1<html><body> 2<style> 3 4body, h1, h2, h3, div, span, p, pre, a { 5 margin: 0; 6 padding: 0; 7 border: 0; 8 font-weight: inherit; 9 font-style: inherit; 10 font-size: 100%; 11 font-family: inherit; 12 vertical-align: baseline; 13} 14 15body { 16 font-size: 13px; 17 padding: 1em; 18} 19 20h1 { 21 font-size: 26px; 22 margin-bottom: 1em; 23} 24 25h2 { 26 font-size: 24px; 27 margin-bottom: 1em; 28} 29 30h3 { 31 font-size: 20px; 32 margin-bottom: 1em; 33 margin-top: 1em; 34} 35 36pre, code { 37 line-height: 1.5; 38 font-family: Monaco, 'DejaVu Sans Mono', 'Bitstream Vera Sans Mono', 'Lucida Console', monospace; 39} 40 41pre { 42 margin-top: 0.5em; 43} 44 45h1, h2, h3, p { 46 font-family: Arial, sans serif; 47} 48 49h1, h2, h3 { 50 border-bottom: solid #CCC 1px; 51} 52 53.toc_element { 54 margin-top: 0.5em; 55} 56 57.firstline { 58 margin-left: 2 em; 59} 60 61.method { 62 margin-top: 1em; 63 border: solid 1px #CCC; 64 padding: 1em; 65 background: #EEE; 66} 67 68.details { 69 font-weight: bold; 70 font-size: 14px; 71} 72 73</style> 74 75<h1><a href="ml_v1.html">Cloud Machine Learning Engine</a> . <a href="ml_v1.projects.html">projects</a> . <a href="ml_v1.projects.jobs.html">jobs</a></h1> 76<h2>Instance Methods</h2> 77<p class="toc_element"> 78 <code><a href="#cancel">cancel(name, body=None, x__xgafv=None)</a></code></p> 79<p class="firstline">Cancels a running job.</p> 80<p class="toc_element"> 81 <code><a href="#create">create(parent, body, x__xgafv=None)</a></code></p> 82<p class="firstline">Creates a training or a batch prediction job.</p> 83<p class="toc_element"> 84 <code><a href="#get">get(name, x__xgafv=None)</a></code></p> 85<p class="firstline">Describes a job.</p> 86<p class="toc_element"> 87 <code><a href="#getIamPolicy">getIamPolicy(resource, x__xgafv=None)</a></code></p> 88<p class="firstline">Gets the access control policy for a resource.</p> 89<p class="toc_element"> 90 <code><a href="#list">list(parent, pageToken=None, x__xgafv=None, pageSize=None, filter=None)</a></code></p> 91<p class="firstline">Lists the jobs in the project.</p> 92<p class="toc_element"> 93 <code><a href="#list_next">list_next(previous_request, previous_response)</a></code></p> 94<p class="firstline">Retrieves the next page of results.</p> 95<p class="toc_element"> 96 <code><a href="#patch">patch(name, body, updateMask=None, x__xgafv=None)</a></code></p> 97<p class="firstline">Updates a specific job resource.</p> 98<p class="toc_element"> 99 <code><a href="#setIamPolicy">setIamPolicy(resource, body, x__xgafv=None)</a></code></p> 100<p class="firstline">Sets the access control policy on the specified resource. Replaces any</p> 101<p class="toc_element"> 102 <code><a href="#testIamPermissions">testIamPermissions(resource, body, x__xgafv=None)</a></code></p> 103<p class="firstline">Returns permissions that a caller has on the specified resource.</p> 104<h3>Method Details</h3> 105<div class="method"> 106 <code class="details" id="cancel">cancel(name, body=None, x__xgafv=None)</code> 107 <pre>Cancels a running job. 108 109Args: 110 name: string, Required. The name of the job to cancel. (required) 111 body: object, The request body. 112 The object takes the form of: 113 114{ # Request message for the CancelJob method. 115 } 116 117 x__xgafv: string, V1 error format. 118 Allowed values 119 1 - v1 error format 120 2 - v2 error format 121 122Returns: 123 An object of the form: 124 125 { # A generic empty message that you can re-use to avoid defining duplicated 126 # empty messages in your APIs. A typical example is to use it as the request 127 # or the response type of an API method. For instance: 128 # 129 # service Foo { 130 # rpc Bar(google.protobuf.Empty) returns (google.protobuf.Empty); 131 # } 132 # 133 # The JSON representation for `Empty` is empty JSON object `{}`. 134 }</pre> 135</div> 136 137<div class="method"> 138 <code class="details" id="create">create(parent, body, x__xgafv=None)</code> 139 <pre>Creates a training or a batch prediction job. 140 141Args: 142 parent: string, Required. The project name. (required) 143 body: object, The request body. (required) 144 The object takes the form of: 145 146{ # Represents a training or prediction job. 147 "errorMessage": "A String", # Output only. The details of a failure or a cancellation. 148 "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. 149 "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. 150 # Only set for hyperparameter tuning jobs. 151 "trials": [ # Results for individual Hyperparameter trials. 152 # Only set for hyperparameter tuning jobs. 153 { # Represents the result of a single hyperparameter tuning trial from a 154 # training job. The TrainingOutput object that is returned on successful 155 # completion of a training job with hyperparameter tuning includes a list 156 # of HyperparameterOutput objects, one for each successful trial. 157 "hyperparameters": { # The hyperparameters given to this trial. 158 "a_key": "A String", 159 }, 160 "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. 161 "trainingStep": "A String", # The global training step for this metric. 162 "objectiveValue": 3.14, # The objective value at this training step. 163 }, 164 "allMetrics": [ # All recorded object metrics for this trial. This field is not currently 165 # populated. 166 { # An observed value of a metric. 167 "trainingStep": "A String", # The global training step for this metric. 168 "objectiveValue": 3.14, # The objective value at this training step. 169 }, 170 ], 171 "isTrialStoppedEarly": True or False, # True if the trial is stopped early. 172 "trialId": "A String", # The trial id for these results. 173 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 174 # Only set for trials of built-in algorithms jobs that have succeeded. 175 "framework": "A String", # Framework on which the built-in algorithm was trained. 176 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 177 # saves the trained model. Only set for successful jobs that don't use 178 # hyperparameter tuning. 179 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 180 # trained. 181 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 182 }, 183 }, 184 ], 185 "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. 186 "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. 187 "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. 188 "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning 189 # trials. See 190 # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) 191 # for more information. Only set for hyperparameter tuning jobs. 192 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 193 # Only set for built-in algorithms jobs. 194 "framework": "A String", # Framework on which the built-in algorithm was trained. 195 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 196 # saves the trained model. Only set for successful jobs that don't use 197 # hyperparameter tuning. 198 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 199 # trained. 200 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 201 }, 202 }, 203 "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. 204 "modelName": "A String", # Use this field if you want to use the default version for the specified 205 # model. The string must use the following format: 206 # 207 # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` 208 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch 209 # prediction. If not set, AI Platform will pick the runtime version used 210 # during the CreateVersion request for this model version, or choose the 211 # latest stable version when model version information is not available 212 # such as when the model is specified by uri. 213 "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for 214 # this job. Please refer to 215 # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) 216 # for information about how to use signatures. 217 # 218 # Defaults to 219 # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) 220 # , which is "serving_default". 221 "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. 222 # The service will buffer batch_size number of records in memory before 223 # invoking one Tensorflow prediction call internally. So take the record 224 # size and memory available into consideration when setting this parameter. 225 "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain 226 # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. 227 "A String", 228 ], 229 "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. 230 # Defaults to 10 if not specified. 231 "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for 232 # the model to use. 233 "outputPath": "A String", # Required. The output Google Cloud Storage location. 234 "dataFormat": "A String", # Required. The format of the input data files. 235 "versionName": "A String", # Use this field if you want to specify a version of the model to use. The 236 # string is formatted the same way as `model_version`, with the addition 237 # of the version information: 238 # 239 # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` 240 "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. 241 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 242 # for AI Platform services. 243 "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. 244 }, 245 "trainingInput": { # Represents input parameters for a training job. When using the # Input parameters to create a training job. 246 # gcloud command to submit your training job, you can specify 247 # the input parameters as command-line arguments and/or in a YAML configuration 248 # file referenced from the --config command-line argument. For 249 # details, see the guide to 250 # <a href="/ml-engine/docs/tensorflow/training-jobs">submitting a training 251 # job</a>. 252 "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 253 # job's worker nodes. 254 # 255 # The supported values are the same as those described in the entry for 256 # `masterType`. 257 # 258 # This value must be consistent with the category of machine type that 259 # `masterType` uses. In other words, both must be AI Platform machine 260 # types or both must be Compute Engine machine types. 261 # 262 # If you use `cloud_tpu` for this value, see special instructions for 263 # [configuring a custom TPU 264 # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). 265 # 266 # This value must be present when `scaleTier` is set to `CUSTOM` and 267 # `workerCount` is greater than zero. 268 "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. 269 # 270 # You should only set `parameterServerConfig.acceleratorConfig` if 271 # `parameterServerConfigType` is set to a Compute Engine machine type. [Learn 272 # about restrictions on accelerator configurations for 273 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 274 # 275 # Set `parameterServerConfig.imageUri` only if you build a custom image for 276 # your parameter server. If `parameterServerConfig.imageUri` has not been 277 # set, AI Platform uses the value of `masterConfig.imageUri`. 278 # Learn more about [configuring custom 279 # containers](/ml-engine/docs/distributed-training-containers). 280 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 281 # [Learn about restrictions on accelerator configurations for 282 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 283 "count": "A String", # The number of accelerators to attach to each machine running the job. 284 "type": "A String", # The type of accelerator to use. 285 }, 286 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 287 # Registry. Learn more about [configuring custom 288 # containers](/ml-engine/docs/distributed-training-containers). 289 }, 290 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. If not 291 # set, AI Platform uses the default stable version, 1.0. For more 292 # information, see the 293 # <a href="/ml-engine/docs/runtime-version-list">runtime version list</a> 294 # and 295 # <a href="/ml-engine/docs/versioning">how to manage runtime versions</a>. 296 "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers 297 # and parameter servers. 298 "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training 299 # job's master worker. 300 # 301 # The following types are supported: 302 # 303 # <dl> 304 # <dt>standard</dt> 305 # <dd> 306 # A basic machine configuration suitable for training simple models with 307 # small to moderate datasets. 308 # </dd> 309 # <dt>large_model</dt> 310 # <dd> 311 # A machine with a lot of memory, specially suited for parameter servers 312 # when your model is large (having many hidden layers or layers with very 313 # large numbers of nodes). 314 # </dd> 315 # <dt>complex_model_s</dt> 316 # <dd> 317 # A machine suitable for the master and workers of the cluster when your 318 # model requires more computation than the standard machine can handle 319 # satisfactorily. 320 # </dd> 321 # <dt>complex_model_m</dt> 322 # <dd> 323 # A machine with roughly twice the number of cores and roughly double the 324 # memory of <i>complex_model_s</i>. 325 # </dd> 326 # <dt>complex_model_l</dt> 327 # <dd> 328 # A machine with roughly twice the number of cores and roughly double the 329 # memory of <i>complex_model_m</i>. 330 # </dd> 331 # <dt>standard_gpu</dt> 332 # <dd> 333 # A machine equivalent to <i>standard</i> that 334 # also includes a single NVIDIA Tesla K80 GPU. See more about 335 # <a href="/ml-engine/docs/tensorflow/using-gpus">using GPUs to 336 # train your model</a>. 337 # </dd> 338 # <dt>complex_model_m_gpu</dt> 339 # <dd> 340 # A machine equivalent to <i>complex_model_m</i> that also includes 341 # four NVIDIA Tesla K80 GPUs. 342 # </dd> 343 # <dt>complex_model_l_gpu</dt> 344 # <dd> 345 # A machine equivalent to <i>complex_model_l</i> that also includes 346 # eight NVIDIA Tesla K80 GPUs. 347 # </dd> 348 # <dt>standard_p100</dt> 349 # <dd> 350 # A machine equivalent to <i>standard</i> that 351 # also includes a single NVIDIA Tesla P100 GPU. 352 # </dd> 353 # <dt>complex_model_m_p100</dt> 354 # <dd> 355 # A machine equivalent to <i>complex_model_m</i> that also includes 356 # four NVIDIA Tesla P100 GPUs. 357 # </dd> 358 # <dt>standard_v100</dt> 359 # <dd> 360 # A machine equivalent to <i>standard</i> that 361 # also includes a single NVIDIA Tesla V100 GPU. 362 # </dd> 363 # <dt>large_model_v100</dt> 364 # <dd> 365 # A machine equivalent to <i>large_model</i> that 366 # also includes a single NVIDIA Tesla V100 GPU. 367 # </dd> 368 # <dt>complex_model_m_v100</dt> 369 # <dd> 370 # A machine equivalent to <i>complex_model_m</i> that 371 # also includes four NVIDIA Tesla V100 GPUs. 372 # </dd> 373 # <dt>complex_model_l_v100</dt> 374 # <dd> 375 # A machine equivalent to <i>complex_model_l</i> that 376 # also includes eight NVIDIA Tesla V100 GPUs. 377 # </dd> 378 # <dt>cloud_tpu</dt> 379 # <dd> 380 # A TPU VM including one Cloud TPU. See more about 381 # <a href="/ml-engine/docs/tensorflow/using-tpus">using TPUs to train 382 # your model</a>. 383 # </dd> 384 # </dl> 385 # 386 # You may also use certain Compute Engine machine types directly in this 387 # field. The following types are supported: 388 # 389 # - `n1-standard-4` 390 # - `n1-standard-8` 391 # - `n1-standard-16` 392 # - `n1-standard-32` 393 # - `n1-standard-64` 394 # - `n1-standard-96` 395 # - `n1-highmem-2` 396 # - `n1-highmem-4` 397 # - `n1-highmem-8` 398 # - `n1-highmem-16` 399 # - `n1-highmem-32` 400 # - `n1-highmem-64` 401 # - `n1-highmem-96` 402 # - `n1-highcpu-16` 403 # - `n1-highcpu-32` 404 # - `n1-highcpu-64` 405 # - `n1-highcpu-96` 406 # 407 # See more about [using Compute Engine machine 408 # types](/ml-engine/docs/tensorflow/machine-types#compute-engine-machine-types). 409 # 410 # You must set this value when `scaleTier` is set to `CUSTOM`. 411 "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. 412 "maxTrials": 42, # Optional. How many training trials should be attempted to optimize 413 # the specified hyperparameters. 414 # 415 # Defaults to one. 416 "goal": "A String", # Required. The type of goal to use for tuning. Available types are 417 # `MAXIMIZE` and `MINIMIZE`. 418 # 419 # Defaults to `MAXIMIZE`. 420 "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter 421 # tuning job. 422 # Uses the default AI Platform hyperparameter tuning 423 # algorithm if unspecified. 424 "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing 425 # the hyperparameter tuning job. You can specify this field to override the 426 # default failing criteria for AI Platform hyperparameter tuning jobs. 427 # 428 # Defaults to zero, which means the service decides when a hyperparameter 429 # job should fail. 430 "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial 431 # early stopping. 432 "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to 433 # continue with. The job id will be used to find the corresponding vizier 434 # study guid and resume the study. 435 "params": [ # Required. The set of parameters to tune. 436 { # Represents a single hyperparameter to optimize. 437 "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 438 # should be unset if type is `CATEGORICAL`. This value should be integers if 439 # type is `INTEGER`. 440 "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. 441 "A String", 442 ], 443 "discreteValues": [ # Required if type is `DISCRETE`. 444 # A list of feasible points. 445 # The list should be in strictly increasing order. For instance, this 446 # parameter might have possible settings of 1.5, 2.5, and 4.0. This list 447 # should not contain more than 1,000 values. 448 3.14, 449 ], 450 "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in 451 # a HyperparameterSpec message. E.g., "learning_rate". 452 "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 453 # should be unset if type is `CATEGORICAL`. This value should be integers if 454 # type is INTEGER. 455 "type": "A String", # Required. The type of the parameter. 456 "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. 457 # Leave unset for categorical parameters. 458 # Some kind of scaling is strongly recommended for real or integral 459 # parameters (e.g., `UNIT_LINEAR_SCALE`). 460 }, 461 ], 462 "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For 463 # current versions of TensorFlow, this tag name should exactly match what is 464 # shown in TensorBoard, including all scopes. For versions of TensorFlow 465 # prior to 0.12, this should be only the tag passed to tf.Summary. 466 # By default, "training/hptuning/metric" will be used. 467 "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. 468 # You can reduce the time it takes to perform hyperparameter tuning by adding 469 # trials in parallel. However, each trail only benefits from the information 470 # gained in completed trials. That means that a trial does not get access to 471 # the results of trials running at the same time, which could reduce the 472 # quality of the overall optimization. 473 # 474 # Each trial will use the same scale tier and machine types. 475 # 476 # Defaults to one. 477 }, 478 "region": "A String", # Required. The Google Compute Engine region to run the training job in. 479 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 480 # for AI Platform services. 481 "args": [ # Optional. Command line arguments to pass to the program. 482 "A String", 483 ], 484 "pythonModule": "A String", # Required. The Python module name to run after installing the packages. 485 "pythonVersion": "A String", # Optional. The version of Python used in training. If not set, the default 486 # version is '2.7'. Python '3.5' is available when `runtime_version` is set 487 # to '1.4' and above. Python '2.7' works with all supported 488 # <a href="/ml-engine/docs/runtime-version-list">runtime versions</a>. 489 "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs 490 # and other data needed for training. This path is passed to your TensorFlow 491 # program as the '--job-dir' command-line argument. The benefit of specifying 492 # this field is that Cloud ML validates the path for use in training. 493 "packageUris": [ # Required. The Google Cloud Storage location of the packages with 494 # the training program and any additional dependencies. 495 # The maximum number of package URIs is 100. 496 "A String", 497 ], 498 "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each 499 # replica in the cluster will be of the type specified in `worker_type`. 500 # 501 # This value can only be used when `scale_tier` is set to `CUSTOM`. If you 502 # set this value, you must also set `worker_type`. 503 # 504 # The default value is zero. 505 "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 506 # job's parameter server. 507 # 508 # The supported values are the same as those described in the entry for 509 # `master_type`. 510 # 511 # This value must be consistent with the category of machine type that 512 # `masterType` uses. In other words, both must be AI Platform machine 513 # types or both must be Compute Engine machine types. 514 # 515 # This value must be present when `scaleTier` is set to `CUSTOM` and 516 # `parameter_server_count` is greater than zero. 517 "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. 518 # 519 # You should only set `workerConfig.acceleratorConfig` if `workerType` is set 520 # to a Compute Engine machine type. [Learn about restrictions on accelerator 521 # configurations for 522 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 523 # 524 # Set `workerConfig.imageUri` only if you build a custom image for your 525 # worker. If `workerConfig.imageUri` has not been set, AI Platform uses 526 # the value of `masterConfig.imageUri`. Learn more about 527 # [configuring custom 528 # containers](/ml-engine/docs/distributed-training-containers). 529 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 530 # [Learn about restrictions on accelerator configurations for 531 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 532 "count": "A String", # The number of accelerators to attach to each machine running the job. 533 "type": "A String", # The type of accelerator to use. 534 }, 535 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 536 # Registry. Learn more about [configuring custom 537 # containers](/ml-engine/docs/distributed-training-containers). 538 }, 539 "maxRunningTime": "A String", # Optional. The maximum job running time. The default is 7 days. 540 "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. 541 # 542 # You should only set `masterConfig.acceleratorConfig` if `masterType` is set 543 # to a Compute Engine machine type. Learn about [restrictions on accelerator 544 # configurations for 545 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 546 # 547 # Set `masterConfig.imageUri` only if you build a custom image. Only one of 548 # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more about 549 # [configuring custom 550 # containers](/ml-engine/docs/distributed-training-containers). 551 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 552 # [Learn about restrictions on accelerator configurations for 553 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 554 "count": "A String", # The number of accelerators to attach to each machine running the job. 555 "type": "A String", # The type of accelerator to use. 556 }, 557 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 558 # Registry. Learn more about [configuring custom 559 # containers](/ml-engine/docs/distributed-training-containers). 560 }, 561 "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training 562 # job. Each replica in the cluster will be of the type specified in 563 # `parameter_server_type`. 564 # 565 # This value can only be used when `scale_tier` is set to `CUSTOM`.If you 566 # set this value, you must also set `parameter_server_type`. 567 # 568 # The default value is zero. 569 }, 570 "jobId": "A String", # Required. The user-specified id of the job. 571 "labels": { # Optional. One or more labels that you can add, to organize your jobs. 572 # Each label is a key-value pair, where both the key and the value are 573 # arbitrary strings that you supply. 574 # For more information, see the documentation on 575 # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. 576 "a_key": "A String", 577 }, 578 "state": "A String", # Output only. The detailed state of a job. 579 "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help 580 # prevent simultaneous updates of a job from overwriting each other. 581 # It is strongly suggested that systems make use of the `etag` in the 582 # read-modify-write cycle to perform job updates in order to avoid race 583 # conditions: An `etag` is returned in the response to `GetJob`, and 584 # systems are expected to put that etag in the request to `UpdateJob` to 585 # ensure that their change will be applied to the same version of the job. 586 "startTime": "A String", # Output only. When the job processing was started. 587 "endTime": "A String", # Output only. When the job processing was completed. 588 "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. 589 "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. 590 "nodeHours": 3.14, # Node hours used by the batch prediction job. 591 "predictionCount": "A String", # The number of generated predictions. 592 "errorCount": "A String", # The number of data instances which resulted in errors. 593 }, 594 "createTime": "A String", # Output only. When the job was created. 595} 596 597 x__xgafv: string, V1 error format. 598 Allowed values 599 1 - v1 error format 600 2 - v2 error format 601 602Returns: 603 An object of the form: 604 605 { # Represents a training or prediction job. 606 "errorMessage": "A String", # Output only. The details of a failure or a cancellation. 607 "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. 608 "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. 609 # Only set for hyperparameter tuning jobs. 610 "trials": [ # Results for individual Hyperparameter trials. 611 # Only set for hyperparameter tuning jobs. 612 { # Represents the result of a single hyperparameter tuning trial from a 613 # training job. The TrainingOutput object that is returned on successful 614 # completion of a training job with hyperparameter tuning includes a list 615 # of HyperparameterOutput objects, one for each successful trial. 616 "hyperparameters": { # The hyperparameters given to this trial. 617 "a_key": "A String", 618 }, 619 "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. 620 "trainingStep": "A String", # The global training step for this metric. 621 "objectiveValue": 3.14, # The objective value at this training step. 622 }, 623 "allMetrics": [ # All recorded object metrics for this trial. This field is not currently 624 # populated. 625 { # An observed value of a metric. 626 "trainingStep": "A String", # The global training step for this metric. 627 "objectiveValue": 3.14, # The objective value at this training step. 628 }, 629 ], 630 "isTrialStoppedEarly": True or False, # True if the trial is stopped early. 631 "trialId": "A String", # The trial id for these results. 632 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 633 # Only set for trials of built-in algorithms jobs that have succeeded. 634 "framework": "A String", # Framework on which the built-in algorithm was trained. 635 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 636 # saves the trained model. Only set for successful jobs that don't use 637 # hyperparameter tuning. 638 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 639 # trained. 640 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 641 }, 642 }, 643 ], 644 "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. 645 "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. 646 "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. 647 "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning 648 # trials. See 649 # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) 650 # for more information. Only set for hyperparameter tuning jobs. 651 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 652 # Only set for built-in algorithms jobs. 653 "framework": "A String", # Framework on which the built-in algorithm was trained. 654 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 655 # saves the trained model. Only set for successful jobs that don't use 656 # hyperparameter tuning. 657 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 658 # trained. 659 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 660 }, 661 }, 662 "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. 663 "modelName": "A String", # Use this field if you want to use the default version for the specified 664 # model. The string must use the following format: 665 # 666 # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` 667 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch 668 # prediction. If not set, AI Platform will pick the runtime version used 669 # during the CreateVersion request for this model version, or choose the 670 # latest stable version when model version information is not available 671 # such as when the model is specified by uri. 672 "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for 673 # this job. Please refer to 674 # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) 675 # for information about how to use signatures. 676 # 677 # Defaults to 678 # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) 679 # , which is "serving_default". 680 "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. 681 # The service will buffer batch_size number of records in memory before 682 # invoking one Tensorflow prediction call internally. So take the record 683 # size and memory available into consideration when setting this parameter. 684 "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain 685 # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. 686 "A String", 687 ], 688 "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. 689 # Defaults to 10 if not specified. 690 "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for 691 # the model to use. 692 "outputPath": "A String", # Required. The output Google Cloud Storage location. 693 "dataFormat": "A String", # Required. The format of the input data files. 694 "versionName": "A String", # Use this field if you want to specify a version of the model to use. The 695 # string is formatted the same way as `model_version`, with the addition 696 # of the version information: 697 # 698 # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` 699 "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. 700 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 701 # for AI Platform services. 702 "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. 703 }, 704 "trainingInput": { # Represents input parameters for a training job. When using the # Input parameters to create a training job. 705 # gcloud command to submit your training job, you can specify 706 # the input parameters as command-line arguments and/or in a YAML configuration 707 # file referenced from the --config command-line argument. For 708 # details, see the guide to 709 # <a href="/ml-engine/docs/tensorflow/training-jobs">submitting a training 710 # job</a>. 711 "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 712 # job's worker nodes. 713 # 714 # The supported values are the same as those described in the entry for 715 # `masterType`. 716 # 717 # This value must be consistent with the category of machine type that 718 # `masterType` uses. In other words, both must be AI Platform machine 719 # types or both must be Compute Engine machine types. 720 # 721 # If you use `cloud_tpu` for this value, see special instructions for 722 # [configuring a custom TPU 723 # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). 724 # 725 # This value must be present when `scaleTier` is set to `CUSTOM` and 726 # `workerCount` is greater than zero. 727 "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. 728 # 729 # You should only set `parameterServerConfig.acceleratorConfig` if 730 # `parameterServerConfigType` is set to a Compute Engine machine type. [Learn 731 # about restrictions on accelerator configurations for 732 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 733 # 734 # Set `parameterServerConfig.imageUri` only if you build a custom image for 735 # your parameter server. If `parameterServerConfig.imageUri` has not been 736 # set, AI Platform uses the value of `masterConfig.imageUri`. 737 # Learn more about [configuring custom 738 # containers](/ml-engine/docs/distributed-training-containers). 739 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 740 # [Learn about restrictions on accelerator configurations for 741 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 742 "count": "A String", # The number of accelerators to attach to each machine running the job. 743 "type": "A String", # The type of accelerator to use. 744 }, 745 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 746 # Registry. Learn more about [configuring custom 747 # containers](/ml-engine/docs/distributed-training-containers). 748 }, 749 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. If not 750 # set, AI Platform uses the default stable version, 1.0. For more 751 # information, see the 752 # <a href="/ml-engine/docs/runtime-version-list">runtime version list</a> 753 # and 754 # <a href="/ml-engine/docs/versioning">how to manage runtime versions</a>. 755 "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers 756 # and parameter servers. 757 "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training 758 # job's master worker. 759 # 760 # The following types are supported: 761 # 762 # <dl> 763 # <dt>standard</dt> 764 # <dd> 765 # A basic machine configuration suitable for training simple models with 766 # small to moderate datasets. 767 # </dd> 768 # <dt>large_model</dt> 769 # <dd> 770 # A machine with a lot of memory, specially suited for parameter servers 771 # when your model is large (having many hidden layers or layers with very 772 # large numbers of nodes). 773 # </dd> 774 # <dt>complex_model_s</dt> 775 # <dd> 776 # A machine suitable for the master and workers of the cluster when your 777 # model requires more computation than the standard machine can handle 778 # satisfactorily. 779 # </dd> 780 # <dt>complex_model_m</dt> 781 # <dd> 782 # A machine with roughly twice the number of cores and roughly double the 783 # memory of <i>complex_model_s</i>. 784 # </dd> 785 # <dt>complex_model_l</dt> 786 # <dd> 787 # A machine with roughly twice the number of cores and roughly double the 788 # memory of <i>complex_model_m</i>. 789 # </dd> 790 # <dt>standard_gpu</dt> 791 # <dd> 792 # A machine equivalent to <i>standard</i> that 793 # also includes a single NVIDIA Tesla K80 GPU. See more about 794 # <a href="/ml-engine/docs/tensorflow/using-gpus">using GPUs to 795 # train your model</a>. 796 # </dd> 797 # <dt>complex_model_m_gpu</dt> 798 # <dd> 799 # A machine equivalent to <i>complex_model_m</i> that also includes 800 # four NVIDIA Tesla K80 GPUs. 801 # </dd> 802 # <dt>complex_model_l_gpu</dt> 803 # <dd> 804 # A machine equivalent to <i>complex_model_l</i> that also includes 805 # eight NVIDIA Tesla K80 GPUs. 806 # </dd> 807 # <dt>standard_p100</dt> 808 # <dd> 809 # A machine equivalent to <i>standard</i> that 810 # also includes a single NVIDIA Tesla P100 GPU. 811 # </dd> 812 # <dt>complex_model_m_p100</dt> 813 # <dd> 814 # A machine equivalent to <i>complex_model_m</i> that also includes 815 # four NVIDIA Tesla P100 GPUs. 816 # </dd> 817 # <dt>standard_v100</dt> 818 # <dd> 819 # A machine equivalent to <i>standard</i> that 820 # also includes a single NVIDIA Tesla V100 GPU. 821 # </dd> 822 # <dt>large_model_v100</dt> 823 # <dd> 824 # A machine equivalent to <i>large_model</i> that 825 # also includes a single NVIDIA Tesla V100 GPU. 826 # </dd> 827 # <dt>complex_model_m_v100</dt> 828 # <dd> 829 # A machine equivalent to <i>complex_model_m</i> that 830 # also includes four NVIDIA Tesla V100 GPUs. 831 # </dd> 832 # <dt>complex_model_l_v100</dt> 833 # <dd> 834 # A machine equivalent to <i>complex_model_l</i> that 835 # also includes eight NVIDIA Tesla V100 GPUs. 836 # </dd> 837 # <dt>cloud_tpu</dt> 838 # <dd> 839 # A TPU VM including one Cloud TPU. See more about 840 # <a href="/ml-engine/docs/tensorflow/using-tpus">using TPUs to train 841 # your model</a>. 842 # </dd> 843 # </dl> 844 # 845 # You may also use certain Compute Engine machine types directly in this 846 # field. The following types are supported: 847 # 848 # - `n1-standard-4` 849 # - `n1-standard-8` 850 # - `n1-standard-16` 851 # - `n1-standard-32` 852 # - `n1-standard-64` 853 # - `n1-standard-96` 854 # - `n1-highmem-2` 855 # - `n1-highmem-4` 856 # - `n1-highmem-8` 857 # - `n1-highmem-16` 858 # - `n1-highmem-32` 859 # - `n1-highmem-64` 860 # - `n1-highmem-96` 861 # - `n1-highcpu-16` 862 # - `n1-highcpu-32` 863 # - `n1-highcpu-64` 864 # - `n1-highcpu-96` 865 # 866 # See more about [using Compute Engine machine 867 # types](/ml-engine/docs/tensorflow/machine-types#compute-engine-machine-types). 868 # 869 # You must set this value when `scaleTier` is set to `CUSTOM`. 870 "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. 871 "maxTrials": 42, # Optional. How many training trials should be attempted to optimize 872 # the specified hyperparameters. 873 # 874 # Defaults to one. 875 "goal": "A String", # Required. The type of goal to use for tuning. Available types are 876 # `MAXIMIZE` and `MINIMIZE`. 877 # 878 # Defaults to `MAXIMIZE`. 879 "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter 880 # tuning job. 881 # Uses the default AI Platform hyperparameter tuning 882 # algorithm if unspecified. 883 "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing 884 # the hyperparameter tuning job. You can specify this field to override the 885 # default failing criteria for AI Platform hyperparameter tuning jobs. 886 # 887 # Defaults to zero, which means the service decides when a hyperparameter 888 # job should fail. 889 "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial 890 # early stopping. 891 "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to 892 # continue with. The job id will be used to find the corresponding vizier 893 # study guid and resume the study. 894 "params": [ # Required. The set of parameters to tune. 895 { # Represents a single hyperparameter to optimize. 896 "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 897 # should be unset if type is `CATEGORICAL`. This value should be integers if 898 # type is `INTEGER`. 899 "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. 900 "A String", 901 ], 902 "discreteValues": [ # Required if type is `DISCRETE`. 903 # A list of feasible points. 904 # The list should be in strictly increasing order. For instance, this 905 # parameter might have possible settings of 1.5, 2.5, and 4.0. This list 906 # should not contain more than 1,000 values. 907 3.14, 908 ], 909 "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in 910 # a HyperparameterSpec message. E.g., "learning_rate". 911 "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 912 # should be unset if type is `CATEGORICAL`. This value should be integers if 913 # type is INTEGER. 914 "type": "A String", # Required. The type of the parameter. 915 "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. 916 # Leave unset for categorical parameters. 917 # Some kind of scaling is strongly recommended for real or integral 918 # parameters (e.g., `UNIT_LINEAR_SCALE`). 919 }, 920 ], 921 "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For 922 # current versions of TensorFlow, this tag name should exactly match what is 923 # shown in TensorBoard, including all scopes. For versions of TensorFlow 924 # prior to 0.12, this should be only the tag passed to tf.Summary. 925 # By default, "training/hptuning/metric" will be used. 926 "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. 927 # You can reduce the time it takes to perform hyperparameter tuning by adding 928 # trials in parallel. However, each trail only benefits from the information 929 # gained in completed trials. That means that a trial does not get access to 930 # the results of trials running at the same time, which could reduce the 931 # quality of the overall optimization. 932 # 933 # Each trial will use the same scale tier and machine types. 934 # 935 # Defaults to one. 936 }, 937 "region": "A String", # Required. The Google Compute Engine region to run the training job in. 938 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 939 # for AI Platform services. 940 "args": [ # Optional. Command line arguments to pass to the program. 941 "A String", 942 ], 943 "pythonModule": "A String", # Required. The Python module name to run after installing the packages. 944 "pythonVersion": "A String", # Optional. The version of Python used in training. If not set, the default 945 # version is '2.7'. Python '3.5' is available when `runtime_version` is set 946 # to '1.4' and above. Python '2.7' works with all supported 947 # <a href="/ml-engine/docs/runtime-version-list">runtime versions</a>. 948 "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs 949 # and other data needed for training. This path is passed to your TensorFlow 950 # program as the '--job-dir' command-line argument. The benefit of specifying 951 # this field is that Cloud ML validates the path for use in training. 952 "packageUris": [ # Required. The Google Cloud Storage location of the packages with 953 # the training program and any additional dependencies. 954 # The maximum number of package URIs is 100. 955 "A String", 956 ], 957 "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each 958 # replica in the cluster will be of the type specified in `worker_type`. 959 # 960 # This value can only be used when `scale_tier` is set to `CUSTOM`. If you 961 # set this value, you must also set `worker_type`. 962 # 963 # The default value is zero. 964 "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 965 # job's parameter server. 966 # 967 # The supported values are the same as those described in the entry for 968 # `master_type`. 969 # 970 # This value must be consistent with the category of machine type that 971 # `masterType` uses. In other words, both must be AI Platform machine 972 # types or both must be Compute Engine machine types. 973 # 974 # This value must be present when `scaleTier` is set to `CUSTOM` and 975 # `parameter_server_count` is greater than zero. 976 "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. 977 # 978 # You should only set `workerConfig.acceleratorConfig` if `workerType` is set 979 # to a Compute Engine machine type. [Learn about restrictions on accelerator 980 # configurations for 981 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 982 # 983 # Set `workerConfig.imageUri` only if you build a custom image for your 984 # worker. If `workerConfig.imageUri` has not been set, AI Platform uses 985 # the value of `masterConfig.imageUri`. Learn more about 986 # [configuring custom 987 # containers](/ml-engine/docs/distributed-training-containers). 988 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 989 # [Learn about restrictions on accelerator configurations for 990 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 991 "count": "A String", # The number of accelerators to attach to each machine running the job. 992 "type": "A String", # The type of accelerator to use. 993 }, 994 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 995 # Registry. Learn more about [configuring custom 996 # containers](/ml-engine/docs/distributed-training-containers). 997 }, 998 "maxRunningTime": "A String", # Optional. The maximum job running time. The default is 7 days. 999 "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. 1000 # 1001 # You should only set `masterConfig.acceleratorConfig` if `masterType` is set 1002 # to a Compute Engine machine type. Learn about [restrictions on accelerator 1003 # configurations for 1004 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 1005 # 1006 # Set `masterConfig.imageUri` only if you build a custom image. Only one of 1007 # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more about 1008 # [configuring custom 1009 # containers](/ml-engine/docs/distributed-training-containers). 1010 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 1011 # [Learn about restrictions on accelerator configurations for 1012 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 1013 "count": "A String", # The number of accelerators to attach to each machine running the job. 1014 "type": "A String", # The type of accelerator to use. 1015 }, 1016 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 1017 # Registry. Learn more about [configuring custom 1018 # containers](/ml-engine/docs/distributed-training-containers). 1019 }, 1020 "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training 1021 # job. Each replica in the cluster will be of the type specified in 1022 # `parameter_server_type`. 1023 # 1024 # This value can only be used when `scale_tier` is set to `CUSTOM`.If you 1025 # set this value, you must also set `parameter_server_type`. 1026 # 1027 # The default value is zero. 1028 }, 1029 "jobId": "A String", # Required. The user-specified id of the job. 1030 "labels": { # Optional. One or more labels that you can add, to organize your jobs. 1031 # Each label is a key-value pair, where both the key and the value are 1032 # arbitrary strings that you supply. 1033 # For more information, see the documentation on 1034 # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. 1035 "a_key": "A String", 1036 }, 1037 "state": "A String", # Output only. The detailed state of a job. 1038 "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help 1039 # prevent simultaneous updates of a job from overwriting each other. 1040 # It is strongly suggested that systems make use of the `etag` in the 1041 # read-modify-write cycle to perform job updates in order to avoid race 1042 # conditions: An `etag` is returned in the response to `GetJob`, and 1043 # systems are expected to put that etag in the request to `UpdateJob` to 1044 # ensure that their change will be applied to the same version of the job. 1045 "startTime": "A String", # Output only. When the job processing was started. 1046 "endTime": "A String", # Output only. When the job processing was completed. 1047 "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. 1048 "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. 1049 "nodeHours": 3.14, # Node hours used by the batch prediction job. 1050 "predictionCount": "A String", # The number of generated predictions. 1051 "errorCount": "A String", # The number of data instances which resulted in errors. 1052 }, 1053 "createTime": "A String", # Output only. When the job was created. 1054 }</pre> 1055</div> 1056 1057<div class="method"> 1058 <code class="details" id="get">get(name, x__xgafv=None)</code> 1059 <pre>Describes a job. 1060 1061Args: 1062 name: string, Required. The name of the job to get the description of. (required) 1063 x__xgafv: string, V1 error format. 1064 Allowed values 1065 1 - v1 error format 1066 2 - v2 error format 1067 1068Returns: 1069 An object of the form: 1070 1071 { # Represents a training or prediction job. 1072 "errorMessage": "A String", # Output only. The details of a failure or a cancellation. 1073 "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. 1074 "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. 1075 # Only set for hyperparameter tuning jobs. 1076 "trials": [ # Results for individual Hyperparameter trials. 1077 # Only set for hyperparameter tuning jobs. 1078 { # Represents the result of a single hyperparameter tuning trial from a 1079 # training job. The TrainingOutput object that is returned on successful 1080 # completion of a training job with hyperparameter tuning includes a list 1081 # of HyperparameterOutput objects, one for each successful trial. 1082 "hyperparameters": { # The hyperparameters given to this trial. 1083 "a_key": "A String", 1084 }, 1085 "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. 1086 "trainingStep": "A String", # The global training step for this metric. 1087 "objectiveValue": 3.14, # The objective value at this training step. 1088 }, 1089 "allMetrics": [ # All recorded object metrics for this trial. This field is not currently 1090 # populated. 1091 { # An observed value of a metric. 1092 "trainingStep": "A String", # The global training step for this metric. 1093 "objectiveValue": 3.14, # The objective value at this training step. 1094 }, 1095 ], 1096 "isTrialStoppedEarly": True or False, # True if the trial is stopped early. 1097 "trialId": "A String", # The trial id for these results. 1098 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 1099 # Only set for trials of built-in algorithms jobs that have succeeded. 1100 "framework": "A String", # Framework on which the built-in algorithm was trained. 1101 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 1102 # saves the trained model. Only set for successful jobs that don't use 1103 # hyperparameter tuning. 1104 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 1105 # trained. 1106 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 1107 }, 1108 }, 1109 ], 1110 "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. 1111 "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. 1112 "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. 1113 "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning 1114 # trials. See 1115 # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) 1116 # for more information. Only set for hyperparameter tuning jobs. 1117 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 1118 # Only set for built-in algorithms jobs. 1119 "framework": "A String", # Framework on which the built-in algorithm was trained. 1120 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 1121 # saves the trained model. Only set for successful jobs that don't use 1122 # hyperparameter tuning. 1123 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 1124 # trained. 1125 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 1126 }, 1127 }, 1128 "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. 1129 "modelName": "A String", # Use this field if you want to use the default version for the specified 1130 # model. The string must use the following format: 1131 # 1132 # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` 1133 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch 1134 # prediction. If not set, AI Platform will pick the runtime version used 1135 # during the CreateVersion request for this model version, or choose the 1136 # latest stable version when model version information is not available 1137 # such as when the model is specified by uri. 1138 "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for 1139 # this job. Please refer to 1140 # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) 1141 # for information about how to use signatures. 1142 # 1143 # Defaults to 1144 # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) 1145 # , which is "serving_default". 1146 "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. 1147 # The service will buffer batch_size number of records in memory before 1148 # invoking one Tensorflow prediction call internally. So take the record 1149 # size and memory available into consideration when setting this parameter. 1150 "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain 1151 # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. 1152 "A String", 1153 ], 1154 "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. 1155 # Defaults to 10 if not specified. 1156 "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for 1157 # the model to use. 1158 "outputPath": "A String", # Required. The output Google Cloud Storage location. 1159 "dataFormat": "A String", # Required. The format of the input data files. 1160 "versionName": "A String", # Use this field if you want to specify a version of the model to use. The 1161 # string is formatted the same way as `model_version`, with the addition 1162 # of the version information: 1163 # 1164 # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` 1165 "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. 1166 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 1167 # for AI Platform services. 1168 "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. 1169 }, 1170 "trainingInput": { # Represents input parameters for a training job. When using the # Input parameters to create a training job. 1171 # gcloud command to submit your training job, you can specify 1172 # the input parameters as command-line arguments and/or in a YAML configuration 1173 # file referenced from the --config command-line argument. For 1174 # details, see the guide to 1175 # <a href="/ml-engine/docs/tensorflow/training-jobs">submitting a training 1176 # job</a>. 1177 "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 1178 # job's worker nodes. 1179 # 1180 # The supported values are the same as those described in the entry for 1181 # `masterType`. 1182 # 1183 # This value must be consistent with the category of machine type that 1184 # `masterType` uses. In other words, both must be AI Platform machine 1185 # types or both must be Compute Engine machine types. 1186 # 1187 # If you use `cloud_tpu` for this value, see special instructions for 1188 # [configuring a custom TPU 1189 # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). 1190 # 1191 # This value must be present when `scaleTier` is set to `CUSTOM` and 1192 # `workerCount` is greater than zero. 1193 "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. 1194 # 1195 # You should only set `parameterServerConfig.acceleratorConfig` if 1196 # `parameterServerConfigType` is set to a Compute Engine machine type. [Learn 1197 # about restrictions on accelerator configurations for 1198 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 1199 # 1200 # Set `parameterServerConfig.imageUri` only if you build a custom image for 1201 # your parameter server. If `parameterServerConfig.imageUri` has not been 1202 # set, AI Platform uses the value of `masterConfig.imageUri`. 1203 # Learn more about [configuring custom 1204 # containers](/ml-engine/docs/distributed-training-containers). 1205 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 1206 # [Learn about restrictions on accelerator configurations for 1207 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 1208 "count": "A String", # The number of accelerators to attach to each machine running the job. 1209 "type": "A String", # The type of accelerator to use. 1210 }, 1211 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 1212 # Registry. Learn more about [configuring custom 1213 # containers](/ml-engine/docs/distributed-training-containers). 1214 }, 1215 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. If not 1216 # set, AI Platform uses the default stable version, 1.0. For more 1217 # information, see the 1218 # <a href="/ml-engine/docs/runtime-version-list">runtime version list</a> 1219 # and 1220 # <a href="/ml-engine/docs/versioning">how to manage runtime versions</a>. 1221 "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers 1222 # and parameter servers. 1223 "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training 1224 # job's master worker. 1225 # 1226 # The following types are supported: 1227 # 1228 # <dl> 1229 # <dt>standard</dt> 1230 # <dd> 1231 # A basic machine configuration suitable for training simple models with 1232 # small to moderate datasets. 1233 # </dd> 1234 # <dt>large_model</dt> 1235 # <dd> 1236 # A machine with a lot of memory, specially suited for parameter servers 1237 # when your model is large (having many hidden layers or layers with very 1238 # large numbers of nodes). 1239 # </dd> 1240 # <dt>complex_model_s</dt> 1241 # <dd> 1242 # A machine suitable for the master and workers of the cluster when your 1243 # model requires more computation than the standard machine can handle 1244 # satisfactorily. 1245 # </dd> 1246 # <dt>complex_model_m</dt> 1247 # <dd> 1248 # A machine with roughly twice the number of cores and roughly double the 1249 # memory of <i>complex_model_s</i>. 1250 # </dd> 1251 # <dt>complex_model_l</dt> 1252 # <dd> 1253 # A machine with roughly twice the number of cores and roughly double the 1254 # memory of <i>complex_model_m</i>. 1255 # </dd> 1256 # <dt>standard_gpu</dt> 1257 # <dd> 1258 # A machine equivalent to <i>standard</i> that 1259 # also includes a single NVIDIA Tesla K80 GPU. See more about 1260 # <a href="/ml-engine/docs/tensorflow/using-gpus">using GPUs to 1261 # train your model</a>. 1262 # </dd> 1263 # <dt>complex_model_m_gpu</dt> 1264 # <dd> 1265 # A machine equivalent to <i>complex_model_m</i> that also includes 1266 # four NVIDIA Tesla K80 GPUs. 1267 # </dd> 1268 # <dt>complex_model_l_gpu</dt> 1269 # <dd> 1270 # A machine equivalent to <i>complex_model_l</i> that also includes 1271 # eight NVIDIA Tesla K80 GPUs. 1272 # </dd> 1273 # <dt>standard_p100</dt> 1274 # <dd> 1275 # A machine equivalent to <i>standard</i> that 1276 # also includes a single NVIDIA Tesla P100 GPU. 1277 # </dd> 1278 # <dt>complex_model_m_p100</dt> 1279 # <dd> 1280 # A machine equivalent to <i>complex_model_m</i> that also includes 1281 # four NVIDIA Tesla P100 GPUs. 1282 # </dd> 1283 # <dt>standard_v100</dt> 1284 # <dd> 1285 # A machine equivalent to <i>standard</i> that 1286 # also includes a single NVIDIA Tesla V100 GPU. 1287 # </dd> 1288 # <dt>large_model_v100</dt> 1289 # <dd> 1290 # A machine equivalent to <i>large_model</i> that 1291 # also includes a single NVIDIA Tesla V100 GPU. 1292 # </dd> 1293 # <dt>complex_model_m_v100</dt> 1294 # <dd> 1295 # A machine equivalent to <i>complex_model_m</i> that 1296 # also includes four NVIDIA Tesla V100 GPUs. 1297 # </dd> 1298 # <dt>complex_model_l_v100</dt> 1299 # <dd> 1300 # A machine equivalent to <i>complex_model_l</i> that 1301 # also includes eight NVIDIA Tesla V100 GPUs. 1302 # </dd> 1303 # <dt>cloud_tpu</dt> 1304 # <dd> 1305 # A TPU VM including one Cloud TPU. See more about 1306 # <a href="/ml-engine/docs/tensorflow/using-tpus">using TPUs to train 1307 # your model</a>. 1308 # </dd> 1309 # </dl> 1310 # 1311 # You may also use certain Compute Engine machine types directly in this 1312 # field. The following types are supported: 1313 # 1314 # - `n1-standard-4` 1315 # - `n1-standard-8` 1316 # - `n1-standard-16` 1317 # - `n1-standard-32` 1318 # - `n1-standard-64` 1319 # - `n1-standard-96` 1320 # - `n1-highmem-2` 1321 # - `n1-highmem-4` 1322 # - `n1-highmem-8` 1323 # - `n1-highmem-16` 1324 # - `n1-highmem-32` 1325 # - `n1-highmem-64` 1326 # - `n1-highmem-96` 1327 # - `n1-highcpu-16` 1328 # - `n1-highcpu-32` 1329 # - `n1-highcpu-64` 1330 # - `n1-highcpu-96` 1331 # 1332 # See more about [using Compute Engine machine 1333 # types](/ml-engine/docs/tensorflow/machine-types#compute-engine-machine-types). 1334 # 1335 # You must set this value when `scaleTier` is set to `CUSTOM`. 1336 "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. 1337 "maxTrials": 42, # Optional. How many training trials should be attempted to optimize 1338 # the specified hyperparameters. 1339 # 1340 # Defaults to one. 1341 "goal": "A String", # Required. The type of goal to use for tuning. Available types are 1342 # `MAXIMIZE` and `MINIMIZE`. 1343 # 1344 # Defaults to `MAXIMIZE`. 1345 "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter 1346 # tuning job. 1347 # Uses the default AI Platform hyperparameter tuning 1348 # algorithm if unspecified. 1349 "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing 1350 # the hyperparameter tuning job. You can specify this field to override the 1351 # default failing criteria for AI Platform hyperparameter tuning jobs. 1352 # 1353 # Defaults to zero, which means the service decides when a hyperparameter 1354 # job should fail. 1355 "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial 1356 # early stopping. 1357 "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to 1358 # continue with. The job id will be used to find the corresponding vizier 1359 # study guid and resume the study. 1360 "params": [ # Required. The set of parameters to tune. 1361 { # Represents a single hyperparameter to optimize. 1362 "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 1363 # should be unset if type is `CATEGORICAL`. This value should be integers if 1364 # type is `INTEGER`. 1365 "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. 1366 "A String", 1367 ], 1368 "discreteValues": [ # Required if type is `DISCRETE`. 1369 # A list of feasible points. 1370 # The list should be in strictly increasing order. For instance, this 1371 # parameter might have possible settings of 1.5, 2.5, and 4.0. This list 1372 # should not contain more than 1,000 values. 1373 3.14, 1374 ], 1375 "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in 1376 # a HyperparameterSpec message. E.g., "learning_rate". 1377 "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 1378 # should be unset if type is `CATEGORICAL`. This value should be integers if 1379 # type is INTEGER. 1380 "type": "A String", # Required. The type of the parameter. 1381 "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. 1382 # Leave unset for categorical parameters. 1383 # Some kind of scaling is strongly recommended for real or integral 1384 # parameters (e.g., `UNIT_LINEAR_SCALE`). 1385 }, 1386 ], 1387 "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For 1388 # current versions of TensorFlow, this tag name should exactly match what is 1389 # shown in TensorBoard, including all scopes. For versions of TensorFlow 1390 # prior to 0.12, this should be only the tag passed to tf.Summary. 1391 # By default, "training/hptuning/metric" will be used. 1392 "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. 1393 # You can reduce the time it takes to perform hyperparameter tuning by adding 1394 # trials in parallel. However, each trail only benefits from the information 1395 # gained in completed trials. That means that a trial does not get access to 1396 # the results of trials running at the same time, which could reduce the 1397 # quality of the overall optimization. 1398 # 1399 # Each trial will use the same scale tier and machine types. 1400 # 1401 # Defaults to one. 1402 }, 1403 "region": "A String", # Required. The Google Compute Engine region to run the training job in. 1404 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 1405 # for AI Platform services. 1406 "args": [ # Optional. Command line arguments to pass to the program. 1407 "A String", 1408 ], 1409 "pythonModule": "A String", # Required. The Python module name to run after installing the packages. 1410 "pythonVersion": "A String", # Optional. The version of Python used in training. If not set, the default 1411 # version is '2.7'. Python '3.5' is available when `runtime_version` is set 1412 # to '1.4' and above. Python '2.7' works with all supported 1413 # <a href="/ml-engine/docs/runtime-version-list">runtime versions</a>. 1414 "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs 1415 # and other data needed for training. This path is passed to your TensorFlow 1416 # program as the '--job-dir' command-line argument. The benefit of specifying 1417 # this field is that Cloud ML validates the path for use in training. 1418 "packageUris": [ # Required. The Google Cloud Storage location of the packages with 1419 # the training program and any additional dependencies. 1420 # The maximum number of package URIs is 100. 1421 "A String", 1422 ], 1423 "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each 1424 # replica in the cluster will be of the type specified in `worker_type`. 1425 # 1426 # This value can only be used when `scale_tier` is set to `CUSTOM`. If you 1427 # set this value, you must also set `worker_type`. 1428 # 1429 # The default value is zero. 1430 "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 1431 # job's parameter server. 1432 # 1433 # The supported values are the same as those described in the entry for 1434 # `master_type`. 1435 # 1436 # This value must be consistent with the category of machine type that 1437 # `masterType` uses. In other words, both must be AI Platform machine 1438 # types or both must be Compute Engine machine types. 1439 # 1440 # This value must be present when `scaleTier` is set to `CUSTOM` and 1441 # `parameter_server_count` is greater than zero. 1442 "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. 1443 # 1444 # You should only set `workerConfig.acceleratorConfig` if `workerType` is set 1445 # to a Compute Engine machine type. [Learn about restrictions on accelerator 1446 # configurations for 1447 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 1448 # 1449 # Set `workerConfig.imageUri` only if you build a custom image for your 1450 # worker. If `workerConfig.imageUri` has not been set, AI Platform uses 1451 # the value of `masterConfig.imageUri`. Learn more about 1452 # [configuring custom 1453 # containers](/ml-engine/docs/distributed-training-containers). 1454 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 1455 # [Learn about restrictions on accelerator configurations for 1456 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 1457 "count": "A String", # The number of accelerators to attach to each machine running the job. 1458 "type": "A String", # The type of accelerator to use. 1459 }, 1460 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 1461 # Registry. Learn more about [configuring custom 1462 # containers](/ml-engine/docs/distributed-training-containers). 1463 }, 1464 "maxRunningTime": "A String", # Optional. The maximum job running time. The default is 7 days. 1465 "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. 1466 # 1467 # You should only set `masterConfig.acceleratorConfig` if `masterType` is set 1468 # to a Compute Engine machine type. Learn about [restrictions on accelerator 1469 # configurations for 1470 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 1471 # 1472 # Set `masterConfig.imageUri` only if you build a custom image. Only one of 1473 # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more about 1474 # [configuring custom 1475 # containers](/ml-engine/docs/distributed-training-containers). 1476 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 1477 # [Learn about restrictions on accelerator configurations for 1478 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 1479 "count": "A String", # The number of accelerators to attach to each machine running the job. 1480 "type": "A String", # The type of accelerator to use. 1481 }, 1482 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 1483 # Registry. Learn more about [configuring custom 1484 # containers](/ml-engine/docs/distributed-training-containers). 1485 }, 1486 "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training 1487 # job. Each replica in the cluster will be of the type specified in 1488 # `parameter_server_type`. 1489 # 1490 # This value can only be used when `scale_tier` is set to `CUSTOM`.If you 1491 # set this value, you must also set `parameter_server_type`. 1492 # 1493 # The default value is zero. 1494 }, 1495 "jobId": "A String", # Required. The user-specified id of the job. 1496 "labels": { # Optional. One or more labels that you can add, to organize your jobs. 1497 # Each label is a key-value pair, where both the key and the value are 1498 # arbitrary strings that you supply. 1499 # For more information, see the documentation on 1500 # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. 1501 "a_key": "A String", 1502 }, 1503 "state": "A String", # Output only. The detailed state of a job. 1504 "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help 1505 # prevent simultaneous updates of a job from overwriting each other. 1506 # It is strongly suggested that systems make use of the `etag` in the 1507 # read-modify-write cycle to perform job updates in order to avoid race 1508 # conditions: An `etag` is returned in the response to `GetJob`, and 1509 # systems are expected to put that etag in the request to `UpdateJob` to 1510 # ensure that their change will be applied to the same version of the job. 1511 "startTime": "A String", # Output only. When the job processing was started. 1512 "endTime": "A String", # Output only. When the job processing was completed. 1513 "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. 1514 "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. 1515 "nodeHours": 3.14, # Node hours used by the batch prediction job. 1516 "predictionCount": "A String", # The number of generated predictions. 1517 "errorCount": "A String", # The number of data instances which resulted in errors. 1518 }, 1519 "createTime": "A String", # Output only. When the job was created. 1520 }</pre> 1521</div> 1522 1523<div class="method"> 1524 <code class="details" id="getIamPolicy">getIamPolicy(resource, x__xgafv=None)</code> 1525 <pre>Gets the access control policy for a resource. 1526Returns an empty policy if the resource exists and does not have a policy 1527set. 1528 1529Args: 1530 resource: string, REQUIRED: The resource for which the policy is being requested. 1531See the operation documentation for the appropriate value for this field. (required) 1532 x__xgafv: string, V1 error format. 1533 Allowed values 1534 1 - v1 error format 1535 2 - v2 error format 1536 1537Returns: 1538 An object of the form: 1539 1540 { # Defines an Identity and Access Management (IAM) policy. It is used to 1541 # specify access control policies for Cloud Platform resources. 1542 # 1543 # 1544 # A `Policy` consists of a list of `bindings`. A `binding` binds a list of 1545 # `members` to a `role`, where the members can be user accounts, Google groups, 1546 # Google domains, and service accounts. A `role` is a named list of permissions 1547 # defined by IAM. 1548 # 1549 # **JSON Example** 1550 # 1551 # { 1552 # "bindings": [ 1553 # { 1554 # "role": "roles/owner", 1555 # "members": [ 1556 # "user:mike@example.com", 1557 # "group:admins@example.com", 1558 # "domain:google.com", 1559 # "serviceAccount:my-other-app@appspot.gserviceaccount.com" 1560 # ] 1561 # }, 1562 # { 1563 # "role": "roles/viewer", 1564 # "members": ["user:sean@example.com"] 1565 # } 1566 # ] 1567 # } 1568 # 1569 # **YAML Example** 1570 # 1571 # bindings: 1572 # - members: 1573 # - user:mike@example.com 1574 # - group:admins@example.com 1575 # - domain:google.com 1576 # - serviceAccount:my-other-app@appspot.gserviceaccount.com 1577 # role: roles/owner 1578 # - members: 1579 # - user:sean@example.com 1580 # role: roles/viewer 1581 # 1582 # 1583 # For a description of IAM and its features, see the 1584 # [IAM developer's guide](https://cloud.google.com/iam/docs). 1585 "bindings": [ # Associates a list of `members` to a `role`. 1586 # `bindings` with no members will result in an error. 1587 { # Associates `members` with a `role`. 1588 "role": "A String", # Role that is assigned to `members`. 1589 # For example, `roles/viewer`, `roles/editor`, or `roles/owner`. 1590 "members": [ # Specifies the identities requesting access for a Cloud Platform resource. 1591 # `members` can have the following values: 1592 # 1593 # * `allUsers`: A special identifier that represents anyone who is 1594 # on the internet; with or without a Google account. 1595 # 1596 # * `allAuthenticatedUsers`: A special identifier that represents anyone 1597 # who is authenticated with a Google account or a service account. 1598 # 1599 # * `user:{emailid}`: An email address that represents a specific Google 1600 # account. For example, `alice@gmail.com` . 1601 # 1602 # 1603 # * `serviceAccount:{emailid}`: An email address that represents a service 1604 # account. For example, `my-other-app@appspot.gserviceaccount.com`. 1605 # 1606 # * `group:{emailid}`: An email address that represents a Google group. 1607 # For example, `admins@example.com`. 1608 # 1609 # 1610 # * `domain:{domain}`: The G Suite domain (primary) that represents all the 1611 # users of that domain. For example, `google.com` or `example.com`. 1612 # 1613 "A String", 1614 ], 1615 "condition": { # Represents an expression text. Example: # The condition that is associated with this binding. 1616 # NOTE: An unsatisfied condition will not allow user access via current 1617 # binding. Different bindings, including their conditions, are examined 1618 # independently. 1619 # 1620 # title: "User account presence" 1621 # description: "Determines whether the request has a user account" 1622 # expression: "size(request.user) > 0" 1623 "description": "A String", # An optional description of the expression. This is a longer text which 1624 # describes the expression, e.g. when hovered over it in a UI. 1625 "expression": "A String", # Textual representation of an expression in 1626 # Common Expression Language syntax. 1627 # 1628 # The application context of the containing message determines which 1629 # well-known feature set of CEL is supported. 1630 "location": "A String", # An optional string indicating the location of the expression for error 1631 # reporting, e.g. a file name and a position in the file. 1632 "title": "A String", # An optional title for the expression, i.e. a short string describing 1633 # its purpose. This can be used e.g. in UIs which allow to enter the 1634 # expression. 1635 }, 1636 }, 1637 ], 1638 "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help 1639 # prevent simultaneous updates of a policy from overwriting each other. 1640 # It is strongly suggested that systems make use of the `etag` in the 1641 # read-modify-write cycle to perform policy updates in order to avoid race 1642 # conditions: An `etag` is returned in the response to `getIamPolicy`, and 1643 # systems are expected to put that etag in the request to `setIamPolicy` to 1644 # ensure that their change will be applied to the same version of the policy. 1645 # 1646 # If no `etag` is provided in the call to `setIamPolicy`, then the existing 1647 # policy is overwritten blindly. 1648 "version": 42, # Deprecated. 1649 "auditConfigs": [ # Specifies cloud audit logging configuration for this policy. 1650 { # Specifies the audit configuration for a service. 1651 # The configuration determines which permission types are logged, and what 1652 # identities, if any, are exempted from logging. 1653 # An AuditConfig must have one or more AuditLogConfigs. 1654 # 1655 # If there are AuditConfigs for both `allServices` and a specific service, 1656 # the union of the two AuditConfigs is used for that service: the log_types 1657 # specified in each AuditConfig are enabled, and the exempted_members in each 1658 # AuditLogConfig are exempted. 1659 # 1660 # Example Policy with multiple AuditConfigs: 1661 # 1662 # { 1663 # "audit_configs": [ 1664 # { 1665 # "service": "allServices" 1666 # "audit_log_configs": [ 1667 # { 1668 # "log_type": "DATA_READ", 1669 # "exempted_members": [ 1670 # "user:foo@gmail.com" 1671 # ] 1672 # }, 1673 # { 1674 # "log_type": "DATA_WRITE", 1675 # }, 1676 # { 1677 # "log_type": "ADMIN_READ", 1678 # } 1679 # ] 1680 # }, 1681 # { 1682 # "service": "fooservice.googleapis.com" 1683 # "audit_log_configs": [ 1684 # { 1685 # "log_type": "DATA_READ", 1686 # }, 1687 # { 1688 # "log_type": "DATA_WRITE", 1689 # "exempted_members": [ 1690 # "user:bar@gmail.com" 1691 # ] 1692 # } 1693 # ] 1694 # } 1695 # ] 1696 # } 1697 # 1698 # For fooservice, this policy enables DATA_READ, DATA_WRITE and ADMIN_READ 1699 # logging. It also exempts foo@gmail.com from DATA_READ logging, and 1700 # bar@gmail.com from DATA_WRITE logging. 1701 "auditLogConfigs": [ # The configuration for logging of each type of permission. 1702 { # Provides the configuration for logging a type of permissions. 1703 # Example: 1704 # 1705 # { 1706 # "audit_log_configs": [ 1707 # { 1708 # "log_type": "DATA_READ", 1709 # "exempted_members": [ 1710 # "user:foo@gmail.com" 1711 # ] 1712 # }, 1713 # { 1714 # "log_type": "DATA_WRITE", 1715 # } 1716 # ] 1717 # } 1718 # 1719 # This enables 'DATA_READ' and 'DATA_WRITE' logging, while exempting 1720 # foo@gmail.com from DATA_READ logging. 1721 "exemptedMembers": [ # Specifies the identities that do not cause logging for this type of 1722 # permission. 1723 # Follows the same format of Binding.members. 1724 "A String", 1725 ], 1726 "logType": "A String", # The log type that this config enables. 1727 }, 1728 ], 1729 "service": "A String", # Specifies a service that will be enabled for audit logging. 1730 # For example, `storage.googleapis.com`, `cloudsql.googleapis.com`. 1731 # `allServices` is a special value that covers all services. 1732 }, 1733 ], 1734 }</pre> 1735</div> 1736 1737<div class="method"> 1738 <code class="details" id="list">list(parent, pageToken=None, x__xgafv=None, pageSize=None, filter=None)</code> 1739 <pre>Lists the jobs in the project. 1740 1741If there are no jobs that match the request parameters, the list 1742request returns an empty response body: {}. 1743 1744Args: 1745 parent: string, Required. The name of the project for which to list jobs. (required) 1746 pageToken: string, Optional. A page token to request the next page of results. 1747 1748You get the token from the `next_page_token` field of the response from 1749the previous call. 1750 x__xgafv: string, V1 error format. 1751 Allowed values 1752 1 - v1 error format 1753 2 - v2 error format 1754 pageSize: integer, Optional. The number of jobs to retrieve per "page" of results. If there 1755are more remaining results than this number, the response message will 1756contain a valid value in the `next_page_token` field. 1757 1758The default value is 20, and the maximum page size is 100. 1759 filter: string, Optional. Specifies the subset of jobs to retrieve. 1760You can filter on the value of one or more attributes of the job object. 1761For example, retrieve jobs with a job identifier that starts with 'census': 1762<p><code>gcloud ai-platform jobs list --filter='jobId:census*'</code> 1763<p>List all failed jobs with names that start with 'rnn': 1764<p><code>gcloud ai-platform jobs list --filter='jobId:rnn* 1765AND state:FAILED'</code> 1766<p>For more examples, see the guide to 1767<a href="/ml-engine/docs/tensorflow/monitor-training">monitoring jobs</a>. 1768 1769Returns: 1770 An object of the form: 1771 1772 { # Response message for the ListJobs method. 1773 "nextPageToken": "A String", # Optional. Pass this token as the `page_token` field of the request for a 1774 # subsequent call. 1775 "jobs": [ # The list of jobs. 1776 { # Represents a training or prediction job. 1777 "errorMessage": "A String", # Output only. The details of a failure or a cancellation. 1778 "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. 1779 "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. 1780 # Only set for hyperparameter tuning jobs. 1781 "trials": [ # Results for individual Hyperparameter trials. 1782 # Only set for hyperparameter tuning jobs. 1783 { # Represents the result of a single hyperparameter tuning trial from a 1784 # training job. The TrainingOutput object that is returned on successful 1785 # completion of a training job with hyperparameter tuning includes a list 1786 # of HyperparameterOutput objects, one for each successful trial. 1787 "hyperparameters": { # The hyperparameters given to this trial. 1788 "a_key": "A String", 1789 }, 1790 "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. 1791 "trainingStep": "A String", # The global training step for this metric. 1792 "objectiveValue": 3.14, # The objective value at this training step. 1793 }, 1794 "allMetrics": [ # All recorded object metrics for this trial. This field is not currently 1795 # populated. 1796 { # An observed value of a metric. 1797 "trainingStep": "A String", # The global training step for this metric. 1798 "objectiveValue": 3.14, # The objective value at this training step. 1799 }, 1800 ], 1801 "isTrialStoppedEarly": True or False, # True if the trial is stopped early. 1802 "trialId": "A String", # The trial id for these results. 1803 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 1804 # Only set for trials of built-in algorithms jobs that have succeeded. 1805 "framework": "A String", # Framework on which the built-in algorithm was trained. 1806 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 1807 # saves the trained model. Only set for successful jobs that don't use 1808 # hyperparameter tuning. 1809 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 1810 # trained. 1811 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 1812 }, 1813 }, 1814 ], 1815 "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. 1816 "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. 1817 "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. 1818 "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning 1819 # trials. See 1820 # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) 1821 # for more information. Only set for hyperparameter tuning jobs. 1822 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 1823 # Only set for built-in algorithms jobs. 1824 "framework": "A String", # Framework on which the built-in algorithm was trained. 1825 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 1826 # saves the trained model. Only set for successful jobs that don't use 1827 # hyperparameter tuning. 1828 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 1829 # trained. 1830 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 1831 }, 1832 }, 1833 "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. 1834 "modelName": "A String", # Use this field if you want to use the default version for the specified 1835 # model. The string must use the following format: 1836 # 1837 # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` 1838 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch 1839 # prediction. If not set, AI Platform will pick the runtime version used 1840 # during the CreateVersion request for this model version, or choose the 1841 # latest stable version when model version information is not available 1842 # such as when the model is specified by uri. 1843 "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for 1844 # this job. Please refer to 1845 # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) 1846 # for information about how to use signatures. 1847 # 1848 # Defaults to 1849 # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) 1850 # , which is "serving_default". 1851 "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. 1852 # The service will buffer batch_size number of records in memory before 1853 # invoking one Tensorflow prediction call internally. So take the record 1854 # size and memory available into consideration when setting this parameter. 1855 "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain 1856 # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. 1857 "A String", 1858 ], 1859 "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. 1860 # Defaults to 10 if not specified. 1861 "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for 1862 # the model to use. 1863 "outputPath": "A String", # Required. The output Google Cloud Storage location. 1864 "dataFormat": "A String", # Required. The format of the input data files. 1865 "versionName": "A String", # Use this field if you want to specify a version of the model to use. The 1866 # string is formatted the same way as `model_version`, with the addition 1867 # of the version information: 1868 # 1869 # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` 1870 "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. 1871 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 1872 # for AI Platform services. 1873 "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. 1874 }, 1875 "trainingInput": { # Represents input parameters for a training job. When using the # Input parameters to create a training job. 1876 # gcloud command to submit your training job, you can specify 1877 # the input parameters as command-line arguments and/or in a YAML configuration 1878 # file referenced from the --config command-line argument. For 1879 # details, see the guide to 1880 # <a href="/ml-engine/docs/tensorflow/training-jobs">submitting a training 1881 # job</a>. 1882 "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 1883 # job's worker nodes. 1884 # 1885 # The supported values are the same as those described in the entry for 1886 # `masterType`. 1887 # 1888 # This value must be consistent with the category of machine type that 1889 # `masterType` uses. In other words, both must be AI Platform machine 1890 # types or both must be Compute Engine machine types. 1891 # 1892 # If you use `cloud_tpu` for this value, see special instructions for 1893 # [configuring a custom TPU 1894 # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). 1895 # 1896 # This value must be present when `scaleTier` is set to `CUSTOM` and 1897 # `workerCount` is greater than zero. 1898 "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. 1899 # 1900 # You should only set `parameterServerConfig.acceleratorConfig` if 1901 # `parameterServerConfigType` is set to a Compute Engine machine type. [Learn 1902 # about restrictions on accelerator configurations for 1903 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 1904 # 1905 # Set `parameterServerConfig.imageUri` only if you build a custom image for 1906 # your parameter server. If `parameterServerConfig.imageUri` has not been 1907 # set, AI Platform uses the value of `masterConfig.imageUri`. 1908 # Learn more about [configuring custom 1909 # containers](/ml-engine/docs/distributed-training-containers). 1910 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 1911 # [Learn about restrictions on accelerator configurations for 1912 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 1913 "count": "A String", # The number of accelerators to attach to each machine running the job. 1914 "type": "A String", # The type of accelerator to use. 1915 }, 1916 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 1917 # Registry. Learn more about [configuring custom 1918 # containers](/ml-engine/docs/distributed-training-containers). 1919 }, 1920 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. If not 1921 # set, AI Platform uses the default stable version, 1.0. For more 1922 # information, see the 1923 # <a href="/ml-engine/docs/runtime-version-list">runtime version list</a> 1924 # and 1925 # <a href="/ml-engine/docs/versioning">how to manage runtime versions</a>. 1926 "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers 1927 # and parameter servers. 1928 "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training 1929 # job's master worker. 1930 # 1931 # The following types are supported: 1932 # 1933 # <dl> 1934 # <dt>standard</dt> 1935 # <dd> 1936 # A basic machine configuration suitable for training simple models with 1937 # small to moderate datasets. 1938 # </dd> 1939 # <dt>large_model</dt> 1940 # <dd> 1941 # A machine with a lot of memory, specially suited for parameter servers 1942 # when your model is large (having many hidden layers or layers with very 1943 # large numbers of nodes). 1944 # </dd> 1945 # <dt>complex_model_s</dt> 1946 # <dd> 1947 # A machine suitable for the master and workers of the cluster when your 1948 # model requires more computation than the standard machine can handle 1949 # satisfactorily. 1950 # </dd> 1951 # <dt>complex_model_m</dt> 1952 # <dd> 1953 # A machine with roughly twice the number of cores and roughly double the 1954 # memory of <i>complex_model_s</i>. 1955 # </dd> 1956 # <dt>complex_model_l</dt> 1957 # <dd> 1958 # A machine with roughly twice the number of cores and roughly double the 1959 # memory of <i>complex_model_m</i>. 1960 # </dd> 1961 # <dt>standard_gpu</dt> 1962 # <dd> 1963 # A machine equivalent to <i>standard</i> that 1964 # also includes a single NVIDIA Tesla K80 GPU. See more about 1965 # <a href="/ml-engine/docs/tensorflow/using-gpus">using GPUs to 1966 # train your model</a>. 1967 # </dd> 1968 # <dt>complex_model_m_gpu</dt> 1969 # <dd> 1970 # A machine equivalent to <i>complex_model_m</i> that also includes 1971 # four NVIDIA Tesla K80 GPUs. 1972 # </dd> 1973 # <dt>complex_model_l_gpu</dt> 1974 # <dd> 1975 # A machine equivalent to <i>complex_model_l</i> that also includes 1976 # eight NVIDIA Tesla K80 GPUs. 1977 # </dd> 1978 # <dt>standard_p100</dt> 1979 # <dd> 1980 # A machine equivalent to <i>standard</i> that 1981 # also includes a single NVIDIA Tesla P100 GPU. 1982 # </dd> 1983 # <dt>complex_model_m_p100</dt> 1984 # <dd> 1985 # A machine equivalent to <i>complex_model_m</i> that also includes 1986 # four NVIDIA Tesla P100 GPUs. 1987 # </dd> 1988 # <dt>standard_v100</dt> 1989 # <dd> 1990 # A machine equivalent to <i>standard</i> that 1991 # also includes a single NVIDIA Tesla V100 GPU. 1992 # </dd> 1993 # <dt>large_model_v100</dt> 1994 # <dd> 1995 # A machine equivalent to <i>large_model</i> that 1996 # also includes a single NVIDIA Tesla V100 GPU. 1997 # </dd> 1998 # <dt>complex_model_m_v100</dt> 1999 # <dd> 2000 # A machine equivalent to <i>complex_model_m</i> that 2001 # also includes four NVIDIA Tesla V100 GPUs. 2002 # </dd> 2003 # <dt>complex_model_l_v100</dt> 2004 # <dd> 2005 # A machine equivalent to <i>complex_model_l</i> that 2006 # also includes eight NVIDIA Tesla V100 GPUs. 2007 # </dd> 2008 # <dt>cloud_tpu</dt> 2009 # <dd> 2010 # A TPU VM including one Cloud TPU. See more about 2011 # <a href="/ml-engine/docs/tensorflow/using-tpus">using TPUs to train 2012 # your model</a>. 2013 # </dd> 2014 # </dl> 2015 # 2016 # You may also use certain Compute Engine machine types directly in this 2017 # field. The following types are supported: 2018 # 2019 # - `n1-standard-4` 2020 # - `n1-standard-8` 2021 # - `n1-standard-16` 2022 # - `n1-standard-32` 2023 # - `n1-standard-64` 2024 # - `n1-standard-96` 2025 # - `n1-highmem-2` 2026 # - `n1-highmem-4` 2027 # - `n1-highmem-8` 2028 # - `n1-highmem-16` 2029 # - `n1-highmem-32` 2030 # - `n1-highmem-64` 2031 # - `n1-highmem-96` 2032 # - `n1-highcpu-16` 2033 # - `n1-highcpu-32` 2034 # - `n1-highcpu-64` 2035 # - `n1-highcpu-96` 2036 # 2037 # See more about [using Compute Engine machine 2038 # types](/ml-engine/docs/tensorflow/machine-types#compute-engine-machine-types). 2039 # 2040 # You must set this value when `scaleTier` is set to `CUSTOM`. 2041 "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. 2042 "maxTrials": 42, # Optional. How many training trials should be attempted to optimize 2043 # the specified hyperparameters. 2044 # 2045 # Defaults to one. 2046 "goal": "A String", # Required. The type of goal to use for tuning. Available types are 2047 # `MAXIMIZE` and `MINIMIZE`. 2048 # 2049 # Defaults to `MAXIMIZE`. 2050 "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter 2051 # tuning job. 2052 # Uses the default AI Platform hyperparameter tuning 2053 # algorithm if unspecified. 2054 "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing 2055 # the hyperparameter tuning job. You can specify this field to override the 2056 # default failing criteria for AI Platform hyperparameter tuning jobs. 2057 # 2058 # Defaults to zero, which means the service decides when a hyperparameter 2059 # job should fail. 2060 "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial 2061 # early stopping. 2062 "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to 2063 # continue with. The job id will be used to find the corresponding vizier 2064 # study guid and resume the study. 2065 "params": [ # Required. The set of parameters to tune. 2066 { # Represents a single hyperparameter to optimize. 2067 "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 2068 # should be unset if type is `CATEGORICAL`. This value should be integers if 2069 # type is `INTEGER`. 2070 "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. 2071 "A String", 2072 ], 2073 "discreteValues": [ # Required if type is `DISCRETE`. 2074 # A list of feasible points. 2075 # The list should be in strictly increasing order. For instance, this 2076 # parameter might have possible settings of 1.5, 2.5, and 4.0. This list 2077 # should not contain more than 1,000 values. 2078 3.14, 2079 ], 2080 "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in 2081 # a HyperparameterSpec message. E.g., "learning_rate". 2082 "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 2083 # should be unset if type is `CATEGORICAL`. This value should be integers if 2084 # type is INTEGER. 2085 "type": "A String", # Required. The type of the parameter. 2086 "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. 2087 # Leave unset for categorical parameters. 2088 # Some kind of scaling is strongly recommended for real or integral 2089 # parameters (e.g., `UNIT_LINEAR_SCALE`). 2090 }, 2091 ], 2092 "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For 2093 # current versions of TensorFlow, this tag name should exactly match what is 2094 # shown in TensorBoard, including all scopes. For versions of TensorFlow 2095 # prior to 0.12, this should be only the tag passed to tf.Summary. 2096 # By default, "training/hptuning/metric" will be used. 2097 "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. 2098 # You can reduce the time it takes to perform hyperparameter tuning by adding 2099 # trials in parallel. However, each trail only benefits from the information 2100 # gained in completed trials. That means that a trial does not get access to 2101 # the results of trials running at the same time, which could reduce the 2102 # quality of the overall optimization. 2103 # 2104 # Each trial will use the same scale tier and machine types. 2105 # 2106 # Defaults to one. 2107 }, 2108 "region": "A String", # Required. The Google Compute Engine region to run the training job in. 2109 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 2110 # for AI Platform services. 2111 "args": [ # Optional. Command line arguments to pass to the program. 2112 "A String", 2113 ], 2114 "pythonModule": "A String", # Required. The Python module name to run after installing the packages. 2115 "pythonVersion": "A String", # Optional. The version of Python used in training. If not set, the default 2116 # version is '2.7'. Python '3.5' is available when `runtime_version` is set 2117 # to '1.4' and above. Python '2.7' works with all supported 2118 # <a href="/ml-engine/docs/runtime-version-list">runtime versions</a>. 2119 "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs 2120 # and other data needed for training. This path is passed to your TensorFlow 2121 # program as the '--job-dir' command-line argument. The benefit of specifying 2122 # this field is that Cloud ML validates the path for use in training. 2123 "packageUris": [ # Required. The Google Cloud Storage location of the packages with 2124 # the training program and any additional dependencies. 2125 # The maximum number of package URIs is 100. 2126 "A String", 2127 ], 2128 "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each 2129 # replica in the cluster will be of the type specified in `worker_type`. 2130 # 2131 # This value can only be used when `scale_tier` is set to `CUSTOM`. If you 2132 # set this value, you must also set `worker_type`. 2133 # 2134 # The default value is zero. 2135 "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 2136 # job's parameter server. 2137 # 2138 # The supported values are the same as those described in the entry for 2139 # `master_type`. 2140 # 2141 # This value must be consistent with the category of machine type that 2142 # `masterType` uses. In other words, both must be AI Platform machine 2143 # types or both must be Compute Engine machine types. 2144 # 2145 # This value must be present when `scaleTier` is set to `CUSTOM` and 2146 # `parameter_server_count` is greater than zero. 2147 "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. 2148 # 2149 # You should only set `workerConfig.acceleratorConfig` if `workerType` is set 2150 # to a Compute Engine machine type. [Learn about restrictions on accelerator 2151 # configurations for 2152 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2153 # 2154 # Set `workerConfig.imageUri` only if you build a custom image for your 2155 # worker. If `workerConfig.imageUri` has not been set, AI Platform uses 2156 # the value of `masterConfig.imageUri`. Learn more about 2157 # [configuring custom 2158 # containers](/ml-engine/docs/distributed-training-containers). 2159 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 2160 # [Learn about restrictions on accelerator configurations for 2161 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2162 "count": "A String", # The number of accelerators to attach to each machine running the job. 2163 "type": "A String", # The type of accelerator to use. 2164 }, 2165 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 2166 # Registry. Learn more about [configuring custom 2167 # containers](/ml-engine/docs/distributed-training-containers). 2168 }, 2169 "maxRunningTime": "A String", # Optional. The maximum job running time. The default is 7 days. 2170 "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. 2171 # 2172 # You should only set `masterConfig.acceleratorConfig` if `masterType` is set 2173 # to a Compute Engine machine type. Learn about [restrictions on accelerator 2174 # configurations for 2175 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2176 # 2177 # Set `masterConfig.imageUri` only if you build a custom image. Only one of 2178 # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more about 2179 # [configuring custom 2180 # containers](/ml-engine/docs/distributed-training-containers). 2181 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 2182 # [Learn about restrictions on accelerator configurations for 2183 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2184 "count": "A String", # The number of accelerators to attach to each machine running the job. 2185 "type": "A String", # The type of accelerator to use. 2186 }, 2187 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 2188 # Registry. Learn more about [configuring custom 2189 # containers](/ml-engine/docs/distributed-training-containers). 2190 }, 2191 "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training 2192 # job. Each replica in the cluster will be of the type specified in 2193 # `parameter_server_type`. 2194 # 2195 # This value can only be used when `scale_tier` is set to `CUSTOM`.If you 2196 # set this value, you must also set `parameter_server_type`. 2197 # 2198 # The default value is zero. 2199 }, 2200 "jobId": "A String", # Required. The user-specified id of the job. 2201 "labels": { # Optional. One or more labels that you can add, to organize your jobs. 2202 # Each label is a key-value pair, where both the key and the value are 2203 # arbitrary strings that you supply. 2204 # For more information, see the documentation on 2205 # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. 2206 "a_key": "A String", 2207 }, 2208 "state": "A String", # Output only. The detailed state of a job. 2209 "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help 2210 # prevent simultaneous updates of a job from overwriting each other. 2211 # It is strongly suggested that systems make use of the `etag` in the 2212 # read-modify-write cycle to perform job updates in order to avoid race 2213 # conditions: An `etag` is returned in the response to `GetJob`, and 2214 # systems are expected to put that etag in the request to `UpdateJob` to 2215 # ensure that their change will be applied to the same version of the job. 2216 "startTime": "A String", # Output only. When the job processing was started. 2217 "endTime": "A String", # Output only. When the job processing was completed. 2218 "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. 2219 "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. 2220 "nodeHours": 3.14, # Node hours used by the batch prediction job. 2221 "predictionCount": "A String", # The number of generated predictions. 2222 "errorCount": "A String", # The number of data instances which resulted in errors. 2223 }, 2224 "createTime": "A String", # Output only. When the job was created. 2225 }, 2226 ], 2227 }</pre> 2228</div> 2229 2230<div class="method"> 2231 <code class="details" id="list_next">list_next(previous_request, previous_response)</code> 2232 <pre>Retrieves the next page of results. 2233 2234Args: 2235 previous_request: The request for the previous page. (required) 2236 previous_response: The response from the request for the previous page. (required) 2237 2238Returns: 2239 A request object that you can call 'execute()' on to request the next 2240 page. Returns None if there are no more items in the collection. 2241 </pre> 2242</div> 2243 2244<div class="method"> 2245 <code class="details" id="patch">patch(name, body, updateMask=None, x__xgafv=None)</code> 2246 <pre>Updates a specific job resource. 2247 2248Currently the only supported fields to update are `labels`. 2249 2250Args: 2251 name: string, Required. The job name. (required) 2252 body: object, The request body. (required) 2253 The object takes the form of: 2254 2255{ # Represents a training or prediction job. 2256 "errorMessage": "A String", # Output only. The details of a failure or a cancellation. 2257 "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. 2258 "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. 2259 # Only set for hyperparameter tuning jobs. 2260 "trials": [ # Results for individual Hyperparameter trials. 2261 # Only set for hyperparameter tuning jobs. 2262 { # Represents the result of a single hyperparameter tuning trial from a 2263 # training job. The TrainingOutput object that is returned on successful 2264 # completion of a training job with hyperparameter tuning includes a list 2265 # of HyperparameterOutput objects, one for each successful trial. 2266 "hyperparameters": { # The hyperparameters given to this trial. 2267 "a_key": "A String", 2268 }, 2269 "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. 2270 "trainingStep": "A String", # The global training step for this metric. 2271 "objectiveValue": 3.14, # The objective value at this training step. 2272 }, 2273 "allMetrics": [ # All recorded object metrics for this trial. This field is not currently 2274 # populated. 2275 { # An observed value of a metric. 2276 "trainingStep": "A String", # The global training step for this metric. 2277 "objectiveValue": 3.14, # The objective value at this training step. 2278 }, 2279 ], 2280 "isTrialStoppedEarly": True or False, # True if the trial is stopped early. 2281 "trialId": "A String", # The trial id for these results. 2282 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 2283 # Only set for trials of built-in algorithms jobs that have succeeded. 2284 "framework": "A String", # Framework on which the built-in algorithm was trained. 2285 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 2286 # saves the trained model. Only set for successful jobs that don't use 2287 # hyperparameter tuning. 2288 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 2289 # trained. 2290 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 2291 }, 2292 }, 2293 ], 2294 "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. 2295 "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. 2296 "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. 2297 "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning 2298 # trials. See 2299 # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) 2300 # for more information. Only set for hyperparameter tuning jobs. 2301 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 2302 # Only set for built-in algorithms jobs. 2303 "framework": "A String", # Framework on which the built-in algorithm was trained. 2304 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 2305 # saves the trained model. Only set for successful jobs that don't use 2306 # hyperparameter tuning. 2307 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 2308 # trained. 2309 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 2310 }, 2311 }, 2312 "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. 2313 "modelName": "A String", # Use this field if you want to use the default version for the specified 2314 # model. The string must use the following format: 2315 # 2316 # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` 2317 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch 2318 # prediction. If not set, AI Platform will pick the runtime version used 2319 # during the CreateVersion request for this model version, or choose the 2320 # latest stable version when model version information is not available 2321 # such as when the model is specified by uri. 2322 "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for 2323 # this job. Please refer to 2324 # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) 2325 # for information about how to use signatures. 2326 # 2327 # Defaults to 2328 # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) 2329 # , which is "serving_default". 2330 "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. 2331 # The service will buffer batch_size number of records in memory before 2332 # invoking one Tensorflow prediction call internally. So take the record 2333 # size and memory available into consideration when setting this parameter. 2334 "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain 2335 # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. 2336 "A String", 2337 ], 2338 "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. 2339 # Defaults to 10 if not specified. 2340 "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for 2341 # the model to use. 2342 "outputPath": "A String", # Required. The output Google Cloud Storage location. 2343 "dataFormat": "A String", # Required. The format of the input data files. 2344 "versionName": "A String", # Use this field if you want to specify a version of the model to use. The 2345 # string is formatted the same way as `model_version`, with the addition 2346 # of the version information: 2347 # 2348 # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` 2349 "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. 2350 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 2351 # for AI Platform services. 2352 "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. 2353 }, 2354 "trainingInput": { # Represents input parameters for a training job. When using the # Input parameters to create a training job. 2355 # gcloud command to submit your training job, you can specify 2356 # the input parameters as command-line arguments and/or in a YAML configuration 2357 # file referenced from the --config command-line argument. For 2358 # details, see the guide to 2359 # <a href="/ml-engine/docs/tensorflow/training-jobs">submitting a training 2360 # job</a>. 2361 "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 2362 # job's worker nodes. 2363 # 2364 # The supported values are the same as those described in the entry for 2365 # `masterType`. 2366 # 2367 # This value must be consistent with the category of machine type that 2368 # `masterType` uses. In other words, both must be AI Platform machine 2369 # types or both must be Compute Engine machine types. 2370 # 2371 # If you use `cloud_tpu` for this value, see special instructions for 2372 # [configuring a custom TPU 2373 # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). 2374 # 2375 # This value must be present when `scaleTier` is set to `CUSTOM` and 2376 # `workerCount` is greater than zero. 2377 "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. 2378 # 2379 # You should only set `parameterServerConfig.acceleratorConfig` if 2380 # `parameterServerConfigType` is set to a Compute Engine machine type. [Learn 2381 # about restrictions on accelerator configurations for 2382 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2383 # 2384 # Set `parameterServerConfig.imageUri` only if you build a custom image for 2385 # your parameter server. If `parameterServerConfig.imageUri` has not been 2386 # set, AI Platform uses the value of `masterConfig.imageUri`. 2387 # Learn more about [configuring custom 2388 # containers](/ml-engine/docs/distributed-training-containers). 2389 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 2390 # [Learn about restrictions on accelerator configurations for 2391 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2392 "count": "A String", # The number of accelerators to attach to each machine running the job. 2393 "type": "A String", # The type of accelerator to use. 2394 }, 2395 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 2396 # Registry. Learn more about [configuring custom 2397 # containers](/ml-engine/docs/distributed-training-containers). 2398 }, 2399 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. If not 2400 # set, AI Platform uses the default stable version, 1.0. For more 2401 # information, see the 2402 # <a href="/ml-engine/docs/runtime-version-list">runtime version list</a> 2403 # and 2404 # <a href="/ml-engine/docs/versioning">how to manage runtime versions</a>. 2405 "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers 2406 # and parameter servers. 2407 "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training 2408 # job's master worker. 2409 # 2410 # The following types are supported: 2411 # 2412 # <dl> 2413 # <dt>standard</dt> 2414 # <dd> 2415 # A basic machine configuration suitable for training simple models with 2416 # small to moderate datasets. 2417 # </dd> 2418 # <dt>large_model</dt> 2419 # <dd> 2420 # A machine with a lot of memory, specially suited for parameter servers 2421 # when your model is large (having many hidden layers or layers with very 2422 # large numbers of nodes). 2423 # </dd> 2424 # <dt>complex_model_s</dt> 2425 # <dd> 2426 # A machine suitable for the master and workers of the cluster when your 2427 # model requires more computation than the standard machine can handle 2428 # satisfactorily. 2429 # </dd> 2430 # <dt>complex_model_m</dt> 2431 # <dd> 2432 # A machine with roughly twice the number of cores and roughly double the 2433 # memory of <i>complex_model_s</i>. 2434 # </dd> 2435 # <dt>complex_model_l</dt> 2436 # <dd> 2437 # A machine with roughly twice the number of cores and roughly double the 2438 # memory of <i>complex_model_m</i>. 2439 # </dd> 2440 # <dt>standard_gpu</dt> 2441 # <dd> 2442 # A machine equivalent to <i>standard</i> that 2443 # also includes a single NVIDIA Tesla K80 GPU. See more about 2444 # <a href="/ml-engine/docs/tensorflow/using-gpus">using GPUs to 2445 # train your model</a>. 2446 # </dd> 2447 # <dt>complex_model_m_gpu</dt> 2448 # <dd> 2449 # A machine equivalent to <i>complex_model_m</i> that also includes 2450 # four NVIDIA Tesla K80 GPUs. 2451 # </dd> 2452 # <dt>complex_model_l_gpu</dt> 2453 # <dd> 2454 # A machine equivalent to <i>complex_model_l</i> that also includes 2455 # eight NVIDIA Tesla K80 GPUs. 2456 # </dd> 2457 # <dt>standard_p100</dt> 2458 # <dd> 2459 # A machine equivalent to <i>standard</i> that 2460 # also includes a single NVIDIA Tesla P100 GPU. 2461 # </dd> 2462 # <dt>complex_model_m_p100</dt> 2463 # <dd> 2464 # A machine equivalent to <i>complex_model_m</i> that also includes 2465 # four NVIDIA Tesla P100 GPUs. 2466 # </dd> 2467 # <dt>standard_v100</dt> 2468 # <dd> 2469 # A machine equivalent to <i>standard</i> that 2470 # also includes a single NVIDIA Tesla V100 GPU. 2471 # </dd> 2472 # <dt>large_model_v100</dt> 2473 # <dd> 2474 # A machine equivalent to <i>large_model</i> that 2475 # also includes a single NVIDIA Tesla V100 GPU. 2476 # </dd> 2477 # <dt>complex_model_m_v100</dt> 2478 # <dd> 2479 # A machine equivalent to <i>complex_model_m</i> that 2480 # also includes four NVIDIA Tesla V100 GPUs. 2481 # </dd> 2482 # <dt>complex_model_l_v100</dt> 2483 # <dd> 2484 # A machine equivalent to <i>complex_model_l</i> that 2485 # also includes eight NVIDIA Tesla V100 GPUs. 2486 # </dd> 2487 # <dt>cloud_tpu</dt> 2488 # <dd> 2489 # A TPU VM including one Cloud TPU. See more about 2490 # <a href="/ml-engine/docs/tensorflow/using-tpus">using TPUs to train 2491 # your model</a>. 2492 # </dd> 2493 # </dl> 2494 # 2495 # You may also use certain Compute Engine machine types directly in this 2496 # field. The following types are supported: 2497 # 2498 # - `n1-standard-4` 2499 # - `n1-standard-8` 2500 # - `n1-standard-16` 2501 # - `n1-standard-32` 2502 # - `n1-standard-64` 2503 # - `n1-standard-96` 2504 # - `n1-highmem-2` 2505 # - `n1-highmem-4` 2506 # - `n1-highmem-8` 2507 # - `n1-highmem-16` 2508 # - `n1-highmem-32` 2509 # - `n1-highmem-64` 2510 # - `n1-highmem-96` 2511 # - `n1-highcpu-16` 2512 # - `n1-highcpu-32` 2513 # - `n1-highcpu-64` 2514 # - `n1-highcpu-96` 2515 # 2516 # See more about [using Compute Engine machine 2517 # types](/ml-engine/docs/tensorflow/machine-types#compute-engine-machine-types). 2518 # 2519 # You must set this value when `scaleTier` is set to `CUSTOM`. 2520 "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. 2521 "maxTrials": 42, # Optional. How many training trials should be attempted to optimize 2522 # the specified hyperparameters. 2523 # 2524 # Defaults to one. 2525 "goal": "A String", # Required. The type of goal to use for tuning. Available types are 2526 # `MAXIMIZE` and `MINIMIZE`. 2527 # 2528 # Defaults to `MAXIMIZE`. 2529 "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter 2530 # tuning job. 2531 # Uses the default AI Platform hyperparameter tuning 2532 # algorithm if unspecified. 2533 "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing 2534 # the hyperparameter tuning job. You can specify this field to override the 2535 # default failing criteria for AI Platform hyperparameter tuning jobs. 2536 # 2537 # Defaults to zero, which means the service decides when a hyperparameter 2538 # job should fail. 2539 "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial 2540 # early stopping. 2541 "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to 2542 # continue with. The job id will be used to find the corresponding vizier 2543 # study guid and resume the study. 2544 "params": [ # Required. The set of parameters to tune. 2545 { # Represents a single hyperparameter to optimize. 2546 "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 2547 # should be unset if type is `CATEGORICAL`. This value should be integers if 2548 # type is `INTEGER`. 2549 "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. 2550 "A String", 2551 ], 2552 "discreteValues": [ # Required if type is `DISCRETE`. 2553 # A list of feasible points. 2554 # The list should be in strictly increasing order. For instance, this 2555 # parameter might have possible settings of 1.5, 2.5, and 4.0. This list 2556 # should not contain more than 1,000 values. 2557 3.14, 2558 ], 2559 "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in 2560 # a HyperparameterSpec message. E.g., "learning_rate". 2561 "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 2562 # should be unset if type is `CATEGORICAL`. This value should be integers if 2563 # type is INTEGER. 2564 "type": "A String", # Required. The type of the parameter. 2565 "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. 2566 # Leave unset for categorical parameters. 2567 # Some kind of scaling is strongly recommended for real or integral 2568 # parameters (e.g., `UNIT_LINEAR_SCALE`). 2569 }, 2570 ], 2571 "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For 2572 # current versions of TensorFlow, this tag name should exactly match what is 2573 # shown in TensorBoard, including all scopes. For versions of TensorFlow 2574 # prior to 0.12, this should be only the tag passed to tf.Summary. 2575 # By default, "training/hptuning/metric" will be used. 2576 "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. 2577 # You can reduce the time it takes to perform hyperparameter tuning by adding 2578 # trials in parallel. However, each trail only benefits from the information 2579 # gained in completed trials. That means that a trial does not get access to 2580 # the results of trials running at the same time, which could reduce the 2581 # quality of the overall optimization. 2582 # 2583 # Each trial will use the same scale tier and machine types. 2584 # 2585 # Defaults to one. 2586 }, 2587 "region": "A String", # Required. The Google Compute Engine region to run the training job in. 2588 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 2589 # for AI Platform services. 2590 "args": [ # Optional. Command line arguments to pass to the program. 2591 "A String", 2592 ], 2593 "pythonModule": "A String", # Required. The Python module name to run after installing the packages. 2594 "pythonVersion": "A String", # Optional. The version of Python used in training. If not set, the default 2595 # version is '2.7'. Python '3.5' is available when `runtime_version` is set 2596 # to '1.4' and above. Python '2.7' works with all supported 2597 # <a href="/ml-engine/docs/runtime-version-list">runtime versions</a>. 2598 "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs 2599 # and other data needed for training. This path is passed to your TensorFlow 2600 # program as the '--job-dir' command-line argument. The benefit of specifying 2601 # this field is that Cloud ML validates the path for use in training. 2602 "packageUris": [ # Required. The Google Cloud Storage location of the packages with 2603 # the training program and any additional dependencies. 2604 # The maximum number of package URIs is 100. 2605 "A String", 2606 ], 2607 "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each 2608 # replica in the cluster will be of the type specified in `worker_type`. 2609 # 2610 # This value can only be used when `scale_tier` is set to `CUSTOM`. If you 2611 # set this value, you must also set `worker_type`. 2612 # 2613 # The default value is zero. 2614 "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 2615 # job's parameter server. 2616 # 2617 # The supported values are the same as those described in the entry for 2618 # `master_type`. 2619 # 2620 # This value must be consistent with the category of machine type that 2621 # `masterType` uses. In other words, both must be AI Platform machine 2622 # types or both must be Compute Engine machine types. 2623 # 2624 # This value must be present when `scaleTier` is set to `CUSTOM` and 2625 # `parameter_server_count` is greater than zero. 2626 "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. 2627 # 2628 # You should only set `workerConfig.acceleratorConfig` if `workerType` is set 2629 # to a Compute Engine machine type. [Learn about restrictions on accelerator 2630 # configurations for 2631 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2632 # 2633 # Set `workerConfig.imageUri` only if you build a custom image for your 2634 # worker. If `workerConfig.imageUri` has not been set, AI Platform uses 2635 # the value of `masterConfig.imageUri`. Learn more about 2636 # [configuring custom 2637 # containers](/ml-engine/docs/distributed-training-containers). 2638 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 2639 # [Learn about restrictions on accelerator configurations for 2640 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2641 "count": "A String", # The number of accelerators to attach to each machine running the job. 2642 "type": "A String", # The type of accelerator to use. 2643 }, 2644 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 2645 # Registry. Learn more about [configuring custom 2646 # containers](/ml-engine/docs/distributed-training-containers). 2647 }, 2648 "maxRunningTime": "A String", # Optional. The maximum job running time. The default is 7 days. 2649 "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. 2650 # 2651 # You should only set `masterConfig.acceleratorConfig` if `masterType` is set 2652 # to a Compute Engine machine type. Learn about [restrictions on accelerator 2653 # configurations for 2654 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2655 # 2656 # Set `masterConfig.imageUri` only if you build a custom image. Only one of 2657 # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more about 2658 # [configuring custom 2659 # containers](/ml-engine/docs/distributed-training-containers). 2660 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 2661 # [Learn about restrictions on accelerator configurations for 2662 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2663 "count": "A String", # The number of accelerators to attach to each machine running the job. 2664 "type": "A String", # The type of accelerator to use. 2665 }, 2666 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 2667 # Registry. Learn more about [configuring custom 2668 # containers](/ml-engine/docs/distributed-training-containers). 2669 }, 2670 "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training 2671 # job. Each replica in the cluster will be of the type specified in 2672 # `parameter_server_type`. 2673 # 2674 # This value can only be used when `scale_tier` is set to `CUSTOM`.If you 2675 # set this value, you must also set `parameter_server_type`. 2676 # 2677 # The default value is zero. 2678 }, 2679 "jobId": "A String", # Required. The user-specified id of the job. 2680 "labels": { # Optional. One or more labels that you can add, to organize your jobs. 2681 # Each label is a key-value pair, where both the key and the value are 2682 # arbitrary strings that you supply. 2683 # For more information, see the documentation on 2684 # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. 2685 "a_key": "A String", 2686 }, 2687 "state": "A String", # Output only. The detailed state of a job. 2688 "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help 2689 # prevent simultaneous updates of a job from overwriting each other. 2690 # It is strongly suggested that systems make use of the `etag` in the 2691 # read-modify-write cycle to perform job updates in order to avoid race 2692 # conditions: An `etag` is returned in the response to `GetJob`, and 2693 # systems are expected to put that etag in the request to `UpdateJob` to 2694 # ensure that their change will be applied to the same version of the job. 2695 "startTime": "A String", # Output only. When the job processing was started. 2696 "endTime": "A String", # Output only. When the job processing was completed. 2697 "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. 2698 "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. 2699 "nodeHours": 3.14, # Node hours used by the batch prediction job. 2700 "predictionCount": "A String", # The number of generated predictions. 2701 "errorCount": "A String", # The number of data instances which resulted in errors. 2702 }, 2703 "createTime": "A String", # Output only. When the job was created. 2704} 2705 2706 updateMask: string, Required. Specifies the path, relative to `Job`, of the field to update. 2707To adopt etag mechanism, include `etag` field in the mask, and include the 2708`etag` value in your job resource. 2709 2710For example, to change the labels of a job, the `update_mask` parameter 2711would be specified as `labels`, `etag`, and the 2712`PATCH` request body would specify the new value, as follows: 2713 { 2714 "labels": { 2715 "owner": "Google", 2716 "color": "Blue" 2717 } 2718 "etag": "33a64df551425fcc55e4d42a148795d9f25f89d4" 2719 } 2720If `etag` matches the one on the server, the labels of the job will be 2721replaced with the given ones, and the server end `etag` will be 2722recalculated. 2723 2724Currently the only supported update masks are `labels` and `etag`. 2725 x__xgafv: string, V1 error format. 2726 Allowed values 2727 1 - v1 error format 2728 2 - v2 error format 2729 2730Returns: 2731 An object of the form: 2732 2733 { # Represents a training or prediction job. 2734 "errorMessage": "A String", # Output only. The details of a failure or a cancellation. 2735 "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. 2736 "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. 2737 # Only set for hyperparameter tuning jobs. 2738 "trials": [ # Results for individual Hyperparameter trials. 2739 # Only set for hyperparameter tuning jobs. 2740 { # Represents the result of a single hyperparameter tuning trial from a 2741 # training job. The TrainingOutput object that is returned on successful 2742 # completion of a training job with hyperparameter tuning includes a list 2743 # of HyperparameterOutput objects, one for each successful trial. 2744 "hyperparameters": { # The hyperparameters given to this trial. 2745 "a_key": "A String", 2746 }, 2747 "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. 2748 "trainingStep": "A String", # The global training step for this metric. 2749 "objectiveValue": 3.14, # The objective value at this training step. 2750 }, 2751 "allMetrics": [ # All recorded object metrics for this trial. This field is not currently 2752 # populated. 2753 { # An observed value of a metric. 2754 "trainingStep": "A String", # The global training step for this metric. 2755 "objectiveValue": 3.14, # The objective value at this training step. 2756 }, 2757 ], 2758 "isTrialStoppedEarly": True or False, # True if the trial is stopped early. 2759 "trialId": "A String", # The trial id for these results. 2760 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 2761 # Only set for trials of built-in algorithms jobs that have succeeded. 2762 "framework": "A String", # Framework on which the built-in algorithm was trained. 2763 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 2764 # saves the trained model. Only set for successful jobs that don't use 2765 # hyperparameter tuning. 2766 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 2767 # trained. 2768 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 2769 }, 2770 }, 2771 ], 2772 "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. 2773 "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. 2774 "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. 2775 "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning 2776 # trials. See 2777 # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) 2778 # for more information. Only set for hyperparameter tuning jobs. 2779 "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. 2780 # Only set for built-in algorithms jobs. 2781 "framework": "A String", # Framework on which the built-in algorithm was trained. 2782 "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job 2783 # saves the trained model. Only set for successful jobs that don't use 2784 # hyperparameter tuning. 2785 "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was 2786 # trained. 2787 "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. 2788 }, 2789 }, 2790 "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. 2791 "modelName": "A String", # Use this field if you want to use the default version for the specified 2792 # model. The string must use the following format: 2793 # 2794 # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` 2795 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch 2796 # prediction. If not set, AI Platform will pick the runtime version used 2797 # during the CreateVersion request for this model version, or choose the 2798 # latest stable version when model version information is not available 2799 # such as when the model is specified by uri. 2800 "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for 2801 # this job. Please refer to 2802 # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) 2803 # for information about how to use signatures. 2804 # 2805 # Defaults to 2806 # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) 2807 # , which is "serving_default". 2808 "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. 2809 # The service will buffer batch_size number of records in memory before 2810 # invoking one Tensorflow prediction call internally. So take the record 2811 # size and memory available into consideration when setting this parameter. 2812 "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain 2813 # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. 2814 "A String", 2815 ], 2816 "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. 2817 # Defaults to 10 if not specified. 2818 "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for 2819 # the model to use. 2820 "outputPath": "A String", # Required. The output Google Cloud Storage location. 2821 "dataFormat": "A String", # Required. The format of the input data files. 2822 "versionName": "A String", # Use this field if you want to specify a version of the model to use. The 2823 # string is formatted the same way as `model_version`, with the addition 2824 # of the version information: 2825 # 2826 # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` 2827 "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. 2828 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 2829 # for AI Platform services. 2830 "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. 2831 }, 2832 "trainingInput": { # Represents input parameters for a training job. When using the # Input parameters to create a training job. 2833 # gcloud command to submit your training job, you can specify 2834 # the input parameters as command-line arguments and/or in a YAML configuration 2835 # file referenced from the --config command-line argument. For 2836 # details, see the guide to 2837 # <a href="/ml-engine/docs/tensorflow/training-jobs">submitting a training 2838 # job</a>. 2839 "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 2840 # job's worker nodes. 2841 # 2842 # The supported values are the same as those described in the entry for 2843 # `masterType`. 2844 # 2845 # This value must be consistent with the category of machine type that 2846 # `masterType` uses. In other words, both must be AI Platform machine 2847 # types or both must be Compute Engine machine types. 2848 # 2849 # If you use `cloud_tpu` for this value, see special instructions for 2850 # [configuring a custom TPU 2851 # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). 2852 # 2853 # This value must be present when `scaleTier` is set to `CUSTOM` and 2854 # `workerCount` is greater than zero. 2855 "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. 2856 # 2857 # You should only set `parameterServerConfig.acceleratorConfig` if 2858 # `parameterServerConfigType` is set to a Compute Engine machine type. [Learn 2859 # about restrictions on accelerator configurations for 2860 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2861 # 2862 # Set `parameterServerConfig.imageUri` only if you build a custom image for 2863 # your parameter server. If `parameterServerConfig.imageUri` has not been 2864 # set, AI Platform uses the value of `masterConfig.imageUri`. 2865 # Learn more about [configuring custom 2866 # containers](/ml-engine/docs/distributed-training-containers). 2867 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 2868 # [Learn about restrictions on accelerator configurations for 2869 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 2870 "count": "A String", # The number of accelerators to attach to each machine running the job. 2871 "type": "A String", # The type of accelerator to use. 2872 }, 2873 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 2874 # Registry. Learn more about [configuring custom 2875 # containers](/ml-engine/docs/distributed-training-containers). 2876 }, 2877 "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. If not 2878 # set, AI Platform uses the default stable version, 1.0. For more 2879 # information, see the 2880 # <a href="/ml-engine/docs/runtime-version-list">runtime version list</a> 2881 # and 2882 # <a href="/ml-engine/docs/versioning">how to manage runtime versions</a>. 2883 "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers 2884 # and parameter servers. 2885 "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training 2886 # job's master worker. 2887 # 2888 # The following types are supported: 2889 # 2890 # <dl> 2891 # <dt>standard</dt> 2892 # <dd> 2893 # A basic machine configuration suitable for training simple models with 2894 # small to moderate datasets. 2895 # </dd> 2896 # <dt>large_model</dt> 2897 # <dd> 2898 # A machine with a lot of memory, specially suited for parameter servers 2899 # when your model is large (having many hidden layers or layers with very 2900 # large numbers of nodes). 2901 # </dd> 2902 # <dt>complex_model_s</dt> 2903 # <dd> 2904 # A machine suitable for the master and workers of the cluster when your 2905 # model requires more computation than the standard machine can handle 2906 # satisfactorily. 2907 # </dd> 2908 # <dt>complex_model_m</dt> 2909 # <dd> 2910 # A machine with roughly twice the number of cores and roughly double the 2911 # memory of <i>complex_model_s</i>. 2912 # </dd> 2913 # <dt>complex_model_l</dt> 2914 # <dd> 2915 # A machine with roughly twice the number of cores and roughly double the 2916 # memory of <i>complex_model_m</i>. 2917 # </dd> 2918 # <dt>standard_gpu</dt> 2919 # <dd> 2920 # A machine equivalent to <i>standard</i> that 2921 # also includes a single NVIDIA Tesla K80 GPU. See more about 2922 # <a href="/ml-engine/docs/tensorflow/using-gpus">using GPUs to 2923 # train your model</a>. 2924 # </dd> 2925 # <dt>complex_model_m_gpu</dt> 2926 # <dd> 2927 # A machine equivalent to <i>complex_model_m</i> that also includes 2928 # four NVIDIA Tesla K80 GPUs. 2929 # </dd> 2930 # <dt>complex_model_l_gpu</dt> 2931 # <dd> 2932 # A machine equivalent to <i>complex_model_l</i> that also includes 2933 # eight NVIDIA Tesla K80 GPUs. 2934 # </dd> 2935 # <dt>standard_p100</dt> 2936 # <dd> 2937 # A machine equivalent to <i>standard</i> that 2938 # also includes a single NVIDIA Tesla P100 GPU. 2939 # </dd> 2940 # <dt>complex_model_m_p100</dt> 2941 # <dd> 2942 # A machine equivalent to <i>complex_model_m</i> that also includes 2943 # four NVIDIA Tesla P100 GPUs. 2944 # </dd> 2945 # <dt>standard_v100</dt> 2946 # <dd> 2947 # A machine equivalent to <i>standard</i> that 2948 # also includes a single NVIDIA Tesla V100 GPU. 2949 # </dd> 2950 # <dt>large_model_v100</dt> 2951 # <dd> 2952 # A machine equivalent to <i>large_model</i> that 2953 # also includes a single NVIDIA Tesla V100 GPU. 2954 # </dd> 2955 # <dt>complex_model_m_v100</dt> 2956 # <dd> 2957 # A machine equivalent to <i>complex_model_m</i> that 2958 # also includes four NVIDIA Tesla V100 GPUs. 2959 # </dd> 2960 # <dt>complex_model_l_v100</dt> 2961 # <dd> 2962 # A machine equivalent to <i>complex_model_l</i> that 2963 # also includes eight NVIDIA Tesla V100 GPUs. 2964 # </dd> 2965 # <dt>cloud_tpu</dt> 2966 # <dd> 2967 # A TPU VM including one Cloud TPU. See more about 2968 # <a href="/ml-engine/docs/tensorflow/using-tpus">using TPUs to train 2969 # your model</a>. 2970 # </dd> 2971 # </dl> 2972 # 2973 # You may also use certain Compute Engine machine types directly in this 2974 # field. The following types are supported: 2975 # 2976 # - `n1-standard-4` 2977 # - `n1-standard-8` 2978 # - `n1-standard-16` 2979 # - `n1-standard-32` 2980 # - `n1-standard-64` 2981 # - `n1-standard-96` 2982 # - `n1-highmem-2` 2983 # - `n1-highmem-4` 2984 # - `n1-highmem-8` 2985 # - `n1-highmem-16` 2986 # - `n1-highmem-32` 2987 # - `n1-highmem-64` 2988 # - `n1-highmem-96` 2989 # - `n1-highcpu-16` 2990 # - `n1-highcpu-32` 2991 # - `n1-highcpu-64` 2992 # - `n1-highcpu-96` 2993 # 2994 # See more about [using Compute Engine machine 2995 # types](/ml-engine/docs/tensorflow/machine-types#compute-engine-machine-types). 2996 # 2997 # You must set this value when `scaleTier` is set to `CUSTOM`. 2998 "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. 2999 "maxTrials": 42, # Optional. How many training trials should be attempted to optimize 3000 # the specified hyperparameters. 3001 # 3002 # Defaults to one. 3003 "goal": "A String", # Required. The type of goal to use for tuning. Available types are 3004 # `MAXIMIZE` and `MINIMIZE`. 3005 # 3006 # Defaults to `MAXIMIZE`. 3007 "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter 3008 # tuning job. 3009 # Uses the default AI Platform hyperparameter tuning 3010 # algorithm if unspecified. 3011 "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing 3012 # the hyperparameter tuning job. You can specify this field to override the 3013 # default failing criteria for AI Platform hyperparameter tuning jobs. 3014 # 3015 # Defaults to zero, which means the service decides when a hyperparameter 3016 # job should fail. 3017 "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial 3018 # early stopping. 3019 "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to 3020 # continue with. The job id will be used to find the corresponding vizier 3021 # study guid and resume the study. 3022 "params": [ # Required. The set of parameters to tune. 3023 { # Represents a single hyperparameter to optimize. 3024 "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 3025 # should be unset if type is `CATEGORICAL`. This value should be integers if 3026 # type is `INTEGER`. 3027 "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. 3028 "A String", 3029 ], 3030 "discreteValues": [ # Required if type is `DISCRETE`. 3031 # A list of feasible points. 3032 # The list should be in strictly increasing order. For instance, this 3033 # parameter might have possible settings of 1.5, 2.5, and 4.0. This list 3034 # should not contain more than 1,000 values. 3035 3.14, 3036 ], 3037 "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in 3038 # a HyperparameterSpec message. E.g., "learning_rate". 3039 "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field 3040 # should be unset if type is `CATEGORICAL`. This value should be integers if 3041 # type is INTEGER. 3042 "type": "A String", # Required. The type of the parameter. 3043 "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. 3044 # Leave unset for categorical parameters. 3045 # Some kind of scaling is strongly recommended for real or integral 3046 # parameters (e.g., `UNIT_LINEAR_SCALE`). 3047 }, 3048 ], 3049 "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For 3050 # current versions of TensorFlow, this tag name should exactly match what is 3051 # shown in TensorBoard, including all scopes. For versions of TensorFlow 3052 # prior to 0.12, this should be only the tag passed to tf.Summary. 3053 # By default, "training/hptuning/metric" will be used. 3054 "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. 3055 # You can reduce the time it takes to perform hyperparameter tuning by adding 3056 # trials in parallel. However, each trail only benefits from the information 3057 # gained in completed trials. That means that a trial does not get access to 3058 # the results of trials running at the same time, which could reduce the 3059 # quality of the overall optimization. 3060 # 3061 # Each trial will use the same scale tier and machine types. 3062 # 3063 # Defaults to one. 3064 }, 3065 "region": "A String", # Required. The Google Compute Engine region to run the training job in. 3066 # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> 3067 # for AI Platform services. 3068 "args": [ # Optional. Command line arguments to pass to the program. 3069 "A String", 3070 ], 3071 "pythonModule": "A String", # Required. The Python module name to run after installing the packages. 3072 "pythonVersion": "A String", # Optional. The version of Python used in training. If not set, the default 3073 # version is '2.7'. Python '3.5' is available when `runtime_version` is set 3074 # to '1.4' and above. Python '2.7' works with all supported 3075 # <a href="/ml-engine/docs/runtime-version-list">runtime versions</a>. 3076 "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs 3077 # and other data needed for training. This path is passed to your TensorFlow 3078 # program as the '--job-dir' command-line argument. The benefit of specifying 3079 # this field is that Cloud ML validates the path for use in training. 3080 "packageUris": [ # Required. The Google Cloud Storage location of the packages with 3081 # the training program and any additional dependencies. 3082 # The maximum number of package URIs is 100. 3083 "A String", 3084 ], 3085 "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each 3086 # replica in the cluster will be of the type specified in `worker_type`. 3087 # 3088 # This value can only be used when `scale_tier` is set to `CUSTOM`. If you 3089 # set this value, you must also set `worker_type`. 3090 # 3091 # The default value is zero. 3092 "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training 3093 # job's parameter server. 3094 # 3095 # The supported values are the same as those described in the entry for 3096 # `master_type`. 3097 # 3098 # This value must be consistent with the category of machine type that 3099 # `masterType` uses. In other words, both must be AI Platform machine 3100 # types or both must be Compute Engine machine types. 3101 # 3102 # This value must be present when `scaleTier` is set to `CUSTOM` and 3103 # `parameter_server_count` is greater than zero. 3104 "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. 3105 # 3106 # You should only set `workerConfig.acceleratorConfig` if `workerType` is set 3107 # to a Compute Engine machine type. [Learn about restrictions on accelerator 3108 # configurations for 3109 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 3110 # 3111 # Set `workerConfig.imageUri` only if you build a custom image for your 3112 # worker. If `workerConfig.imageUri` has not been set, AI Platform uses 3113 # the value of `masterConfig.imageUri`. Learn more about 3114 # [configuring custom 3115 # containers](/ml-engine/docs/distributed-training-containers). 3116 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 3117 # [Learn about restrictions on accelerator configurations for 3118 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 3119 "count": "A String", # The number of accelerators to attach to each machine running the job. 3120 "type": "A String", # The type of accelerator to use. 3121 }, 3122 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 3123 # Registry. Learn more about [configuring custom 3124 # containers](/ml-engine/docs/distributed-training-containers). 3125 }, 3126 "maxRunningTime": "A String", # Optional. The maximum job running time. The default is 7 days. 3127 "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. 3128 # 3129 # You should only set `masterConfig.acceleratorConfig` if `masterType` is set 3130 # to a Compute Engine machine type. Learn about [restrictions on accelerator 3131 # configurations for 3132 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 3133 # 3134 # Set `masterConfig.imageUri` only if you build a custom image. Only one of 3135 # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more about 3136 # [configuring custom 3137 # containers](/ml-engine/docs/distributed-training-containers). 3138 "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. 3139 # [Learn about restrictions on accelerator configurations for 3140 # training.](/ml-engine/docs/tensorflow/using-gpus#compute-engine-machine-types-with-gpu) 3141 "count": "A String", # The number of accelerators to attach to each machine running the job. 3142 "type": "A String", # The type of accelerator to use. 3143 }, 3144 "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container 3145 # Registry. Learn more about [configuring custom 3146 # containers](/ml-engine/docs/distributed-training-containers). 3147 }, 3148 "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training 3149 # job. Each replica in the cluster will be of the type specified in 3150 # `parameter_server_type`. 3151 # 3152 # This value can only be used when `scale_tier` is set to `CUSTOM`.If you 3153 # set this value, you must also set `parameter_server_type`. 3154 # 3155 # The default value is zero. 3156 }, 3157 "jobId": "A String", # Required. The user-specified id of the job. 3158 "labels": { # Optional. One or more labels that you can add, to organize your jobs. 3159 # Each label is a key-value pair, where both the key and the value are 3160 # arbitrary strings that you supply. 3161 # For more information, see the documentation on 3162 # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. 3163 "a_key": "A String", 3164 }, 3165 "state": "A String", # Output only. The detailed state of a job. 3166 "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help 3167 # prevent simultaneous updates of a job from overwriting each other. 3168 # It is strongly suggested that systems make use of the `etag` in the 3169 # read-modify-write cycle to perform job updates in order to avoid race 3170 # conditions: An `etag` is returned in the response to `GetJob`, and 3171 # systems are expected to put that etag in the request to `UpdateJob` to 3172 # ensure that their change will be applied to the same version of the job. 3173 "startTime": "A String", # Output only. When the job processing was started. 3174 "endTime": "A String", # Output only. When the job processing was completed. 3175 "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. 3176 "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. 3177 "nodeHours": 3.14, # Node hours used by the batch prediction job. 3178 "predictionCount": "A String", # The number of generated predictions. 3179 "errorCount": "A String", # The number of data instances which resulted in errors. 3180 }, 3181 "createTime": "A String", # Output only. When the job was created. 3182 }</pre> 3183</div> 3184 3185<div class="method"> 3186 <code class="details" id="setIamPolicy">setIamPolicy(resource, body, x__xgafv=None)</code> 3187 <pre>Sets the access control policy on the specified resource. Replaces any 3188existing policy. 3189 3190Args: 3191 resource: string, REQUIRED: The resource for which the policy is being specified. 3192See the operation documentation for the appropriate value for this field. (required) 3193 body: object, The request body. (required) 3194 The object takes the form of: 3195 3196{ # Request message for `SetIamPolicy` method. 3197 "policy": { # Defines an Identity and Access Management (IAM) policy. It is used to # REQUIRED: The complete policy to be applied to the `resource`. The size of 3198 # the policy is limited to a few 10s of KB. An empty policy is a 3199 # valid policy but certain Cloud Platform services (such as Projects) 3200 # might reject them. 3201 # specify access control policies for Cloud Platform resources. 3202 # 3203 # 3204 # A `Policy` consists of a list of `bindings`. A `binding` binds a list of 3205 # `members` to a `role`, where the members can be user accounts, Google groups, 3206 # Google domains, and service accounts. A `role` is a named list of permissions 3207 # defined by IAM. 3208 # 3209 # **JSON Example** 3210 # 3211 # { 3212 # "bindings": [ 3213 # { 3214 # "role": "roles/owner", 3215 # "members": [ 3216 # "user:mike@example.com", 3217 # "group:admins@example.com", 3218 # "domain:google.com", 3219 # "serviceAccount:my-other-app@appspot.gserviceaccount.com" 3220 # ] 3221 # }, 3222 # { 3223 # "role": "roles/viewer", 3224 # "members": ["user:sean@example.com"] 3225 # } 3226 # ] 3227 # } 3228 # 3229 # **YAML Example** 3230 # 3231 # bindings: 3232 # - members: 3233 # - user:mike@example.com 3234 # - group:admins@example.com 3235 # - domain:google.com 3236 # - serviceAccount:my-other-app@appspot.gserviceaccount.com 3237 # role: roles/owner 3238 # - members: 3239 # - user:sean@example.com 3240 # role: roles/viewer 3241 # 3242 # 3243 # For a description of IAM and its features, see the 3244 # [IAM developer's guide](https://cloud.google.com/iam/docs). 3245 "bindings": [ # Associates a list of `members` to a `role`. 3246 # `bindings` with no members will result in an error. 3247 { # Associates `members` with a `role`. 3248 "role": "A String", # Role that is assigned to `members`. 3249 # For example, `roles/viewer`, `roles/editor`, or `roles/owner`. 3250 "members": [ # Specifies the identities requesting access for a Cloud Platform resource. 3251 # `members` can have the following values: 3252 # 3253 # * `allUsers`: A special identifier that represents anyone who is 3254 # on the internet; with or without a Google account. 3255 # 3256 # * `allAuthenticatedUsers`: A special identifier that represents anyone 3257 # who is authenticated with a Google account or a service account. 3258 # 3259 # * `user:{emailid}`: An email address that represents a specific Google 3260 # account. For example, `alice@gmail.com` . 3261 # 3262 # 3263 # * `serviceAccount:{emailid}`: An email address that represents a service 3264 # account. For example, `my-other-app@appspot.gserviceaccount.com`. 3265 # 3266 # * `group:{emailid}`: An email address that represents a Google group. 3267 # For example, `admins@example.com`. 3268 # 3269 # 3270 # * `domain:{domain}`: The G Suite domain (primary) that represents all the 3271 # users of that domain. For example, `google.com` or `example.com`. 3272 # 3273 "A String", 3274 ], 3275 "condition": { # Represents an expression text. Example: # The condition that is associated with this binding. 3276 # NOTE: An unsatisfied condition will not allow user access via current 3277 # binding. Different bindings, including their conditions, are examined 3278 # independently. 3279 # 3280 # title: "User account presence" 3281 # description: "Determines whether the request has a user account" 3282 # expression: "size(request.user) > 0" 3283 "description": "A String", # An optional description of the expression. This is a longer text which 3284 # describes the expression, e.g. when hovered over it in a UI. 3285 "expression": "A String", # Textual representation of an expression in 3286 # Common Expression Language syntax. 3287 # 3288 # The application context of the containing message determines which 3289 # well-known feature set of CEL is supported. 3290 "location": "A String", # An optional string indicating the location of the expression for error 3291 # reporting, e.g. a file name and a position in the file. 3292 "title": "A String", # An optional title for the expression, i.e. a short string describing 3293 # its purpose. This can be used e.g. in UIs which allow to enter the 3294 # expression. 3295 }, 3296 }, 3297 ], 3298 "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help 3299 # prevent simultaneous updates of a policy from overwriting each other. 3300 # It is strongly suggested that systems make use of the `etag` in the 3301 # read-modify-write cycle to perform policy updates in order to avoid race 3302 # conditions: An `etag` is returned in the response to `getIamPolicy`, and 3303 # systems are expected to put that etag in the request to `setIamPolicy` to 3304 # ensure that their change will be applied to the same version of the policy. 3305 # 3306 # If no `etag` is provided in the call to `setIamPolicy`, then the existing 3307 # policy is overwritten blindly. 3308 "version": 42, # Deprecated. 3309 "auditConfigs": [ # Specifies cloud audit logging configuration for this policy. 3310 { # Specifies the audit configuration for a service. 3311 # The configuration determines which permission types are logged, and what 3312 # identities, if any, are exempted from logging. 3313 # An AuditConfig must have one or more AuditLogConfigs. 3314 # 3315 # If there are AuditConfigs for both `allServices` and a specific service, 3316 # the union of the two AuditConfigs is used for that service: the log_types 3317 # specified in each AuditConfig are enabled, and the exempted_members in each 3318 # AuditLogConfig are exempted. 3319 # 3320 # Example Policy with multiple AuditConfigs: 3321 # 3322 # { 3323 # "audit_configs": [ 3324 # { 3325 # "service": "allServices" 3326 # "audit_log_configs": [ 3327 # { 3328 # "log_type": "DATA_READ", 3329 # "exempted_members": [ 3330 # "user:foo@gmail.com" 3331 # ] 3332 # }, 3333 # { 3334 # "log_type": "DATA_WRITE", 3335 # }, 3336 # { 3337 # "log_type": "ADMIN_READ", 3338 # } 3339 # ] 3340 # }, 3341 # { 3342 # "service": "fooservice.googleapis.com" 3343 # "audit_log_configs": [ 3344 # { 3345 # "log_type": "DATA_READ", 3346 # }, 3347 # { 3348 # "log_type": "DATA_WRITE", 3349 # "exempted_members": [ 3350 # "user:bar@gmail.com" 3351 # ] 3352 # } 3353 # ] 3354 # } 3355 # ] 3356 # } 3357 # 3358 # For fooservice, this policy enables DATA_READ, DATA_WRITE and ADMIN_READ 3359 # logging. It also exempts foo@gmail.com from DATA_READ logging, and 3360 # bar@gmail.com from DATA_WRITE logging. 3361 "auditLogConfigs": [ # The configuration for logging of each type of permission. 3362 { # Provides the configuration for logging a type of permissions. 3363 # Example: 3364 # 3365 # { 3366 # "audit_log_configs": [ 3367 # { 3368 # "log_type": "DATA_READ", 3369 # "exempted_members": [ 3370 # "user:foo@gmail.com" 3371 # ] 3372 # }, 3373 # { 3374 # "log_type": "DATA_WRITE", 3375 # } 3376 # ] 3377 # } 3378 # 3379 # This enables 'DATA_READ' and 'DATA_WRITE' logging, while exempting 3380 # foo@gmail.com from DATA_READ logging. 3381 "exemptedMembers": [ # Specifies the identities that do not cause logging for this type of 3382 # permission. 3383 # Follows the same format of Binding.members. 3384 "A String", 3385 ], 3386 "logType": "A String", # The log type that this config enables. 3387 }, 3388 ], 3389 "service": "A String", # Specifies a service that will be enabled for audit logging. 3390 # For example, `storage.googleapis.com`, `cloudsql.googleapis.com`. 3391 # `allServices` is a special value that covers all services. 3392 }, 3393 ], 3394 }, 3395 "updateMask": "A String", # OPTIONAL: A FieldMask specifying which fields of the policy to modify. Only 3396 # the fields in the mask will be modified. If no mask is provided, the 3397 # following default mask is used: 3398 # paths: "bindings, etag" 3399 # This field is only used by Cloud IAM. 3400 } 3401 3402 x__xgafv: string, V1 error format. 3403 Allowed values 3404 1 - v1 error format 3405 2 - v2 error format 3406 3407Returns: 3408 An object of the form: 3409 3410 { # Defines an Identity and Access Management (IAM) policy. It is used to 3411 # specify access control policies for Cloud Platform resources. 3412 # 3413 # 3414 # A `Policy` consists of a list of `bindings`. A `binding` binds a list of 3415 # `members` to a `role`, where the members can be user accounts, Google groups, 3416 # Google domains, and service accounts. A `role` is a named list of permissions 3417 # defined by IAM. 3418 # 3419 # **JSON Example** 3420 # 3421 # { 3422 # "bindings": [ 3423 # { 3424 # "role": "roles/owner", 3425 # "members": [ 3426 # "user:mike@example.com", 3427 # "group:admins@example.com", 3428 # "domain:google.com", 3429 # "serviceAccount:my-other-app@appspot.gserviceaccount.com" 3430 # ] 3431 # }, 3432 # { 3433 # "role": "roles/viewer", 3434 # "members": ["user:sean@example.com"] 3435 # } 3436 # ] 3437 # } 3438 # 3439 # **YAML Example** 3440 # 3441 # bindings: 3442 # - members: 3443 # - user:mike@example.com 3444 # - group:admins@example.com 3445 # - domain:google.com 3446 # - serviceAccount:my-other-app@appspot.gserviceaccount.com 3447 # role: roles/owner 3448 # - members: 3449 # - user:sean@example.com 3450 # role: roles/viewer 3451 # 3452 # 3453 # For a description of IAM and its features, see the 3454 # [IAM developer's guide](https://cloud.google.com/iam/docs). 3455 "bindings": [ # Associates a list of `members` to a `role`. 3456 # `bindings` with no members will result in an error. 3457 { # Associates `members` with a `role`. 3458 "role": "A String", # Role that is assigned to `members`. 3459 # For example, `roles/viewer`, `roles/editor`, or `roles/owner`. 3460 "members": [ # Specifies the identities requesting access for a Cloud Platform resource. 3461 # `members` can have the following values: 3462 # 3463 # * `allUsers`: A special identifier that represents anyone who is 3464 # on the internet; with or without a Google account. 3465 # 3466 # * `allAuthenticatedUsers`: A special identifier that represents anyone 3467 # who is authenticated with a Google account or a service account. 3468 # 3469 # * `user:{emailid}`: An email address that represents a specific Google 3470 # account. For example, `alice@gmail.com` . 3471 # 3472 # 3473 # * `serviceAccount:{emailid}`: An email address that represents a service 3474 # account. For example, `my-other-app@appspot.gserviceaccount.com`. 3475 # 3476 # * `group:{emailid}`: An email address that represents a Google group. 3477 # For example, `admins@example.com`. 3478 # 3479 # 3480 # * `domain:{domain}`: The G Suite domain (primary) that represents all the 3481 # users of that domain. For example, `google.com` or `example.com`. 3482 # 3483 "A String", 3484 ], 3485 "condition": { # Represents an expression text. Example: # The condition that is associated with this binding. 3486 # NOTE: An unsatisfied condition will not allow user access via current 3487 # binding. Different bindings, including their conditions, are examined 3488 # independently. 3489 # 3490 # title: "User account presence" 3491 # description: "Determines whether the request has a user account" 3492 # expression: "size(request.user) > 0" 3493 "description": "A String", # An optional description of the expression. This is a longer text which 3494 # describes the expression, e.g. when hovered over it in a UI. 3495 "expression": "A String", # Textual representation of an expression in 3496 # Common Expression Language syntax. 3497 # 3498 # The application context of the containing message determines which 3499 # well-known feature set of CEL is supported. 3500 "location": "A String", # An optional string indicating the location of the expression for error 3501 # reporting, e.g. a file name and a position in the file. 3502 "title": "A String", # An optional title for the expression, i.e. a short string describing 3503 # its purpose. This can be used e.g. in UIs which allow to enter the 3504 # expression. 3505 }, 3506 }, 3507 ], 3508 "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help 3509 # prevent simultaneous updates of a policy from overwriting each other. 3510 # It is strongly suggested that systems make use of the `etag` in the 3511 # read-modify-write cycle to perform policy updates in order to avoid race 3512 # conditions: An `etag` is returned in the response to `getIamPolicy`, and 3513 # systems are expected to put that etag in the request to `setIamPolicy` to 3514 # ensure that their change will be applied to the same version of the policy. 3515 # 3516 # If no `etag` is provided in the call to `setIamPolicy`, then the existing 3517 # policy is overwritten blindly. 3518 "version": 42, # Deprecated. 3519 "auditConfigs": [ # Specifies cloud audit logging configuration for this policy. 3520 { # Specifies the audit configuration for a service. 3521 # The configuration determines which permission types are logged, and what 3522 # identities, if any, are exempted from logging. 3523 # An AuditConfig must have one or more AuditLogConfigs. 3524 # 3525 # If there are AuditConfigs for both `allServices` and a specific service, 3526 # the union of the two AuditConfigs is used for that service: the log_types 3527 # specified in each AuditConfig are enabled, and the exempted_members in each 3528 # AuditLogConfig are exempted. 3529 # 3530 # Example Policy with multiple AuditConfigs: 3531 # 3532 # { 3533 # "audit_configs": [ 3534 # { 3535 # "service": "allServices" 3536 # "audit_log_configs": [ 3537 # { 3538 # "log_type": "DATA_READ", 3539 # "exempted_members": [ 3540 # "user:foo@gmail.com" 3541 # ] 3542 # }, 3543 # { 3544 # "log_type": "DATA_WRITE", 3545 # }, 3546 # { 3547 # "log_type": "ADMIN_READ", 3548 # } 3549 # ] 3550 # }, 3551 # { 3552 # "service": "fooservice.googleapis.com" 3553 # "audit_log_configs": [ 3554 # { 3555 # "log_type": "DATA_READ", 3556 # }, 3557 # { 3558 # "log_type": "DATA_WRITE", 3559 # "exempted_members": [ 3560 # "user:bar@gmail.com" 3561 # ] 3562 # } 3563 # ] 3564 # } 3565 # ] 3566 # } 3567 # 3568 # For fooservice, this policy enables DATA_READ, DATA_WRITE and ADMIN_READ 3569 # logging. It also exempts foo@gmail.com from DATA_READ logging, and 3570 # bar@gmail.com from DATA_WRITE logging. 3571 "auditLogConfigs": [ # The configuration for logging of each type of permission. 3572 { # Provides the configuration for logging a type of permissions. 3573 # Example: 3574 # 3575 # { 3576 # "audit_log_configs": [ 3577 # { 3578 # "log_type": "DATA_READ", 3579 # "exempted_members": [ 3580 # "user:foo@gmail.com" 3581 # ] 3582 # }, 3583 # { 3584 # "log_type": "DATA_WRITE", 3585 # } 3586 # ] 3587 # } 3588 # 3589 # This enables 'DATA_READ' and 'DATA_WRITE' logging, while exempting 3590 # foo@gmail.com from DATA_READ logging. 3591 "exemptedMembers": [ # Specifies the identities that do not cause logging for this type of 3592 # permission. 3593 # Follows the same format of Binding.members. 3594 "A String", 3595 ], 3596 "logType": "A String", # The log type that this config enables. 3597 }, 3598 ], 3599 "service": "A String", # Specifies a service that will be enabled for audit logging. 3600 # For example, `storage.googleapis.com`, `cloudsql.googleapis.com`. 3601 # `allServices` is a special value that covers all services. 3602 }, 3603 ], 3604 }</pre> 3605</div> 3606 3607<div class="method"> 3608 <code class="details" id="testIamPermissions">testIamPermissions(resource, body, x__xgafv=None)</code> 3609 <pre>Returns permissions that a caller has on the specified resource. 3610If the resource does not exist, this will return an empty set of 3611permissions, not a NOT_FOUND error. 3612 3613Note: This operation is designed to be used for building permission-aware 3614UIs and command-line tools, not for authorization checking. This operation 3615may "fail open" without warning. 3616 3617Args: 3618 resource: string, REQUIRED: The resource for which the policy detail is being requested. 3619See the operation documentation for the appropriate value for this field. (required) 3620 body: object, The request body. (required) 3621 The object takes the form of: 3622 3623{ # Request message for `TestIamPermissions` method. 3624 "permissions": [ # The set of permissions to check for the `resource`. Permissions with 3625 # wildcards (such as '*' or 'storage.*') are not allowed. For more 3626 # information see 3627 # [IAM Overview](https://cloud.google.com/iam/docs/overview#permissions). 3628 "A String", 3629 ], 3630 } 3631 3632 x__xgafv: string, V1 error format. 3633 Allowed values 3634 1 - v1 error format 3635 2 - v2 error format 3636 3637Returns: 3638 An object of the form: 3639 3640 { # Response message for `TestIamPermissions` method. 3641 "permissions": [ # A subset of `TestPermissionsRequest.permissions` that the caller is 3642 # allowed. 3643 "A String", 3644 ], 3645 }</pre> 3646</div> 3647 3648</body></html>