rasa-pro 3.14.0.dev3__py3-none-any.whl → 3.14.0.dev5__py3-none-any.whl

rasa/shared/core/slots.py

@@ -273,10 +273,14 @@ class Slot(ABC):
         try:
             return rasa.shared.utils.common.class_from_module_path(type_name)
         except (ImportError, AttributeError):
+            known_types = [
+                cls.type_name for cls in rasa.shared.utils.common.all_subclasses(Slot)
+            ]
             raise InvalidSlotTypeException(
                 f"Failed to find slot type, '{type_name}' is neither a known type nor "
                 f"user-defined. If you are creating your own slot type, make "
                 f"sure its module path is correct. "
+                f"Known types: {', '.join(known_types)} "
                 f"You can find all build in types at {DOCS_URL_SLOTS}"
             )
 

rasa/shared/importers/importer.py

@@ -207,6 +207,12 @@ class TrainingDataImporter(ABC):
                 )
             ]
 
+        return TrainingDataImporter.wrap_in_builtins(importers)
+
+    @staticmethod
+    def wrap_in_builtins(
+        importers: List["TrainingDataImporter"],
+    ) -> "TrainingDataImporter":
         return LanguageImporter(
             E2EImporter(
                 FlowSyncImporter(ResponsesSyncImporter(CombinedDataImporter(importers)))

rasa/shared/importers/utils.py

@@ -1,9 +1,37 @@
-from typing import Iterable, List, Optional, Text
+from typing import Any, Dict, Iterable, List, Optional, Text
+
+from pydantic import BaseModel, Field
 
 from rasa.shared.core.domain import Domain
 from rasa.shared.core.flows import FlowsList
+from rasa.shared.core.flows.yaml_flows_io import KEY_FLOWS, get_flows_as_json
 from rasa.shared.core.training_data.structures import StoryGraph
+from rasa.shared.importers.importer import TrainingDataImporter
+from rasa.shared.nlu.training_data.formats.rasa_yaml import RasaYAMLWriter
 from rasa.shared.nlu.training_data.training_data import TrainingData
+from rasa.utils.json_utils import extract_values
+
+
+class CALMUserData(BaseModel):
+    """All pieces that will be uploaded to Rasa Studio."""
+
+    flows: Dict[str, Any] = Field(default_factory=dict)
+    domain: Dict[str, Any] = Field(default_factory=dict)
+    config: Dict[str, Any] = Field(default_factory=dict)
+    endpoints: Dict[str, Any] = Field(default_factory=dict)
+    nlu: Dict[str, Any] = Field(default_factory=dict)
+
+
+DOMAIN_KEYS = [
+    "version",
+    "actions",
+    "responses",
+    "slots",
+    "intents",
+    "entities",
+    "forms",
+    "session_config",
+]
 
 
 def training_data_from_paths(paths: Iterable[Text], language: Text) -> TrainingData:
@@ -34,3 +62,51 @@ def flows_from_paths(files: List[Text]) -> FlowsList:
         )
     flows.validate()
     return flows
+
+
+def extract_calm_import_parts_from_importer(
+    importer: TrainingDataImporter,
+    config: Optional[Dict[str, Any]] = None,
+    endpoints: Optional[Dict[str, Any]] = None,
+) -> CALMUserData:
+    """Extracts CALMUserData from a TrainingDataImporter.
+
+    Args:
+        importer: The training data importer
+        data_paths: The path(s) to the training data for flows
+        config: Optional config dict, if not provided will use importer.get_config()
+        endpoints: Optional endpoints dict, defaults to empty dict
+
+    Returns:
+        CALMUserData containing flows, domain, config, endpoints, and nlu data
+    """
+    # Extract config
+    if config is None:
+        config = importer.get_config()
+
+    # Extract domain
+    domain_from_files = importer.get_user_domain().as_dict()
+    domain = extract_values(domain_from_files, DOMAIN_KEYS)
+
+    # Extract flows
+    flows = importer.get_user_flows()
+    flows_dict = {KEY_FLOWS: get_flows_as_json(flows)}
+
+    # Extract NLU data
+    nlu_data = importer.get_nlu_data()
+    nlu_examples = nlu_data.filter_training_examples(
+        lambda ex: ex.get("intent") in nlu_data.intents
+    )
+    nlu_dict = RasaYAMLWriter().training_data_to_dict(nlu_examples)
+
+    # Use provided endpoints or default to empty dict
+    if endpoints is None:
+        endpoints = {}
+
+    return CALMUserData(
+        flows=flows_dict or {},
+        domain=domain or {},
+        config=config or {},
+        endpoints=endpoints or {},
+        nlu=nlu_dict or {},
+    )

rasa/shared/providers/_utils.py

@@ -1,87 +1,103 @@
 from typing import Any, Dict, Optional
 
+import boto3
 import structlog
-from litellm import validate_environment
+from botocore.exceptions import BotoCoreError, ClientError
 
 from rasa.shared.constants import (
     API_BASE_CONFIG_KEY,
     API_VERSION_CONFIG_KEY,
     AWS_ACCESS_KEY_ID_CONFIG_KEY,
-    AWS_ACCESS_KEY_ID_ENV_VAR,
+    AWS_BEDROCK_PROVIDER,
     AWS_REGION_NAME_CONFIG_KEY,
-    AWS_REGION_NAME_ENV_VAR,
+    AWS_SAGEMAKER_CHAT_PROVIDER,
+    AWS_SAGEMAKER_PROVIDER,
     AWS_SECRET_ACCESS_KEY_CONFIG_KEY,
-    AWS_SECRET_ACCESS_KEY_ENV_VAR,
     AWS_SESSION_TOKEN_CONFIG_KEY,
-    AWS_SESSION_TOKEN_ENV_VAR,
     AZURE_API_BASE_ENV_VAR,
     AZURE_API_VERSION_ENV_VAR,
     DEPLOYMENT_CONFIG_KEY,
 )
 from rasa.shared.exceptions import ProviderClientValidationError
-from rasa.shared.providers.embedding._base_litellm_embedding_client import (
-    _VALIDATE_ENVIRONMENT_MISSING_KEYS_KEY,
-)
+from rasa.shared.utils.io import resolve_environment_variables
 
 structlogger = structlog.get_logger()
 
 
 def validate_aws_setup_for_litellm_clients(
-    litellm_model_name: str, litellm_call_kwargs: dict, source_log: str
+    litellm_model_name: str, litellm_call_kwargs: Dict, source_log: str, provider: str
 ) -> None:
-    """Validates the AWS setup for LiteLLM clients to ensure all required
-    environment variables or corresponding call kwargs are set.
+    """Validates the AWS setup for LiteLLM clients to ensure credentials are set.
 
     Args:
         litellm_model_name (str): The name of the LiteLLM model being validated.
         litellm_call_kwargs (dict): Additional keyword arguments passed to the client,
             which may include configuration values for AWS credentials.
         source_log (str): The source log identifier for structured logging.
+        provider (str): The provider for which the validation is being performed.
 
     Raises:
         ProviderClientValidationError: If any required AWS environment variable
             or corresponding configuration key is missing.
     """
-
-    # Mapping of environment variable names to their corresponding config keys
-    envs_to_args = {
-        AWS_ACCESS_KEY_ID_ENV_VAR: AWS_ACCESS_KEY_ID_CONFIG_KEY,
-        AWS_SECRET_ACCESS_KEY_ENV_VAR: AWS_SECRET_ACCESS_KEY_CONFIG_KEY,
-        AWS_REGION_NAME_ENV_VAR: AWS_REGION_NAME_CONFIG_KEY,
-        AWS_SESSION_TOKEN_ENV_VAR: AWS_SESSION_TOKEN_CONFIG_KEY,
-    }
-
-    # Validate the environment setup for the model
-    validation_info = validate_environment(litellm_model_name)
-    missing_environment_variables = validation_info.get(
-        _VALIDATE_ENVIRONMENT_MISSING_KEYS_KEY, []
+    # expand environment variables if referenced in the config
+    resolved_litellm_call_kwargs: Dict = resolve_environment_variables(
+        litellm_call_kwargs
+    )  # type: ignore[assignment]
+
+    # boto3 only accepts bedrock and sagemaker as valid clients
+    # therefore we need to convert the provider name if it is defined
+    # as sagemaker_chat
+    provider = (
+        AWS_SAGEMAKER_PROVIDER if provider == AWS_SAGEMAKER_CHAT_PROVIDER else provider
     )
-    # Filter out missing environment variables that have been set trough arguments
-    # in extra parameters
-    missing_environment_variables = [
-        missing_env_var
-        for missing_env_var in missing_environment_variables
-        if litellm_call_kwargs.get(envs_to_args.get(missing_env_var)) is None
-    ]
 
-    if missing_environment_variables:
-        missing_environment_details = [
-            (
-                f"'{missing_env_var}' environment variable or "
-                f"'{envs_to_args.get(missing_env_var)}' config key"
-            )
-            for missing_env_var in missing_environment_variables
+    # if the AWS credentials are defined in the endpoints yaml model config,
+    # either as referenced secret env vars or direct values, we need to pass them
+    # to the boto3 client to ensure that the client can connect to the AWS service.
+    additional_kwargs: Dict[str, Any] = {}
+    if AWS_ACCESS_KEY_ID_CONFIG_KEY in resolved_litellm_call_kwargs:
+        additional_kwargs[AWS_ACCESS_KEY_ID_CONFIG_KEY] = resolved_litellm_call_kwargs[
+            AWS_ACCESS_KEY_ID_CONFIG_KEY
         ]
+    if AWS_SECRET_ACCESS_KEY_CONFIG_KEY in resolved_litellm_call_kwargs:
+        additional_kwargs[AWS_SECRET_ACCESS_KEY_CONFIG_KEY] = (
+            resolved_litellm_call_kwargs[AWS_SECRET_ACCESS_KEY_CONFIG_KEY]
+        )
+    if AWS_SESSION_TOKEN_CONFIG_KEY in resolved_litellm_call_kwargs:
+        additional_kwargs[AWS_SESSION_TOKEN_CONFIG_KEY] = resolved_litellm_call_kwargs[
+            AWS_SESSION_TOKEN_CONFIG_KEY
+        ]
+    if AWS_REGION_NAME_CONFIG_KEY in resolved_litellm_call_kwargs:
+        additional_kwargs["region_name"] = resolved_litellm_call_kwargs[
+            AWS_REGION_NAME_CONFIG_KEY
+        ]
+
+    try:
+        # We are using the boto3 client because it can discover the AWS credentials
+        # from the environment variables, credentials file, or IAM roles.
+        # This is necessary to ensure that the client can connect to the AWS service.
+        aws_client = boto3.client(provider, **additional_kwargs)
+
+        # Using different method calls available to different AWS clients
+        # to test the connection
+        if provider == AWS_SAGEMAKER_PROVIDER:
+            aws_client.list_models()
+        elif provider == AWS_BEDROCK_PROVIDER:
+            aws_client.get_model_invocation_logging_configuration()
+
+    except (ClientError, BotoCoreError) as exc:
         event_info = (
-            f"The following environment variables or configuration keys are "
-            f"missing: "
-            f"{', '.join(missing_environment_details)}. "
-            f"These settings are required for API calls."
+            f"Failed to validate AWS setup for LiteLLM clients: {exc}. "
+            f"Ensure that you are using one of the available authentication methods:"
+            f"credentials file, environment variables, or IAM roles. "
+            f"Also, ensure that the AWS region is set correctly. "
         )
         structlogger.error(
-            f"{source_log}.validate_aws_environment_variables",
+            f"{source_log}.validate_aws_credentials_for_litellm_clients",
             event_info=event_info,
-            missing_environment_variables=missing_environment_variables,
+            exception=str(exc),
+            model_name=litellm_model_name,
         )
         raise ProviderClientValidationError(event_info)
 

rasa/shared/providers/embedding/default_litellm_embedding_client.py

@@ -37,6 +37,7 @@ class DefaultLiteLLMEmbeddingClient(_BaseLiteLLMEmbeddingClient):
 
     @classmethod
     def from_config(cls, config: Dict[str, Any]) -> "DefaultLiteLLMEmbeddingClient":
+        """Creates a DefaultLiteLLMEmbeddingClient instance from a config dict."""
         default_config = DefaultLiteLLMClientConfig.from_dict(config)
         return cls(
             model=default_config.model,
@@ -121,6 +122,7 @@ class DefaultLiteLLMEmbeddingClient(_BaseLiteLLMEmbeddingClient):
                 self._litellm_model_name,
                 self._litellm_extra_parameters,
                 "default_litellm_embedding_client",
+                provider=self.provider,
             )
         else:
             super().validate_client_setup()

rasa/shared/providers/llm/_base_litellm_client.py

@@ -21,7 +21,7 @@ from rasa.shared.providers._ssl_verification_utils import (
     ensure_ssl_certificates_for_litellm_non_openai_based_clients,
     ensure_ssl_certificates_for_litellm_openai_based_clients,
 )
-from rasa.shared.providers.llm.llm_response import LLMResponse, LLMToolCall, LLMUsage
+from rasa.shared.providers.llm.llm_response import LLMResponse, LLMUsage
 from rasa.shared.utils.io import resolve_environment_variables, suppress_logs
 
 structlogger = structlog.get_logger()
@@ -126,9 +126,7 @@ class _BaseLiteLLMClient:
             raise ProviderClientValidationError(event_info)
 
     @suppress_logs(log_level=logging.WARNING)
-    def completion(
-        self, messages: Union[List[dict], List[str], str], **kwargs: Any
-    ) -> LLMResponse:
+    def completion(self, messages: Union[List[dict], List[str], str]) -> LLMResponse:
         """Synchronously generate completions for given list of messages.
 
         Args:
@@ -140,7 +138,6 @@ class _BaseLiteLLMClient:
                 - a list of messages. Each message is a string and will be formatted
                     as a user message.
                 - a single message as a string which will be formatted as user message.
-            **kwargs: Additional parameters to pass to the completion call.
 
         Returns:
             List of message completions.
@@ -150,19 +147,15 @@ class _BaseLiteLLMClient:
         """
         try:
             formatted_messages = self._get_formatted_messages(messages)
-            arguments = cast(
-                Dict[str, Any], resolve_environment_variables(self._completion_fn_args)
-            )
-            response = completion(
-                messages=formatted_messages, **{**arguments, **kwargs}
-            )
+            arguments = resolve_environment_variables(self._completion_fn_args)
+            response = completion(messages=formatted_messages, **arguments)
             return self._format_response(response)
         except Exception as e:
             raise ProviderClientAPIException(e)
 
     @suppress_logs(log_level=logging.WARNING)
     async def acompletion(
-        self, messages: Union[List[dict], List[str], str], **kwargs: Any
+        self, messages: Union[List[dict], List[str], str]
     ) -> LLMResponse:
         """Asynchronously generate completions for given list of messages.
 
@@ -175,7 +168,6 @@ class _BaseLiteLLMClient:
                 - a list of messages. Each message is a string and will be formatted
                     as a user message.
                 - a single message as a string which will be formatted as user message.
-            **kwargs: Additional parameters to pass to the completion call.
 
         Returns:
             List of message completions.
@@ -185,12 +177,8 @@ class _BaseLiteLLMClient:
         """
         try:
             formatted_messages = self._get_formatted_messages(messages)
-            arguments = cast(
-                Dict[str, Any], resolve_environment_variables(self._completion_fn_args)
-            )
-            response = await acompletion(
-                messages=formatted_messages, **{**arguments, **kwargs}
-            )
+            arguments = resolve_environment_variables(self._completion_fn_args)
+            response = await acompletion(messages=formatted_messages, **arguments)
             return self._format_response(response)
         except Exception as e:
             message = ""
@@ -258,32 +246,12 @@ class _BaseLiteLLMClient:
                 else 0
             )
             formatted_response.usage = LLMUsage(prompt_tokens, completion_tokens)
-
-        # Extract tool calls from all choices
-        formatted_response.tool_calls = self._extract_tool_calls(response)
-
         structlogger.debug(
             "base_litellm_client.formatted_response",
             formatted_response=formatted_response.to_dict(),
         )
         return formatted_response
 
-    def _extract_tool_calls(self, response: Any) -> List[LLMToolCall]:
-        """Extract tool calls from response choices.
-
-        Args:
-            response: List of response choices from LiteLLM
-
-        Returns:
-            List of LLMToolCall objects, empty if no tool calls found
-        """
-        return [
-            LLMToolCall.from_litellm(tool_call)
-            for choice in response.choices
-            if choice.message.tool_calls
-            for tool_call in choice.message.tool_calls
-        ]
-
     def _format_text_completion_response(self, response: Any) -> LLMResponse:
         """Parses the LiteLLM text completion response to Rasa format."""
         formatted_response = LLMResponse(

rasa/shared/providers/llm/default_litellm_llm_client.py

@@ -39,6 +39,7 @@ class DefaultLiteLLMClient(_BaseLiteLLMClient):
 
     @classmethod
     def from_config(cls, config: Dict[str, Any]) -> DefaultLiteLLMClient:
+        """Creates a DefaultLiteLLMClient instance from a configuration dictionary."""
         default_config = DefaultLiteLLMClientConfig.from_dict(config)
         return cls(
             model=default_config.model,
@@ -110,6 +111,7 @@ class DefaultLiteLLMClient(_BaseLiteLLMClient):
                 self._litellm_model_name,
                 self._litellm_extra_parameters,
                 "default_litellm_llm_client",
+                provider=self.provider,
             )
         else:
             super().validate_client_setup()

rasa/shared/providers/llm/litellm_router_llm_client.py

@@ -122,9 +122,7 @@ class LiteLLMRouterLLMClient(_BaseLiteLLMRouterClient, _BaseLiteLLMClient):
             raise ProviderClientAPIException(e)
 
     @suppress_logs(log_level=logging.WARNING)
-    def completion(
-        self, messages: Union[List[dict], List[str], str], **kwargs: Any
-    ) -> LLMResponse:
+    def completion(self, messages: Union[List[dict], List[str], str]) -> LLMResponse:
         """
         Synchronously generate completions for given list of messages.
 
@@ -142,7 +140,6 @@ class LiteLLMRouterLLMClient(_BaseLiteLLMRouterClient, _BaseLiteLLMClient):
                 - a list of messages. Each message is a string and will be formatted
                     as a user message.
                 - a single message as a string which will be formatted as user message.
-            **kwargs: Additional parameters to pass to the completion call.
         Returns:
             List of message completions.
         Raises:
@@ -153,7 +150,7 @@ class LiteLLMRouterLLMClient(_BaseLiteLLMRouterClient, _BaseLiteLLMClient):
         try:
             formatted_messages = self._format_messages(messages)
             response = self.router_client.completion(
-                messages=formatted_messages, **{**self._completion_fn_args, **kwargs}
+                messages=formatted_messages, **self._completion_fn_args
             )
             return self._format_response(response)
         except Exception as e:
@@ -161,7 +158,7 @@ class LiteLLMRouterLLMClient(_BaseLiteLLMRouterClient, _BaseLiteLLMClient):
 
     @suppress_logs(log_level=logging.WARNING)
     async def acompletion(
-        self, messages: Union[List[dict], List[str], str], **kwargs: Any
+        self, messages: Union[List[dict], List[str], str]
     ) -> LLMResponse:
         """
         Asynchronously generate completions for given list of messages.
@@ -180,7 +177,6 @@ class LiteLLMRouterLLMClient(_BaseLiteLLMRouterClient, _BaseLiteLLMClient):
                 - a list of messages. Each message is a string and will be formatted
                     as a user message.
                 - a single message as a string which will be formatted as user message.
-            **kwargs: Additional parameters to pass to the completion call.
         Returns:
             List of message completions.
         Raises:
@@ -191,7 +187,7 @@ class LiteLLMRouterLLMClient(_BaseLiteLLMRouterClient, _BaseLiteLLMClient):
         try:
             formatted_messages = self._format_messages(messages)
             response = await self.router_client.acompletion(
-                messages=formatted_messages, **{**self._completion_fn_args, **kwargs}
+                messages=formatted_messages, **self._completion_fn_args
             )
             return self._format_response(response)
         except Exception as e:

rasa/shared/providers/llm/llm_client.py

@@ -1,6 +1,6 @@
 from __future__ import annotations
 
-from typing import Any, Dict, List, Protocol, Union, runtime_checkable
+from typing import Dict, List, Protocol, Union, runtime_checkable
 
 from rasa.shared.providers.llm.llm_response import LLMResponse
 
@@ -32,9 +32,7 @@ class LLMClient(Protocol):
         """
         ...
 
-    def completion(
-        self, messages: Union[List[dict], List[str], str], **kwargs: Any
-    ) -> LLMResponse:
+    def completion(self, messages: Union[List[dict], List[str], str]) -> LLMResponse:
         """
         Synchronously generate completions for given list of messages.
 
@@ -50,14 +48,13 @@ class LLMClient(Protocol):
                 - a list of messages. Each message is a string and will be formatted
                     as a user message.
                 - a single message as a string which will be formatted as user message.
-            **kwargs: Additional parameters to pass to the completion call.
         Returns:
             LLMResponse
         """
         ...
 
     async def acompletion(
-        self, messages: Union[List[dict], List[str], str], **kwargs: Any
+        self, messages: Union[List[dict], List[str], str]
     ) -> LLMResponse:
         """
         Asynchronously generate completions for given list of messages.
@@ -74,7 +71,6 @@ class LLMClient(Protocol):
                 - a list of messages. Each message is a string and will be formatted
                     as a user message.
                 - a single message as a string which will be formatted as user message.
-            **kwargs: Additional parameters to pass to the completion call.
         Returns:
             LLMResponse
         """

rasa/shared/providers/llm/llm_response.py

@@ -1,14 +1,9 @@
 import functools
-import json
 import time
 from dataclasses import asdict, dataclass, field
 from typing import Any, Awaitable, Callable, Dict, List, Optional, Text, Union
 
 import structlog
-from litellm.utils import ChatCompletionMessageToolCall
-from pydantic import BaseModel
-
-from rasa.shared.constants import KEY_TOOL_CALLS
 
 structlogger = structlog.get_logger()
 
@@ -43,37 +38,6 @@ class LLMUsage:
         return asdict(self)
 
 
-class LLMToolCall(BaseModel):
-    """A class representing a response from an LLM tool call."""
-
-    id: str
-    """The ID of the tool call."""
-
-    tool_name: str
-    """The name of the tool that was called."""
-
-    tool_args: Dict[str, Any]
-    """The arguments passed to the tool call."""
-
-    type: str = "function"
-    """The type of the tool call."""
-
-    @classmethod
-    def from_dict(cls, data: Dict[Text, Any]) -> "LLMToolCall":
-        """Creates an LLMToolResponse from a dictionary."""
-        return cls(**data)
-
-    @classmethod
-    def from_litellm(cls, data: ChatCompletionMessageToolCall) -> "LLMToolCall":
-        """Creates an LLMToolResponse from a dictionary."""
-        return cls(
-            id=data.id,
-            tool_name=data.function.name,
-            tool_args=json.loads(data.function.arguments),
-            type=data.type,
-        )
-
-
 @dataclass
 class LLMResponse:
     id: str
@@ -98,22 +62,12 @@ class LLMResponse:
     latency: Optional[float] = None
     """Optional field to store the latency of the LLM API call."""
 
-    tool_calls: Optional[List[LLMToolCall]] = None
-    """The list of tool calls the model generated for the input prompt."""
-
     @classmethod
     def from_dict(cls, data: Dict[Text, Any]) -> "LLMResponse":
         """Creates an LLMResponse from a dictionary."""
         usage_data = data.get("usage", {})
         usage_obj = LLMUsage.from_dict(usage_data) if usage_data else None
 
-        tool_calls_data = data.get(KEY_TOOL_CALLS, [])
-        tool_calls_obj = (
-            [LLMToolCall.from_dict(tool) for tool in tool_calls_data]
-            if tool_calls_data
-            else None
-        )
-
         return cls(
             id=data["id"],
             choices=data["choices"],
@@ -122,7 +76,6 @@ class LLMResponse:
             usage=usage_obj,
             additional_info=data.get("additional_info"),
             latency=data.get("latency"),
-            tool_calls=tool_calls_obj,
         )
 
     @classmethod
@@ -139,8 +92,6 @@ class LLMResponse:
         result = asdict(self)
         if self.usage:
             result["usage"] = self.usage.to_dict()
-        if self.tool_calls:
-            result[KEY_TOOL_CALLS] = [tool.model_dump() for tool in self.tool_calls]
         return result
 
 

rasa/shared/providers/llm/self_hosted_llm_client.py

@@ -237,7 +237,7 @@ class SelfHostedLLMClient(_BaseLiteLLMClient):
             raise ProviderClientAPIException(e)
 
     async def acompletion(
-        self, messages: Union[List[dict], List[str], str], **kwargs: Any
+        self, messages: Union[List[dict], List[str], str]
     ) -> LLMResponse:
         """Asynchronous completion of the model with the given messages.
 
@@ -255,18 +255,15 @@ class SelfHostedLLMClient(_BaseLiteLLMClient):
                 - a list of messages. Each message is a string and will be formatted
                     as a user message.
                 - a single message as a string which will be formatted as user message.
-            **kwargs: Additional parameters to pass to the completion call.
 
         Returns:
             The completion response.
         """
         if self._use_chat_completions_endpoint:
-            return await super().acompletion(messages, **kwargs)
+            return await super().acompletion(messages)
         return await self._atext_completion(messages)
 
-    def completion(
-        self, messages: Union[List[dict], List[str], str], **kwargs: Any
-    ) -> LLMResponse:
+    def completion(self, messages: Union[List[dict], List[str], str]) -> LLMResponse:
         """Completion of the model with the given messages.
 
         Method overrides the base class method to call the appropriate
@@ -276,13 +273,12 @@ class SelfHostedLLMClient(_BaseLiteLLMClient):
 
         Args:
             messages: The messages to be used for completion.
-            **kwargs: Additional parameters to pass to the completion call.
 
         Returns:
             The completion response.
         """
         if self._use_chat_completions_endpoint:
-            return super().completion(messages, **kwargs)
+            return super().completion(messages)
         return self._text_completion(messages)
 
     @staticmethod