Upload 3 files

.gitattributes

@@ -53,3 +53,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.jpg filter=lfs diff=lfs merge=lfs -text
 *.jpeg filter=lfs diff=lfs merge=lfs -text
 *.webp filter=lfs diff=lfs merge=lfs -text
+train.jsonl filter=lfs diff=lfs merge=lfs -text

cn_data.py

@@ -0,0 +1,101 @@
+import pandas as pd
+from huggingface_hub import hf_hub_url
+import datasets
+import os
+
+_VERSION = datasets.Version("0.0.2")
+
+_DESCRIPTION = "TODO"
+_HOMEPAGE = "TODO"
+_LICENSE = "TODO"
+_CITATION = "TODO"
+
+_FEATURES = datasets.Features(
+    {
+        "target": datasets.Image(),
+        "source": datasets.Image(),
+        "caption": datasets.Value("string"),
+    },
+)
+
+METADATA_URL = hf_hub_url(
+    "kmaksatk/cn_data",
+    filename="train.jsonl",
+    repo_type="dataset",
+)
+
+IMAGES_URL = hf_hub_url(
+    "kmaksatk/cn_data",
+    filename="images.zip",
+    repo_type="dataset",
+)
+
+CONDITIONING_IMAGES_URL = hf_hub_url(
+    "kmaksatk/cn_data",
+    filename="conditioning_images.zip",
+    repo_type="dataset",
+)
+
+_DEFAULT_CONFIG = datasets.BuilderConfig(name="default", version=_VERSION)
+
+
+class Fill50k(datasets.GeneratorBasedBuilder):
+    BUILDER_CONFIGS = [_DEFAULT_CONFIG]
+    DEFAULT_CONFIG_NAME = "default"
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=_FEATURES,
+            supervised_keys=None,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+            citation=_CITATION,
+        )
+
+    def _split_generators(self, dl_manager):
+        metadata_path = dl_manager.download(METADATA_URL)
+        images_dir = dl_manager.download_and_extract(IMAGES_URL)
+        conditioning_images_dir = dl_manager.download_and_extract(
+            CONDITIONING_IMAGES_URL
+        )
+
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                # These kwargs will be passed to _generate_examples
+                gen_kwargs={
+                    "metadata_path": metadata_path,
+                    "images_dir": images_dir,
+                    "conditioning_images_dir": conditioning_images_dir,
+                },
+            ),
+        ]
+
+    def _generate_examples(self, metadata_path, images_dir, conditioning_images_dir):
+        metadata = pd.read_json(metadata_path, lines=True)
+
+        for _, row in metadata.iterrows():
+            text = row["caption"]
+
+            image_path = row["target"]
+            image_path = os.path.join(images_dir, image_path)
+            image = open(image_path, "rb").read()
+
+            conditioning_image_path = row["source"]
+            conditioning_image_path = os.path.join(
+                conditioning_images_dir, row["source"]
+            )
+            conditioning_image = open(conditioning_image_path, "rb").read()
+
+            yield row["target"], {
+                "text": text,
+                "image": {
+                    "path": image_path,
+                    "bytes": image,
+                },
+                "conditioning_image": {
+                    "path": conditioning_image_path,
+                    "bytes": conditioning_image,
+                },
+            }

conditioning_images.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ce71537f3da1f8a62174332eb3d835b2b72a180deb30627b8935cc2f18675dc
+size 1311971989

train.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e55b260b273752a520ee7d35978cacb3a4ea0e2d29d2303e960af83807c9ab0f
+size 20942737