Update mmod2.py

mmod2.py

@@ -47,35 +47,47 @@ class Mmod2(datasets.GeneratorBasedBuilder):
         # task_templates=[datasets.ImageClassification(image_column="image", label_column="label")],
     )
 
-  def _split_generators(self, dl_manager):
-    # archive_path = dl_manager.download(_BASE_URL)
-    print("0000000")
-    split_metadata_paths = dl_manager.download(_METADATA_URLS)
-    # print(f"{split_metadata_paths = }")
-    self.data_dir = dl_manager.download_and_extract(_BASE_URL)
-    with open(split_metadata_paths[0], encoding="utf-8") as f:
-      train_files = set(f.read().split("\n"))
+  # def _split_generators(self, dl_manager):
+  #   # archive_path = dl_manager.download(_BASE_URL)
+  #   print("0000000")
+  #   split_metadata_paths = dl_manager.download(_METADATA_URLS)
+  #   # print(f"{split_metadata_paths = }")
+  #   self.data_dir = dl_manager.download_and_extract(_BASE_URL)
+  #   with open(split_metadata_paths[0], encoding="utf-8") as f:
+  #     train_files = set(f.read().split("\n"))
+
+  #   with open(split_metadata_paths[1], encoding="utf-8") as f:
+  #     validation_files = set(f.read().split("\n"))
+
+  #   return [
+  #     datasets.SplitGenerator(
+  #       name=datasets.Split.TRAIN,
+  #       gen_kwargs={
+  #         "split_key": "train",
+  #         "images": train_files,
+  #         "metadata_path": split_metadata_paths[0],
+  #       },
+  #     ),
+  #     datasets.SplitGenerator(
+  #       name=datasets.Split.VALIDATION,
+  #       gen_kwargs={
+  #         "split_key": "validation",
+  #         "images": validation_files,
+  #         "metadata_path": split_metadata_paths[1],
+  #       },
+  #     ),
+  #   ]
 
-    with open(split_metadata_paths[1], encoding="utf-8") as f:
-      validation_files = set(f.read().split("\n"))
+  def _split_generators(self, dl_manager):
+    archive = dl_manager.download(_DATA_URL)
 
     return [
-      datasets.SplitGenerator(
-        name=datasets.Split.TRAIN,
-        gen_kwargs={
-          "split_key": "train",
-          "images": train_files,
-          "metadata_path": split_metadata_paths[0],
-        },
-      ),
-      datasets.SplitGenerator(
-        name=datasets.Split.VALIDATION,
-        gen_kwargs={
-          "split_key": "validation",
-          "images": validation_files,
-          "metadata_path": split_metadata_paths[1],
-        },
-      ),
+        datasets.SplitGenerator(
+            name=datasets.Split.TRAIN, gen_kwargs={"images": dl_manager.iter_archive(archive), "split": "train"}
+        ),
+        datasets.SplitGenerator(
+            name=datasets.Split.VALIDATION, gen_kwargs={"images": dl_manager.iter_archive(archive), "split": "validation"}
+        ),
     ]
 
   def _generate_examples(self, images, metadata_path=None, split_key="train"):