guidellm 0.4.0a21__py3-none-any.whl → 0.4.0a169__py3-none-any.whl

guidellm-0.4.0a169.dist-info/RECORD

@@ -0,0 +1,95 @@
+guidellm/__init__.py,sha256=1zl-PT9IZJvDfdLSMviPLzhVE3_ZXpizmc9s7UWa6kQ,1206
+guidellm/__main__.py,sha256=uU5K-QV7rHBARdSTwsNRAPLVoTVT5NQ6DoHx7jssZyc,20554
+guidellm/logger.py,sha256=6qGOeff8hOJF6p57Zietq6qr64N7E40CJSQSQcUFgKc,2912
+guidellm/settings.py,sha256=C4miDtWaI5lJ4NBXxfuUitt5-6_FCzZPzM1Bjie9XoA,7283
+guidellm/version.py,sha256=NIzyWA7lNdSpf2MtPJuOjvW5h6E9nGDea2G4nGFDbgY,127
+guidellm/backends/__init__.py,sha256=Ou-SHPHLoHYfRFkaf_LocNxUUHI_DzmXXRGITyCasac,1101
+guidellm/backends/backend.py,sha256=Qz7z3s6rWwdYGVC-CbPvSFKWSsPiJsSFKBFyfvBG6rY,3385
+guidellm/backends/openai.py,sha256=0At-DMt3_kD0Qa788j_HJ1BcFiUh-SFReoYyaTy8k1g,13007
+guidellm/backends/response_handlers.py,sha256=jwoAGzy-BXxYzMroq08dEUA5ooWoscRpGdU5I4cCEUw,17136
+guidellm/benchmark/__init__.py,sha256=E9g3x0Peopsapw6Pkk9AAthUv3tIOaH59Ai6_92xnzM,2337
+guidellm/benchmark/benchmarker.py,sha256=wZNStFlAcygFBPPA5aee9RipXr-jq-GiH8dUUrijwvo,6500
+guidellm/benchmark/entrypoints.py,sha256=U_0JvZYCkQ1Z-qkF8tN_JeuhuGn8mINScwi819bdzl8,17962
+guidellm/benchmark/output.py,sha256=jHa7u6wTh_YSBdO2oDo47079KReXHZ-AKB6zgo8SMvg,27308
+guidellm/benchmark/profile.py,sha256=RvQdmVLNLU-V8U8xIXm3vH8tY7Hp_4jNPQe5dombj8g,24007
+guidellm/benchmark/progress.py,sha256=oZqZZ_vInmifBNd490ZTgcCjaGy2_slViEABSWDJgHI,25976
+guidellm/benchmark/schemas.py,sha256=eC6WXTXoY3-6Q4SSBLs7VtgfLSilwP9FhsHdRH-6fr4,81249
+guidellm/benchmark/scenarios/__init__.py,sha256=SmaYf8hfByJU4LVJ7pZKNxJPYBObl7UKpoaJEmLPdTI,1276
+guidellm/benchmark/scenarios/chat.json,sha256=4H_ByPCv_9azHn6iTxCY3FfpoUtlbShDPdNyzDwHJVQ,226
+guidellm/benchmark/scenarios/rag.json,sha256=BIpifJoAtWgB3NRRYK51ZuCH4Zvh1OeBFanB7vcxS-E,231
+guidellm/data/__init__.py,sha256=0-7B9vBgY6QHT4svxnBhGN4DoA4jE_9HZxOgndyOiUM,743
+guidellm/data/collators.py,sha256=j4OLGqwbt4sNnTqE8iSbe11qmgJBHnyWjipNeEz1SCk,445
+guidellm/data/loaders.py,sha256=4JxpnOXD-Oju9TcyBumW2JRLFqMo8-EIHN5DpKZJmCA,5413
+guidellm/data/processor.py,sha256=HjSdXKJ67Y2tqcxNoP4LWkefNDLzuN_5nM8MXW_RVec,854
+guidellm/data/schemas.py,sha256=6KUv2OPW_f369SzEhyPBXPDMGoQtJzEPJalBCpp3f84,280
+guidellm/data/deserializers/__init__.py,sha256=NhfbgebOun2FgWTNPydto2f3LNSTlOqdaxeFpDvQUhY,1608
+guidellm/data/deserializers/deserializer.py,sha256=SOCtXik1fVeS7yHgNFMihkq3RmqA-klgmLJULVmSjFU,5007
+guidellm/data/deserializers/file.py,sha256=PzFOJcPuXrUM-OK6EbDIKhzEXDPp7X4xfe_wHZxXqKw,7445
+guidellm/data/deserializers/huggingface.py,sha256=uk2WBH9WJ8csaRrw6wXQ5Xiz_pj_fN5c4jDV9pWf1F0,2954
+guidellm/data/deserializers/memory.py,sha256=F6o2JwIUgcZHdeRkT051AS76i6wWlIw-XGH09_pOqDs,6670
+guidellm/data/deserializers/synthetic.py,sha256=Gcx39gwW0ZvFQSplMfT4ULXfGAVp2v7YiByKmRH5C7Y,12188
+guidellm/data/preprocessors/__init__.py,sha256=khp1-m5EqJ6I40qFAYVv71LncrEXzKBmRocxQG5-ZuE,757
+guidellm/data/preprocessors/formatters.py,sha256=F5BHtJZ6PdmevS9LI6e9TJPwUKnuSsZbt7qS8n2H_eM,14078
+guidellm/data/preprocessors/mappers.py,sha256=7UBdRF2cdADqPbsri_1Mv3FhsQLJtUoIe_lSBV9owEQ,6715
+guidellm/data/preprocessors/preprocessor.py,sha256=OqzAaabEG0v3VSxBfKEVda3ZnTmVjn4OesB2AOC9rhA,812
+guidellm/data/utils/__init__.py,sha256=A8cJqUC7UVSr3upkkD6PzONGQUkjSV9HhMp31NJbI7E,125
+guidellm/data/utils/dataset.py,sha256=ZaKlgGM_L6gcHghAfo5vG0NaHzPtpDWHddD1KX7E7_c,2323
+guidellm/extras/__init__.py,sha256=bNtt6CNDhwMM5XlL1q74j_df-1xoXavTShB05LjDYMw,96
+guidellm/extras/audio.py,sha256=ECDK5IFFBhfae1UQrOGGQCE_7wSCuTySo-TThpm4WfU,6421
+guidellm/extras/vision.py,sha256=hU8e7ryUnMZOT6_utR9GKhayvCXYPljeSwCx8S4-nIQ,7691
+guidellm/mock_server/__init__.py,sha256=oRvGpE8a2U9CUdGnza4GDbShT96NfjOW-cAoh0xDR84,183
+guidellm/mock_server/config.py,sha256=t67sJjFV1aO0YMVrRJEm5ysFe1SzYCm-XVSlgmALPdA,2988
+guidellm/mock_server/models.py,sha256=7CPhbQNSLhRJ4XlAXRIHn0_9yQ8IrGWgoImK-8oUHnk,18946
+guidellm/mock_server/server.py,sha256=Nsc6eADcPRMlMmVZhVA1XML9AYfhR75T90YAH50va1Q,6178
+guidellm/mock_server/utils.py,sha256=NYhLj2dJ4EfC2UQIhgzhr_LInAUi_lYqth7A5QK2Djw,10336
+guidellm/mock_server/handlers/__init__.py,sha256=GX2KD41Uc3H-b338mI4Cf1tK-TZcpTKqKdZH9CPgy8Y,698
+guidellm/mock_server/handlers/chat_completions.py,sha256=xrC0OnPDa5V393u2QmUCIWXVgzqc72YiiyxSMwyZv18,10386
+guidellm/mock_server/handlers/completions.py,sha256=BGTI9tJ1PrldT-Nzz_e7KjEtPFpc0NmLvr3nF-tEYKk,10024
+guidellm/mock_server/handlers/tokenizer.py,sha256=OJAILmsk1tvYfHmdP6iuTf8Fg2gDm2_JyPZH-U-pxFE,5243
+guidellm/preprocess/__init__.py,sha256=6mRs1atYwYkdX4txez_bEVk-_nCDsNt5Wo20eWZ24jA,112
+guidellm/preprocess/dataset.py,sha256=msPfS_--kgEtREgoAVvmyxvUN4aonu6hN9YbS41FJuI,12082
+guidellm/presentation/__init__.py,sha256=tkkHf8ZB_rH-rqnAGY21NuOw2GQq3WCGix_whMqeUSs,483
+guidellm/presentation/builder.py,sha256=eSKsUUx7RbmsqgMRKxcaMuYqquanzS5moc8Uv9TI3Z8,905
+guidellm/presentation/data_models.py,sha256=1CjeHaevj2r4oHLcBQeDY0BQ9VXTol46KULr3F-ps48,7424
+guidellm/presentation/injector.py,sha256=mDo0hvrh4NE2c4RZK5GoegtzWVcvw_zEpvm5sRy7xGE,1834
+guidellm/scheduler/__init__.py,sha256=dj-RZDd5B6H6mt7dPAYGDayqa2k6ngqwfKgRUeCkwDw,2511
+guidellm/scheduler/constraints.py,sha256=a_cDJtmREU1rpbJ1UYxsdH9xW5fnOYd_uLj3VbfStb0,39590
+guidellm/scheduler/environments.py,sha256=W_kp2dICO7Z-NWoji9MwgDu0HjPDhUJNGXyll_LL45Y,8829
+guidellm/scheduler/scheduler.py,sha256=7y3PMAmkNwEYQt-I5ZPkilZZIaw1ad3Hs7z6ZgtQVBw,6986
+guidellm/scheduler/schemas.py,sha256=iVxT0GaDK8q8ruuCL5D4046WrCf7X4v2KeZWfP8gdng,9461
+guidellm/scheduler/strategies.py,sha256=LfYogQCubT3eCKKEUQBWcsBh19kg8J5_9NZz77YAlGc,17796
+guidellm/scheduler/worker.py,sha256=4-RBw1znsR_Fv4O4agT1XEHETp7Fz9TOaOMBRIBm7-U,17157
+guidellm/scheduler/worker_group.py,sha256=czOxgbWJ0BKBqa0HOdqt09-n-xR4VUkiatJHsOCbJio,28597
+guidellm/schemas/__init__.py,sha256=4odN5dEqgRQaxsPpYLnyls0JAlDoEhbWnbYTnPZspN8,879
+guidellm/schemas/info.py,sha256=A_LVqpQteCQvf6XXPi02m1pYX2-vOtCEswLLTZfa9_Y,5678
+guidellm/schemas/request.py,sha256=YoOV6zmHFF9LoNZdF8810H9H3d9BEoNXufUzqTteH7g,7920
+guidellm/schemas/response.py,sha256=ROesx1rDI7g1jRoGbZjmGROazBxcT-3NVwQTIhwI2O4,4578
+guidellm/schemas/stats.py,sha256=4FPdMtoAVv-vQMLTaWYgIcqof7z6_nLHxYGV_lD1L9g,7507
+guidellm/utils/__init__.py,sha256=XGBV3fdETLihLn97_Sd0KM1B4hneoe3d1Oh0nMKObv8,3040
+guidellm/utils/auto_importer.py,sha256=rkraMx815TasixoFn0bwtp--7V7TxuEvfZUVFB8V5L0,3658
+guidellm/utils/cli.py,sha256=oMqLEw2pk9aCYQEuwmSsTO349Mil5Q6CKwocW3akmLo,4343
+guidellm/utils/colors.py,sha256=D0IGz8A346-Pt5qgnP3S5uV-VgngJoXbfToVCOna41k,175
+guidellm/utils/console.py,sha256=IC9vZ0PpwW9SxReZA3BGXyNRK20tdV0FDNuUmbMwUlE,4382
+guidellm/utils/default_group.py,sha256=iZ47bwRcUCxkX04Zdg0qpmqKtFg4P7lt5_hpw1CnKkA,4167
+guidellm/utils/dict.py,sha256=oogh34_NznFEn1L6NKY2RDVBm7TUK9LOZfMc-rquNw8,673
+guidellm/utils/encoding.py,sha256=hf__oDU-iQYKg_FDqYptyV4BxVC6RseIn3OBgaZXBzE,27821
+guidellm/utils/functions.py,sha256=Q4fLePI12yboNb6mx9p5oYC92PPGUZ2o1lBQT690-ro,4316
+guidellm/utils/hf_datasets.py,sha256=M4uESvC08SzOekQUzUFRcnS49qBYJCAubElqURucU8w,992
+guidellm/utils/hf_transformers.py,sha256=EXOctGUB-ZkwS1yrhIpSOK9IGJTaxXXZO0kUR1qXhp4,992
+guidellm/utils/imports.py,sha256=Ch7TCnsvvLW-2ExAERxj2DUG1Dthl4KefYj1bYSAlTs,179
+guidellm/utils/messaging.py,sha256=OLGC6aunhulC0-aKj6aI5VKlRoXQETulcO-XGqjlQg8,45566
+guidellm/utils/mixins.py,sha256=i48rD2FVm-2qXKq7ENCTnvYZgr0IXOrVSJpkh-8-a_Q,4202
+guidellm/utils/pydantic_utils.py,sha256=sP9mYVZyQpj4OoCGdJTwyKY_vasDYsKSesafZ5RJNBw,14742
+guidellm/utils/random.py,sha256=rDy1lpJ9vYMM59DYgKeHT8IG_I7fnjoHjNfD8QIF03k,1273
+guidellm/utils/registry.py,sha256=1yS3_4s4Zkvq0L2bnouTIwxhUT2nZwJ021LcVuRooaE,7644
+guidellm/utils/singleton.py,sha256=yjpUPAtRRvOh63Ubg9ivjQjLdwsExOQM6U1nSQvZuTc,5026
+guidellm/utils/statistics.py,sha256=KzUYm4fVNVtDd6FRCRBnqYmFcea-9n0JKCAZyqeZLM8,40006
+guidellm/utils/synchronous.py,sha256=rRkWwbDf1ty607KUhDKsqV4HcdKU5o0-1s5hwdG-Hak,5209
+guidellm/utils/text.py,sha256=0K8yUEB4gzztevxzuiMXossSoHhvzcHoKqRhQYQdOrg,11644
+guidellm/utils/typing.py,sha256=jt0o7SRbDhnvrifR3l4hN8oL3uJNxl8aMnvaoABb-MU,1235
+guidellm-0.4.0a169.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+guidellm-0.4.0a169.dist-info/METADATA,sha256=J9ycRV41DAnX_ZNLS54X6A8DrNlGzujtwfpLJqom6DE,21923
+guidellm-0.4.0a169.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+guidellm-0.4.0a169.dist-info/entry_points.txt,sha256=DzLFEg47fF7qY1b-9laPz9jg0KSKJ1_D9TbF93kLz_E,51
+guidellm-0.4.0a169.dist-info/top_level.txt,sha256=EXRGjnvFtL6MeZTe0tnHRMYcEWUW3vEqoG2zO7vFOtk,9
+guidellm-0.4.0a169.dist-info/RECORD,,

guidellm/backend/__init__.py

@@ -1,23 +0,0 @@
-from .backend import (
-    Backend,
-    BackendType,
-)
-from .openai import CHAT_COMPLETIONS_PATH, TEXT_COMPLETIONS_PATH, OpenAIHTTPBackend
-from .response import (
-    RequestArgs,
-    ResponseSummary,
-    StreamingResponseType,
-    StreamingTextResponse,
-)
-
-__all__ = [
-    "CHAT_COMPLETIONS_PATH",
-    "TEXT_COMPLETIONS_PATH",
-    "Backend",
-    "BackendType",
-    "OpenAIHTTPBackend",
-    "RequestArgs",
-    "ResponseSummary",
-    "StreamingResponseType",
-    "StreamingTextResponse",
-]

guidellm/backend/backend.py

@@ -1,259 +0,0 @@
-from abc import ABC, abstractmethod
-from collections.abc import AsyncGenerator
-from pathlib import Path
-from typing import Any, Literal, Optional, Union
-
-from loguru import logger
-from PIL import Image
-
-from guidellm.backend.response import ResponseSummary, StreamingTextResponse
-from guidellm.config import settings
-
-__all__ = [
-    "Backend",
-    "BackendType",
-]
-
-
-BackendType = Literal["openai_http"]
-
-
-class Backend(ABC):
-    """
-    Abstract base class for generative AI backends.
-
-    This class provides a common interface for creating and interacting with different
-    generative AI backends. Subclasses should implement the abstract methods to
-    define specific backend behavior.
-
-    :cvar _registry: A registration dictionary that maps BackendType to backend classes.
-    :param type_: The type of the backend.
-    """
-
-    _registry: dict[BackendType, "type[Backend]"] = {}
-
-    @classmethod
-    def register(cls, backend_type: BackendType):
-        """
-        A decorator to register a backend class in the backend registry.
-
-        :param backend_type: The type of backend to register.
-        :type backend_type: BackendType
-        :return: The decorated backend class.
-        :rtype: Type[Backend]
-        """
-        if backend_type in cls._registry:
-            raise ValueError(f"Backend type already registered: {backend_type}")
-
-        if not issubclass(cls, Backend):
-            raise TypeError("Only subclasses of Backend can be registered")
-
-        def inner_wrapper(wrapped_class: type["Backend"]):
-            cls._registry[backend_type] = wrapped_class
-            logger.info("Registered backend type: {}", backend_type)
-            return wrapped_class
-
-        return inner_wrapper
-
-    @classmethod
-    def create(cls, type_: BackendType, **kwargs) -> "Backend":
-        """
-        Factory method to create a backend instance based on the backend type.
-
-        :param type_: The type of backend to create.
-        :type type_: BackendType
-        :param kwargs: Additional arguments for backend initialization.
-        :return: An instance of a subclass of Backend.
-        :rtype: Backend
-        :raises ValueError: If the backend type is not registered.
-        """
-
-        logger.info("Creating backend of type {}", type_)
-
-        if type_ not in cls._registry:
-            err = ValueError(f"Unsupported backend type: {type_}")
-            logger.error("{}", err)
-            raise err
-
-        return Backend._registry[type_](**kwargs)
-
-    def __init__(self, type_: BackendType):
-        self._type = type_
-
-    @property
-    def type_(self) -> BackendType:
-        """
-        :return: The type of the backend.
-        """
-        return self._type
-
-    @property
-    @abstractmethod
-    def target(self) -> str:
-        """
-        :return: The target location for the backend.
-        """
-        ...
-
-    @property
-    @abstractmethod
-    def model(self) -> Optional[str]:
-        """
-        :return: The model used for the backend requests.
-        """
-        ...
-
-    @property
-    @abstractmethod
-    def info(self) -> dict[str, Any]:
-        """
-        :return: The information about the backend.
-        """
-        ...
-
-    @abstractmethod
-    async def reset(self) -> None:
-        """
-        Reset the connection object. This is useful for backends that
-        reuse connections or have state that needs to be cleared.
-        """
-        ...
-
-    async def validate(self):
-        """
-        Handle final setup and validate the backend is ready for use.
-        If not successful, raises the appropriate exception.
-        """
-        logger.info("{} validating backend {}", self.__class__.__name__, self.type_)
-        await self.check_setup()
-        models = await self.available_models()
-        if not models:
-            raise ValueError("No models available for the backend")
-
-        # Use the preferred route defined in the global settings when performing the
-        # validation request. This avoids calling an unavailable endpoint (ie
-        # /v1/completions) when the deployment only supports the chat completions
-        # endpoint.
-        if settings.preferred_route == "chat_completions":
-            async for _ in self.chat_completions(  # type: ignore[attr-defined]
-                content="Test connection", output_token_count=1
-            ):
-                pass
-        else:
-            async for _ in self.text_completions(  # type: ignore[attr-defined]
-                prompt="Test connection", output_token_count=1
-            ):
-                pass
-
-        await self.reset()
-
-    @abstractmethod
-    async def check_setup(self):
-        """
-        Check the setup for the backend.
-        If unsuccessful, raises the appropriate exception.
-
-        :raises ValueError: If the setup check fails.
-        """
-        ...
-
-    @abstractmethod
-    async def prepare_multiprocessing(self):
-        """
-        Prepare the backend for use in a multiprocessing environment.
-        This is useful for backends that have instance state that can not
-        be shared across processes and should be cleared out and re-initialized
-        for each new process.
-        """
-        ...
-
-    @abstractmethod
-    async def available_models(self) -> list[str]:
-        """
-        Get the list of available models for the backend.
-
-        :return: The list of available models.
-        :rtype: List[str]
-        """
-        ...
-
-    @abstractmethod
-    async def text_completions(
-        self,
-        prompt: Union[str, list[str]],
-        request_id: Optional[str] = None,
-        prompt_token_count: Optional[int] = None,
-        output_token_count: Optional[int] = None,
-        **kwargs,
-    ) -> AsyncGenerator[Union[StreamingTextResponse, ResponseSummary], None]:
-        """
-        Generate text only completions for the given prompt.
-        Does not support multiple modalities, complicated chat interfaces,
-        or chat templates. Specifically, it requests with only the prompt.
-
-        :param prompt: The prompt (or list of prompts) to generate a completion for.
-            If a list is supplied, these are concatenated and run through the model
-            for a single prompt.
-        :param request_id: The unique identifier for the request, if any.
-            Added to logging statements and the response for tracking purposes.
-        :param prompt_token_count: The number of tokens measured in the prompt, if any.
-            Returned in the response stats for later analysis, if applicable.
-        :param output_token_count: If supplied, the number of tokens to enforce
-            generation of for the output for this request.
-        :param kwargs: Additional keyword arguments to pass with the request.
-        :return: An async generator that yields a StreamingTextResponse for start,
-            a StreamingTextResponse for each received iteration,
-            and a ResponseSummary for the final response.
-        """
-        ...
-
-    @abstractmethod
-    async def chat_completions(
-        self,
-        content: Union[
-            str,
-            list[Union[str, dict[str, Union[str, dict[str, str]]], Path, Image.Image]],
-            Any,
-        ],
-        request_id: Optional[str] = None,
-        prompt_token_count: Optional[int] = None,
-        output_token_count: Optional[int] = None,
-        raw_content: bool = False,
-        **kwargs,
-    ) -> AsyncGenerator[Union[StreamingTextResponse, ResponseSummary], None]:
-        """
-        Generate chat completions for the given content.
-        Supports multiple modalities, complicated chat interfaces, and chat templates.
-        Specifically, it requests with the content, which can be any combination of
-        text, images, and audio provided the target model supports it,
-        and returns the output text. Additionally, any chat templates
-        for the model are applied within the backend.
-
-        :param content: The content (or list of content) to generate a completion for.
-            This supports any combination of text, images, and audio (model dependent).
-            Supported text only request examples:
-                content="Sample prompt", content=["Sample prompt", "Second prompt"],
-                content=[{"type": "text", "value": "Sample prompt"}.
-            Supported text and image request examples:
-                content=["Describe the image", PIL.Image.open("image.jpg")],
-                content=["Describe the image", Path("image.jpg")],
-                content=["Describe the image", {"type": "image_url",
-                    "image_url": {"url": f"data:image/jpeg;base64,{base64_image}"}].
-            Supported text and audio request examples:
-                content=["Transcribe the audio", Path("audio.wav")],
-                content=["Transcribe the audio", {"type": "input_audio",
-                "input_audio": {"data": f"{base64_bytes}", "format": "wav}].
-            Additionally, if raw_content=True then the content is passed directly to the
-            backend without any processing.
-        :param request_id: The unique identifier for the request, if any.
-            Added to logging statements and the response for tracking purposes.
-        :param prompt_token_count: The number of tokens measured in the prompt, if any.
-            Returned in the response stats for later analysis, if applicable.
-        :param output_token_count: If supplied, the number of tokens to enforce
-            generation of for the output for this request.
-        :param kwargs: Additional keyword arguments to pass with the request.
-        :return: An async generator that yields a StreamingTextResponse for start,
-            a StreamingTextResponse for each received iteration,
-            and a ResponseSummary for the final response.
-        """
-        ...