Harmonize HF environment variables + deprecate use_auth_token (#6066)

* Harmonize HF environment variables + deprecate use_auth_token

* fix import

* fix

Author: Wauplin

docs/source/en/using-diffusers/push_to_hub.md

@@ -174,10 +174,4 @@ Set `private=True` in the [`~diffusers.utils.PushToHubMixin.push_to_hub`] functi
 controlnet.push_to_hub("my-controlnet-model-private", private=True)
 ```
 
-Private repositories are only visible to you, and other users won't be able to clone the repository and your repository won't appear in search results. Even if a user has the URL to your private repository, they'll receive a `404 - Sorry, we can't find the page you are looking for.`
-
-To load a model, scheduler, or pipeline from private or gated repositories, set `use_auth_token=True`:
-
-```py
-model = ControlNetModel.from_pretrained("your-namespace/my-controlnet-model-private", use_auth_token=True)
-```
+Private repositories are only visible to you, and other users won't be able to clone the repository and your repository won't appear in search results. Even if a user has the URL to your private repository, they'll receive a `404 - Sorry, we can't find the page you are looking for`. You must be [logged in](https://huggingface.co/docs/huggingface_hub/quick-start#login) to load a model from a private repository.

examples/community/README.md

@@ -512,7 +512,6 @@ device = torch.device('cpu' if not has_cuda else 'cuda')
 pipe = DiffusionPipeline.from_pretrained(
     "CompVis/stable-diffusion-v1-4",
         safety_checker=None,
-    use_auth_token=True,
     custom_pipeline="imagic_stable_diffusion",
     scheduler = DDIMScheduler(beta_start=0.00085, beta_end=0.012, beta_schedule="scaled_linear", clip_sample=False, set_alpha_to_one=False)
 ).to(device)
@@ -552,7 +551,6 @@ device = th.device('cpu' if not has_cuda else 'cuda')
 
 pipe = DiffusionPipeline.from_pretrained(
     "CompVis/stable-diffusion-v1-4",
-    use_auth_token=True,
     custom_pipeline="seed_resize_stable_diffusion"
 ).to(device)
 
@@ -588,7 +586,6 @@ generator = th.Generator("cuda").manual_seed(0)
 
 pipe = DiffusionPipeline.from_pretrained(
     "CompVis/stable-diffusion-v1-4",
-    use_auth_token=True,
     custom_pipeline="/home/mark/open_source/diffusers/examples/community/"
 ).to(device)
 
@@ -607,7 +604,6 @@ image.save('./seed_resize/seed_resize_{w}_{h}_image.png'.format(w=width, h=heigh
 
 pipe_compare = DiffusionPipeline.from_pretrained(
     "CompVis/stable-diffusion-v1-4",
-    use_auth_token=True,
     custom_pipeline="/home/mark/open_source/diffusers/examples/community/"
 ).to(device)
 

examples/community/checkpoint_merger.py

@@ -5,10 +5,11 @@
 import safetensors.torch
 import torch
 from huggingface_hub import snapshot_download
+from huggingface_hub.utils import validate_hf_hub_args
 
 from diffusers import DiffusionPipeline, __version__
 from diffusers.schedulers.scheduling_utils import SCHEDULER_CONFIG_NAME
-from diffusers.utils import CONFIG_NAME, DIFFUSERS_CACHE, ONNX_WEIGHTS_NAME, WEIGHTS_NAME
+from diffusers.utils import CONFIG_NAME, ONNX_WEIGHTS_NAME, WEIGHTS_NAME
 
 
 class CheckpointMergerPipeline(DiffusionPipeline):
@@ -57,6 +58,7 @@ def _remove_meta_keys(self, config_dict: Dict):
         return (temp_dict, meta_keys)
 
     @torch.no_grad()
+    @validate_hf_hub_args
     def merge(self, pretrained_model_name_or_path_list: List[Union[str, os.PathLike]], **kwargs):
         """
         Returns a new pipeline object of the class 'DiffusionPipeline' with the merged checkpoints(weights) of the models passed
@@ -69,7 +71,7 @@ def merge(self, pretrained_model_name_or_path_list: List[Union[str, os.PathLike]
             **kwargs:
                 Supports all the default DiffusionPipeline.get_config_dict kwargs viz..
 
-                cache_dir, resume_download, force_download, proxies, local_files_only, use_auth_token, revision, torch_dtype, device_map.
+                cache_dir, resume_download, force_download, proxies, local_files_only, token, revision, torch_dtype, device_map.
 
                 alpha - The interpolation parameter. Ranges from 0 to 1.  It affects the ratio in which the checkpoints are merged. A 0.8 alpha
                     would mean that the first model checkpoints would affect the final result far less than an alpha of 0.2
@@ -81,12 +83,12 @@ def merge(self, pretrained_model_name_or_path_list: List[Union[str, os.PathLike]
 
         """
         # Default kwargs from DiffusionPipeline
-        cache_dir = kwargs.pop("cache_dir", DIFFUSERS_CACHE)
+        cache_dir = kwargs.pop("cache_dir", None)
         resume_download = kwargs.pop("resume_download", False)
         force_download = kwargs.pop("force_download", False)
         proxies = kwargs.pop("proxies", None)
         local_files_only = kwargs.pop("local_files_only", False)
-        use_auth_token = kwargs.pop("use_auth_token", None)
+        token = kwargs.pop("token", None)
         revision = kwargs.pop("revision", None)
         torch_dtype = kwargs.pop("torch_dtype", None)
         device_map = kwargs.pop("device_map", None)
@@ -123,7 +125,7 @@ def merge(self, pretrained_model_name_or_path_list: List[Union[str, os.PathLike]
                 force_download=force_download,
                 proxies=proxies,
                 local_files_only=local_files_only,
-                use_auth_token=use_auth_token,
+                token=token,
                 revision=revision,
             )
             config_dicts.append(config_dict)
@@ -159,7 +161,7 @@ def merge(self, pretrained_model_name_or_path_list: List[Union[str, os.PathLike]
                     resume_download=resume_download,
                     proxies=proxies,
                     local_files_only=local_files_only,
-                    use_auth_token=use_auth_token,
+                    token=token,
                     revision=revision,
                     allow_patterns=allow_patterns,
                     user_agent=user_agent,

examples/community/stable_diffusion_tensorrt_img2img.py

@@ -28,6 +28,7 @@
 import tensorrt as trt
 import torch
 from huggingface_hub import snapshot_download
+from huggingface_hub.utils import validate_hf_hub_args
 from onnx import shape_inference
 from polygraphy import cuda
 from polygraphy.backend.common import bytes_from_path
@@ -50,7 +51,7 @@
     StableDiffusionSafetyChecker,
 )
 from diffusers.schedulers import DDIMScheduler
-from diffusers.utils import DIFFUSERS_CACHE, logging
+from diffusers.utils import logging
 
 
 """
@@ -778,12 +779,13 @@ def __loadModels(self):
             self.models["vae_encoder"] = make_VAEEncoder(self.vae, **models_args)
 
     @classmethod
+    @validate_hf_hub_args
     def set_cached_folder(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], **kwargs):
-        cache_dir = kwargs.pop("cache_dir", DIFFUSERS_CACHE)
+        cache_dir = kwargs.pop("cache_dir", None)
         resume_download = kwargs.pop("resume_download", False)
         proxies = kwargs.pop("proxies", None)
         local_files_only = kwargs.pop("local_files_only", False)
-        use_auth_token = kwargs.pop("use_auth_token", None)
+        token = kwargs.pop("token", None)
         revision = kwargs.pop("revision", None)
 
         cls.cached_folder = (
@@ -795,7 +797,7 @@ def set_cached_folder(cls, pretrained_model_name_or_path: Optional[Union[str, os
                 resume_download=resume_download,
                 proxies=proxies,
                 local_files_only=local_files_only,
-                use_auth_token=use_auth_token,
+                token=token,
                 revision=revision,
             )
         )

examples/community/stable_diffusion_tensorrt_inpaint.py

@@ -28,6 +28,7 @@
 import tensorrt as trt
 import torch
 from huggingface_hub import snapshot_download
+from huggingface_hub.utils import validate_hf_hub_args
 from onnx import shape_inference
 from polygraphy import cuda
 from polygraphy.backend.common import bytes_from_path
@@ -51,7 +52,7 @@
 )
 from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_inpaint import prepare_mask_and_masked_image
 from diffusers.schedulers import DDIMScheduler
-from diffusers.utils import DIFFUSERS_CACHE, logging
+from diffusers.utils import logging
 
 
 """
@@ -779,12 +780,13 @@ def __loadModels(self):
             self.models["vae_encoder"] = make_VAEEncoder(self.vae, **models_args)
 
     @classmethod
+    @validate_hf_hub_args
     def set_cached_folder(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], **kwargs):
-        cache_dir = kwargs.pop("cache_dir", DIFFUSERS_CACHE)
+        cache_dir = kwargs.pop("cache_dir", None)
         resume_download = kwargs.pop("resume_download", False)
         proxies = kwargs.pop("proxies", None)
         local_files_only = kwargs.pop("local_files_only", False)
-        use_auth_token = kwargs.pop("use_auth_token", None)
+        token = kwargs.pop("token", None)
         revision = kwargs.pop("revision", None)
 
         cls.cached_folder = (
@@ -796,7 +798,7 @@ def set_cached_folder(cls, pretrained_model_name_or_path: Optional[Union[str, os
                 resume_download=resume_download,
                 proxies=proxies,
                 local_files_only=local_files_only,
-                use_auth_token=use_auth_token,
+                token=token,
                 revision=revision,
             )
         )

examples/community/stable_diffusion_tensorrt_txt2img.py

@@ -27,6 +27,7 @@
 import tensorrt as trt
 import torch
 from huggingface_hub import snapshot_download
+from huggingface_hub.utils import validate_hf_hub_args
 from onnx import shape_inference
 from polygraphy import cuda
 from polygraphy.backend.common import bytes_from_path
@@ -49,7 +50,7 @@
     StableDiffusionSafetyChecker,
 )
 from diffusers.schedulers import DDIMScheduler
-from diffusers.utils import DIFFUSERS_CACHE, logging
+from diffusers.utils import logging
 
 
 """
@@ -691,12 +692,13 @@ def __loadModels(self):
             self.models["vae"] = make_VAE(self.vae, **models_args)
 
     @classmethod
+    @validate_hf_hub_args
     def set_cached_folder(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], **kwargs):
-        cache_dir = kwargs.pop("cache_dir", DIFFUSERS_CACHE)
+        cache_dir = kwargs.pop("cache_dir", None)
         resume_download = kwargs.pop("resume_download", False)
         proxies = kwargs.pop("proxies", None)
         local_files_only = kwargs.pop("local_files_only", False)
-        use_auth_token = kwargs.pop("use_auth_token", None)
+        token = kwargs.pop("token", None)
         revision = kwargs.pop("revision", None)
 
         cls.cached_folder = (
@@ -708,7 +710,7 @@ def set_cached_folder(cls, pretrained_model_name_or_path: Optional[Union[str, os
                 resume_download=resume_download,
                 proxies=proxies,
                 local_files_only=local_files_only,
-                use_auth_token=use_auth_token,
+                token=token,
                 revision=revision,
             )
         )

examples/consistency_distillation/train_lcm_distill_lora_sd_wds.py

@@ -423,7 +423,7 @@ def import_model_class_from_model_name_or_path(
     pretrained_model_name_or_path: str, revision: str, subfolder: str = "text_encoder"
 ):
     text_encoder_config = PretrainedConfig.from_pretrained(
-        pretrained_model_name_or_path, subfolder=subfolder, revision=revision, use_auth_token=True
+        pretrained_model_name_or_path, subfolder=subfolder, revision=revision
     )
     model_class = text_encoder_config.architectures[0]
 

examples/consistency_distillation/train_lcm_distill_lora_sdxl_wds.py

@@ -397,7 +397,7 @@ def import_model_class_from_model_name_or_path(
     pretrained_model_name_or_path: str, revision: str, subfolder: str = "text_encoder"
 ):
     text_encoder_config = PretrainedConfig.from_pretrained(
-        pretrained_model_name_or_path, subfolder=subfolder, revision=revision, use_auth_token=True
+        pretrained_model_name_or_path, subfolder=subfolder, revision=revision
     )
     model_class = text_encoder_config.architectures[0]
 

examples/consistency_distillation/train_lcm_distill_sd_wds.py

@@ -400,7 +400,7 @@ def import_model_class_from_model_name_or_path(
     pretrained_model_name_or_path: str, revision: str, subfolder: str = "text_encoder"
 ):
     text_encoder_config = PretrainedConfig.from_pretrained(
-        pretrained_model_name_or_path, subfolder=subfolder, revision=revision, use_auth_token=True
+        pretrained_model_name_or_path, subfolder=subfolder, revision=revision
     )
     model_class = text_encoder_config.architectures[0]
 

examples/consistency_distillation/train_lcm_distill_sdxl_wds.py

@@ -419,7 +419,7 @@ def import_model_class_from_model_name_or_path(
     pretrained_model_name_or_path: str, revision: str, subfolder: str = "text_encoder"
 ):
     text_encoder_config = PretrainedConfig.from_pretrained(
-        pretrained_model_name_or_path, subfolder=subfolder, revision=revision, use_auth_token=True
+        pretrained_model_name_or_path, subfolder=subfolder, revision=revision
     )
     model_class = text_encoder_config.architectures[0]
 

examples/research_projects/controlnet/train_controlnet_webdataset.py

@@ -420,7 +420,7 @@ def import_model_class_from_model_name_or_path(
     pretrained_model_name_or_path: str, revision: str, subfolder: str = "text_encoder"
 ):
     text_encoder_config = PretrainedConfig.from_pretrained(
-        pretrained_model_name_or_path, subfolder=subfolder, revision=revision, use_auth_token=True
+        pretrained_model_name_or_path, subfolder=subfolder, revision=revision
     )
     model_class = text_encoder_config.architectures[0]
 
@@ -975,7 +975,7 @@ def main(args):
         revision=args.revision,
     )
     unet = UNet2DConditionModel.from_pretrained(
-        args.pretrained_model_name_or_path, subfolder="unet", revision=args.revision, use_auth_token=True
+        args.pretrained_model_name_or_path, subfolder="unet", revision=args.revision
     )
 
     if args.controlnet_model_name_or_path:

src/diffusers/commands/fp16_safetensors.py

@@ -19,6 +19,7 @@
 
 import glob
 import json
+import warnings
 from argparse import ArgumentParser, Namespace
 from importlib import import_module
 
@@ -32,12 +33,12 @@
 
 
 def conversion_command_factory(args: Namespace):
-    return FP16SafetensorsCommand(
-        args.ckpt_id,
-        args.fp16,
-        args.use_safetensors,
-        args.use_auth_token,
-    )
+    if args.use_auth_token:
+        warnings.warn(
+            "The `--use_auth_token` flag is deprecated and will be removed in a future version. Authentication is now"
+            " handled automatically if user is logged in."
+        )
+    return FP16SafetensorsCommand(args.ckpt_id, args.fp16, args.use_safetensors)
 
 
 class FP16SafetensorsCommand(BaseDiffusersCLICommand):
@@ -62,7 +63,7 @@ def register_subcommand(parser: ArgumentParser):
         )
         conversion_parser.set_defaults(func=conversion_command_factory)
 
-    def __init__(self, ckpt_id: str, fp16: bool, use_safetensors: bool, use_auth_token: bool):
+    def __init__(self, ckpt_id: str, fp16: bool, use_safetensors: bool):
         self.logger = logging.get_logger("diffusers-cli/fp16_safetensors")
         self.ckpt_id = ckpt_id
         self.local_ckpt_dir = f"/tmp/{ckpt_id}"
@@ -75,8 +76,6 @@ def __init__(self, ckpt_id: str, fp16: bool, use_safetensors: bool, use_auth_tok
                 "When `use_safetensors` and `fp16` both are False, then this command is of no use."
             )
 
-        self.use_auth_token = use_auth_token
-
     def run(self):
         if version.parse(huggingface_hub.__version__) < version.parse("0.9.0"):
             raise ImportError(
@@ -87,7 +86,7 @@ def run(self):
             from huggingface_hub import create_commit
             from huggingface_hub._commit_api import CommitOperationAdd
 
-        model_index = hf_hub_download(repo_id=self.ckpt_id, filename="model_index.json", token=self.use_auth_token)
+        model_index = hf_hub_download(repo_id=self.ckpt_id, filename="model_index.json")
         with open(model_index, "r") as f:
             pipeline_class_name = json.load(f)["_class_name"]
         pipeline_class = getattr(import_module("diffusers"), pipeline_class_name)
@@ -96,7 +95,7 @@ def run(self):
         # Load the appropriate pipeline. We could have use `DiffusionPipeline`
         # here, but just to avoid any rough edge cases.
         pipeline = pipeline_class.from_pretrained(
-            self.ckpt_id, torch_dtype=torch.float16 if self.fp16 else torch.float32, use_auth_token=self.use_auth_token
+            self.ckpt_id, torch_dtype=torch.float16 if self.fp16 else torch.float32
         )
         pipeline.save_pretrained(
             self.local_ckpt_dir,