guidellm 0.4.0a173__py3-none-any.whl → 0.4.0a186__py3-none-any.whl

guidellm/__main__.py

@@ -33,7 +33,7 @@ from pydantic import ValidationError
 try:
     import uvloop
 except ImportError:
-    uvloop = None # type: ignore[assignment] # Optional dependency
+    uvloop = None  # type: ignore[assignment] # Optional dependency
 
 from guidellm.backends import BackendType
 from guidellm.benchmark import (
@@ -116,6 +116,7 @@ def benchmark():
 )
 @click.option(
     "--scenario",
+    "-c",
     type=cli_tools.Union(
         click.Path(
             exists=True,
@@ -392,8 +393,10 @@ def run(**kwargs):
     disable_progress = kwargs.pop("disable_progress", False)
 
     try:
+        # Only set CLI args that differ from click defaults
+        new_kwargs = cli_tools.set_if_not_default(click.get_current_context(), **kwargs)
         args = BenchmarkGenerativeTextArgs.create(
-            scenario=kwargs.pop("scenario", None), **kwargs
+            scenario=new_kwargs.pop("scenario", None), **new_kwargs
         )
     except ValidationError as err:
         # Translate pydantic valdation error to click argument error

guidellm/benchmark/schemas.py

@@ -23,7 +23,15 @@ from pathlib import Path
 from typing import Any, ClassVar, Literal, TypeVar, cast
 
 import yaml
-from pydantic import ConfigDict, Field, computed_field, model_serializer
+from pydantic import (
+    ConfigDict,
+    Field,
+    ValidationError,
+    ValidatorFunctionWrapHandler,
+    computed_field,
+    field_validator,
+    model_serializer,
+)
 from torch.utils.data import Sampler
 from transformers import PreTrainedTokenizerBase
 
@@ -1142,7 +1150,8 @@ class GenerativeMetrics(StandardBaseDict):
         )
         request_duration = (
             (request_end_time - request_start_time)
-            if request_end_time and request_start_time else None
+            if request_end_time and request_start_time
+            else None
         )
 
         # Always track concurrency
@@ -1669,7 +1678,7 @@ class GenerativeBenchmark(Benchmark, StandardBaseDict):
         estimated_state: EstimatedBenchmarkState,
         scheduler_state: SchedulerState,
         profile: Profile,
-        requests: Iterable,
+        requests: Iterable,  # noqa: ARG003
         backend: BackendInterface,
         environment: Environment,
         strategy: SchedulingStrategy,
@@ -1818,8 +1827,6 @@ class BenchmarkGenerativeTextArgs(StandardBaseModel):
         else:
             return factory({})  # type: ignore[call-arg] # Confirmed correct at runtime by code above
 
-
-
     model_config = ConfigDict(
         extra="ignore",
         use_enum_values=True,
@@ -1838,7 +1845,7 @@ class BenchmarkGenerativeTextArgs(StandardBaseModel):
     profile: StrategyType | ProfileType | Profile = Field(
         default="sweep", description="Benchmark profile or scheduling strategy type"
     )
-    rate: float | list[float] | None = Field(
+    rate: list[float] | None = Field(
         default=None, description="Request rate(s) for rate-based scheduling"
     )
     # Backend configuration
@@ -1931,6 +1938,26 @@ class BenchmarkGenerativeTextArgs(StandardBaseModel):
         default=None, description="Maximum global error rate (0-1) before stopping"
     )
 
+    @field_validator("data", "data_args", "rate", mode="wrap")
+    @classmethod
+    def single_to_list(
+        cls, value: Any, handler: ValidatorFunctionWrapHandler
+    ) -> list[Any]:
+        """
+        Ensures field is always a list.
+
+        :param value: Input value for the 'data' field
+        :return: List of data sources
+        """
+        try:
+            return handler(value)
+        except ValidationError as err:
+            # If validation fails, try wrapping the value in a list
+            if err.errors()[0]["type"] == "list_type":
+                return handler([value])
+            else:
+                raise
+
     @model_serializer
     def serialize_model(self):
         """

guidellm/data/deserializers/synthetic.py

@@ -9,7 +9,7 @@ from typing import Any
 import yaml
 from datasets import Features, IterableDataset, Value
 from faker import Faker
-from pydantic import ConfigDict, Field, model_validator
+from pydantic import ConfigDict, Field, ValidationError, model_validator
 from transformers import PreTrainedTokenizerBase
 
 from guidellm.data.deserializers.deserializer import (
@@ -242,6 +242,10 @@ class SyntheticTextDatasetDeserializer(DatasetDeserializer):
         if (config := self._load_config_str(data)) is not None:
             return self(config, processor_factory, random_seed, **data_kwargs)
 
+        # Try to parse dict-like data directly
+        if (config := self._load_config_dict(data)) is not None:
+            return self(config, processor_factory, random_seed, **data_kwargs)
+
         if not isinstance(data, SyntheticTextDatasetConfig):
             raise DataNotSupportedError(
                 "Unsupported data for SyntheticTextDatasetDeserializer, "
@@ -266,6 +270,15 @@ class SyntheticTextDatasetDeserializer(DatasetDeserializer):
             ),
         )
 
+    def _load_config_dict(self, data: Any) -> SyntheticTextDatasetConfig | None:
+        if not isinstance(data, dict | list):
+            return None
+
+        try:
+            return SyntheticTextDatasetConfig.model_validate(data)
+        except ValidationError:
+            return None
+
     def _load_config_file(self, data: Any) -> SyntheticTextDatasetConfig | None:
         if (not isinstance(data, str) and not isinstance(data, Path)) or (
             not Path(data).is_file()

{guidellm-0.4.0a173.dist-info → guidellm-0.4.0a186.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: guidellm
-Version: 0.4.0a173
+Version: 0.4.0a186
 Summary: Guidance platform for deploying and managing large language models.
 Author: Red Hat
 License: Apache-2.0

{guidellm-0.4.0a173.dist-info → guidellm-0.4.0a186.dist-info}/RECORD

@@ -1,5 +1,5 @@
 guidellm/__init__.py,sha256=1zl-PT9IZJvDfdLSMviPLzhVE3_ZXpizmc9s7UWa6kQ,1206
-guidellm/__main__.py,sha256=uU5K-QV7rHBARdSTwsNRAPLVoTVT5NQ6DoHx7jssZyc,20554
+guidellm/__main__.py,sha256=4Z8Rl6yEM8iHyKPEbVviCVQ29EYPc-rB_ZRTqzQfKDM,20722
 guidellm/logger.py,sha256=6qGOeff8hOJF6p57Zietq6qr64N7E40CJSQSQcUFgKc,2912
 guidellm/settings.py,sha256=C4miDtWaI5lJ4NBXxfuUitt5-6_FCzZPzM1Bjie9XoA,7283
 guidellm/version.py,sha256=NIzyWA7lNdSpf2MtPJuOjvW5h6E9nGDea2G4nGFDbgY,127
@@ -13,7 +13,7 @@ guidellm/benchmark/entrypoints.py,sha256=U_0JvZYCkQ1Z-qkF8tN_JeuhuGn8mINScwi819b
 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/schemas.py,sha256=WiOVAw48-Edcrb0kNxMMiE0vjlYki1foI3izsp2iVXE,81999
 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
@@ -27,7 +27,7 @@ guidellm/data/deserializers/deserializer.py,sha256=SOCtXik1fVeS7yHgNFMihkq3RmqA-
 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/deserializers/synthetic.py,sha256=2WOE8pURABFyou0vy5h1HNMlA2tIOErZ9EIjS1Du2S8,12686
 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
@@ -87,9 +87,9 @@ guidellm/utils/statistics.py,sha256=KzUYm4fVNVtDd6FRCRBnqYmFcea-9n0JKCAZyqeZLM8,
 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-0.4.0a173.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-guidellm-0.4.0a173.dist-info/METADATA,sha256=tgoCM_1UQSUh_WYNTeUKrZehj6HJUVJnwDzkmCRhEMY,21923
-guidellm-0.4.0a173.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-guidellm-0.4.0a173.dist-info/entry_points.txt,sha256=DzLFEg47fF7qY1b-9laPz9jg0KSKJ1_D9TbF93kLz_E,51
-guidellm-0.4.0a173.dist-info/top_level.txt,sha256=EXRGjnvFtL6MeZTe0tnHRMYcEWUW3vEqoG2zO7vFOtk,9
-guidellm-0.4.0a173.dist-info/RECORD,,
+guidellm-0.4.0a186.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+guidellm-0.4.0a186.dist-info/METADATA,sha256=U1y28unlMVgQ1WzN_4Euq17kt75TrHQQzvQFfgUPtzU,21923
+guidellm-0.4.0a186.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+guidellm-0.4.0a186.dist-info/entry_points.txt,sha256=DzLFEg47fF7qY1b-9laPz9jg0KSKJ1_D9TbF93kLz_E,51
+guidellm-0.4.0a186.dist-info/top_level.txt,sha256=EXRGjnvFtL6MeZTe0tnHRMYcEWUW3vEqoG2zO7vFOtk,9
+guidellm-0.4.0a186.dist-info/RECORD,,