Use repo_type instead of deprecated datasets repo IDs (#19202)

* Use repo_type instead of deprecated datasets repo IDs

* Add missing one in doc

src/transformers/models/segformer/convert_segformer_original_to_pytorch.py

@@ -128,7 +128,7 @@ def convert_segformer_checkpoint(model_name, checkpoint_path, pytorch_dump_folde
     encoder_only = False
 
     # set attributes based on model_name
-    repo_id = "datasets/huggingface/label-files"
+    repo_id = "huggingface/label-files"
     if "segformer" in model_name:
         size = model_name[len("segformer.") : len("segformer.") + 2]
         if "ade" in model_name:
@@ -151,7 +151,7 @@ def convert_segformer_checkpoint(model_name, checkpoint_path, pytorch_dump_folde
         raise ValueError(f"Model {model_name} not supported")
 
     # set config attributes
-    id2label = json.load(open(hf_hub_download(repo_id, filename), "r"))
+    id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
     id2label = {int(k): v for k, v in id2label.items()}
     config.id2label = id2label
     config.label2id = {v: k for k, v in id2label.items()}

src/transformers/models/swin/convert_swin_timm_to_pytorch.py

@@ -39,9 +39,9 @@ def get_swin_config(swin_name):
         num_classes = 21841
     else:
         num_classes = 1000
-        repo_id = "datasets/huggingface/label-files"
+        repo_id = "huggingface/label-files"
         filename = "imagenet-1k-id2label.json"
-        id2label = json.load(open(hf_hub_download(repo_id, filename), "r"))
+        id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
         id2label = {int(k): v for k, v in id2label.items()}
         config.id2label = id2label
         config.label2id = {v: k for k, v in id2label.items()}

src/transformers/models/swinv2/convert_swinv2_timm_to_pytorch.py

@@ -63,18 +63,18 @@ def get_swinv2_config(swinv2_name):
 
     if ("22k" in swinv2_name) and ("to" not in swinv2_name):
         num_classes = 21841
-        repo_id = "datasets/huggingface/label-files"
+        repo_id = "huggingface/label-files"
         filename = "imagenet-22k-id2label.json"
-        id2label = json.load(open(hf_hub_download(repo_id, filename), "r"))
+        id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
         id2label = {int(k): v for k, v in id2label.items()}
         config.id2label = id2label
         config.label2id = {v: k for k, v in id2label.items()}
 
     else:
         num_classes = 1000
-        repo_id = "datasets/huggingface/label-files"
+        repo_id = "huggingface/label-files"
         filename = "imagenet-1k-id2label.json"
-        id2label = json.load(open(hf_hub_download(repo_id, filename), "r"))
+        id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
         id2label = {int(k): v for k, v in id2label.items()}
         config.id2label = id2label
         config.label2id = {v: k for k, v in id2label.items()}

src/transformers/models/van/convert_van_to_pytorch.py

@@ -168,9 +168,9 @@ def convert_weights_and_push(save_directory: Path, model_name: str = None, push_
     filename = "imagenet-1k-id2label.json"
     num_labels = 1000
 
-    repo_id = "datasets/huggingface/label-files"
+    repo_id = "huggingface/label-files"
     num_labels = num_labels
-    id2label = json.load(open(hf_hub_download(repo_id, filename), "r"))
+    id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
     id2label = {int(k): v for k, v in id2label.items()}
 
     id2label = id2label

src/transformers/models/videomae/convert_videomae_to_pytorch.py

@@ -47,7 +47,7 @@ def get_videomae_config(model_name):
         config.use_mean_pooling = False
 
     if "finetuned" in model_name:
-        repo_id = "datasets/huggingface/label-files"
+        repo_id = "huggingface/label-files"
         if "kinetics" in model_name:
             config.num_labels = 400
             filename = "kinetics400-id2label.json"
@@ -56,7 +56,7 @@ def get_videomae_config(model_name):
             filename = "something-something-v2-id2label.json"
         else:
             raise ValueError("Model name should either contain 'kinetics' or 'ssv2' in case it's fine-tuned.")
-        id2label = json.load(open(hf_hub_download(repo_id, filename), "r"))
+        id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
         id2label = {int(k): v for k, v in id2label.items()}
         config.id2label = id2label
         config.label2id = {v: k for k, v in id2label.items()}
@@ -145,7 +145,9 @@ def convert_state_dict(orig_state_dict, config):
 # We will verify our results on a video of eating spaghetti
 # Frame indices used: [164 168 172 176 181 185 189 193 198 202 206 210 215 219 223 227]
 def prepare_video():
-    file = hf_hub_download(repo_id="datasets/hf-internal-testing/spaghetti-video", filename="eating_spaghetti.npy")
+    file = hf_hub_download(
+        repo_id="hf-internal-testing/spaghetti-video", filename="eating_spaghetti.npy", repo_type="dataset"
+    )
     video = np.load(file)
     return list(video)
 

src/transformers/models/vilt/convert_vilt_original_to_pytorch.py

@@ -180,9 +180,9 @@ def convert_vilt_checkpoint(checkpoint_url, pytorch_dump_folder_path):
     if "vqa" in checkpoint_url:
         vqa_model = True
         config.num_labels = 3129
-        repo_id = "datasets/huggingface/label-files"
+        repo_id = "huggingface/label-files"
         filename = "vqa2-id2label.json"
-        id2label = json.load(open(hf_hub_download(repo_id, filename), "r"))
+        id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
         id2label = {int(k): v for k, v in id2label.items()}
         config.id2label = id2label
         config.label2id = {v: k for k, v in id2label.items()}

src/transformers/models/vit/convert_dino_to_pytorch.py

@@ -142,9 +142,9 @@ def convert_vit_checkpoint(model_name, pytorch_dump_folder_path, base_model=True
     # set labels if required
     if not base_model:
         config.num_labels = 1000
-        repo_id = "datasets/huggingface/label-files"
+        repo_id = "huggingface/label-files"
         filename = "imagenet-1k-id2label.json"
-        id2label = json.load(open(hf_hub_download(repo_id, filename), "r"))
+        id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
         id2label = {int(k): v for k, v in id2label.items()}
         config.id2label = id2label
         config.label2id = {v: k for k, v in id2label.items()}

src/transformers/models/vit/convert_vit_timm_to_pytorch.py

@@ -147,9 +147,9 @@ def convert_vit_checkpoint(vit_name, pytorch_dump_folder_path):
         config.image_size = int(vit_name[-9:-6])
     else:
         config.num_labels = 1000
-        repo_id = "datasets/huggingface/label-files"
+        repo_id = "huggingface/label-files"
         filename = "imagenet-1k-id2label.json"
-        id2label = json.load(open(hf_hub_download(repo_id, filename), "r"))
+        id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
         id2label = {int(k): v for k, v in id2label.items()}
         config.id2label = id2label
         config.label2id = {v: k for k, v in id2label.items()}

src/transformers/models/x_clip/convert_x_clip_original_pytorch_to_hf.py

@@ -207,8 +207,9 @@ def prepare_video(num_frames):
     elif num_frames == 32:
         filename = "eating_spaghetti_32_frames.npy"
     file = hf_hub_download(
-        repo_id="datasets/hf-internal-testing/spaghetti-video",
+        repo_id="hf-internal-testing/spaghetti-video",
         filename=filename,
+        repo_type="dataset",
     )
     video = np.load(file)
     return list(video)

src/transformers/models/yolos/convert_yolos_to_pytorch.py

@@ -57,9 +57,9 @@ def get_yolos_config(yolos_name):
         config.image_size = [800, 1344]
 
     config.num_labels = 91
-    repo_id = "datasets/huggingface/label-files"
+    repo_id = "huggingface/label-files"
     filename = "coco-detection-id2label.json"
-    id2label = json.load(open(hf_hub_download(repo_id, filename), "r"))
+    id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
     id2label = {int(k): v for k, v in id2label.items()}
     config.id2label = id2label
     config.label2id = {v: k for k, v in id2label.items()}

tests/models/videomae/test_modeling_videomae.py

@@ -342,7 +342,9 @@ class VideoMAEModelTest(ModelTesterMixin, unittest.TestCase):
 # We will verify our results on a video of eating spaghetti
 # Frame indices used: [164 168 172 176 181 185 189 193 198 202 206 210 215 219 223 227]
 def prepare_video():
-    file = hf_hub_download(repo_id="datasets/hf-internal-testing/spaghetti-video", filename="eating_spaghetti.npy")
+    file = hf_hub_download(
+        repo_id="hf-internal-testing/spaghetti-video", filename="eating_spaghetti.npy", repo_type="dataset"
+    )
     video = np.load(file)
     return list(video)
 

tests/models/x_clip/test_modeling_x_clip.py

@@ -633,7 +633,7 @@ class XCLIPModelTest(ModelTesterMixin, unittest.TestCase):
 # We will verify our results on a spaghetti video
 def prepare_video():
     file = hf_hub_download(
-        repo_id="datasets/hf-internal-testing/spaghetti-video", filename="eating_spaghetti_8_frames.npy"
+        repo_id="hf-internal-testing/spaghetti-video", filename="eating_spaghetti_8_frames.npy", repo_type="dataset"
     )
     video = np.load(file)
     return list(video)