use two classes for text-generation and conversational

src/huggingface_hub/inference/_providers/__init__.py

@@ -12,7 +12,7 @@
 from .hyperbolic import HyperbolicTextGenerationTask, HyperbolicTextToImageTask
 from .replicate import ReplicateTask, ReplicateTextToSpeechTask
 from .sambanova import SambanovaConversationalTask
-from .together import TogetherTextGenerationTask, TogetherTextToImageTask
+from .together import TogetherConversationalTask, TogetherTextGenerationTask, TogetherTextToImageTask
 
 
 PROVIDER_T = Literal[
@@ -78,8 +78,8 @@
     },
     "together": {
         "text-to-image": TogetherTextToImageTask(),
-        "conversational": TogetherTextGenerationTask("conversational"),
-        "text-generation": TogetherTextGenerationTask("text-generation"),
+        "conversational": TogetherConversationalTask(),
+        "text-generation": TogetherTextGenerationTask(),
     },
 }
 

src/huggingface_hub/inference/_providers/_common.py

@@ -180,23 +180,36 @@ def _prepare_payload_as_bytes(
         return None
 
 
+class BaseConversationalTask(TaskProviderHelper):
+    """
+    Base class for conversational (chat completion) tasks.
+    The schema follows the OpenAI API format defined here: https://platform.openai.com/docs/api-reference/chat
+    """
+
+    def __init__(self, provider: str, base_url: str):
+        super().__init__(provider=provider, base_url=base_url, task="conversational")
+
+    def _prepare_route(self, mapped_model: str) -> str:
+        return "/v1/chat/completions"
+
+    def _prepare_payload_as_dict(self, inputs: Any, parameters: Dict, mapped_model: str) -> Optional[Dict]:
+        return {"messages": inputs, **filter_none(parameters), "model": mapped_model}
+
+
 class BaseTextGenerationTask(TaskProviderHelper):
-    def __init__(self, provider: str, base_url: str, task: str):
-        super().__init__(provider=provider, base_url=base_url, task=task)
+    """
+    Base class for text-generation (completion) tasks.
+    The schema follows the OpenAI API format defined here: https://platform.openai.com/docs/api-reference/completions
+    """
+
+    def __init__(self, provider: str, base_url: str):
+        super().__init__(provider=provider, base_url=base_url, task="text-generation")
 
     def _prepare_route(self, mapped_model: str) -> str:
-        if self.task == "conversational":
-            return "/v1/chat/completions"
-        elif self.task == "text-generation":
-            return "/v1/completions"
-        raise ValueError(f"Unsupported task '{self.task}' for {self.provider}.")
+        return "/v1/completions"
 
     def _prepare_payload_as_dict(self, inputs: Any, parameters: Dict, mapped_model: str) -> Optional[Dict]:
-        if self.task == "conversational":
-            return {"messages": inputs, **filter_none(parameters), "model": mapped_model}
-        elif self.task == "text-generation":
-            return {"prompt": inputs, **filter_none(parameters), "model": mapped_model}
-        raise ValueError(f"Unsupported task '{self.task}' for {self.provider}.")
+        return {"prompt": inputs, **filter_none(parameters), "model": mapped_model}
 
 
 @lru_cache(maxsize=None)

src/huggingface_hub/inference/_providers/fireworks_ai.py

@@ -1,6 +1,6 @@
-from ._common import BaseTextGenerationTask
+from ._common import BaseConversationalTask
 
 
-class FireworksAIConversationalTask(BaseTextGenerationTask):
+class FireworksAIConversationalTask(BaseConversationalTask):
     def __init__(self):
-        super().__init__(provider="fireworks-ai", base_url="https://api.fireworks.ai/inference", task="conversational")
+        super().__init__(provider="fireworks-ai", base_url="https://api.fireworks.ai/inference")

src/huggingface_hub/inference/_providers/hyperbolic.py

@@ -2,7 +2,7 @@
 from typing import Any, Dict, Optional, Union
 
 from huggingface_hub.inference._common import _as_dict
-from huggingface_hub.inference._providers._common import BaseTextGenerationTask, TaskProviderHelper, filter_none
+from huggingface_hub.inference._providers._common import BaseConversationalTask, TaskProviderHelper, filter_none
 
 
 class HyperbolicTextToImageTask(TaskProviderHelper):
@@ -30,12 +30,14 @@ def get_response(self, response: Union[bytes, Dict]) -> Any:
         return base64.b64decode(response_dict["images"][0]["image"])
 
 
-class HyperbolicTextGenerationTask(BaseTextGenerationTask):
-    def __init__(self, task: str):
-        super().__init__(provider="hyperbolic", base_url="https://api.hyperbolic.xyz", task=task)
+class HyperbolicTextGenerationTask(BaseConversationalTask):
+    """
+    Special case for Hyperbolic, where text-generation task is handled as a conversational task.
+    """
 
-    def _prepare_route(self, mapped_model: str) -> str:
-        # For Hyperbolic, the route is the same for text-generation and conversational tasks
-        if self.task in ("text-generation", "conversational"):
-            return "/v1/chat/completions"
-        raise ValueError(f"Unsupported task '{self.task}' for Hyperbolic.")
+    def __init__(self, task: str):
+        super().__init__(
+            provider="hyperbolic",
+            base_url="https://api.hyperbolic.xyz",
+        )
+        self.task = task

src/huggingface_hub/inference/_providers/sambanova.py

@@ -1,6 +1,6 @@
-from huggingface_hub.inference._providers._common import BaseTextGenerationTask
+from huggingface_hub.inference._providers._common import BaseConversationalTask
 
 
-class SambanovaConversationalTask(BaseTextGenerationTask):
+class SambanovaConversationalTask(BaseConversationalTask):
     def __init__(self):
-        super().__init__(provider="sambanova", base_url="https://api.sambanova.ai", task="conversational")
+        super().__init__(provider="sambanova", base_url="https://api.sambanova.ai")

src/huggingface_hub/inference/_providers/together.py

@@ -3,7 +3,12 @@
 from typing import Any, Dict, Optional, Union
 
 from huggingface_hub.inference._common import _as_dict
-from huggingface_hub.inference._providers._common import BaseTextGenerationTask, TaskProviderHelper, filter_none
+from huggingface_hub.inference._providers._common import (
+    BaseConversationalTask,
+    BaseTextGenerationTask,
+    TaskProviderHelper,
+    filter_none,
+)
 
 
 class TogetherTask(TaskProviderHelper, ABC):
@@ -23,8 +28,13 @@ def _prepare_route(self, mapped_model: str) -> str:
 
 
 class TogetherTextGenerationTask(BaseTextGenerationTask):
-    def __init__(self, task: str):
-        super().__init__(provider="together", base_url="https://api.together.xyz", task=task)
+    def __init__(self):
+        super().__init__(provider="together", base_url="https://api.together.xyz")
+
+
+class TogetherConversationalTask(BaseConversationalTask):
+    def __init__(self):
+        super().__init__(provider="together", base_url="https://api.together.xyz")
 
 
 class TogetherTextToImageTask(TogetherTask):

tests/test_inference_providers.py

@@ -22,7 +22,11 @@
 )
 from huggingface_hub.inference._providers.replicate import ReplicateTask, ReplicateTextToSpeechTask
 from huggingface_hub.inference._providers.sambanova import SambanovaConversationalTask
-from huggingface_hub.inference._providers.together import TogetherTextGenerationTask, TogetherTextToImageTask
+from huggingface_hub.inference._providers.together import (
+    TogetherConversationalTask,
+    TogetherTextGenerationTask,
+    TogetherTextToImageTask,
+)
 
 
 class TestFalAIProvider:
@@ -233,9 +237,6 @@ def test_prepare_route(self):
         helper = HyperbolicTextGenerationTask("conversational")
         assert helper._prepare_route("username/repo_name") == "/v1/chat/completions"
 
-        with pytest.raises(ValueError, match="Unsupported task 'invalid-task' for Hyperbolic."):
-            HyperbolicTextGenerationTask("invalid-task")._prepare_route("username/repo_name")
-
     def test_prepare_payload_conversational(self):
         """Test payload preparation for conversational task."""
         helper = HyperbolicTextGenerationTask("conversational")
@@ -357,17 +358,17 @@ def test_prepare_payload_as_dict(self):
 
 class TestTogetherProvider:
     def test_prepare_route(self):
-        helper = TogetherTextGenerationTask("text-generation")
+        helper = TogetherTextGenerationTask()
         assert helper._prepare_route("username/repo_name") == "/v1/completions"
 
-        helper = TogetherTextGenerationTask("conversational")
+        helper = TogetherConversationalTask()
         assert helper._prepare_route("username/repo_name") == "/v1/chat/completions"
 
         helper = TogetherTextToImageTask()
         assert helper._prepare_route("username/repo_name") == "/v1/images/generations"
 
     def test_prepare_payload_as_dict_conversational(self):
-        helper = TogetherTextGenerationTask("conversational")
+        helper = TogetherConversationalTask()
         payload = helper._prepare_payload_as_dict(
             [{"role": "user", "content": "Hello!"}], {}, "meta-llama/Llama-3.1-8B-Instruct"
         )