Fix URLs for data (again)
Browse files
unsplash25k_image_embeddings.py
CHANGED
|
@@ -34,8 +34,8 @@ _LICENSE = ""
|
|
| 34 |
# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
|
| 35 |
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
|
| 36 |
_URLS = {
|
| 37 |
-
"image_ids": "https://huggingface.co/datasets/JLD/unsplash25k-image-embeddings/main/data/image_ids.feather.zstd",
|
| 38 |
-
"embeddings": "https://huggingface.co/datasets/JLD/unsplash25k-image-embeddings/main/data/unsplash_embeddings.safetensors"
|
| 39 |
}
|
| 40 |
|
| 41 |
class Unsplash25kImageEmbeddingsDataset(datasets.GeneratorBasedBuilder):
|
|
@@ -70,7 +70,6 @@ class Unsplash25kImageEmbeddingsDataset(datasets.GeneratorBasedBuilder):
|
|
| 70 |
def _split_generators(self, dl_manager):
|
| 71 |
embeddings_path = dl_manager.download(_URLS["embeddings"])
|
| 72 |
image_ids_path = dl_manager.download(_URLS["image_ids"])
|
| 73 |
-
|
| 74 |
return [
|
| 75 |
datasets.SplitGenerator(
|
| 76 |
name=datasets.Split.ALL,
|
|
|
|
| 34 |
# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
|
| 35 |
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
|
| 36 |
_URLS = {
|
| 37 |
+
"image_ids": "https://huggingface.co/datasets/JLD/unsplash25k-image-embeddings/resolve/main/data/image_ids.feather.zstd",
|
| 38 |
+
"embeddings": "https://huggingface.co/datasets/JLD/unsplash25k-image-embeddings/resolve/main/data/unsplash_embeddings.safetensors"
|
| 39 |
}
|
| 40 |
|
| 41 |
class Unsplash25kImageEmbeddingsDataset(datasets.GeneratorBasedBuilder):
|
|
|
|
| 70 |
def _split_generators(self, dl_manager):
|
| 71 |
embeddings_path = dl_manager.download(_URLS["embeddings"])
|
| 72 |
image_ids_path = dl_manager.download(_URLS["image_ids"])
|
|
|
|
| 73 |
return [
|
| 74 |
datasets.SplitGenerator(
|
| 75 |
name=datasets.Split.ALL,
|