PyTorchDistribution الفصل
تكوين توزيع PyTorch.
- توريث
-
azure.ai.ml.entities._job.distribution.DistributionConfigurationPyTorchDistribution
الدالمنشئ
PyTorchDistribution(*, process_count_per_instance: int | None = None, **kwargs: Any)
معلمات الكلمة الأساسية فقط
الاسم | الوصف |
---|---|
process_count_per_instance
|
عدد العمليات لكل عقدة. |
أمثلة
تكوين CommandComponent باستخدام PyTorchDistribution.
from azure.ai.ml import PyTorchDistribution
from azure.ai.ml.entities import CommandComponent
component = CommandComponent(
name="microsoftsamples_torch",
description="This is the PyTorch command component",
inputs={
"component_in_number": {"description": "A number", "type": "number", "default": 10.99},
"component_in_path": {"description": "A path", "type": "uri_folder"},
},
outputs={"component_out_path": {"type": "uri_folder"}},
command="echo Hello World & echo ${{inputs.component_in_number}} & echo ${{inputs.component_in_path}} "
"& echo ${{outputs.component_out_path}}",
environment="AzureML-sklearn-1.0-ubuntu20.04-py38-cpu:33",
distribution=PyTorchDistribution(
process_count_per_instance=2,
),
instance_count=2,
)
المتغيرات
الاسم | الوصف |
---|---|
type
|
يحدد نوع التوزيع. تعيين تلقائيا إلى "pytorch" لهذه الفئة. |
التعاون معنا على GitHub
يمكن العثور على مصدر هذا المحتوى على GitHub حيث يمكنك أيضاً إضافة مشاكل وطلبات سحب ومراجعتها. للحصول على معلومات إضافية، اطلع على دليل المساهم لدينا.
Azure SDK for Python
الملاحظات
https://aka.ms/ContentUserFeedback.
قريبًا: خلال عام 2024، سنتخلص تدريجيًا من GitHub Issues بوصفها آلية إرسال ملاحظات للمحتوى ونستبدلها بنظام ملاحظات جديد. لمزيد من المعلومات، راجعإرسال الملاحظات وعرضها المتعلقة بـ