arize 8.0.0b1__py3-none-any.whl → 8.0.0b4__py3-none-any.whl

arize/__init__.py

@@ -2,6 +2,7 @@
 
 import logging
 from collections.abc import Mapping
+from typing import Literal, cast
 
 from arize._generated.api_client import models
 from arize.client import ArizeClient
@@ -37,7 +38,7 @@ def make_to_df(field_name: str) -> object:
         json_normalize: bool = False,
         convert_dtypes: bool = True,
     ) -> object:
-        """Convert a list of objects to a pandas DataFrame.
+        """Convert a list of objects to a :class:`pandas.DataFrame`.
 
         Behavior:
           - If an item is a Pydantic v2 model, use `.model_dump(by_alias=...)`.
@@ -82,7 +83,11 @@ def make_to_df(field_name: str) -> object:
 
         # Drop None/NaN columns if requested
         if exclude_none in ("any", "all", True):
-            drop_how = "all" if exclude_none is True else exclude_none
+            drop_how: Literal["any", "all"] = (
+                "all"
+                if exclude_none is True
+                else cast("Literal['any', 'all']", exclude_none)
+            )
             df.dropna(axis=1, how=drop_how, inplace=True)
 
         if convert_dtypes:
@@ -92,6 +97,8 @@ def make_to_df(field_name: str) -> object:
     return to_df
 
 
+# Monkey-patch convenience methods onto generated response models
+# Type ignore comments needed: mypy can't verify runtime attribute additions
 models.DatasetsList200Response.to_df = make_to_df("datasets")  # type: ignore[attr-defined]
 models.DatasetsExamplesList200Response.to_df = make_to_df("examples")  # type: ignore[attr-defined]
 models.ExperimentsList200Response.to_df = make_to_df("experiments")  # type: ignore[attr-defined]

arize/_client_factory.py

@@ -0,0 +1,50 @@
+"""Factory for creating and caching the generated OpenAPI client."""
+
+from __future__ import annotations
+
+import threading
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from arize._generated.api_client.api_client import ApiClient
+    from arize.config import SDKConfiguration
+
+
+class GeneratedClientFactory:
+    """Factory for creating and caching generated OpenAPI clients.
+
+    This factory is owned by ArizeClient and provides thread-safe lazy
+    initialization of the OpenAPI client used by various subclients.
+    """
+
+    def __init__(self, sdk_config: SDKConfiguration) -> None:
+        """Initialize the factory.
+
+        Args:
+            sdk_config: SDK configuration containing API settings.
+        """
+        self._sdk_config = sdk_config
+        self._client: ApiClient | None = None
+        self._lock = threading.Lock()
+
+    def get_client(self) -> ApiClient:
+        """Get or create the generated OpenAPI client instance.
+
+        Returns:
+            The shared generated API client instance.
+        """
+        if self._client is not None:
+            return self._client
+
+        with self._lock:
+            if self._client is not None:
+                return self._client
+
+            # Import lazily to avoid extra dependencies at config time
+            from arize._generated import api_client as gen
+
+            cfg = gen.Configuration(host=self._sdk_config.api_url)
+            if self._sdk_config.api_key:
+                cfg.access_token = self._sdk_config.api_key
+            self._client = gen.ApiClient(cfg)
+            return self._client

arize/_exporter/client.py

@@ -1,4 +1,3 @@
-# type: ignore[pb2]
 import logging
 from dataclasses import dataclass
 from datetime import datetime
@@ -41,7 +40,7 @@ class ArizeExportClient:
         batch_id: str = "",
         include_actuals: bool = False,
         stream_chunk_size: int | None = None,
-    ) -> object:
+    ) -> pd.DataFrame:
         """Exports data of a specific model in the Arize platform to a pandas dataframe.
 
         The export covers a defined time interval and model environment, and can
@@ -237,6 +236,22 @@ class ArizeExportClient:
         columns: list | None = None,
         stream_chunk_size: int | None = None,
     ) -> tuple[flight.FlightStreamReader | None, int]:
+        # Validate inputs first before creating logging context
+        validate_input_type(space_id, "space_id", str)
+        validate_input_type(model_id, "model_id", str)
+        validate_input_type(environment, "environment", Environments)
+        validate_input_type(include_actuals, "include_actuals", bool)
+        validate_input_type(start_time, "start_time", datetime)
+        validate_input_type(end_time, "end_time", datetime)
+        validate_input_type(model_version, "model_version", str)
+        validate_input_type(batch_id, "batch_id", str)
+        validate_input_type(where, "where", str)
+        validate_input_type(columns, "columns", list, allow_none=True)
+        validate_input_type(
+            stream_chunk_size, "stream_chunk_size", int, allow_none=True
+        )
+        validate_start_end_time(start_time, end_time)
+
         # Bind common context for this operation
         log = CtxAdapter(
             logger,
@@ -258,20 +273,6 @@ class ArizeExportClient:
             },
         )
         log.debug("Getting stream reader...")
-        validate_input_type(space_id, "space_id", str)
-        validate_input_type(model_id, "model_id", str)
-        validate_input_type(environment, "environment", Environments)
-        validate_input_type(include_actuals, "include_actuals", bool)
-        validate_input_type(start_time, "start_time", datetime)
-        validate_input_type(end_time, "end_time", datetime)
-        validate_input_type(model_version, "model_version", str)
-        validate_input_type(batch_id, "batch_id", str)
-        validate_input_type(where, "where", str)
-        validate_input_type(columns, "columns", list, allow_none=True)
-        validate_input_type(
-            stream_chunk_size, "stream_chunk_size", int, allow_none=True
-        )
-        validate_start_end_time(start_time, end_time)
 
         # Create query descriptor
         query_descriptor = flight_pb2.RecordQueryDescriptor(
@@ -300,7 +301,7 @@ class ArizeExportClient:
         try:
             flight_info = self.flight_client.get_flight_info(
                 flight.FlightDescriptor.for_command(
-                    json_format.MessageToJson(query_descriptor)  # type: ignore
+                    json_format.MessageToJson(query_descriptor)
                 ),
             )
             logger.info("Fetching data...")

arize/_exporter/parsers/tracing_data_parser.py

@@ -1,5 +1,7 @@
 import json
 import logging
+from collections.abc import Callable
+from typing import Any, TypeGuard
 
 import numpy as np
 import pandas as pd
@@ -28,7 +30,10 @@ logger = logging.getLogger(__name__)
 # of the error is on the data; It should not prevent a user from continuing to use the data
 class OtelTracingDataTransformer:
     def _apply_column_transformation(
-        self, df: pd.DataFrame, col_name: str, transform_func: object
+        self,
+        df: pd.DataFrame,
+        col_name: str,
+        transform_func: Callable[[Any], Any],
     ) -> str | None:
         """Apply a transformation to a column and return error message if it fails."""
         try:
@@ -89,7 +94,7 @@ class OtelTracingDataTransformer:
             if col.name in df.columns
         ]
         for col_name in dirty_string_column_names:
-            df[col_name] = df[col_name].apply(self._clean_json_string)
+            df[col_name] = df[col_name].apply(self._clean_json_string)  # type: ignore[arg-type]
 
         # Convert timestamp columns to datetime objects
         timestamp_column_names: list[str] = [
@@ -102,7 +107,7 @@ class OtelTracingDataTransformer:
         ]
         for col_name in timestamp_column_names:
             df[col_name] = df[col_name].apply(
-                self._convert_timestamp_to_datetime
+                self._convert_timestamp_to_datetime  # type: ignore[arg-type]
             )
 
         for err in errors:
@@ -138,7 +143,7 @@ class OtelTracingDataTransformer:
             return None
         return None
 
-    def _is_non_empty_string(self, value: object) -> bool:
+    def _is_non_empty_string(self, value: object) -> TypeGuard[str]:
         return isinstance(value, str) and value != ""
 
     def _deserialize_json_string_to_dict(self, value: str) -> object:

arize/_exporter/validation.py

@@ -35,7 +35,7 @@ def validate_input_value(
     if input in choices:
         return
     raise ValueError(
-        f"{input_name} is {input}, but must be one of {', '.join(choices)}"
+        f"{input_name} is {input}, but must be one of {', '.join(str(c) for c in choices)}"
     )
 
 

arize/_flight/client.py

@@ -1,11 +1,10 @@
-# type: ignore[pb2]
 from __future__ import annotations
 
 import base64
 import logging
 from dataclasses import dataclass, field
 from enum import Enum
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, TypeAlias
 
 from google.protobuf import json_format
 from pyarrow import flight
@@ -20,15 +19,15 @@ from arize.version import __version__
 
 if TYPE_CHECKING:
     import types
-    from collections.abc import Iterable
+    from collections.abc import Iterator
 
     import pandas as pd
     import pyarrow as pa
 
 
-BytesPair = tuple[bytes, bytes]
-Headers = list[BytesPair]
-FlightPostArrowFileResponse = (
+BytesPair: TypeAlias = tuple[bytes, bytes]
+Headers: TypeAlias = list[BytesPair]
+FlightPostArrowFileResponse: TypeAlias = (
     flight_pb2.WriteSpanEvaluationResponse
     | flight_pb2.WriteSpanAnnotationResponse
     | flight_pb2.WriteSpanAttributesMetadataResponse
@@ -90,7 +89,11 @@ class ArizeFlightClient:
     # ---------- Connection management ----------
 
     def _ensure_client(self) -> flight.FlightClient:
-        """Lazily initialize and return the underlying Flight client connection."""
+        """Lazily initialize and return the underlying Flight client connection.
+
+        Returns:
+            flight.FlightClient: The initialized Apache Arrow Flight client.
+        """
         client = object.__getattribute__(self, "_client")
         if client is not None:
             return client
@@ -135,7 +138,11 @@ class ArizeFlightClient:
     # ---------- methods simple passthrough wrappers ----------
 
     def get_flight_info(self, *args: object, **kwargs: object) -> object:
-        """Get flight information. Passthrough to underlying Flight client with auth options."""
+        """Get flight information. Passthrough to underlying Flight client with auth options.
+
+        Returns:
+            object: FlightInfo object containing metadata about the requested data stream.
+        """
         client = self._ensure_client()
         kwargs.setdefault("options", self.call_options)
         return client.get_flight_info(*args, **kwargs)
@@ -146,6 +153,9 @@ class ArizeFlightClient:
         """Retrieve data stream via Flight DoGet.
 
         Passthrough to underlying Flight client with auth options.
+
+        Returns:
+            flight.FlightStreamReader: A stream reader for retrieving Arrow record batches.
         """
         client = self._ensure_client()
         kwargs.setdefault("options", self.call_options)
@@ -153,10 +163,15 @@ class ArizeFlightClient:
 
     def do_put(
         self, *args: object, **kwargs: object
-    ) -> [flight.FlightStreamWriter, flight.FlightMetadataReader]:
+    ) -> tuple[flight.FlightStreamWriter, flight.FlightMetadataReader]:
         """Upload data stream via Flight DoPut.
 
         Passthrough to underlying Flight client with auth options.
+
+        Returns:
+            tuple[flight.FlightStreamWriter, flight.FlightMetadataReader]: A tuple containing
+                a stream writer for uploading Arrow record batches and a metadata reader for
+                receiving server responses.
         """
         client = self._ensure_client()
         kwargs.setdefault("options", self.call_options)
@@ -164,10 +179,13 @@ class ArizeFlightClient:
 
     def do_action(
         self, *args: object, **kwargs: object
-    ) -> Iterable[flight.Result]:
+    ) -> Iterator[flight.Result]:
         """Execute an action via Flight DoAction.
 
         Passthrough to underlying Flight client with auth options.
+
+        Returns:
+            Iterable[flight.Result]: An iterable of Result objects from the action execution.
         """
         client = self._ensure_client()
         kwargs.setdefault("options", self.call_options)
@@ -210,6 +228,10 @@ class ArizeFlightClient:
             FlightRequestType.ANNOTATION,
             FlightRequestType.METADATA,
         ):
+            if project_name is None:
+                raise ValueError(
+                    f"project_name is required for {request_type.name} request type"
+                )
             proto_schema = get_pb_schema_tracing(project_name=project_name)
             base64_schema = base64.b64encode(proto_schema.SerializeToString())
             pa_schema = append_to_pyarrow_metadata(
@@ -260,8 +282,6 @@ class ArizeFlightClient:
             case FlightRequestType.LOG_EXPERIMENT_DATA:
                 res = flight_pb2.PostExperimentDataResponse()
                 res.ParseFromString(flight_response.to_pybytes())
-            case _:
-                raise ValueError(f"Unsupported request_type: {request_type}")
         return res
 
     # ---------- dataset methods ----------
@@ -271,7 +291,7 @@ class ArizeFlightClient:
         space_id: str,
         dataset_name: str,
         pa_table: pa.Table,
-    ) -> str:
+    ) -> str | None:
         """Create a new dataset via Flight DoPut.
 
         Args:
@@ -333,8 +353,8 @@ class ArizeFlightClient:
                 latest version.
 
         Returns:
-            A pandas DataFrame containing the dataset examples with JSON string columns
-            converted to dict objects.
+            :class:`pandas.DataFrame`: A pandas DataFrame containing the dataset examples
+                with JSON string columns converted to dict objects.
 
         Raises:
             RuntimeError: If the Flight request fails.
@@ -374,8 +394,8 @@ class ArizeFlightClient:
             experiment_id: Experiment ID to retrieve runs from.
 
         Returns:
-            A pandas DataFrame containing the experiment runs with JSON string columns
-            converted to dict objects.
+            :class:`pandas.DataFrame`: A pandas DataFrame containing the experiment runs
+                with JSON string columns converted to dict objects.
 
         Raises:
             RuntimeError: If the Flight request fails.

arize/_generated/api_client/api/datasets_api.py

@@ -940,7 +940,7 @@ class DatasetsApi:
         self,
         dataset_id: Annotated[StrictStr, Field(description="The unique identifier of the dataset")],
         dataset_version_id: Annotated[Optional[StrictStr], Field(description="The unique identifier of the dataset version")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=10000, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         _request_timeout: Union[
             None,
             Annotated[StrictFloat, Field(gt=0)],
@@ -1021,7 +1021,7 @@ class DatasetsApi:
         self,
         dataset_id: Annotated[StrictStr, Field(description="The unique identifier of the dataset")],
         dataset_version_id: Annotated[Optional[StrictStr], Field(description="The unique identifier of the dataset version")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=10000, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         _request_timeout: Union[
             None,
             Annotated[StrictFloat, Field(gt=0)],
@@ -1102,7 +1102,7 @@ class DatasetsApi:
         self,
         dataset_id: Annotated[StrictStr, Field(description="The unique identifier of the dataset")],
         dataset_version_id: Annotated[Optional[StrictStr], Field(description="The unique identifier of the dataset version")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=10000, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         _request_timeout: Union[
             None,
             Annotated[StrictFloat, Field(gt=0)],
@@ -1861,7 +1861,7 @@ class DatasetsApi:
     def datasets_list(
         self,
         space_id: Annotated[Optional[StrictStr], Field(description="Filter search results to a particular space ID")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         cursor: Annotated[Optional[StrictStr], Field(description="Opaque pagination cursor returned from a previous response (`pagination.next_cursor`). Treat it as an unreadable token; do not attempt to parse or construct it. ")] = None,
         _request_timeout: Union[
             None,
@@ -1941,7 +1941,7 @@ class DatasetsApi:
     def datasets_list_with_http_info(
         self,
         space_id: Annotated[Optional[StrictStr], Field(description="Filter search results to a particular space ID")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         cursor: Annotated[Optional[StrictStr], Field(description="Opaque pagination cursor returned from a previous response (`pagination.next_cursor`). Treat it as an unreadable token; do not attempt to parse or construct it. ")] = None,
         _request_timeout: Union[
             None,
@@ -2021,7 +2021,7 @@ class DatasetsApi:
     def datasets_list_without_preload_content(
         self,
         space_id: Annotated[Optional[StrictStr], Field(description="Filter search results to a particular space ID")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         cursor: Annotated[Optional[StrictStr], Field(description="Opaque pagination cursor returned from a previous response (`pagination.next_cursor`). Treat it as an unreadable token; do not attempt to parse or construct it. ")] = None,
         _request_timeout: Union[
             None,

arize/_generated/api_client/api/experiments_api.py

@@ -898,7 +898,7 @@ class ExperimentsApi:
     def experiments_list(
         self,
         dataset_id: Annotated[Optional[StrictStr], Field(description="Filter experiments to a particular dataset ID")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         cursor: Annotated[Optional[StrictStr], Field(description="Opaque pagination cursor returned from a previous response (`pagination.next_cursor`). Treat it as an unreadable token; do not attempt to parse or construct it. ")] = None,
         _request_timeout: Union[
             None,
@@ -978,7 +978,7 @@ class ExperimentsApi:
     def experiments_list_with_http_info(
         self,
         dataset_id: Annotated[Optional[StrictStr], Field(description="Filter experiments to a particular dataset ID")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         cursor: Annotated[Optional[StrictStr], Field(description="Opaque pagination cursor returned from a previous response (`pagination.next_cursor`). Treat it as an unreadable token; do not attempt to parse or construct it. ")] = None,
         _request_timeout: Union[
             None,
@@ -1058,7 +1058,7 @@ class ExperimentsApi:
     def experiments_list_without_preload_content(
         self,
         dataset_id: Annotated[Optional[StrictStr], Field(description="Filter experiments to a particular dataset ID")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         cursor: Annotated[Optional[StrictStr], Field(description="Opaque pagination cursor returned from a previous response (`pagination.next_cursor`). Treat it as an unreadable token; do not attempt to parse or construct it. ")] = None,
         _request_timeout: Union[
             None,
@@ -1211,7 +1211,7 @@ class ExperimentsApi:
     def experiments_runs_list(
         self,
         experiment_id: Annotated[StrictStr, Field(description="The unique identifier of the experiment")],
-        limit: Annotated[Optional[Annotated[int, Field(le=10000, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         _request_timeout: Union[
             None,
             Annotated[StrictFloat, Field(gt=0)],
@@ -1288,7 +1288,7 @@ class ExperimentsApi:
     def experiments_runs_list_with_http_info(
         self,
         experiment_id: Annotated[StrictStr, Field(description="The unique identifier of the experiment")],
-        limit: Annotated[Optional[Annotated[int, Field(le=10000, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         _request_timeout: Union[
             None,
             Annotated[StrictFloat, Field(gt=0)],
@@ -1365,7 +1365,7 @@ class ExperimentsApi:
     def experiments_runs_list_without_preload_content(
         self,
         experiment_id: Annotated[StrictStr, Field(description="The unique identifier of the experiment")],
-        limit: Annotated[Optional[Annotated[int, Field(le=10000, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         _request_timeout: Union[
             None,
             Annotated[StrictFloat, Field(gt=0)],

arize/_generated/api_client/api/projects_api.py

@@ -891,7 +891,7 @@ class ProjectsApi:
     def projects_list(
         self,
         space_id: Annotated[Optional[StrictStr], Field(description="Filter search results to a particular space ID")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         cursor: Annotated[Optional[StrictStr], Field(description="Opaque pagination cursor returned from a previous response (`pagination.next_cursor`). Treat it as an unreadable token; do not attempt to parse or construct it. ")] = None,
         _request_timeout: Union[
             None,
@@ -970,7 +970,7 @@ class ProjectsApi:
     def projects_list_with_http_info(
         self,
         space_id: Annotated[Optional[StrictStr], Field(description="Filter search results to a particular space ID")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         cursor: Annotated[Optional[StrictStr], Field(description="Opaque pagination cursor returned from a previous response (`pagination.next_cursor`). Treat it as an unreadable token; do not attempt to parse or construct it. ")] = None,
         _request_timeout: Union[
             None,
@@ -1049,7 +1049,7 @@ class ProjectsApi:
     def projects_list_without_preload_content(
         self,
         space_id: Annotated[Optional[StrictStr], Field(description="Filter search results to a particular space ID")] = None,
-        limit: Annotated[Optional[Annotated[int, Field(le=500, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum items to return")] = None,
         cursor: Annotated[Optional[StrictStr], Field(description="Opaque pagination cursor returned from a previous response (`pagination.next_cursor`). Treat it as an unreadable token; do not attempt to parse or construct it. ")] = None,
         _request_timeout: Union[
             None,

arize/_lazy.py

@@ -1,6 +1,7 @@
 # src/arize/_lazy.py
 from __future__ import annotations
 
+import inspect
 import logging
 import sys
 import threading
@@ -8,6 +9,8 @@ from importlib import import_module
 from typing import TYPE_CHECKING, ClassVar
 
 if TYPE_CHECKING:
+    import types
+
     from arize.config import SDKConfiguration
 
 logger = logging.getLogger(__name__)
@@ -22,6 +25,11 @@ class LazySubclientsMixin:
         self._lazy_cache: dict[str, object] = {}
         self._lazy_lock = threading.Lock()
 
+        # Add generated client factory
+        from arize._client_factory import GeneratedClientFactory
+
+        self._gen_client_factory = GeneratedClientFactory(sdk_config)
+
     def __getattr__(self, name: str) -> object:
         subs = self._SUBCLIENTS
         if name not in subs:
@@ -41,12 +49,18 @@ class LazySubclientsMixin:
             module = _dynamic_import(module_path)
             klass = getattr(module, class_name)
 
-            # Pass sdk_config if the child accepts it; otherwise construct bare.
-            try:
-                instance = klass(sdk_config=self.sdk_config)
-            except TypeError:
-                instance = klass()
-
+            # Determine which parameters this subclient needs
+            # and build kwargs accordingly
+            sig = inspect.signature(klass.__init__)
+            kwargs: dict[str, object] = {}
+            if "sdk_config" in sig.parameters:
+                kwargs["sdk_config"] = self.sdk_config
+            if "generated_client" in sig.parameters:
+                kwargs["generated_client"] = (
+                    self._gen_client_factory.get_client()
+                )
+
+            instance = klass(**kwargs)
             self._lazy_cache[name] = instance
             return instance
 
@@ -58,7 +72,14 @@ class OptionalDependencyError(ImportError): ...
 
 
 def _can_import(module_name: str) -> bool:
-    """Check if a module can be imported without raising an exception."""
+    """Check if a module can be imported without raising an exception.
+
+    Args:
+        module_name: The fully qualified module name to check (e.g., 'numpy', 'sklearn.preprocessing').
+
+    Returns:
+        bool: True if the module can be imported successfully, False otherwise.
+    """
     try:
         import_module(module_name)
     except Exception:
@@ -72,6 +93,18 @@ def require(
     required: tuple[str, ...],
     pkgname: str = "arize",
 ) -> None:
+    """Ensure required optional dependencies are installed, raising an error if missing.
+
+    Args:
+        extra_key: The extras group key for pip install (e.g., 'mimic', 'embeddings').
+            Used in the error message to guide users.
+        required: Tuple of required module names to check for availability.
+        pkgname: The package name for installation instructions. Defaults to 'arize'.
+
+    Raises:
+        OptionalDependencyError: If any of the required modules cannot be imported.
+            The error message includes pip install instructions with the extras group.
+    """
     if not required:
         return
     missing = [p for p in required if not _can_import(p)]
@@ -82,8 +115,24 @@ def require(
         )
 
 
-def _dynamic_import(modname: str, retries: int = 2) -> object:
-    def _attempt_import(remaining_attempts: int) -> object:
+def _dynamic_import(modname: str, retries: int = 2) -> types.ModuleType:
+    """Dynamically import a module with retry logic and sys.modules cleanup on failure.
+
+    Args:
+        modname: The fully qualified module name to import.
+        retries: Number of import attempts to make. Must be > 0. Defaults to 2.
+
+    Returns:
+        types.ModuleType: The successfully imported module.
+
+    Raises:
+        ValueError: If retries is <= 0.
+        ModuleNotFoundError: If the module cannot be found after all retry attempts.
+        ImportError: If the module import fails after all retry attempts.
+        KeyError: If a key error occurs during import after all retry attempts.
+    """
+
+    def _attempt_import(remaining_attempts: int) -> types.ModuleType:
         try:
             return import_module(modname)
         except (ModuleNotFoundError, ImportError, KeyError):
@@ -92,4 +141,6 @@ def _dynamic_import(modname: str, retries: int = 2) -> object:
                 raise
             return _attempt_import(remaining_attempts - 1)
 
-    return _attempt_import(retries) if retries > 0 else None
+    if retries <= 0:
+        raise ValueError(f"retries must be > 0, got {retries}")
+    return _attempt_import(retries)