+            datasets.SplitGenerator(name="test_unseen", gen_kwargs={"filepath": data_path, "split": "test_unseen"}),
+            datasets.SplitGenerator(name="pretrain", gen_kwargs={"filepath": data_path, "split": "pretrain"}),
+        ]
+
+    def _generate_examples(self, filepath, split):
+        with open(filepath, encoding="utf-8") as f:
+            reader = csv.DictReader(f)
+            idx = 0
+            for row in reader:
+                if row["split"] == split:
+                    yield idx, {
+                        "processid": row["processid"],
+                        "sampleid": row["sampleid"],
+                        "dna_bin": row["dna_bin"],
+                        "phylum": row["phylum"],
+                        "class": row["class"],
+                        "order": row["order"],
+                        "family": row["family"],
+                        "genus": row["genus"],
+                        "species": row["species"],
+                        "dna_barcode": row["dna_barcode"],
+                        "split": row["split"],
+                    }
+                    idx += 1