Type TPUClusterResolver
Namespace tensorflow.distribute.cluster_resolver
Parent ClusterResolver
Interfaces ITPUClusterResolver
Cluster Resolver for Google Cloud TPUs. This is an implementation of cluster resolvers for the Google Cloud TPU
service. As Cloud TPUs are in alpha, you will need to specify a API definition
file for this to consume, in addition to a list of Cloud TPUs in your Google
Cloud Platform project. TPUClusterResolver supports the following distinct environments:
Google Compute Engine
Google Kubernetes Engine
Google internal
Methods
Properties
Public instance methods
string get_job_name()
object get_job_name_dyn()
string get_master()
object get_master_dyn()
Public static methods
TPUClusterResolver NewDyn(object tpu, object zone, object project, ImplicitContainer<T> job_name, object coordinator_name, object coordinator_address, ImplicitContainer<T> credentials, object service, object discovery_url)
Creates a new TPUClusterResolver object. The ClusterResolver will then use the parameters to query the Cloud TPU APIs
for the IP addresses and ports of each Cloud TPU listed.
Parameters
-
object
tpu - A string corresponding to the TPU to use. If the string is an empty string, the string 'local', or a string that begins with 'grpc://' or '/bns', then it is assumed to not correspond with a Cloud TPU and will instead be passed as the session master and no ClusterSpec propagation will be done. In the future, this may also support a list of strings when multiple Cloud TPUs are used.
-
object
zone - Zone where the TPUs are located. If omitted or empty, we will assume that the zone of the TPU is the same as the zone of the GCE VM, which we will try to discover from the GCE metadata service.
-
object
project - Name of the GCP project containing Cloud TPUs. If omitted or empty, we will try to discover the project name of the GCE VM from the GCE metadata service.
-
ImplicitContainer<T>
job_name - Name of the TensorFlow job the TPUs belong to.
-
object
coordinator_name - The name to use for the coordinator. Set to None if the coordinator should not be included in the computed ClusterSpec.
-
object
coordinator_address - The address of the coordinator (typically an ip:port pair). If set to None, a TF server will be started. If coordinator_name is None, a TF server will not be started even if coordinator_address is None.
-
ImplicitContainer<T>
credentials - GCE Credentials. If None, then we use default credentials from the oauth2client
-
object
service - The GCE API object returned by the googleapiclient.discovery function. If you specify a custom service object, then the credentials parameter will be ignored.
-
object
discovery_url - A URL template that points to the location of the discovery service. It should have two parameters {api} and {apiVersion} that when filled in produce an absolute URL to the discovery document for that service. The environment variable 'TPU_API_DISCOVERY_URL' will override this.
Public properties
string environment get;
Returns the current environment which TensorFlow is running in.
object environment_dyn get;
Returns the current environment which TensorFlow is running in.