huggingface-hub 0.28.0rc5__py3-none-any.whl → 0.29.0__py3-none-any.whl

huggingface_hub/inference/_providers/replicate.py

@@ -1,114 +1,32 @@
 from typing import Any, Dict, Optional, Union
 
-from huggingface_hub import constants
-from huggingface_hub.inference._common import RequestParameters, TaskProviderHelper, _as_dict
-from huggingface_hub.utils import build_hf_headers, get_session, get_token, logging
+from huggingface_hub.inference._common import _as_dict
+from huggingface_hub.inference._providers._common import TaskProviderHelper, filter_none
+from huggingface_hub.utils import get_session
 
 
-logger = logging.get_logger(__name__)
-
-
-BASE_URL = "https://api.replicate.com"
-
-SUPPORTED_MODELS = {
-    "text-to-image": {
-        "black-forest-labs/FLUX.1-dev": "black-forest-labs/flux-dev",
-        "black-forest-labs/FLUX.1-schnell": "black-forest-labs/flux-schnell",
-        "ByteDance/Hyper-SD": "bytedance/hyper-flux-16step:382cf8959fb0f0d665b26e7e80b8d6dc3faaef1510f14ce017e8c732bb3d1eb7",
-        "ByteDance/SDXL-Lightning": "bytedance/sdxl-lightning-4step:5599ed30703defd1d160a25a63321b4dec97101d98b4674bcc56e41f62f35637",
-        "playgroundai/playground-v2.5-1024px-aesthetic": "playgroundai/playground-v2.5-1024px-aesthetic:a45f82a1382bed5c7aeb861dac7c7d191b0fdf74d8d57c4a0e6ed7d4d0bf7d24",
-        "stabilityai/stable-diffusion-3.5-large-turbo": "stability-ai/stable-diffusion-3.5-large-turbo",
-        "stabilityai/stable-diffusion-3.5-large": "stability-ai/stable-diffusion-3.5-large",
-        "stabilityai/stable-diffusion-3.5-medium": "stability-ai/stable-diffusion-3.5-medium",
-        "stabilityai/stable-diffusion-xl-base-1.0": "stability-ai/sdxl:7762fd07cf82c948538e41f63f77d685e02b063e37e496e96eefd46c929f9bdc",
-    },
-    "text-to-speech": {
-        "OuteAI/OuteTTS-0.3-500M": "jbilcke/oute-tts:39a59319327b27327fa3095149c5a746e7f2aee18c75055c3368237a6503cd26",
-    },
-    "text-to-video": {
-        "genmo/mochi-1-preview": "genmoai/mochi-1:1944af04d098ef69bed7f9d335d102e652203f268ec4aaa2d836f6217217e460",
-    },
-}
-
-
-def _build_url(base_url: str, model: str) -> str:
-    if ":" in model:
-        return f"{base_url}/v1/predictions"
-    return f"{base_url}/v1/models/{model}/predictions"
+_PROVIDER = "replicate"
+_BASE_URL = "https://api.replicate.com"
 
 
 class ReplicateTask(TaskProviderHelper):
     def __init__(self, task: str):
-        self.task = task
-
-    def prepare_request(
-        self,
-        *,
-        inputs: Any,
-        parameters: Dict[str, Any],
-        headers: Dict,
-        model: Optional[str],
-        api_key: Optional[str],
-        extra_payload: Optional[Dict[str, Any]] = None,
-    ) -> RequestParameters:
-        if api_key is None:
-            api_key = get_token()
-        if api_key is None:
-            raise ValueError(
-                "You must provide an api_key to work with Replicate API or log in with `huggingface-cli login`."
-            )
-
-        # Route to the proxy if the api_key is a HF TOKEN
-        if api_key.startswith("hf_"):
-            base_url = constants.INFERENCE_PROXY_TEMPLATE.format(provider="replicate")
-            logger.info("Calling Replicate provider through Hugging Face proxy.")
-        else:
-            base_url = BASE_URL
-            logger.info("Calling Replicate provider directly.")
-        mapped_model = self._map_model(model)
-        url = _build_url(base_url, mapped_model)
-
-        headers = {
-            **build_hf_headers(token=api_key),
-            **headers,
-            "Prefer": "wait",
-        }
-
-        payload = self._prepare_payload(inputs, parameters=parameters, model=mapped_model)
-
-        return RequestParameters(
-            url=url,
-            task=self.task,
-            model=mapped_model,
-            json=payload,
-            data=None,
-            headers=headers,
-        )
-
-    def _map_model(self, model: Optional[str]) -> str:
-        if model is None:
-            raise ValueError("Please provide a model available on Replicate.")
-        if self.task not in SUPPORTED_MODELS:
-            raise ValueError(f"Task {self.task} not supported with Replicate.")
-        mapped_model = SUPPORTED_MODELS[self.task].get(model)
-        if mapped_model is None:
-            raise ValueError(f"Model {model} is not supported with Replicate for task {self.task}.")
-        return mapped_model
-
-    def _prepare_payload(
-        self,
-        inputs: Any,
-        parameters: Dict[str, Any],
-        model: str,
-    ) -> Dict[str, Any]:
-        payload: Dict[str, Any] = {
-            "input": {
-                "prompt": inputs,
-                **{k: v for k, v in parameters.items() if v is not None},
-            }
-        }
-        if ":" in model:
-            version = model.split(":", 1)[1]
+        super().__init__(provider=_PROVIDER, base_url=_BASE_URL, task=task)
+
+    def _prepare_headers(self, headers: Dict, api_key: str) -> Dict:
+        headers = super()._prepare_headers(headers, api_key)
+        headers["Prefer"] = "wait"
+        return headers
+
+    def _prepare_route(self, mapped_model: str) -> str:
+        if ":" in mapped_model:
+            return "/v1/predictions"
+        return f"/v1/models/{mapped_model}/predictions"
+
+    def _prepare_payload_as_dict(self, inputs: Any, parameters: Dict, mapped_model: str) -> Optional[Dict]:
+        payload: Dict[str, Any] = {"input": {"prompt": inputs, **filter_none(parameters)}}
+        if ":" in mapped_model:
+            version = mapped_model.split(":", 1)[1]
             payload["version"] = version
         return payload
 
@@ -129,20 +47,7 @@ class ReplicateTextToSpeechTask(ReplicateTask):
     def __init__(self):
         super().__init__("text-to-speech")
 
-    def _prepare_payload(
-        self,
-        inputs: Any,
-        parameters: Dict[str, Any],
-        model: str,
-    ) -> Dict[str, Any]:
-        # The following payload might work only for a subset of text-to-speech Replicate models.
-        payload: Dict[str, Any] = {
-            "input": {
-                "inputs": inputs,
-                **{k: v for k, v in parameters.items() if v is not None},
-            },
-        }
-        if ":" in model:
-            version = model.split(":", 1)[1]
-            payload["version"] = version
+    def _prepare_payload_as_dict(self, inputs: Any, parameters: Dict, mapped_model: str) -> Optional[Dict]:
+        payload: Dict = super()._prepare_payload_as_dict(inputs, parameters, mapped_model)  # type: ignore[assignment]
+        payload["input"]["text"] = payload["input"].pop("prompt")  # rename "prompt" to "text" for TTS
         return payload

huggingface_hub/inference/_providers/sambanova.py

@@ -1,89 +1,6 @@
-from typing import Any, Dict, Optional, Union
+from huggingface_hub.inference._providers._common import BaseConversationalTask
 
-from huggingface_hub import constants
-from huggingface_hub.inference._common import RequestParameters, TaskProviderHelper
-from huggingface_hub.utils import build_hf_headers, get_token, logging
 
-
-logger = logging.get_logger(__name__)
-
-
-BASE_URL = "https://api.sambanova.ai"
-
-SUPPORTED_MODELS = {
-    "conversational": {
-        "Qwen/Qwen2.5-Coder-32B-Instruct": "Qwen2.5-Coder-32B-Instruct",
-        "Qwen/Qwen2.5-72B-Instruct": "Qwen2.5-72B-Instruct",
-        "Qwen/QwQ-32B-Preview": "QwQ-32B-Preview",
-        "meta-llama/Llama-3.3-70B-Instruct": "Meta-Llama-3.3-70B-Instruct",
-        "meta-llama/Llama-3.2-1B": "Meta-Llama-3.2-1B-Instruct",
-        "meta-llama/Llama-3.2-3B": "Meta-Llama-3.2-3B-Instruct",
-        "meta-llama/Llama-3.2-11B-Vision-Instruct": "Llama-3.2-11B-Vision-Instruct",
-        "meta-llama/Llama-3.2-90B-Vision-Instruct": "Llama-3.2-90B-Vision-Instruct",
-        "meta-llama/Llama-3.1-8B-Instruct": "Meta-Llama-3.1-8B-Instruct",
-        "meta-llama/Llama-3.1-70B-Instruct": "Meta-Llama-3.1-70B-Instruct",
-        "meta-llama/Llama-3.1-405B-Instruct": "Meta-Llama-3.1-405B-Instruct",
-        "meta-llama/Llama-Guard-3-8B": "Meta-Llama-Guard-3-8B",
-    },
-}
-
-
-class SambanovaConversationalTask(TaskProviderHelper):
+class SambanovaConversationalTask(BaseConversationalTask):
     def __init__(self):
-        # TODO: adapt in a base class when supporting multiple tasks
-        self.task = "conversational"
-
-    def prepare_request(
-        self,
-        *,
-        inputs: Any,
-        parameters: Dict[str, Any],
-        headers: Dict,
-        model: Optional[str],
-        api_key: Optional[str],
-        extra_payload: Optional[Dict[str, Any]] = None,
-    ) -> RequestParameters:
-        if api_key is None:
-            api_key = get_token()
-        if api_key is None:
-            raise ValueError(
-                "You must provide an api_key to work with Sambanova API or log in with `huggingface-cli login`."
-            )
-
-        # Route to the proxy if the api_key is a HF TOKEN
-        if api_key.startswith("hf_"):
-            base_url = constants.INFERENCE_PROXY_TEMPLATE.format(provider="sambanova")
-            logger.info("Calling Sambanova provider through Hugging Face proxy.")
-        else:
-            base_url = BASE_URL
-            logger.info("Calling Sambanova provider directly.")
-        headers = {**build_hf_headers(token=api_key), **headers}
-
-        mapped_model = self._map_model(model)
-        payload = {
-            "messages": inputs,
-            **{k: v for k, v in parameters.items() if v is not None},
-            "model": mapped_model,
-        }
-
-        return RequestParameters(
-            url=f"{base_url}/v1/chat/completions",
-            task=self.task,
-            model=mapped_model,
-            json=payload,
-            data=None,
-            headers=headers,
-        )
-
-    def _map_model(self, model: Optional[str]) -> str:
-        if model is None:
-            raise ValueError("Please provide a model available on Sambanova.")
-        if self.task not in SUPPORTED_MODELS:
-            raise ValueError(f"Task {self.task} not supported with Sambanova.")
-        mapped_model = SUPPORTED_MODELS[self.task].get(model)
-        if mapped_model is None:
-            raise ValueError(f"Model {model} is not supported with Sambanova for task {self.task}.")
-        return mapped_model
-
-    def get_response(self, response: Union[bytes, Dict]) -> Any:
-        return response
+        super().__init__(provider="sambanova", base_url="https://api.sambanova.ai")

huggingface_hub/inference/_providers/together.py

@@ -1,152 +1,58 @@
 import base64
-from abc import ABC, abstractmethod
+from abc import ABC
 from typing import Any, Dict, Optional, Union
 
-from huggingface_hub import constants
-from huggingface_hub.inference._common import RequestParameters, TaskProviderHelper, _as_dict
-from huggingface_hub.utils import build_hf_headers, get_token, logging
-
-
-logger = logging.get_logger(__name__)
-
-
-BASE_URL = "https://api.together.xyz"
-
-SUPPORTED_MODELS = {
-    "conversational": {
-        "databricks/dbrx-instruct": "databricks/dbrx-instruct",
-        "deepseek-ai/DeepSeek-R1": "deepseek-ai/DeepSeek-R1",
-        "deepseek-ai/DeepSeek-V3": "deepseek-ai/DeepSeek-V3",
-        "deepseek-ai/deepseek-llm-67b-chat": "deepseek-ai/deepseek-llm-67b-chat",
-        "google/gemma-2-9b-it": "google/gemma-2-9b-it",
-        "google/gemma-2b-it": "google/gemma-2-27b-it",
-        "meta-llama/Llama-2-13b-chat-hf": "meta-llama/Llama-2-13b-chat-hf",
-        "meta-llama/Llama-2-7b-chat-hf": "meta-llama/Llama-2-7b-chat-hf",
-        "meta-llama/Llama-3.2-11B-Vision-Instruct": "meta-llama/Llama-Vision-Free",
-        "meta-llama/Llama-3.2-3B-Instruct": "meta-llama/Llama-3.2-3B-Instruct-Turbo",
-        "meta-llama/Llama-3.2-90B-Vision-Instruct": "meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo",
-        "meta-llama/Llama-3.3-70B-Instruct": "meta-llama/Llama-3.3-70B-Instruct-Turbo",
-        "meta-llama/Meta-Llama-3-70B-Instruct": "meta-llama/Llama-3-70b-chat-hf",
-        "meta-llama/Meta-Llama-3-8B-Instruct": "meta-llama/Meta-Llama-3-8B-Instruct-Turbo",
-        "meta-llama/Meta-Llama-3.1-405B-Instruct": "meta-llama/Llama-3.2-11B-Vision-Instruct-Turbo",
-        "meta-llama/Meta-Llama-3.1-70B-Instruct": "meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo",
-        "meta-llama/Meta-Llama-3.1-8B-Instruct": "meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
-        "microsoft/WizardLM-2-8x22B": "microsoft/WizardLM-2-8x22B",
-        "mistralai/Mistral-7B-Instruct-v0.3": "mistralai/Mistral-7B-Instruct-v0.3",
-        "mistralai/Mixtral-8x22B-Instruct-v0.1": "mistralai/Mixtral-8x22B-Instruct-v0.1",
-        "mistralai/Mixtral-8x7B-Instruct-v0.1": "mistralai/Mixtral-8x7B-Instruct-v0.1",
-        "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO": "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
-        "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF": "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF",
-        "Qwen/Qwen2-72B-Instruct": "Qwen/Qwen2-72B-Instruct",
-        "Qwen/Qwen2.5-72B-Instruct": "Qwen/Qwen2.5-72B-Instruct-Turbo",
-        "Qwen/Qwen2.5-7B-Instruct": "Qwen/Qwen2.5-7B-Instruct-Turbo",
-        "Qwen/Qwen2.5-Coder-32B-Instruct": "Qwen/Qwen2.5-Coder-32B-Instruct",
-        "Qwen/QwQ-32B-Preview": "Qwen/QwQ-32B-Preview",
-        "scb10x/llama-3-typhoon-v1.5-8b-instruct": "scb10x/scb10x-llama3-typhoon-v1-5-8b-instruct",
-        "scb10x/llama-3-typhoon-v1.5x-70b-instruct-awq": "scb10x/scb10x-llama3-typhoon-v1-5x-4f316",
-    },
-    "text-generation": {
-        "meta-llama/Llama-2-70b-hf": "meta-llama/Llama-2-70b-hf",
-        "meta-llama/Meta-Llama-3-8B": "meta-llama/Meta-Llama-3-8B",
-        "mistralai/Mixtral-8x7B-v0.1": "mistralai/Mixtral-8x7B-v0.1",
-    },
-    "text-to-image": {
-        "black-forest-labs/FLUX.1-Canny-dev": "black-forest-labs/FLUX.1-canny",
-        "black-forest-labs/FLUX.1-Depth-dev": "black-forest-labs/FLUX.1-depth",
-        "black-forest-labs/FLUX.1-dev": "black-forest-labs/FLUX.1-dev",
-        "black-forest-labs/FLUX.1-Redux-dev": "black-forest-labs/FLUX.1-redux",
-        "black-forest-labs/FLUX.1-schnell": "black-forest-labs/FLUX.1-pro",
-        "stabilityai/stable-diffusion-xl-base-1.0": "stabilityai/stable-diffusion-xl-base-1.0",
-    },
-}
-
-
-PER_TASK_ROUTES = {
-    "conversational": "v1/chat/completions",
-    "text-generation": "v1/completions",
-    "text-to-image": "v1/images/generations",
-}
+from huggingface_hub.inference._common import _as_dict
+from huggingface_hub.inference._providers._common import (
+    BaseConversationalTask,
+    BaseTextGenerationTask,
+    TaskProviderHelper,
+    filter_none,
+)
+
+
+_PROVIDER = "together"
+_BASE_URL = "https://api.together.xyz"
 
 
 class TogetherTask(TaskProviderHelper, ABC):
     """Base class for Together API tasks."""
 
     def __init__(self, task: str):
-        self.task = task
-
-    def prepare_request(
-        self,
-        *,
-        inputs: Any,
-        parameters: Dict[str, Any],
-        headers: Dict,
-        model: Optional[str],
-        api_key: Optional[str],
-        extra_payload: Optional[Dict[str, Any]] = None,
-    ) -> RequestParameters:
-        if api_key is None:
-            api_key = get_token()
-        if api_key is None:
-            raise ValueError(
-                "You must provide an api_key to work with Together API or log in with `huggingface-cli login`."
-            )
-        headers = {**build_hf_headers(token=api_key), **headers}
-
-        # Route to the proxy if the api_key is a HF TOKEN
-        if api_key.startswith("hf_"):
-            base_url = constants.INFERENCE_PROXY_TEMPLATE.format(provider="together")
-            logger.info("Calling Together provider through Hugging Face proxy.")
-        else:
-            base_url = BASE_URL
-            logger.info("Calling Together provider directly.")
-        mapped_model = self._map_model(model)
-        if "model" in parameters:
-            parameters["model"] = mapped_model
-        payload = self._prepare_payload(inputs, parameters=parameters)
-
-        return RequestParameters(
-            url=f"{base_url}/{PER_TASK_ROUTES[self.task]}",
-            task=self.task,
-            model=mapped_model,
-            json=payload,
-            data=None,
-            headers=headers,
-        )
-
-    def _map_model(self, model: Optional[str]) -> str:
-        if model is None:
-            raise ValueError("Please provide a model available on Together.")
-        if self.task not in SUPPORTED_MODELS:
-            raise ValueError(f"Task {self.task} not supported with Together.")
-        mapped_model = SUPPORTED_MODELS[self.task].get(model)
-        if mapped_model is None:
-            raise ValueError(f"Model {model} is not supported with Together for task {self.task}.")
-        return mapped_model
+        super().__init__(provider=_PROVIDER, base_url=_BASE_URL, task=task)
 
-    def get_response(self, response: Union[bytes, Dict]) -> Any:
-        return response
+    def _prepare_route(self, mapped_model: str) -> str:
+        if self.task == "text-to-image":
+            return "/v1/images/generations"
+        elif self.task == "conversational":
+            return "/v1/chat/completions"
+        elif self.task == "text-generation":
+            return "/v1/completions"
+        raise ValueError(f"Unsupported task '{self.task}' for Together API.")
 
-    @abstractmethod
-    def _prepare_payload(self, inputs: Any, parameters: Dict[str, Any]) -> Dict[str, Any]: ...
 
+class TogetherTextGenerationTask(BaseTextGenerationTask):
+    def __init__(self):
+        super().__init__(provider=_PROVIDER, base_url=_BASE_URL)
 
-class TogetherTextGenerationTask(TogetherTask):
-    # Handle both "text-generation" and "conversational"
-    def _prepare_payload(self, inputs: Any, parameters: Dict[str, Any]) -> Dict[str, Any]:
-        return {"messages": inputs, **{k: v for k, v in parameters.items() if v is not None}}
+
+class TogetherConversationalTask(BaseConversationalTask):
+    def __init__(self):
+        super().__init__(provider=_PROVIDER, base_url=_BASE_URL)
 
 
 class TogetherTextToImageTask(TogetherTask):
     def __init__(self):
         super().__init__("text-to-image")
 
-    def _prepare_payload(self, inputs: Any, parameters: Dict[str, Any]) -> Dict[str, Any]:
-        payload = {
-            "prompt": inputs,
-            "response_format": "base64",
-            **{k: v for k, v in parameters.items() if v is not None},
-        }
-        return payload
+    def _prepare_payload_as_dict(self, inputs: Any, parameters: Dict, mapped_model: str) -> Optional[Dict]:
+        parameters = filter_none(parameters)
+        if "num_inference_steps" in parameters:
+            parameters["steps"] = parameters.pop("num_inference_steps")
+        if "guidance_scale" in parameters:
+            parameters["guidance"] = parameters.pop("guidance_scale")
+
+        return {"prompt": inputs, "response_format": "base64", **parameters, "model": mapped_model}
 
     def get_response(self, response: Union[bytes, Dict]) -> Any:
         response_dict = _as_dict(response)

huggingface_hub/utils/_headers.py

@@ -213,6 +213,11 @@ def _http_user_agent(
     elif isinstance(user_agent, str):
         ua += "; " + user_agent
 
+    # Retrieve user-agent origin headers from environment variable
+    origin = constants.HF_HUB_USER_AGENT_ORIGIN
+    if origin is not None:
+        ua += "; origin/" + origin
+
     return _deduplicate_user_agent(ua)
 
 

huggingface_hub/utils/_hf_folder.py

@@ -14,7 +14,6 @@
 # limitations under the License.
 """Contain helper class to retrieve/store token from/to local cache."""
 
-import warnings
 from pathlib import Path
 from typing import Optional
 
@@ -23,10 +22,6 @@ from ._auth import get_token
 
 
 class HfFolder:
-    path_token = Path(constants.HF_TOKEN_PATH)
-    # Private attribute. Will be removed in v0.15
-    _old_path_token = Path(constants._OLD_HF_TOKEN_PATH)
-
     # TODO: deprecate when adapted in transformers/datasets/gradio
     # @_deprecate_method(version="1.0", message="Use `huggingface_hub.login` instead.")
     @classmethod
@@ -41,8 +36,9 @@ class HfFolder:
             token (`str`):
                 The token to save to the [`HfFolder`]
         """
-        cls.path_token.parent.mkdir(parents=True, exist_ok=True)
-        cls.path_token.write_text(token)
+        path_token = Path(constants.HF_TOKEN_PATH)
+        path_token.parent.mkdir(parents=True, exist_ok=True)
+        path_token.write_text(token)
 
     # TODO: deprecate when adapted in transformers/datasets/gradio
     # @_deprecate_method(version="1.0", message="Use `huggingface_hub.get_token` instead.")
@@ -57,12 +53,6 @@ class HfFolder:
         Returns:
             `str` or `None`: The token, `None` if it doesn't exist.
         """
-        # 0. Check if token exist in old path but not new location
-        try:
-            cls._copy_to_new_path_and_warn()
-        except Exception:  # if not possible (e.g. PermissionError), do not raise
-            pass
-
         return get_token()
 
     # TODO: deprecate when adapted in transformers/datasets/gradio
@@ -73,24 +63,6 @@ class HfFolder:
         Deletes the token from storage. Does not fail if token does not exist.
         """
         try:
-            cls.path_token.unlink()
-        except FileNotFoundError:
-            pass
-
-        try:
-            cls._old_path_token.unlink()
+            Path(constants.HF_TOKEN_PATH).unlink()
         except FileNotFoundError:
             pass
-
-    @classmethod
-    def _copy_to_new_path_and_warn(cls):
-        if cls._old_path_token.exists() and not cls.path_token.exists():
-            cls.save_token(cls._old_path_token.read_text())
-            warnings.warn(
-                f"A token has been found in `{cls._old_path_token}`. This is the old"
-                " path where tokens were stored. The new location is"
-                f" `{cls.path_token}` which is configurable using `HF_HOME` environment"
-                " variable. Your token has been copied to this new location. You can"
-                " now safely delete the old token file manually or use"
-                " `huggingface-cli logout`."
-            )

huggingface_hub/utils/_http.py

@@ -22,7 +22,8 @@ import time
 import uuid
 from functools import lru_cache
 from http import HTTPStatus
-from typing import Callable, Optional, Tuple, Type, Union
+from shlex import quote
+from typing import Any, Callable, List, Optional, Tuple, Type, Union
 
 import requests
 from requests import HTTPError, Response
@@ -82,13 +83,15 @@ class UniqueRequestIdAdapter(HTTPAdapter):
             request.headers[X_AMZN_TRACE_ID] = request.headers.get(X_REQUEST_ID) or str(uuid.uuid4())
 
         # Add debug log
-        has_token = str(request.headers.get("authorization", "")).startswith("Bearer hf_")
+        has_token = len(str(request.headers.get("authorization", ""))) > 0
         logger.debug(
             f"Request {request.headers[X_AMZN_TRACE_ID]}: {request.method} {request.url} (authenticated: {has_token})"
         )
 
     def send(self, request: PreparedRequest, *args, **kwargs) -> Response:
         """Catch any RequestException to append request id to the error message for debugging."""
+        if constants.HF_DEBUG:
+            logger.debug(f"Send: {_curlify(request)}")
         try:
             return super().send(request, *args, **kwargs)
         except requests.RequestException as e:
@@ -434,6 +437,7 @@ def hf_raise_for_status(response: Response, endpoint_name: Optional[str] = None)
 
         elif error_code == "RepoNotFound" or (
             response.status_code == 401
+            and error_message != "Invalid credentials in Authorization header"
             and response.request is not None
             and response.request.url is not None
             and REPO_API_REGEX.search(response.request.url) is not None
@@ -549,3 +553,82 @@ def _format(error_type: Type[HfHubHTTPError], custom_message: str, response: Res
 
     # Return
     return error_type(final_error_message.strip(), response=response, server_message=server_message or None)
+
+
+def _curlify(request: requests.PreparedRequest) -> str:
+    """Convert a `requests.PreparedRequest` into a curl command (str).
+
+    Used for debug purposes only.
+
+    Implementation vendored from https://github.com/ofw/curlify/blob/master/curlify.py.
+    MIT License Copyright (c) 2016 Egor.
+    """
+    parts: List[Tuple[Any, Any]] = [
+        ("curl", None),
+        ("-X", request.method),
+    ]
+
+    for k, v in sorted(request.headers.items()):
+        if k.lower() == "authorization":
+            v = "<TOKEN>"  # Hide authorization header, no matter its value (can be Bearer, Key, etc.)
+        parts += [("-H", "{0}: {1}".format(k, v))]
+
+    if request.body:
+        body = request.body
+        if isinstance(body, bytes):
+            body = body.decode("utf-8", errors="ignore")
+        if len(body) > 1000:
+            body = body[:1000] + " ... [truncated]"
+        parts += [("-d", body.replace("\n", ""))]
+
+    parts += [(None, request.url)]
+
+    flat_parts = []
+    for k, v in parts:
+        if k:
+            flat_parts.append(quote(k))
+        if v:
+            flat_parts.append(quote(v))
+
+    return " ".join(flat_parts)
+
+
+# Regex to parse HTTP Range header
+RANGE_REGEX = re.compile(r"^\s*bytes\s*=\s*(\d*)\s*-\s*(\d*)\s*$", re.IGNORECASE)
+
+
+def _adjust_range_header(original_range: Optional[str], resume_size: int) -> Optional[str]:
+    """
+    Adjust HTTP Range header to account for resume position.
+    """
+    if not original_range:
+        return f"bytes={resume_size}-"
+
+    if "," in original_range:
+        raise ValueError(f"Multiple ranges detected - {original_range!r}, not supported yet.")
+
+    match = RANGE_REGEX.match(original_range)
+    if not match:
+        raise RuntimeError(f"Invalid range format - {original_range!r}.")
+    start, end = match.groups()
+
+    if not start:
+        if not end:
+            raise RuntimeError(f"Invalid range format - {original_range!r}.")
+
+        new_suffix = int(end) - resume_size
+        new_range = f"bytes=-{new_suffix}"
+        if new_suffix <= 0:
+            raise RuntimeError(f"Empty new range - {new_range!r}.")
+        return new_range
+
+    start = int(start)
+    new_start = start + resume_size
+    if end:
+        end = int(end)
+        new_range = f"bytes={new_start}-{end}"
+        if new_start > end:
+            raise RuntimeError(f"Empty new range - {new_range!r}.")
+        return new_range
+
+    return f"bytes={new_start}-"

huggingface_hub/utils/_typing.py

@@ -50,7 +50,7 @@ def is_jsonable(obj: Any) -> bool:
         if isinstance(obj, (list, tuple)):
             return all(is_jsonable(item) for item in obj)
         if isinstance(obj, dict):
-            return all(isinstance(key, str) and is_jsonable(value) for key, value in obj.items())
+            return all(isinstance(key, _JSON_SERIALIZABLE_TYPES) and is_jsonable(value) for key, value in obj.items())
         if hasattr(obj, "__json__"):
             return True
         return False

huggingface_hub/utils/logging.py

@@ -28,6 +28,8 @@ from logging import (
 )
 from typing import Optional
 
+from .. import constants
+
 
 log_levels = {
     "debug": logging.DEBUG,
@@ -180,3 +182,7 @@ def enable_propagation() -> None:
 
 
 _configure_library_root_logger()
+
+if constants.HF_DEBUG:
+    # If `HF_DEBUG` environment variable is set, set the verbosity of `huggingface_hub` logger to `DEBUG`.
+    set_verbosity_debug()