huggingface-hub 0.32.5__py3-none-any.whl → 0.33.0__py3-none-any.whl

huggingface_hub/__init__.py

@@ -46,7 +46,7 @@ import sys
 from typing import TYPE_CHECKING
 
 
-__version__ = "0.32.5"
+__version__ = "0.33.0"
 
 # Alphabetical order of definitions is ensured in tests
 # WARNING: any comment added in this dictionary definition will be lost when

huggingface_hub/_local_folder.py

@@ -173,6 +173,7 @@ class LocalUploadFileMetadata:
 
                 if self.upload_mode is not None:
                     f.write(self.upload_mode)
+                f.write("\n")
 
                 if self.remote_oid is not None:
                     f.write(self.remote_oid)

huggingface_hub/_snapshot_download.py

@@ -1,6 +1,6 @@
 import os
 from pathlib import Path
-from typing import Dict, Iterable, List, Literal, Optional, Union
+from typing import Dict, Iterable, List, Literal, Optional, Type, Union
 
 import requests
 from tqdm.auto import tqdm as base_tqdm
@@ -44,7 +44,7 @@ def snapshot_download(
     allow_patterns: Optional[Union[List[str], str]] = None,
     ignore_patterns: Optional[Union[List[str], str]] = None,
     max_workers: int = 8,
-    tqdm_class: Optional[base_tqdm] = None,
+    tqdm_class: Optional[Type[base_tqdm]] = None,
     headers: Optional[Dict[str, str]] = None,
     endpoint: Optional[str] = None,
     # Deprecated args

huggingface_hub/file_download.py

@@ -1706,7 +1706,7 @@ def _download_to_tmp_and_move(
             _check_disk_space(expected_size, destination_path.parent)
 
         if xet_file_data is not None and is_xet_available():
-            logger.info("Xet Storage is enabled for this repo. Downloading file from Xet Storage..")
+            logger.debug("Xet Storage is enabled for this repo. Downloading file from Xet Storage..")
             xet_get(
                 incomplete_path=incomplete_path,
                 xet_file_data=xet_file_data,

huggingface_hub/hf_api.py

@@ -28,6 +28,7 @@ from functools import wraps
 from itertools import islice
 from pathlib import Path
 from typing import (
+    TYPE_CHECKING,
     Any,
     BinaryIO,
     Callable,
@@ -38,6 +39,7 @@ from typing import (
     Literal,
     Optional,
     Tuple,
+    Type,
     TypeVar,
     Union,
     overload,
@@ -134,8 +136,11 @@ from .utils._typing import CallableT
 from .utils.endpoint_helpers import _is_emission_within_threshold
 
 
+if TYPE_CHECKING:
+    from .inference._providers import PROVIDER_T
+
 R = TypeVar("R")  # Return type
-CollectionItemType_T = Literal["model", "dataset", "space", "paper"]
+CollectionItemType_T = Literal["model", "dataset", "space", "paper", "collection"]
 
 ExpandModelProperty_T = Literal[
     "author",
@@ -708,21 +713,26 @@ class RepoFolder:
 
 @dataclass
 class InferenceProviderMapping:
-    hf_model_id: str
-    status: Literal["live", "staging"]
-    provider_id: str
+    provider: "PROVIDER_T"  # Provider name
+    hf_model_id: str  # ID of the model on the Hugging Face Hub
+    provider_id: str  # ID of the model on the provider's side
+    status: Literal["error", "live", "staging"]
     task: str
 
     adapter: Optional[str] = None
     adapter_weights_path: Optional[str] = None
+    type: Optional[Literal["single-model", "tag-filter"]] = None
 
     def __init__(self, **kwargs):
+        self.provider = kwargs.pop("provider")
         self.hf_model_id = kwargs.pop("hf_model_id")
-        self.status = kwargs.pop("status")
         self.provider_id = kwargs.pop("providerId")
+        self.status = kwargs.pop("status")
         self.task = kwargs.pop("task")
+
         self.adapter = kwargs.pop("adapter", None)
         self.adapter_weights_path = kwargs.pop("adapterWeightsPath", None)
+        self.type = kwargs.pop("type", None)
         self.__dict__.update(**kwargs)
 
 
@@ -764,12 +774,10 @@ class ModelInfo:
             If so, whether there is manual or automatic approval.
         gguf (`Dict`, *optional*):
             GGUF information of the model.
-        inference (`Literal["cold", "frozen", "warm"]`, *optional*):
-            Status of the model on the inference API.
-            Warm models are available for immediate use. Cold models will be loaded on first inference call.
-            Frozen models are not available in Inference API.
-        inference_provider_mapping (`Dict`, *optional*):
-            Model's inference provider mapping.
+        inference (`Literal["warm"]`, *optional*):
+            Status of the model on Inference Providers. Warm if the model is served by at least one provider.
+        inference_provider_mapping (`List[InferenceProviderMapping]`, *optional*):
+            A list of [`InferenceProviderMapping`] ordered after the user's provider order.
         likes (`int`):
             Number of likes of the model.
         library_name (`str`, *optional*):
@@ -814,8 +822,8 @@ class ModelInfo:
     downloads_all_time: Optional[int]
     gated: Optional[Literal["auto", "manual", False]]
     gguf: Optional[Dict]
-    inference: Optional[Literal["warm", "cold", "frozen"]]
-    inference_provider_mapping: Optional[Dict[str, InferenceProviderMapping]]
+    inference: Optional[Literal["warm"]]
+    inference_provider_mapping: Optional[List[InferenceProviderMapping]]
     likes: Optional[int]
     library_name: Optional[str]
     tags: Optional[List[str]]
@@ -851,14 +859,25 @@ class ModelInfo:
         self.gguf = kwargs.pop("gguf", None)
 
         self.inference = kwargs.pop("inference", None)
-        self.inference_provider_mapping = kwargs.pop("inferenceProviderMapping", None)
-        if self.inference_provider_mapping:
-            self.inference_provider_mapping = {
-                provider: InferenceProviderMapping(
-                    **{**value, "hf_model_id": self.id}
-                )  # little hack to simplify Inference Providers logic
-                for provider, value in self.inference_provider_mapping.items()
-            }
+
+        # little hack to simplify Inference Providers logic and make it backward and forward compatible
+        # right now, API returns a dict on model_info and a list on list_models. Let's harmonize to list.
+        mapping = kwargs.pop("inferenceProviderMapping", None)
+        if isinstance(mapping, list):
+            self.inference_provider_mapping = [
+                InferenceProviderMapping(**{**value, "hf_model_id": self.id}) for value in mapping
+            ]
+        elif isinstance(mapping, dict):
+            self.inference_provider_mapping = [
+                InferenceProviderMapping(**{**value, "hf_model_id": self.id, "provider": provider})
+                for provider, value in mapping.items()
+            ]
+        elif mapping is None:
+            self.inference_provider_mapping = None
+        else:
+            raise ValueError(
+                f"Unexpected type for `inferenceProviderMapping`. Expecting `dict` or `list`. Got {mapping}."
+            )
 
         self.tags = kwargs.pop("tags", None)
         self.pipeline_tag = kwargs.pop("pipeline_tag", None)
@@ -1169,16 +1188,16 @@ class SpaceInfo:
 @dataclass
 class CollectionItem:
     """
-    Contains information about an item of a Collection (model, dataset, Space or paper).
+    Contains information about an item of a Collection (model, dataset, Space, paper or collection).
 
     Attributes:
         item_object_id (`str`):
             Unique ID of the item in the collection.
         item_id (`str`):
-            ID of the underlying object on the Hub. Can be either a repo_id or a paper id
-            e.g. `"jbilcke-hf/ai-comic-factory"`, `"2307.09288"`.
+            ID of the underlying object on the Hub. Can be either a repo_id, a paper id or a collection slug.
+            e.g. `"jbilcke-hf/ai-comic-factory"`, `"2307.09288"`, `"celinah/cerebras-function-calling-682607169c35fbfa98b30b9a"`.
         item_type (`str`):
-            Type of the underlying object. Can be one of `"model"`, `"dataset"`, `"space"` or `"paper"`.
+            Type of the underlying object. Can be one of `"model"`, `"dataset"`, `"space"`, `"paper"` or `"collection"`.
         position (`int`):
             Position of the item in the collection.
         note (`str`, *optional*):
@@ -1192,10 +1211,20 @@ class CollectionItem:
     note: Optional[str] = None
 
     def __init__(
-        self, _id: str, id: str, type: CollectionItemType_T, position: int, note: Optional[Dict] = None, **kwargs
+        self,
+        _id: str,
+        id: str,
+        type: CollectionItemType_T,
+        position: int,
+        note: Optional[Dict] = None,
+        **kwargs,
     ) -> None:
         self.item_object_id: str = _id  # id in database
         self.item_id: str = id  # repo_id or paper id
+        # if the item is a collection, override item_id with the slug
+        slug = kwargs.get("slug")
+        if slug is not None:
+            self.item_id = slug  # collection slug
         self.item_type: CollectionItemType_T = type
         self.position: int = position
         self.note: str = note["text"] if note is not None else None
@@ -1825,7 +1854,8 @@ class HfApi:
         filter: Union[str, Iterable[str], None] = None,
         author: Optional[str] = None,
         gated: Optional[bool] = None,
-        inference: Optional[Literal["cold", "frozen", "warm"]] = None,
+        inference: Optional[Literal["warm"]] = None,
+        inference_provider: Optional[Union[Literal["all"], "PROVIDER_T", List["PROVIDER_T"]]] = None,
         library: Optional[Union[str, List[str]]] = None,
         language: Optional[Union[str, List[str]]] = None,
         model_name: Optional[str] = None,
@@ -1859,10 +1889,11 @@ class HfApi:
                 A boolean to filter models on the Hub that are gated or not. By default, all models are returned.
                 If `gated=True` is passed, only gated models are returned.
                 If `gated=False` is passed, only non-gated models are returned.
-            inference (`Literal["cold", "frozen", "warm"]`, *optional*):
-                A string to filter models on the Hub by their state on the Inference API.
-                Warm models are available for immediate use. Cold models will be loaded on first inference call.
-                Frozen models are not available in Inference API.
+            inference (`Literal["warm"]`, *optional*):
+                If "warm", filter models on the Hub currently served by at least one provider.
+            inference_provider (`Literal["all"]` or `str`, *optional*):
+                A string to filter models on the Hub that are served by a specific provider.
+                Pass `"all"` to get all models served by at least one provider.
             library (`str` or `List`, *optional*):
                 A string or list of strings of foundational libraries models were
                 originally trained from, such as pytorch, tensorflow, or allennlp.
@@ -1922,7 +1953,7 @@ class HfApi:
         Returns:
             `Iterable[ModelInfo]`: an iterable of [`huggingface_hub.hf_api.ModelInfo`] objects.
 
-        Example usage with the `filter` argument:
+        Example:
 
         ```python
         >>> from huggingface_hub import HfApi
@@ -1932,24 +1963,19 @@ class HfApi:
         # List all models
         >>> api.list_models()
 
-        # List only the text classification models
+        # List text classification models
         >>> api.list_models(filter="text-classification")
 
-        # List only models from the AllenNLP library
-        >>> api.list_models(filter="allennlp")
-        ```
-
-        Example usage with the `search` argument:
+        # List models from the KerasHub library
+        >>> api.list_models(filter="keras-hub")
 
-        ```python
-        >>> from huggingface_hub import HfApi
-
-        >>> api = HfApi()
+        # List models served by Cohere
+        >>> api.list_models(inference_provider="cohere")
 
-        # List all models with "bert" in their name
+        # List models with "bert" in their name
         >>> api.list_models(search="bert")
 
-        # List all models with "bert" in their name made by google
+        # List models with "bert" in their name and pushed by google
         >>> api.list_models(search="bert", author="google")
         ```
         """
@@ -1992,6 +2018,8 @@ class HfApi:
             params["gated"] = gated
         if inference is not None:
             params["inference"] = inference
+        if inference_provider is not None:
+            params["inference_provider"] = inference_provider
         if pipeline_tag:
             params["pipeline_tag"] = pipeline_tag
         search_list = []
@@ -4482,7 +4510,7 @@ class HfApi:
             isinstance(addition.path_or_fileobj, io.BufferedIOBase) for addition in new_lfs_additions_to_upload
         )
         if xet_enabled and not has_buffered_io_data and is_xet_available():
-            logger.info("Uploading files using Xet Storage..")
+            logger.debug("Uploading files using Xet Storage..")
             _upload_xet_files(**upload_kwargs, create_pr=create_pr)  # type: ignore [arg-type]
         else:
             if xet_enabled and is_xet_available():
@@ -5523,7 +5551,7 @@ class HfApi:
         allow_patterns: Optional[Union[List[str], str]] = None,
         ignore_patterns: Optional[Union[List[str], str]] = None,
         max_workers: int = 8,
-        tqdm_class: Optional[base_tqdm] = None,
+        tqdm_class: Optional[Type[base_tqdm]] = None,
         # Deprecated args
         local_dir_use_symlinks: Union[bool, Literal["auto"]] = "auto",
         resume_download: Optional[bool] = None,

huggingface_hub/inference/_client.py

@@ -134,7 +134,7 @@ class InferenceClient:
             path will be appended to the base URL (see the [TGI Messages API](https://huggingface.co/docs/text-generation-inference/en/messages_api)
             documentation for details). When passing a URL as `model`, the client will not append any suffix path to it.
         provider (`str`, *optional*):
-            Name of the provider to use for inference. Can be `"black-forest-labs"`, `"cerebras"`, `"cohere"`, `"fal-ai"`, `"fireworks-ai"`, `"hf-inference"`, `"hyperbolic"`, `"nebius"`, `"novita"`, `"nscale"`, `"openai"`, `"replicate"`, "sambanova"` or `"together"`.
+            Name of the provider to use for inference. Can be `"black-forest-labs"`, `"cerebras"`, `"cohere"`, `"fal-ai"`, `"featherless-ai"`, `"fireworks-ai"`, `"groq"`, `"hf-inference"`, `"hyperbolic"`, `"nebius"`, `"novita"`, `"nscale"`, `"openai"`, `"replicate"`, "sambanova"` or `"together"`.
             Defaults to "auto" i.e. the first of the providers available for the model, sorted by the user's order in https://hf.co/settings/inference-providers.
             If model is a URL or `base_url` is passed, then `provider` is not used.
         token (`str`, *optional*):
@@ -1685,9 +1685,8 @@ class InferenceClient:
         model_id = model or self.model
         provider_helper = get_provider_helper(self.provider, task="table-question-answering", model=model_id)
         request_parameters = provider_helper.prepare_request(
-            inputs=None,
+            inputs={"query": query, "table": table},
             parameters={"model": model, "padding": padding, "sequential": sequential, "truncation": truncation},
-            extra_payload={"query": query, "table": table},
             headers=self.headers,
             model=model_id,
             api_key=self.token,
@@ -3196,7 +3195,7 @@ class InferenceClient:
         return ZeroShotImageClassificationOutputElement.parse_obj_as_list(response)
 
     @_deprecate_method(
-        version="0.33.0",
+        version="0.35.0",
         message=(
             "HF Inference API is getting revamped and will only support warm models in the future (no cold start allowed)."
             " Use `HfApi.list_models(..., inference_provider='...')` to list warm models per provider."
@@ -3386,7 +3385,7 @@ class InferenceClient:
         return response.status_code == 200
 
     @_deprecate_method(
-        version="0.33.0",
+        version="0.35.0",
         message=(
             "HF Inference API is getting revamped and will only support warm models in the future (no cold start allowed)."
             " Use `HfApi.model_info` to get the model status both with HF Inference API and external providers."

huggingface_hub/inference/_generated/_async_client.py

@@ -122,7 +122,7 @@ class AsyncInferenceClient:
             path will be appended to the base URL (see the [TGI Messages API](https://huggingface.co/docs/text-generation-inference/en/messages_api)
             documentation for details). When passing a URL as `model`, the client will not append any suffix path to it.
         provider (`str`, *optional*):
-            Name of the provider to use for inference. Can be `"black-forest-labs"`, `"cerebras"`, `"cohere"`, `"fal-ai"`, `"fireworks-ai"`, `"hf-inference"`, `"hyperbolic"`, `"nebius"`, `"novita"`, `"nscale"`, `"openai"`, `"replicate"`, "sambanova"` or `"together"`.
+            Name of the provider to use for inference. Can be `"black-forest-labs"`, `"cerebras"`, `"cohere"`, `"fal-ai"`, `"featherless-ai"`, `"fireworks-ai"`, `"groq"`, `"hf-inference"`, `"hyperbolic"`, `"nebius"`, `"novita"`, `"nscale"`, `"openai"`, `"replicate"`, "sambanova"` or `"together"`.
             Defaults to "auto" i.e. the first of the providers available for the model, sorted by the user's order in https://hf.co/settings/inference-providers.
             If model is a URL or `base_url` is passed, then `provider` is not used.
         token (`str`, *optional*):
@@ -1737,9 +1737,8 @@ class AsyncInferenceClient:
         model_id = model or self.model
         provider_helper = get_provider_helper(self.provider, task="table-question-answering", model=model_id)
         request_parameters = provider_helper.prepare_request(
-            inputs=None,
+            inputs={"query": query, "table": table},
             parameters={"model": model, "padding": padding, "sequential": sequential, "truncation": truncation},
-            extra_payload={"query": query, "table": table},
             headers=self.headers,
             model=model_id,
             api_key=self.token,
@@ -3260,7 +3259,7 @@ class AsyncInferenceClient:
         return ZeroShotImageClassificationOutputElement.parse_obj_as_list(response)
 
     @_deprecate_method(
-        version="0.33.0",
+        version="0.35.0",
         message=(
             "HF Inference API is getting revamped and will only support warm models in the future (no cold start allowed)."
             " Use `HfApi.list_models(..., inference_provider='...')` to list warm models per provider."
@@ -3496,7 +3495,7 @@ class AsyncInferenceClient:
             return response.status == 200
 
     @_deprecate_method(
-        version="0.33.0",
+        version="0.35.0",
         message=(
             "HF Inference API is getting revamped and will only support warm models in the future (no cold start allowed)."
             " Use `HfApi.model_info` to get the model status both with HF Inference API and external providers."

huggingface_hub/inference/_mcp/mcp_client.py

@@ -310,7 +310,19 @@ class MCPClient:
         # Process tool calls one by one
         for tool_call in final_tool_calls.values():
             function_name = tool_call.function.name
-            function_args = json.loads(tool_call.function.arguments or "{}")
+            try:
+                function_args = json.loads(tool_call.function.arguments or "{}")
+            except json.JSONDecodeError as err:
+                tool_message = {
+                    "role": "tool",
+                    "tool_call_id": tool_call.id,
+                    "name": function_name,
+                    "content": f"Invalid JSON generated by the model: {err}",
+                }
+                tool_message_as_obj = ChatCompletionInputMessage.parse_obj_as_instance(tool_message)
+                messages.append(tool_message_as_obj)
+                yield tool_message_as_obj
+                continue  # move to next tool call
 
             tool_message = {"role": "tool", "tool_call_id": tool_call.id, "content": "", "name": function_name}
 
@@ -324,11 +336,13 @@ class MCPClient:
             # Execute tool call with the appropriate session
             session = self.sessions.get(function_name)
             if session is not None:
-                result = await session.call_tool(function_name, function_args)
-                tool_message["content"] = format_result(result)
+                try:
+                    result = await session.call_tool(function_name, function_args)
+                    tool_message["content"] = format_result(result)
+                except Exception as err:
+                    tool_message["content"] = f"Error: MCP tool call failed with error message: {err}"
             else:
-                error_msg = f"Error: No session found for tool: {function_name}"
-                tool_message["content"] = error_msg
+                tool_message["content"] = f"Error: No session found for tool: {function_name}"
 
             # Yield tool message
             tool_message_as_obj = ChatCompletionInputMessage.parse_obj_as_instance(tool_message)

huggingface_hub/inference/_providers/__init__.py

@@ -1,5 +1,9 @@
 from typing import Dict, Literal, Optional, Union
 
+from huggingface_hub.inference._providers.featherless_ai import (
+    FeatherlessConversationalTask,
+    FeatherlessTextGenerationTask,
+)
 from huggingface_hub.utils import logging
 
 from ._common import TaskProviderHelper, _fetch_inference_provider_mapping
@@ -13,6 +17,7 @@ from .fal_ai import (
     FalAITextToVideoTask,
 )
 from .fireworks_ai import FireworksAIConversationalTask
+from .groq import GroqConversationalTask
 from .hf_inference import (
     HFInferenceBinaryInputTask,
     HFInferenceConversational,
@@ -42,7 +47,9 @@ PROVIDER_T = Literal[
     "cerebras",
     "cohere",
     "fal-ai",
+    "featherless-ai",
     "fireworks-ai",
+    "groq",
     "hf-inference",
     "hyperbolic",
     "nebius",
@@ -72,9 +79,16 @@ PROVIDERS: Dict[PROVIDER_T, Dict[str, TaskProviderHelper]] = {
         "text-to-speech": FalAITextToSpeechTask(),
         "text-to-video": FalAITextToVideoTask(),
     },
+    "featherless-ai": {
+        "conversational": FeatherlessConversationalTask(),
+        "text-generation": FeatherlessTextGenerationTask(),
+    },
     "fireworks-ai": {
         "conversational": FireworksAIConversationalTask(),
     },
+    "groq": {
+        "conversational": GroqConversationalTask(),
+    },
     "hf-inference": {
         "text-to-image": HFInferenceTask("text-to-image"),
         "conversational": HFInferenceConversational(),
@@ -174,7 +188,7 @@ def get_provider_helper(
         if model is None:
             raise ValueError("Specifying a model is required when provider is 'auto'")
         provider_mapping = _fetch_inference_provider_mapping(model)
-        provider = next(iter(provider_mapping))
+        provider = next(iter(provider_mapping)).provider
 
     provider_tasks = PROVIDERS.get(provider)  # type: ignore
     if provider_tasks is None:

huggingface_hub/inference/_providers/_common.py

@@ -1,5 +1,5 @@
 from functools import lru_cache
-from typing import Any, Dict, Optional, Union
+from typing import Any, Dict, List, Optional, Union
 
 from huggingface_hub import constants
 from huggingface_hub.hf_api import InferenceProviderMapping
@@ -9,6 +9,7 @@ from huggingface_hub.utils import build_hf_headers, get_token, logging
 
 logger = logging.get_logger(__name__)
 
+
 # Dev purposes only.
 # If you want to try to run inference for a new model locally before it's registered on huggingface.co
 # for a given Inference Provider, you can add it to the following dictionary.
@@ -24,6 +25,7 @@ HARDCODED_MODEL_INFERENCE_MAPPING: Dict[str, Dict[str, InferenceProviderMapping]
     "cohere": {},
     "fal-ai": {},
     "fireworks-ai": {},
+    "groq": {},
     "hf-inference": {},
     "hyperbolic": {},
     "nebius": {},
@@ -124,7 +126,12 @@ class TaskProviderHelper:
         if HARDCODED_MODEL_INFERENCE_MAPPING.get(self.provider, {}).get(model):
             return HARDCODED_MODEL_INFERENCE_MAPPING[self.provider][model]
 
-        provider_mapping = _fetch_inference_provider_mapping(model).get(self.provider)
+        provider_mapping = None
+        for mapping in _fetch_inference_provider_mapping(model):
+            if mapping.provider == self.provider:
+                provider_mapping = mapping
+                break
+
         if provider_mapping is None:
             raise ValueError(f"Model {model} is not supported by provider {self.provider}.")
 
@@ -138,6 +145,11 @@ class TaskProviderHelper:
             logger.warning(
                 f"Model {model} is in staging mode for provider {self.provider}. Meant for test purposes only."
             )
+        if provider_mapping.status == "error":
+            logger.warning(
+                f"Our latest automated health check on model '{model}' for provider '{self.provider}' did not complete successfully.  "
+                "Inference call might fail."
+            )
         return provider_mapping
 
     def _prepare_headers(self, headers: Dict, api_key: str) -> Dict:
@@ -236,7 +248,7 @@ class BaseTextGenerationTask(TaskProviderHelper):
 
 
 @lru_cache(maxsize=None)
-def _fetch_inference_provider_mapping(model: str) -> Dict:
+def _fetch_inference_provider_mapping(model: str) -> List["InferenceProviderMapping"]:
     """
     Fetch provider mappings for a model from the Hub.
     """

huggingface_hub/inference/_providers/featherless_ai.py

@@ -0,0 +1,38 @@
+from typing import Any, Dict, Optional, Union
+
+from huggingface_hub.hf_api import InferenceProviderMapping
+from huggingface_hub.inference._common import RequestParameters, _as_dict
+
+from ._common import BaseConversationalTask, BaseTextGenerationTask, filter_none
+
+
+_PROVIDER = "featherless-ai"
+_BASE_URL = "https://api.featherless.ai"
+
+
+class FeatherlessTextGenerationTask(BaseTextGenerationTask):
+    def __init__(self):
+        super().__init__(provider=_PROVIDER, base_url=_BASE_URL)
+
+    def _prepare_payload_as_dict(
+        self, inputs: Any, parameters: Dict, provider_mapping_info: InferenceProviderMapping
+    ) -> Optional[Dict]:
+        params = filter_none(parameters.copy())
+        params["max_tokens"] = params.pop("max_new_tokens", None)
+
+        return {"prompt": inputs, **params, "model": provider_mapping_info.provider_id}
+
+    def get_response(self, response: Union[bytes, Dict], request_params: Optional[RequestParameters] = None) -> Any:
+        output = _as_dict(response)["choices"][0]
+        return {
+            "generated_text": output["text"],
+            "details": {
+                "finish_reason": output.get("finish_reason"),
+                "seed": output.get("seed"),
+            },
+        }
+
+
+class FeatherlessConversationalTask(BaseConversationalTask):
+    def __init__(self):
+        super().__init__(provider=_PROVIDER, base_url=_BASE_URL)

huggingface_hub/inference/_providers/groq.py

@@ -0,0 +1,9 @@
+from ._common import BaseConversationalTask
+
+
+class GroqConversationalTask(BaseConversationalTask):
+    def __init__(self):
+        super().__init__(provider="groq", base_url="https://api.groq.com")
+
+    def _prepare_route(self, mapped_model: str, api_key: str) -> str:
+        return "/openai/v1/chat/completions"

huggingface_hub/inference/_providers/hf_inference.py

@@ -26,7 +26,9 @@ class HFInferenceTask(TaskProviderHelper):
 
     def _prepare_mapping_info(self, model: Optional[str]) -> InferenceProviderMapping:
         if model is not None and model.startswith(("http://", "https://")):
-            return InferenceProviderMapping(providerId=model, hf_model_id=model, task=self.task, status="live")
+            return InferenceProviderMapping(
+                provider="hf-inference", providerId=model, hf_model_id=model, task=self.task, status="live"
+            )
         model_id = model if model is not None else _fetch_recommended_models().get(self.task)
         if model_id is None:
             raise ValueError(
@@ -34,7 +36,9 @@ class HFInferenceTask(TaskProviderHelper):
                 " explicitly. Visit https://huggingface.co/tasks for more info."
             )
         _check_supported_task(model_id, self.task)
-        return InferenceProviderMapping(providerId=model_id, hf_model_id=model_id, task=self.task, status="live")
+        return InferenceProviderMapping(
+            provider="hf-inference", providerId=model_id, hf_model_id=model_id, task=self.task, status="live"
+        )
 
     def _prepare_url(self, api_key: str, mapped_model: str) -> str:
         # hf-inference provider can handle URLs (e.g. Inference Endpoints or TGI deployment)

huggingface_hub/inference/_providers/openai.py

@@ -20,4 +20,6 @@ class OpenAIConversationalTask(BaseConversationalTask):
     def _prepare_mapping_info(self, model: Optional[str]) -> InferenceProviderMapping:
         if model is None:
             raise ValueError("Please provide an OpenAI model ID, e.g. `gpt-4o` or `o1`.")
-        return InferenceProviderMapping(providerId=model, task="conversational", status="live", hf_model_id=model)
+        return InferenceProviderMapping(
+            provider="openai", providerId=model, task="conversational", status="live", hf_model_id=model
+        )

{huggingface_hub-0.32.5.dist-info → huggingface_hub-0.33.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: huggingface-hub
-Version: 0.32.5
+Version: 0.33.0
 Summary: Client library to download and publish models, datasets and other repos on the huggingface.co hub
 Home-page: https://github.com/huggingface/huggingface_hub
 Author: Hugging Face, Inc.

{huggingface_hub-0.32.5.dist-info → huggingface_hub-0.33.0.dist-info}/RECORD

@@ -1,11 +1,11 @@
-huggingface_hub/__init__.py,sha256=KPUAhpuW3bgEl-TXncI_3l_Vt3qwj4KLyt9Rw8a0LGs,50644
+huggingface_hub/__init__.py,sha256=jv6NeJ3utSer-d4lltLZJ1kWnseFXgt7KBH-BnKY2K8,50644
 huggingface_hub/_commit_api.py,sha256=ZbmuIhFdF8B3F_cvGtxorka7MmIQOk8oBkCtYltnCvI,39456
 huggingface_hub/_commit_scheduler.py,sha256=tfIoO1xWHjTJ6qy6VS6HIoymDycFPg0d6pBSZprrU2U,14679
 huggingface_hub/_inference_endpoints.py,sha256=qXR0utAYRaEWTI8EXzAsDpVDcYpp8bJPEBbcOxRS52E,17413
-huggingface_hub/_local_folder.py,sha256=7Uce_z51D7ZZ58GF7eUOtcq1cCuYQMOEF-W4p85iQTo,16885
+huggingface_hub/_local_folder.py,sha256=9NkNGsyEfTtopfhXbicS2TFIcm9lAzLFqItzYy2h0D4,16915
 huggingface_hub/_login.py,sha256=ssf4viT5BhHI2ZidnSuAZcrwSxzaLOrf8xgRVKuvu_A,20298
 huggingface_hub/_oauth.py,sha256=YNbSSZCNZLiCqwMoYboSAfI3XjEsbyAADJcwgRAdhBc,18802
-huggingface_hub/_snapshot_download.py,sha256=5BvLNm_1DgdBNXNWOP8omK-9unIyqNSVpDNmxfClxFk,16078
+huggingface_hub/_snapshot_download.py,sha256=6XR6z_BWVP484pUX6hzX8JgsqIrKMFGDBqT97qArPS4,16090
 huggingface_hub/_space_api.py,sha256=jb6rF8qLtjaNU12D-8ygAPM26xDiHCu8CHXHowhGTmg,5470
 huggingface_hub/_tensorboard_logger.py,sha256=ZkYcAUiRC8RGL214QUYtp58O8G5tn-HF6DCWha9imcA,8358
 huggingface_hub/_upload_large_folder.py,sha256=elY5Rv2YVJECVpdZ9PM1zdO8kG-jmi8DifLOa7aC3EU,24178
@@ -16,8 +16,8 @@ huggingface_hub/constants.py,sha256=1RdXbeORR-21auyKLsLbOJDIC9Cd70tYEAVWzP64BJc,
 huggingface_hub/dataclasses.py,sha256=sgPdEi2UDprhNPP2PPkiSlzsHdC1WcpwVTLwlHAEcr0,17224
 huggingface_hub/errors.py,sha256=D7Lw0Jjrf8vfmD0B26LEvg-JWkU8Zq0KDPJOzFY4QLw,11201
 huggingface_hub/fastai_utils.py,sha256=DpeH9d-6ut2k_nCAAwglM51XmRmgfbRe2SPifpVL5Yk,16745
-huggingface_hub/file_download.py,sha256=R0nIB8_zG9nE0imhDmsB62r3aGLrUD3iqBhnMtNRv14,78541
-huggingface_hub/hf_api.py,sha256=qZiemvNF7ltFNcfu5JTYyARzfMfrQhW3eVmfP4uoLFI,443010
+huggingface_hub/file_download.py,sha256=qXPRmGRTv1qAA_QwU7CHYusFGCME32ox1yQ6X62_5O8,78542
+huggingface_hub/hf_api.py,sha256=LSteoR6ndvbkS1GortgAlfIZh2HNTZBOhiKUJ4pwgdY,444347
 huggingface_hub/hf_file_system.py,sha256=U6IY_QLNzZfvpsbvKEiakOBS2U6cduZw5t0x8wBPUn4,47531
 huggingface_hub/hub_mixin.py,sha256=LpbggOPIlr7L2QVi3DOfWsGYsde9OMlwxT5LZfcSdSQ,38115
 huggingface_hub/inference_api.py,sha256=b4-NhPSn9b44nYKV8tDKXodmE4JVdEymMWL4CVGkzlE,8323
@@ -43,10 +43,10 @@ huggingface_hub/commands/upload_large_folder.py,sha256=P-EO44JWVl39Ax4b0E0Z873d0
 huggingface_hub/commands/user.py,sha256=_4rjCrP84KqtqCMn-r3YWLuGLrnklOWTdJFVTNFMLuU,7096
 huggingface_hub/commands/version.py,sha256=vfCJn7GO1m-DtDmbdsty8_RTVtnZ7lX6MJsx0Bf4e-s,1266
 huggingface_hub/inference/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-huggingface_hub/inference/_client.py,sha256=9XhzTsC-87iGfRLW0grvGp53f4tE2QvvNCmOsrd1vJU,161538
+huggingface_hub/inference/_client.py,sha256=HQgPsM6LSYXULCzq6z-S0utdijbswhwcwV2kuq4YxkE,161536
 huggingface_hub/inference/_common.py,sha256=iwCkq2fWE1MVoPTeeXN7UN5FZi7g5fZ3K8PHSOCi5dU,14591
 huggingface_hub/inference/_generated/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-huggingface_hub/inference/_generated/_async_client.py,sha256=1rFYv_FcjaDW4F7L-QIYgkbWeFKgW6x3UdcZp1HYhZk,167698
+huggingface_hub/inference/_generated/_async_client.py,sha256=D3Kv_EId7tMoU6ED41_2SHLrw2jlJIW2q0vHx0SxTpk,167696
 huggingface_hub/inference/_generated/types/__init__.py,sha256=qI8Eu9WcBcKhVkLli6YniGHpfiJ9MLqtzmwXX35E7bA,6443
 huggingface_hub/inference/_generated/types/audio_classification.py,sha256=Jg3mzfGhCSH6CfvVvgJSiFpkz6v4nNA0G4LJXacEgNc,1573
 huggingface_hub/inference/_generated/types/audio_to_audio.py,sha256=2Ep4WkePL7oJwcp5nRJqApwviumGHbft9HhXE9XLHj4,891
@@ -85,22 +85,24 @@ huggingface_hub/inference/_mcp/_cli_hacks.py,sha256=cMZirVFe4N0EM9Nzzs9aEmzUBUEB
 huggingface_hub/inference/_mcp/agent.py,sha256=azX9_lsFjNlgsEvRYdKgsmOmpNReWIcbuMeIVWc852k,4264
 huggingface_hub/inference/_mcp/cli.py,sha256=9IKItC1XJ4yzQAKP1iZwpYL1BA56bem2AQlKlB0SGdc,9251
 huggingface_hub/inference/_mcp/constants.py,sha256=tE_V6qcvsmvVoJa4eg04jhoTR2Cx1cNHieY2ENrm1_M,2511
-huggingface_hub/inference/_mcp/mcp_client.py,sha256=jLIw_fkIPCGYbKI-T2C5UkYmCECJRGNGGw7gnVmXQ_s,14130
+huggingface_hub/inference/_mcp/mcp_client.py,sha256=qefnsJOv2B7YS9k68pLRrsEZXF_N09yjuP_5S_kBv68,14839
 huggingface_hub/inference/_mcp/types.py,sha256=JPK7rC9j-abot8pN3xw1UbSv9S2OBSRStjl_cidWs1Q,1247
 huggingface_hub/inference/_mcp/utils.py,sha256=VsRWl0fuSZDS0zNT9n7FOMSlzA0UBbP8p8xWKWDt2Pc,4093
-huggingface_hub/inference/_providers/__init__.py,sha256=IrLTMERrbRuPiVdBQEMK9TMvXrsGId4-u2ucMkG-vTU,7671
-huggingface_hub/inference/_providers/_common.py,sha256=Octgz-PbHw62iW3Oa8rF7rxvBJR0ZmL4ouv3NoX-weE,10131
+huggingface_hub/inference/_providers/__init__.py,sha256=rOaUL8zXKazYMgnPMDxEN7Y3nZwaKsA0gkILLWN1HLg,8116
+huggingface_hub/inference/_providers/_common.py,sha256=V4oDtWGeihEOSBbcn1zWvYmOAyG7ZVZvvAUVE_RFA2Q,10578
 huggingface_hub/inference/_providers/black_forest_labs.py,sha256=wO7qgRyNyrIKlZtvL3vJEbS4-D19kfoXZk6PDh1dTis,2842
 huggingface_hub/inference/_providers/cerebras.py,sha256=QOJ-1U-os7uE7p6eUnn_P_APq-yQhx28be7c3Tq2EuA,210
 huggingface_hub/inference/_providers/cohere.py,sha256=O3tC-qIUL91mx_mE8bOHCtDWcQuKOUauhUoXSUBUCZ8,1253
 huggingface_hub/inference/_providers/fal_ai.py,sha256=gGWPsvQIsuk3kTIXHwpOqA0R1ZsPEo5MYc7OwUoFjxY,7162
+huggingface_hub/inference/_providers/featherless_ai.py,sha256=QxBz-32O4PztxixrIjrfKuTOzvfqyUi-cVsw0Hf_zlY,1382
 huggingface_hub/inference/_providers/fireworks_ai.py,sha256=Id226ITfPkOcFMFzly3MW9l-dZl9l4qizL4JEHWkBFk,1215
-huggingface_hub/inference/_providers/hf_inference.py,sha256=ZfVB6KIWvYA4aQ_R6A0OqTsLXwvBMYVxny8DKz1DVIQ,8406
+huggingface_hub/inference/_providers/groq.py,sha256=JTk2JV4ZOlaohho7zLAFQtk92kGVsPmLJ1hmzcwsqvQ,315
+huggingface_hub/inference/_providers/hf_inference.py,sha256=LRQ_FzmDOCOY0NWIbDHM87DnyE1aVFZuxJleAiAnSvM,8508
 huggingface_hub/inference/_providers/hyperbolic.py,sha256=OQIBi2j3aNvuaSQ8BUK1K1PVeRXdrxc80G-6YmBa-ns,1985
 huggingface_hub/inference/_providers/nebius.py,sha256=VJpTF2JZ58rznc9wxdk-57vwF8sV2vESw_WkXjXqCho,3580
 huggingface_hub/inference/_providers/novita.py,sha256=HGVC8wPraRQUuI5uBoye1Y4Wqe4X116B71GhhbWy5yM,2514
 huggingface_hub/inference/_providers/nscale.py,sha256=qWUsWinQmUbNUqehyKn34tVoWehu8gd-OZ2F4uj2SWM,1802
-huggingface_hub/inference/_providers/openai.py,sha256=2TJPEwcbq1DKPYKB8roJKnMDiXTcCEquSqGPmibc6tQ,1048
+huggingface_hub/inference/_providers/openai.py,sha256=GCVYeNdjWIgpQQ7E_Xv8IebmdhTi0S6WfFosz3nLtps,1089
 huggingface_hub/inference/_providers/replicate.py,sha256=zFQnnAaNmRruqTvZUG_8It8xkKePHLGKRomSkwjrUuk,3157
 huggingface_hub/inference/_providers/sambanova.py,sha256=Unt3H3jr_kgI9vzRjmmW1DFyoEuPkKCcgIIloiOj3j8,2037
 huggingface_hub/inference/_providers/together.py,sha256=KHF19CS3qXS7G1-CwcMiD8Z5wzPKEKi4F2DzqAthbBE,3439
@@ -139,9 +141,9 @@ huggingface_hub/utils/insecure_hashlib.py,sha256=iAaepavFZ5Dhfa5n8KozRfQprKmvcjS
 huggingface_hub/utils/logging.py,sha256=0A8fF1yh3L9Ka_bCDX2ml4U5Ht0tY8Dr3JcbRvWFuwo,4909
 huggingface_hub/utils/sha.py,sha256=OFnNGCba0sNcT2gUwaVCJnldxlltrHHe0DS_PCpV3C4,2134
 huggingface_hub/utils/tqdm.py,sha256=xAKcyfnNHsZ7L09WuEM5Ew5-MDhiahLACbbN2zMmcLs,10671
-huggingface_hub-0.32.5.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-huggingface_hub-0.32.5.dist-info/METADATA,sha256=M-4wlI4JyB4h2UR7CXNVvKr-sMarVD1JZhAU_ue_nt0,14777
-huggingface_hub-0.32.5.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
-huggingface_hub-0.32.5.dist-info/entry_points.txt,sha256=uelw0-fu0kd-CxIuOsR1bsjLIFnAaMQ6AIqluJYDhQw,184
-huggingface_hub-0.32.5.dist-info/top_level.txt,sha256=8KzlQJAY4miUvjAssOAJodqKOw3harNzuiwGQ9qLSSk,16
-huggingface_hub-0.32.5.dist-info/RECORD,,
+huggingface_hub-0.33.0.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+huggingface_hub-0.33.0.dist-info/METADATA,sha256=W36QzG9PlzKu3fSJWL_qWux7SIG-kb0OECNBHNiSWsE,14777
+huggingface_hub-0.33.0.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
+huggingface_hub-0.33.0.dist-info/entry_points.txt,sha256=uelw0-fu0kd-CxIuOsR1bsjLIFnAaMQ6AIqluJYDhQw,184
+huggingface_hub-0.33.0.dist-info/top_level.txt,sha256=8KzlQJAY4miUvjAssOAJodqKOw3harNzuiwGQ9qLSSk,16
+huggingface_hub-0.33.0.dist-info/RECORD,,