Q fSrSSKJr SSKJr SSKJr SSKJr SSKJr SSK J r SSK J r SS K J r SS K Jr SS KJr SS KJr SS KJr SSKJr SSKJr SSKJr SSKJr SSKJr SSKrSrSrSrSrSr Sr!Sr"\RF"S\RH"5RJRLSSRO\ \!54SSRO\"54S S!.S"S#S$9r(\RF"S%\RH"5RRRTS&S'S(.S)S#S$9r+S*S+S,S-S.S/S0.r,\RF"S1\RH"5RZR\\,S2SS39r/"S4S5\05r1S6r2S7r3S8r4S9r5S:r6S;r7S<r8S=r9S>r:S?r;SIS@jrSJSCjr?SDr@SKSEjrASFrBSGrCSHrDg)Lzml-engine jobs command code.)absolute_import)division)unicode_literals) exceptions)jobs)stream)flags) jobs_prep) log_utils) arg_utils) labels_util)execution_utils)log) properties) resources)yaml)resource_printerNzIhttps://console.cloud.google.com/mlengine/jobs/{job_id}?project={project}z[https://console.cloud.google.com/logs?resource=ml_job%2Fjob_id%2F{job_id}&project={project}zAyaml(jobId,state,startTime.date(tz=LOCAL),endTime.date(tz=LOCAL)) z=https://www.tensorflow.org/guide/datasets#consuming_text_datazYhttps://cloud.google.com/ai-platform/prediction/docs/overview#batch_prediction_input_datazAhttps://www.tensorflow.org/guide/datasets#consuming_tfrecord_dataz --data-formattextzCText and JSON files; for text files, see {}, for JSON files, see {}z tf-recordzTFRecord files; see {})ztf-record-gzipzGZIP-compressed TFRecord files.)TEXT TF_RECORDTF_RECORD_GZIPzData format of the input files.T)custom_mappingshelp_strrequiredz--accelerator-type)znvidia-tesla-k80zNVIDIA Tesla K80 GPU)znvidia-tesla-p100zNVIDIA Tesla P100 GPU.)NVIDIA_TESLA_K80NVIDIA_TESLA_P100z$The available types of accelerators.)basiczSingle worker instance. This tier is suitable for learning how to use AI Platform, and for experimenting with new models using small datasets.)z standard-1z)Many workers and a few parameter servers.)z premium-1z4Large number of workers with many parameter servers.)z basic-gpuz"Single worker instance with a GPU.)z basic-tpuz(Single worker instance with a Cloud TPU.)customayCUSTOM tier is not a set tier, but rather enables you to use your own cluster specification. When you use this tier, set values to configure your processing cluster according to these guidelines (using the `--config` flag): * You _must_ set `TrainingInput.masterType` to specify the type of machine to use for your master node. This is the only required setting. * You _may_ set `TrainingInput.workerCount` to specify the number of workers to use. If you specify one or more workers, you _must_ also set `TrainingInput.workerType` to specify the type of machine to use for your worker nodes. * You _may_ set `TrainingInput.parameterServerCount` to specify the number of parameter servers to use. If you specify one or more parameter servers, you _must_ also set `TrainingInput.parameterServerType` to specify the type of machine to use for your parameter servers. Note that all of your workers must use the same machine type, which can be different from your parameter server type and master type. Your parameter servers must likewise use the same machine type, which can be different from your worker type and master type.)BASIC STANDARD_1 PREMIUM_1 BASIC_GPU BASIC_TPUCUSTOMz --scale-tierzUSpecify the machine types, the number of replicas for workers, and parameter servers.)rrdefaultcb\rSrSrSrS SjrSrSr\S Sj5r Sr g) TrainingCustomInputServerConfigzzCData class for passing custom server config for training job input.NcX@lX0lXPlX`lXplXlXlXlXlXl Xl Xl Xl UUl UUlXlX lUUlgN)master_image_urimaster_machine_typemaster_accelerator_typemaster_accelerator_countparameter_machine_typeparameter_machine_countparameter_image_uriparameter_accelerator_typeparameter_accelerator_counttpu_tf_versionworker_machine_typeworker_machine_countworker_image_uriwork_accelerator_typework_accelerator_countruntime_version scale_tieruse_chief_in_tf_config)selfr;r<r-r,r.r/r0r1r2r3r4r5r6r7r8r9r:r=s 5lib/googlecloudsdk/command_lib/ml_engine/jobs_util.py__init__(TrainingCustomInputServerConfig.__init__}su&-2#: $<!"8#: 2&@#'B$(2 4,!6D"8D* O"8DcUR(a'UR(a[R"S5eUR(aAURR S:Xa'UR (d[R"S5eg)z9Validate that custom config parameters are set correctly.z=Only one of --master-image-uri, --runtime-version can be set.r%zC--master-machine-type is required if scale-tier is set to `CUSTOM`.T)r,r;r ArgumentErrorr<namer-r>s r?ValidateConfig.TrainingCustomInputServerConfig.ValidateConfigsm !5!5   !A BB 4??//8;  % %!!#CD D rBc URURURS.S.URURUR UR S.S.URURURURURS.URS.URURURS. $)z=Return a mapping of object fields to apitools message fields.)counttype)imageUriacceleratorConfig)rLrM tpuTfVersion) masterConfig masterTypeparameterServerConfigparameterServerCountparameterServerType workerConfig workerCount workerTypeuseChiefInTfConfig)r,r/r.r-r2r4r3r1r0r8r:r9r5r7r6r=rFs r? GetFieldMap+TrainingCustomInputServerConfig.GetFieldMaps--6644" ..009977"" !% < <#::--4422"!//  00.."999 rBcvURnU(dSUR(aB[R"UR5nUR S05R SS5n[ 5R U5nU"S0SU_SUR_SUR_SUR_SUR(aURR S 5OS_S UR(aURR S 5OS_S UR_S UR_SUR_SUR(aURR S 5OS_SUR(aURR S 5OS_SU(a UROS_SUR _SUR"_SUR$_SUR&(aUR&R S 5OS_SUR&(aUR&R S 5OS_SUR(_6$)z>Build TrainingCustomInputServerConfig from argparse.Namespace. trainingInput scaleTierNr<r;r-r,r.rKr/rJr0r1r2r3r4r5r6r7r8r9r:r=)r<configr load_pathgetScaleTierFlagMapGetEnumForChoicer;r-r,master_acceleratorparameter_server_machine_typeparameter_server_countparameter_server_image_uriparameter_server_acceleratorr5r6 worker_countr8worker_acceleratorr=)clsargssupport_tpu_tf_versiontierdata parsed_tiers r?FromArgs(TrainingCustomInputServerConfig.FromArgss! ??D  ~~dkk*xx,00dC"$55d;K  < <,, <!44 <..  < %)$;$;"&!8!8!rBcrU(aU(a [S5eU(a[R"S5 U$)aReturn, based on the command line arguments, whether we should stream logs. Both arguments cannot be set (they're mutually exclusive flags) and the default is False. Args: asyncronous: bool, the value of the --async flag. stream_logs: bool, the value of the --stream-logs flag. Returns: bool, whether to stream the logs Raises: ValueError: if both asyncronous and stream_logs are True. z---async and --stream-logs cannot both be set.zThe --async flag is deprecated, as the default behavior is to submit the job asynchronously; it can be omitted. For synchronous behavior, please pass --stream-logs. ) ValueErrorrwarning) asyncronous stream_logss r? GetStreamLogsrBs5 [ D EEKKIJ rBcV[R"XRR5$r+)r ParseCreateArgs job_class LabelsValue)rrks r?ParseCreateLabelsr^s  $ $T+@+@+L+L MMrBc [RRRR SS9n[ R "X1US9n[ R"UUUS9n[R"SRU55 URRnU(aU"U5OSnUR!UUUUUU(aUR#5OSUUU U U U UUUUS9n[$R&R)[RR*R,R SS9S S 9nUR/UU5nU (d[1UR2SS 9 U$[1UR2S S 9 [4R6"[8R:"UR25[RR<R>RA5[B[8RD"UR25S9n[FRH"[8RJ[RLS9n[NRP"5 URS[8RT"URW555 SSS5 [$R&R)UR2S[RR*R,Rj0SS9nUR U5nU$![ Ra [R"S5ef=f![ Ra [R"S 5ef=f![Xal [RZRSS5 [RZRS[\RUR2UR_5S95 GNC[`Rban[RZRSSRRf"U555 [Rh"SSS9 [RZRS[\RUR2UR_5S95 SnAGNSnAff=f!,(df  GN =f)zSubmit a training job.Tr)staging_bucketrjob_dir)packages package_pathstaging_locationzYIf local packages are provided, the `--staging-bucket` or `--job-dir` flag must be given.zUsing {0} as trainer urisN)path module_namejob_name trainer_uriregionrr< user_argsr;networkservice_accountpython_versionlabelskms_keycustom_train_server_configenable_web_accesszeIf `--package-path` is not specified, at least one Python package must be specified via `--packages`.rrrFr)outzReceived keyboard interrupt. rzPolling logs failed: {} zFailure details:)exc_inforrr)6rrcomputerrr GetStagingLocationUploadPythonPackagesNoStagingLocationErrorr rDrdebugrtraining_input_classScaleTierValueValuesEnumBuildTrainingJobToUrlrrrrrCreaterjobIdrrr r ml_enginerGetIntrrrPrinter LOG_FORMATerrrRaisesKeyboardInterruptrrrKeyboardInterruptrrNamer HttpErrorsix text_typeinfor)rrrrrrr<r^rr;rrrrrrrrrrrurisscale_tier_enumrrprinterrrs r?SubmitTrainingrbs(    $ $ + + / / / >&11#+  ) )!) +D)) ' . .t 4544MM/.8z*d*/  & & #* ''%#=+! ' -C,""(($$(($(7) + ;,#  4@ J 5A!!""399-!((22CCJJL*22399= ?+  $ $Y%9%9), 2'..0 N mmI,,[-B-B-DEF1    $ $ iiJ--22::DD E# % %'  # *K  ) )+    * +++6  ) )/    . ///4 N jj78 jj)00 9D9I9I9K1MN   N jj3::3==;MNO hh!D1 jj)00 9D9I9I9K1MNNN10sVK$+5L=Q6?3M$+L+L=A2Q32Q65Q3B Q.(Q6.Q33Q66 RcNU(aU(a[R"S5egg)Nz,`--version` cannot be set with `--model-dir`)r rD) model_dirversions r?_ValidateSubmitPredictionArgsrs 7   L MMYrBc^[X$5 [RR[R R RRSS9SS9nURUUUUUUUUU U U U U U[RU5S9n[URSS9 URUU5$)zSubmit a prediction job.Trrr)rr model_name version_name input_paths data_format output_pathrr;max_worker_count batch_sizesignature_nameraccelerator_countaccelerator_typer)rrrrrrrrrBuildBatchPredictionJobr}rbrrr)rrrmodelrrrrrr;rrrrrrrs r?SubmitPredictionr s 3""(($$(($(7) + ++ %'# )'889IJ , #"cii>   K --rBcU(am[USS5(aF[URSS5(a[R"5$[R"5$[R "5$g)zGet summary table format for an ml job resource. Args: job: job resource to build summary output for. Returns: dynamic format string for resource output. r[Fhyperparametersr)getattrr[r GetHPTrainingJobSummaryGetStandardTrainingJobSummaryGetPredictJobSummaryrs r?GetSummaryFormatrs\ sOU++ ""$5u = =,,..  0 0 22  ' ' )) rBcl^^UU4Sjn[R"UTRRU5$)Nc:>TRT5R$r+)rr)clientrsr? GetLabels$ParseUpdateLabels..GetLabelss ::g  % %%rB)r ProcessUpdateArgsLazyrr)rrrkrs`` r?ParseUpdateLabelsrs.&  * * F   ( () 55rBc^[TR5n[XT5nURX#5$![R a= [ U4SjS55(de[RRS5 gf=f)z Update a job.c3F># UHnTRU5v M g7fr+) IsSpecified).0argrks r? Update..s*C1Bt$$1Bs!) update_labels clear_labels remove_labelszNo update to perform.N) rrrPatchrNoFieldsSpecifiedErroranyrrr)rrkr labels_updates ` r?Updater&s dhh '#K$?-   W 44 $ $ C1BC C C JJ,-  s5ABB)T)NNNNNNNNNNNNNNNNN)NNNNNNNNNNNNNN)Erv __future__rrrapitools.base.pyr googlecloudsdk.api_lib.ml_enginergooglecloudsdk.command_lib.logsr$googlecloudsdk.command_lib.ml_enginer r r $googlecloudsdk.command_lib.util.apisr $googlecloudsdk.command_lib.util.argsr googlecloudsdk.corerrrrrgooglecloudsdk.core.resourcerrrr JOB_FORMATr_TEXT_FILE_URL_JSON_FILE_URL_TF_RECORD_URLChoiceEnumMapperGetMessagesModuleGoogleCloudMlV1PredictionInputDataFormatValueValuesEnumrrz GoogleCloudMlV1AcceleratorConfigTypeValueValuesEnumr}_SCALE_TIER_CHOICESGoogleCloudMlV1TrainingInputrrobjectr(r{r~rarrrrrrrrrrrrr rrr&r]rBr?r=sD#&''126:::</#*)$9 $ E P )9-"+!;!;$$%>%>"F>>BE".55nEG> / " --&&':':HL4  PNJDJ<(88!!":":'' yEIIMBF>B .B&5 rB