Update dataset_viewer.py

dataset_viewer.py

@@ -36,24 +36,6 @@ class SyntheticImageCaptionPairDataset(datasets.GeneratorBasedBuilder):
             citation=_CITATION,
         )
 
-    # def _split_generators(self, dl_manager):
-    #     images_archive = dl_manager.download(f"{_REPO}/images.tgz")
-    #     image_iters = dl_manager.iter_archive(images_archive)
-    #     # Download and load the JSON file containing descriptions
-    #     descriptions_file = dl_manager.download(f"{_REPO}/sampled_blip_captions.json")
-    #     with open(descriptions_file, "r") as f:
-    #         descriptions = json.load(f)
-    #
-    #     return [
-    #         datasets.SplitGenerator(
-    #             name=datasets.Split.TRAIN,
-    #             gen_kwargs={
-    #                 "images": image_iters,
-    #                 "descriptions": descriptions
-    #             }
-    #         ),
-    #     ]
-
     def _generate_examples(self, **kwargs):
         """This function returns the examples in the raw (text) form."""
         images_archive = kwargs['dl_manager'].download(f"{_REPO}/images.tgz")
@@ -67,8 +49,8 @@ class SyntheticImageCaptionPairDataset(datasets.GeneratorBasedBuilder):
             image_id = os.path.splitext(os.path.basename(filepath))[0]
 
             # Get caption corresponding to image ID
-            description = descriptions.get(image_id, "No description available")
+            caption = descriptions.get(image_id, "No description available")
             yield idx, {
-                "image": {"path": filepath, "bytes": image.read()},
+                "image_id": image_id,
                 "blip-2 caption": descriptions[image_id],
             }