Share via

DatacacheConsumptionConfig Class

  • Reference

Note

This is an experimental class, and may change at any time. Please see https://aka.ms/azuremlexperimental for more information.

Represent how to deliver the datacache to the compute target.

Represent how to deliver the datacache to the compute target.

Inheritance
builtins.object
DatacacheConsumptionConfig

Constructor

DatacacheConsumptionConfig(datacache_store, dataset, mode='mount', replica_count=None, path_on_compute=None, **kwargs)

Parameters

Name Description
datacache_store
Required
DatacacheStore

The datacachestore to be used to cache the dataset.
dataset
Required
FileDataset

The dataset that needs to be cached.
mode
Required
str

Defines how the datacache should be delivered to the compute target. Only mount mode is supported currently.
replica_count
Required
<xref:Int>, Optional, <xref:Defaults to 1>

Number of replicas to be used by the job.
path_on_compute
Required
str, Optional

The target path on the compute to make the data available at.
datacache_store
Required
DatacacheStore

The datacachestore to be used to cache the dataset.
dataset
Required
FileDataset

The dataset that needs to be cached.
mode
Required
str

Defines how the datacache should be delivered to the compute target. Only mount mode is supported currently.
replica_count
Required
<xref:Int>, Optional

Number of replicas to be used by the training job.
path_on_compute
Required
str, Optional

The target path on the compute to make the data available at.

Methods

as_mount

Set the datacache mode to mount from requested number of replicas.

In the submitted run, required replicas will be mounted to local path on the compute target. The local path on the compute can be retrieved from the arguments in the run.

as_mount

Set the datacache mode to mount from requested number of replicas.

In the submitted run, required replicas will be mounted to local path on the compute target. The local path on the compute can be retrieved from the arguments in the run.

as_mount(path_on_compute=None, replica_count=None)

Parameters

Name Description
path_on_compute
str, Optional

The target path on the compute to make the data available at.

Default value: None
replica_count
<xref:Int>

Number of replicas

Default value: None