| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 1 | <html><body> |
| 2 | <style> |
| 3 | |
| 4 | body, 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 | |
| 15 | body { |
| 16 | font-size: 13px; |
| 17 | padding: 1em; |
| 18 | } |
| 19 | |
| 20 | h1 { |
| 21 | font-size: 26px; |
| 22 | margin-bottom: 1em; |
| 23 | } |
| 24 | |
| 25 | h2 { |
| 26 | font-size: 24px; |
| 27 | margin-bottom: 1em; |
| 28 | } |
| 29 | |
| 30 | h3 { |
| 31 | font-size: 20px; |
| 32 | margin-bottom: 1em; |
| 33 | margin-top: 1em; |
| 34 | } |
| 35 | |
| 36 | pre, code { |
| 37 | line-height: 1.5; |
| 38 | font-family: Monaco, 'DejaVu Sans Mono', 'Bitstream Vera Sans Mono', 'Lucida Console', monospace; |
| 39 | } |
| 40 | |
| 41 | pre { |
| 42 | margin-top: 0.5em; |
| 43 | } |
| 44 | |
| 45 | h1, h2, h3, p { |
| 46 | font-family: Arial, sans serif; |
| 47 | } |
| 48 | |
| 49 | h1, 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 | |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 75 | <h1><a href="ml_v1.html">AI Platform Training & Prediction API</a> . <a href="ml_v1.projects.html">projects</a> . <a href="ml_v1.projects.jobs.html">jobs</a></h1> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 76 | <h2>Instance Methods</h2> |
| 77 | <p class="toc_element"> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 78 | <code><a href="#cancel">cancel(name, body=None, x__xgafv=None)</a></code></p> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 79 | <p class="firstline">Cancels a running job.</p> |
| 80 | <p class="toc_element"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 81 | <code><a href="#create">create(parent, body=None, x__xgafv=None)</a></code></p> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 82 | <p class="firstline">Creates a training or a batch prediction job.</p> |
| 83 | <p class="toc_element"> |
| Thomas Coffee | 2f24537 | 2017-03-27 10:39:26 -0700 | [diff] [blame] | 84 | <code><a href="#get">get(name, x__xgafv=None)</a></code></p> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 85 | <p class="firstline">Describes a job.</p> |
| 86 | <p class="toc_element"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 87 | <code><a href="#getIamPolicy">getIamPolicy(resource, options_requestedPolicyVersion=None, x__xgafv=None)</a></code></p> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 88 | <p class="firstline">Gets the access control policy for a resource.</p> |
| 89 | <p class="toc_element"> |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 90 | <code><a href="#list">list(parent, pageToken=None, pageSize=None, filter=None, x__xgafv=None)</a></code></p> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 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> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 95 | <p class="toc_element"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 96 | <code><a href="#patch">patch(name, body=None, updateMask=None, x__xgafv=None)</a></code></p> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 97 | <p class="firstline">Updates a specific job resource.</p> |
| 98 | <p class="toc_element"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 99 | <code><a href="#setIamPolicy">setIamPolicy(resource, body=None, x__xgafv=None)</a></code></p> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 100 | <p class="firstline">Sets the access control policy on the specified resource. Replaces any</p> |
| 101 | <p class="toc_element"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 102 | <code><a href="#testIamPermissions">testIamPermissions(resource, body=None, x__xgafv=None)</a></code></p> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 103 | <p class="firstline">Returns permissions that a caller has on the specified resource.</p> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 104 | <h3>Method Details</h3> |
| 105 | <div class="method"> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 106 | <code class="details" id="cancel">cancel(name, body=None, x__xgafv=None)</code> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 107 | <pre>Cancels a running job. |
| 108 | |
| 109 | Args: |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 110 | name: string, Required. The name of the job to cancel. (required) |
| 111 | body: object, The request body. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 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 | |
| 122 | Returns: |
| 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"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 138 | <code class="details" id="create">create(parent, body=None, x__xgafv=None)</code> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 139 | <pre>Creates a training or a batch prediction job. |
| 140 | |
| 141 | Args: |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 142 | parent: string, Required. The project name. (required) |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 143 | body: object, The request body. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 144 | The object takes the form of: |
| 145 | |
| 146 | { # Represents a training or prediction job. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 147 | "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 148 | # prevent simultaneous updates of a job from overwriting each other. |
| 149 | # It is strongly suggested that systems make use of the `etag` in the |
| 150 | # read-modify-write cycle to perform job updates in order to avoid race |
| 151 | # conditions: An `etag` is returned in the response to `GetJob`, and |
| 152 | # systems are expected to put that etag in the request to `UpdateJob` to |
| 153 | # ensure that their change will be applied to the same version of the job. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 154 | "trainingInput": { # Represents input parameters for a training job. When using the gcloud command # Input parameters to create a training job. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 155 | # to submit your training job, you can specify the input parameters as |
| 156 | # command-line arguments and/or in a YAML configuration file referenced from |
| 157 | # the --config command-line argument. For details, see the guide to [submitting |
| 158 | # a training job](/ai-platform/training/docs/training-jobs). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 159 | "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. |
| 160 | # |
| 161 | # You should only set `parameterServerConfig.acceleratorConfig` if |
| 162 | # `parameterServerType` is set to a Compute Engine machine type. [Learn |
| 163 | # about restrictions on accelerator configurations for |
| 164 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 165 | # |
| 166 | # Set `parameterServerConfig.imageUri` only if you build a custom image for |
| 167 | # your parameter server. If `parameterServerConfig.imageUri` has not been |
| 168 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 169 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 170 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 171 | # the one used in the custom container. This field is required if the replica |
| 172 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 173 | # this field. This must be a [runtime version that currently supports |
| 174 | # training with |
| 175 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 176 | # |
| 177 | # Note that the version of TensorFlow included in a runtime version may |
| 178 | # differ from the numbering of the runtime version itself, because it may |
| 179 | # have a different [patch |
| 180 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 181 | # In this field, you must specify the runtime version (TensorFlow minor |
| 182 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 183 | # specify `1.x`. |
| 184 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 185 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 186 | # If not provided, the docker image's ENTRYPOINT is used. |
| 187 | # It cannot be set if custom container image is |
| 188 | # not provided. |
| 189 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 190 | # both cannot be set at the same time. |
| 191 | "A String", |
| 192 | ], |
| 193 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 194 | # Registry. Learn more about [configuring custom |
| 195 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 196 | "containerArgs": [ # Arguments to the entrypoint command. |
| 197 | # The following rules apply for container_command and container_args: |
| 198 | # - If you do not supply command or args: |
| 199 | # The defaults defined in the Docker image are used. |
| 200 | # - If you supply a command but no args: |
| 201 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 202 | # are ignored. Your command is run without any arguments. |
| 203 | # - If you supply only args: |
| 204 | # The default Entrypoint defined in the Docker image is run with the args |
| 205 | # that you supplied. |
| 206 | # - If you supply a command and args: |
| 207 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 208 | # are ignored. Your command is run with your args. |
| 209 | # It cannot be set if custom container image is |
| 210 | # not provided. |
| 211 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 212 | # both cannot be set at the same time. |
| 213 | "A String", |
| 214 | ], |
| 215 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 216 | # [Learn about restrictions on accelerator configurations for |
| 217 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 218 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 219 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 220 | # [accelerators for online |
| 221 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 222 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 223 | "type": "A String", # The type of accelerator to use. |
| 224 | }, |
| 225 | }, |
| 226 | "encryptionConfig": { # Represents a custom encryption key configuration that can be applied to # Optional. Options for using customer-managed encryption keys (CMEK) to |
| 227 | # protect resources created by a training job, instead of using Google's |
| 228 | # default encryption. If this is set, then all resources created by the |
| 229 | # training job will be encrypted with the customer-managed encryption key |
| 230 | # that you specify. |
| 231 | # |
| 232 | # [Learn how and when to use CMEK with AI Platform |
| 233 | # Training](/ai-platform/training/docs/cmek). |
| 234 | # a resource. |
| 235 | "kmsKeyName": "A String", # The Cloud KMS resource identifier of the customer-managed encryption key |
| 236 | # used to protect a resource, such as a training job. It has the following |
| 237 | # format: |
| 238 | # `projects/{PROJECT_ID}/locations/{REGION}/keyRings/{KEY_RING_NAME}/cryptoKeys/{KEY_NAME}` |
| 239 | }, |
| 240 | "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. |
| 241 | "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For |
| 242 | # current versions of TensorFlow, this tag name should exactly match what is |
| 243 | # shown in TensorBoard, including all scopes. For versions of TensorFlow |
| 244 | # prior to 0.12, this should be only the tag passed to tf.Summary. |
| 245 | # By default, "training/hptuning/metric" will be used. |
| 246 | "params": [ # Required. The set of parameters to tune. |
| 247 | { # Represents a single hyperparameter to optimize. |
| 248 | "type": "A String", # Required. The type of the parameter. |
| 249 | "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. |
| 250 | "A String", |
| 251 | ], |
| 252 | "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in |
| 253 | # a HyperparameterSpec message. E.g., "learning_rate". |
| 254 | "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 255 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 256 | # type is INTEGER. |
| 257 | "discreteValues": [ # Required if type is `DISCRETE`. |
| 258 | # A list of feasible points. |
| 259 | # The list should be in strictly increasing order. For instance, this |
| 260 | # parameter might have possible settings of 1.5, 2.5, and 4.0. This list |
| 261 | # should not contain more than 1,000 values. |
| 262 | 3.14, |
| 263 | ], |
| 264 | "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. |
| 265 | # Leave unset for categorical parameters. |
| 266 | # Some kind of scaling is strongly recommended for real or integral |
| 267 | # parameters (e.g., `UNIT_LINEAR_SCALE`). |
| 268 | "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 269 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 270 | # type is `INTEGER`. |
| 271 | }, |
| 272 | ], |
| 273 | "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial |
| 274 | # early stopping. |
| 275 | "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to |
| 276 | # continue with. The job id will be used to find the corresponding vizier |
| 277 | # study guid and resume the study. |
| 278 | "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. |
| 279 | # You can reduce the time it takes to perform hyperparameter tuning by adding |
| 280 | # trials in parallel. However, each trail only benefits from the information |
| 281 | # gained in completed trials. That means that a trial does not get access to |
| 282 | # the results of trials running at the same time, which could reduce the |
| 283 | # quality of the overall optimization. |
| 284 | # |
| 285 | # Each trial will use the same scale tier and machine types. |
| 286 | # |
| 287 | # Defaults to one. |
| 288 | "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing |
| 289 | # the hyperparameter tuning job. You can specify this field to override the |
| 290 | # default failing criteria for AI Platform hyperparameter tuning jobs. |
| 291 | # |
| 292 | # Defaults to zero, which means the service decides when a hyperparameter |
| 293 | # job should fail. |
| 294 | "goal": "A String", # Required. The type of goal to use for tuning. Available types are |
| 295 | # `MAXIMIZE` and `MINIMIZE`. |
| 296 | # |
| 297 | # Defaults to `MAXIMIZE`. |
| 298 | "maxTrials": 42, # Optional. How many training trials should be attempted to optimize |
| 299 | # the specified hyperparameters. |
| 300 | # |
| 301 | # Defaults to one. |
| 302 | "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter |
| 303 | # tuning job. |
| 304 | # Uses the default AI Platform hyperparameter tuning |
| 305 | # algorithm if unspecified. |
| 306 | }, |
| 307 | "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. |
| 308 | # |
| 309 | # You should only set `workerConfig.acceleratorConfig` if `workerType` is set |
| 310 | # to a Compute Engine machine type. [Learn about restrictions on accelerator |
| 311 | # configurations for |
| 312 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 313 | # |
| 314 | # Set `workerConfig.imageUri` only if you build a custom image for your |
| 315 | # worker. If `workerConfig.imageUri` has not been set, AI Platform uses |
| 316 | # the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 317 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 318 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 319 | # the one used in the custom container. This field is required if the replica |
| 320 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 321 | # this field. This must be a [runtime version that currently supports |
| 322 | # training with |
| 323 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 324 | # |
| 325 | # Note that the version of TensorFlow included in a runtime version may |
| 326 | # differ from the numbering of the runtime version itself, because it may |
| 327 | # have a different [patch |
| 328 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 329 | # In this field, you must specify the runtime version (TensorFlow minor |
| 330 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 331 | # specify `1.x`. |
| 332 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 333 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 334 | # If not provided, the docker image's ENTRYPOINT is used. |
| 335 | # It cannot be set if custom container image is |
| 336 | # not provided. |
| 337 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 338 | # both cannot be set at the same time. |
| 339 | "A String", |
| 340 | ], |
| 341 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 342 | # Registry. Learn more about [configuring custom |
| 343 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 344 | "containerArgs": [ # Arguments to the entrypoint command. |
| 345 | # The following rules apply for container_command and container_args: |
| 346 | # - If you do not supply command or args: |
| 347 | # The defaults defined in the Docker image are used. |
| 348 | # - If you supply a command but no args: |
| 349 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 350 | # are ignored. Your command is run without any arguments. |
| 351 | # - If you supply only args: |
| 352 | # The default Entrypoint defined in the Docker image is run with the args |
| 353 | # that you supplied. |
| 354 | # - If you supply a command and args: |
| 355 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 356 | # are ignored. Your command is run with your args. |
| 357 | # It cannot be set if custom container image is |
| 358 | # not provided. |
| 359 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 360 | # both cannot be set at the same time. |
| 361 | "A String", |
| 362 | ], |
| 363 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 364 | # [Learn about restrictions on accelerator configurations for |
| 365 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 366 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 367 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 368 | # [accelerators for online |
| 369 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 370 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 371 | "type": "A String", # The type of accelerator to use. |
| 372 | }, |
| 373 | }, |
| 374 | "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training |
| 375 | # job. Each replica in the cluster will be of the type specified in |
| 376 | # `parameter_server_type`. |
| 377 | # |
| 378 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 379 | # set this value, you must also set `parameter_server_type`. |
| 380 | # |
| 381 | # The default value is zero. |
| 382 | "packageUris": [ # Required. The Google Cloud Storage location of the packages with |
| 383 | # the training program and any additional dependencies. |
| 384 | # The maximum number of package URIs is 100. |
| 385 | "A String", |
| 386 | ], |
| 387 | "evaluatorCount": "A String", # Optional. The number of evaluator replicas to use for the training job. |
| 388 | # Each replica in the cluster will be of the type specified in |
| 389 | # `evaluator_type`. |
| 390 | # |
| 391 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 392 | # set this value, you must also set `evaluator_type`. |
| 393 | # |
| 394 | # The default value is zero. |
| 395 | "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 396 | # job's master worker. You must specify this field when `scaleTier` is set to |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 397 | # `CUSTOM`. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 398 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 399 | # You can use certain Compute Engine machine types directly in this field. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 400 | # The following types are supported: |
| 401 | # |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 402 | # - `n1-standard-4` |
| 403 | # - `n1-standard-8` |
| 404 | # - `n1-standard-16` |
| 405 | # - `n1-standard-32` |
| 406 | # - `n1-standard-64` |
| 407 | # - `n1-standard-96` |
| 408 | # - `n1-highmem-2` |
| 409 | # - `n1-highmem-4` |
| 410 | # - `n1-highmem-8` |
| 411 | # - `n1-highmem-16` |
| 412 | # - `n1-highmem-32` |
| 413 | # - `n1-highmem-64` |
| 414 | # - `n1-highmem-96` |
| 415 | # - `n1-highcpu-16` |
| 416 | # - `n1-highcpu-32` |
| 417 | # - `n1-highcpu-64` |
| 418 | # - `n1-highcpu-96` |
| 419 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 420 | # Learn more about [using Compute Engine machine |
| 421 | # types](/ml-engine/docs/machine-types#compute-engine-machine-types). |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 422 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 423 | # Alternatively, you can use the following legacy machine types: |
| 424 | # |
| 425 | # - `standard` |
| 426 | # - `large_model` |
| 427 | # - `complex_model_s` |
| 428 | # - `complex_model_m` |
| 429 | # - `complex_model_l` |
| 430 | # - `standard_gpu` |
| 431 | # - `complex_model_m_gpu` |
| 432 | # - `complex_model_l_gpu` |
| 433 | # - `standard_p100` |
| 434 | # - `complex_model_m_p100` |
| 435 | # - `standard_v100` |
| 436 | # - `large_model_v100` |
| 437 | # - `complex_model_m_v100` |
| 438 | # - `complex_model_l_v100` |
| 439 | # |
| 440 | # Learn more about [using legacy machine |
| 441 | # types](/ml-engine/docs/machine-types#legacy-machine-types). |
| 442 | # |
| 443 | # Finally, if you want to use a TPU for training, specify `cloud_tpu` in this |
| 444 | # field. Learn more about the [special configuration options for training |
| 445 | # with |
| 446 | # TPUs](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 447 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. You must |
| 448 | # either specify this field or specify `masterConfig.imageUri`. |
| 449 | # |
| 450 | # For more information, see the [runtime version |
| 451 | # list](/ai-platform/training/docs/runtime-version-list) and learn [how to |
| 452 | # manage runtime versions](/ai-platform/training/docs/versioning). |
| 453 | "evaluatorType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 454 | # job's evaluator nodes. |
| 455 | # |
| 456 | # The supported values are the same as those described in the entry for |
| 457 | # `masterType`. |
| 458 | # |
| 459 | # This value must be consistent with the category of machine type that |
| 460 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 461 | # types or both must be legacy machine types. |
| 462 | # |
| 463 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 464 | # `evaluatorCount` is greater than zero. |
| 465 | "region": "A String", # Required. The region to run the training job in. See the [available |
| 466 | # regions](/ai-platform/training/docs/regions) for AI Platform Training. |
| 467 | "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 468 | # job's worker nodes. |
| 469 | # |
| 470 | # The supported values are the same as those described in the entry for |
| 471 | # `masterType`. |
| 472 | # |
| 473 | # This value must be consistent with the category of machine type that |
| 474 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 475 | # types or both must be legacy machine types. |
| 476 | # |
| 477 | # If you use `cloud_tpu` for this value, see special instructions for |
| 478 | # [configuring a custom TPU |
| 479 | # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| 480 | # |
| 481 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 482 | # `workerCount` is greater than zero. |
| 483 | "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 484 | # job's parameter server. |
| 485 | # |
| 486 | # The supported values are the same as those described in the entry for |
| 487 | # `master_type`. |
| 488 | # |
| 489 | # This value must be consistent with the category of machine type that |
| 490 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 491 | # types or both must be legacy machine types. |
| 492 | # |
| 493 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 494 | # `parameter_server_count` is greater than zero. |
| 495 | "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. |
| 496 | # |
| 497 | # You should only set `masterConfig.acceleratorConfig` if `masterType` is set |
| 498 | # to a Compute Engine machine type. Learn about [restrictions on accelerator |
| 499 | # configurations for |
| 500 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 501 | # |
| 502 | # Set `masterConfig.imageUri` only if you build a custom image. Only one of |
| 503 | # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more |
| 504 | # about [configuring custom |
| 505 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 506 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 507 | # the one used in the custom container. This field is required if the replica |
| 508 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 509 | # this field. This must be a [runtime version that currently supports |
| 510 | # training with |
| 511 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 512 | # |
| 513 | # Note that the version of TensorFlow included in a runtime version may |
| 514 | # differ from the numbering of the runtime version itself, because it may |
| 515 | # have a different [patch |
| 516 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 517 | # In this field, you must specify the runtime version (TensorFlow minor |
| 518 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 519 | # specify `1.x`. |
| 520 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 521 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 522 | # If not provided, the docker image's ENTRYPOINT is used. |
| 523 | # It cannot be set if custom container image is |
| 524 | # not provided. |
| 525 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 526 | # both cannot be set at the same time. |
| 527 | "A String", |
| 528 | ], |
| 529 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 530 | # Registry. Learn more about [configuring custom |
| 531 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 532 | "containerArgs": [ # Arguments to the entrypoint command. |
| 533 | # The following rules apply for container_command and container_args: |
| 534 | # - If you do not supply command or args: |
| 535 | # The defaults defined in the Docker image are used. |
| 536 | # - If you supply a command but no args: |
| 537 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 538 | # are ignored. Your command is run without any arguments. |
| 539 | # - If you supply only args: |
| 540 | # The default Entrypoint defined in the Docker image is run with the args |
| 541 | # that you supplied. |
| 542 | # - If you supply a command and args: |
| 543 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 544 | # are ignored. Your command is run with your args. |
| 545 | # It cannot be set if custom container image is |
| 546 | # not provided. |
| 547 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 548 | # both cannot be set at the same time. |
| 549 | "A String", |
| 550 | ], |
| 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.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 554 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 555 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 556 | # [accelerators for online |
| 557 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 558 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 559 | "type": "A String", # The type of accelerator to use. |
| 560 | }, |
| 561 | }, |
| 562 | "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers |
| 563 | # and parameter servers. |
| 564 | "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 565 | # and other data needed for training. This path is passed to your TensorFlow |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 566 | # program as the '--job-dir' command-line argument. The benefit of specifying |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 567 | # this field is that Cloud ML validates the path for use in training. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 568 | "pythonVersion": "A String", # Optional. The version of Python used in training. You must either specify |
| 569 | # this field or specify `masterConfig.imageUri`. |
| 570 | # |
| 571 | # The following Python versions are available: |
| 572 | # |
| 573 | # * Python '3.7' is available when `runtime_version` is set to '1.15' or |
| 574 | # later. |
| 575 | # * Python '3.5' is available when `runtime_version` is set to a version |
| 576 | # from '1.4' to '1.14'. |
| 577 | # * Python '2.7' is available when `runtime_version` is set to '1.15' or |
| 578 | # earlier. |
| 579 | # |
| 580 | # Read more about the Python versions available for [each runtime |
| 581 | # version](/ml-engine/docs/runtime-version-list). |
| 582 | "network": "A String", # Optional. The full name of the Google Compute Engine |
| 583 | # [network](/compute/docs/networks-and-firewalls#networks) to which the Job |
| 584 | # is peered. For example, projects/12345/global/networks/myVPC. Format is of |
| 585 | # the form projects/{project}/global/networks/{network}. Where {project} is a |
| 586 | # project number, as in '12345', and {network} is network name.". |
| 587 | # |
| 588 | # Private services access must already be configured for the network. If left |
| 589 | # unspecified, the Job is not peered with any network. Learn more - |
| 590 | # Connecting Job to user network over private |
| 591 | # IP. |
| 592 | "scheduling": { # All parameters related to scheduling of training jobs. # Optional. Scheduling options for a training job. |
| 593 | "maxWaitTime": "A String", |
| 594 | "maxRunningTime": "A String", # Optional. The maximum job running time, expressed in seconds. The field can |
| 595 | # contain up to nine fractional digits, terminated by `s`. If not specified, |
| 596 | # this field defaults to `604800s` (seven days). |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 597 | # |
| 598 | # If the training job is still running after this duration, AI Platform |
| 599 | # Training cancels it. |
| 600 | # |
| 601 | # For example, if you want to ensure your job runs for no more than 2 hours, |
| 602 | # set this field to `7200s` (2 hours * 60 minutes / hour * 60 seconds / |
| 603 | # minute). |
| 604 | # |
| 605 | # If you submit your training job using the `gcloud` tool, you can [provide |
| 606 | # this field in a `config.yaml` |
| 607 | # file](/ai-platform/training/docs/training-jobs#formatting_your_configuration_parameters). |
| 608 | # For example: |
| 609 | # |
| 610 | # ```yaml |
| 611 | # trainingInput: |
| 612 | # ... |
| 613 | # scheduling: |
| 614 | # maxRunningTime: 7200s |
| 615 | # ... |
| 616 | # ``` |
| 617 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 618 | "evaluatorConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for evaluators. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 619 | # |
| 620 | # You should only set `evaluatorConfig.acceleratorConfig` if |
| 621 | # `evaluatorType` is set to a Compute Engine machine type. [Learn |
| 622 | # about restrictions on accelerator configurations for |
| 623 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 624 | # |
| 625 | # Set `evaluatorConfig.imageUri` only if you build a custom image for |
| 626 | # your evaluator. If `evaluatorConfig.imageUri` has not been |
| 627 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 628 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 629 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 630 | # the one used in the custom container. This field is required if the replica |
| 631 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 632 | # this field. This must be a [runtime version that currently supports |
| 633 | # training with |
| 634 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 635 | # |
| 636 | # Note that the version of TensorFlow included in a runtime version may |
| 637 | # differ from the numbering of the runtime version itself, because it may |
| 638 | # have a different [patch |
| 639 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 640 | # In this field, you must specify the runtime version (TensorFlow minor |
| 641 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 642 | # specify `1.x`. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 643 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 644 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 645 | # If not provided, the docker image's ENTRYPOINT is used. |
| 646 | # It cannot be set if custom container image is |
| 647 | # not provided. |
| 648 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 649 | # both cannot be set at the same time. |
| 650 | "A String", |
| 651 | ], |
| 652 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 653 | # Registry. Learn more about [configuring custom |
| 654 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 655 | "containerArgs": [ # Arguments to the entrypoint command. |
| 656 | # The following rules apply for container_command and container_args: |
| 657 | # - If you do not supply command or args: |
| 658 | # The defaults defined in the Docker image are used. |
| 659 | # - If you supply a command but no args: |
| 660 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 661 | # are ignored. Your command is run without any arguments. |
| 662 | # - If you supply only args: |
| 663 | # The default Entrypoint defined in the Docker image is run with the args |
| 664 | # that you supplied. |
| 665 | # - If you supply a command and args: |
| 666 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 667 | # are ignored. Your command is run with your args. |
| 668 | # It cannot be set if custom container image is |
| 669 | # not provided. |
| 670 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 671 | # both cannot be set at the same time. |
| 672 | "A String", |
| 673 | ], |
| 674 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 675 | # [Learn about restrictions on accelerator configurations for |
| 676 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 677 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 678 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 679 | # [accelerators for online |
| 680 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 681 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 682 | "type": "A String", # The type of accelerator to use. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 683 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 684 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 685 | "useChiefInTfConfig": True or False, # Optional. Use `chief` instead of `master` in the `TF_CONFIG` environment |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 686 | # variable when training with a custom container. Defaults to `false`. [Learn |
| 687 | # more about this |
| 688 | # field.](/ai-platform/training/docs/distributed-training-details#chief-versus-master) |
| 689 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 690 | # This field has no effect for training jobs that don't use a custom |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 691 | # container. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 692 | "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 693 | # replica in the cluster will be of the type specified in `worker_type`. |
| 694 | # |
| 695 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 696 | # set this value, you must also set `worker_type`. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 697 | # |
| 698 | # The default value is zero. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 699 | "pythonModule": "A String", # Required. The Python module name to run after installing the packages. |
| 700 | "args": [ # Optional. Command-line arguments passed to the training application when it |
| 701 | # starts. If your job uses a custom container, then the arguments are passed |
| 702 | # to the container's <a class="external" target="_blank" |
| 703 | # href="https://docs.docker.com/engine/reference/builder/#entrypoint"> |
| 704 | # `ENTRYPOINT`</a> command. |
| 705 | "A String", |
| 706 | ], |
| 707 | }, |
| 708 | "state": "A String", # Output only. The detailed state of a job. |
| 709 | "jobId": "A String", # Required. The user-specified id of the job. |
| 710 | "endTime": "A String", # Output only. When the job processing was completed. |
| 711 | "startTime": "A String", # Output only. When the job processing was started. |
| 712 | "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. |
| 713 | "errorCount": "A String", # The number of data instances which resulted in errors. |
| 714 | "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. |
| 715 | "nodeHours": 3.14, # Node hours used by the batch prediction job. |
| 716 | "predictionCount": "A String", # The number of generated predictions. |
| 717 | }, |
| 718 | "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. |
| 719 | "trials": [ # Results for individual Hyperparameter trials. |
| 720 | # Only set for hyperparameter tuning jobs. |
| 721 | { # Represents the result of a single hyperparameter tuning trial from a |
| 722 | # training job. The TrainingOutput object that is returned on successful |
| 723 | # completion of a training job with hyperparameter tuning includes a list |
| 724 | # of HyperparameterOutput objects, one for each successful trial. |
| 725 | "allMetrics": [ # All recorded object metrics for this trial. This field is not currently |
| 726 | # populated. |
| 727 | { # An observed value of a metric. |
| 728 | "objectiveValue": 3.14, # The objective value at this training step. |
| 729 | "trainingStep": "A String", # The global training step for this metric. |
| 730 | }, |
| 731 | ], |
| 732 | "hyperparameters": { # The hyperparameters given to this trial. |
| 733 | "a_key": "A String", |
| 734 | }, |
| 735 | "trialId": "A String", # The trial id for these results. |
| 736 | "endTime": "A String", # Output only. End time for the trial. |
| 737 | "isTrialStoppedEarly": True or False, # True if the trial is stopped early. |
| 738 | "startTime": "A String", # Output only. Start time for the trial. |
| 739 | "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. |
| 740 | "objectiveValue": 3.14, # The objective value at this training step. |
| 741 | "trainingStep": "A String", # The global training step for this metric. |
| 742 | }, |
| 743 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 744 | # Only set for trials of built-in algorithms jobs that have succeeded. |
| 745 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 746 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 747 | # saves the trained model. Only set for successful jobs that don't use |
| 748 | # hyperparameter tuning. |
| 749 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 750 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 751 | # trained. |
| 752 | }, |
| 753 | "state": "A String", # Output only. The detailed state of the trial. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 754 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 755 | ], |
| 756 | "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning |
| 757 | # trials. See |
| 758 | # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) |
| 759 | # for more information. Only set for hyperparameter tuning jobs. |
| 760 | "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. |
| 761 | # Only set for hyperparameter tuning jobs. |
| 762 | "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. |
| 763 | "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. |
| 764 | "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. |
| 765 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 766 | # Only set for built-in algorithms jobs. |
| 767 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 768 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 769 | # saves the trained model. Only set for successful jobs that don't use |
| 770 | # hyperparameter tuning. |
| 771 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 772 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 773 | # trained. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 774 | }, |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 775 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 776 | "createTime": "A String", # Output only. When the job was created. |
| 777 | "labels": { # Optional. One or more labels that you can add, to organize your jobs. |
| 778 | # Each label is a key-value pair, where both the key and the value are |
| 779 | # arbitrary strings that you supply. |
| 780 | # For more information, see the documentation on |
| 781 | # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. |
| 782 | "a_key": "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 783 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 784 | "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. |
| 785 | "outputPath": "A String", # Required. The output Google Cloud Storage location. |
| 786 | "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. |
| 787 | "dataFormat": "A String", # Required. The format of the input data files. |
| 788 | "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. |
| 789 | # The service will buffer batch_size number of records in memory before |
| 790 | # invoking one Tensorflow prediction call internally. So take the record |
| 791 | # size and memory available into consideration when setting this parameter. |
| 792 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch |
| 793 | # prediction. If not set, AI Platform will pick the runtime version used |
| 794 | # during the CreateVersion request for this model version, or choose the |
| 795 | # latest stable version when model version information is not available |
| 796 | # such as when the model is specified by uri. |
| 797 | "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain |
| 798 | # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. |
| 799 | "A String", |
| 800 | ], |
| 801 | "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. |
| 802 | # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> |
| 803 | # for AI Platform services. |
| 804 | "versionName": "A String", # Use this field if you want to specify a version of the model to use. The |
| 805 | # string is formatted the same way as `model_version`, with the addition |
| 806 | # of the version information: |
| 807 | # |
| 808 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` |
| 809 | "modelName": "A String", # Use this field if you want to use the default version for the specified |
| 810 | # model. The string must use the following format: |
| 811 | # |
| 812 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` |
| 813 | "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for |
| 814 | # the model to use. |
| 815 | "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. |
| 816 | # Defaults to 10 if not specified. |
| 817 | "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for |
| 818 | # this job. Please refer to |
| 819 | # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) |
| 820 | # for information about how to use signatures. |
| 821 | # |
| 822 | # Defaults to |
| 823 | # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) |
| 824 | # , which is "serving_default". |
| 825 | }, |
| 826 | "errorMessage": "A String", # Output only. The details of a failure or a cancellation. |
| 827 | } |
| 828 | |
| 829 | x__xgafv: string, V1 error format. |
| 830 | Allowed values |
| 831 | 1 - v1 error format |
| 832 | 2 - v2 error format |
| 833 | |
| 834 | Returns: |
| 835 | An object of the form: |
| 836 | |
| 837 | { # Represents a training or prediction job. |
| 838 | "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help |
| 839 | # prevent simultaneous updates of a job from overwriting each other. |
| 840 | # It is strongly suggested that systems make use of the `etag` in the |
| 841 | # read-modify-write cycle to perform job updates in order to avoid race |
| 842 | # conditions: An `etag` is returned in the response to `GetJob`, and |
| 843 | # systems are expected to put that etag in the request to `UpdateJob` to |
| 844 | # ensure that their change will be applied to the same version of the job. |
| 845 | "trainingInput": { # Represents input parameters for a training job. When using the gcloud command # Input parameters to create a training job. |
| 846 | # to submit your training job, you can specify the input parameters as |
| 847 | # command-line arguments and/or in a YAML configuration file referenced from |
| 848 | # the --config command-line argument. For details, see the guide to [submitting |
| 849 | # a training job](/ai-platform/training/docs/training-jobs). |
| 850 | "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. |
| 851 | # |
| 852 | # You should only set `parameterServerConfig.acceleratorConfig` if |
| 853 | # `parameterServerType` is set to a Compute Engine machine type. [Learn |
| 854 | # about restrictions on accelerator configurations for |
| 855 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 856 | # |
| 857 | # Set `parameterServerConfig.imageUri` only if you build a custom image for |
| 858 | # your parameter server. If `parameterServerConfig.imageUri` has not been |
| 859 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 860 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 861 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 862 | # the one used in the custom container. This field is required if the replica |
| 863 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 864 | # this field. This must be a [runtime version that currently supports |
| 865 | # training with |
| 866 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 867 | # |
| 868 | # Note that the version of TensorFlow included in a runtime version may |
| 869 | # differ from the numbering of the runtime version itself, because it may |
| 870 | # have a different [patch |
| 871 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 872 | # In this field, you must specify the runtime version (TensorFlow minor |
| 873 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 874 | # specify `1.x`. |
| 875 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 876 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 877 | # If not provided, the docker image's ENTRYPOINT is used. |
| 878 | # It cannot be set if custom container image is |
| 879 | # not provided. |
| 880 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 881 | # both cannot be set at the same time. |
| 882 | "A String", |
| 883 | ], |
| 884 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 885 | # Registry. Learn more about [configuring custom |
| 886 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 887 | "containerArgs": [ # Arguments to the entrypoint command. |
| 888 | # The following rules apply for container_command and container_args: |
| 889 | # - If you do not supply command or args: |
| 890 | # The defaults defined in the Docker image are used. |
| 891 | # - If you supply a command but no args: |
| 892 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 893 | # are ignored. Your command is run without any arguments. |
| 894 | # - If you supply only args: |
| 895 | # The default Entrypoint defined in the Docker image is run with the args |
| 896 | # that you supplied. |
| 897 | # - If you supply a command and args: |
| 898 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 899 | # are ignored. Your command is run with your args. |
| 900 | # It cannot be set if custom container image is |
| 901 | # not provided. |
| 902 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 903 | # both cannot be set at the same time. |
| 904 | "A String", |
| 905 | ], |
| 906 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 907 | # [Learn about restrictions on accelerator configurations for |
| 908 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 909 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 910 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 911 | # [accelerators for online |
| 912 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 913 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 914 | "type": "A String", # The type of accelerator to use. |
| 915 | }, |
| 916 | }, |
| 917 | "encryptionConfig": { # Represents a custom encryption key configuration that can be applied to # Optional. Options for using customer-managed encryption keys (CMEK) to |
| 918 | # protect resources created by a training job, instead of using Google's |
| 919 | # default encryption. If this is set, then all resources created by the |
| 920 | # training job will be encrypted with the customer-managed encryption key |
| 921 | # that you specify. |
| 922 | # |
| 923 | # [Learn how and when to use CMEK with AI Platform |
| 924 | # Training](/ai-platform/training/docs/cmek). |
| 925 | # a resource. |
| 926 | "kmsKeyName": "A String", # The Cloud KMS resource identifier of the customer-managed encryption key |
| 927 | # used to protect a resource, such as a training job. It has the following |
| 928 | # format: |
| 929 | # `projects/{PROJECT_ID}/locations/{REGION}/keyRings/{KEY_RING_NAME}/cryptoKeys/{KEY_NAME}` |
| 930 | }, |
| 931 | "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. |
| 932 | "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For |
| 933 | # current versions of TensorFlow, this tag name should exactly match what is |
| 934 | # shown in TensorBoard, including all scopes. For versions of TensorFlow |
| 935 | # prior to 0.12, this should be only the tag passed to tf.Summary. |
| 936 | # By default, "training/hptuning/metric" will be used. |
| 937 | "params": [ # Required. The set of parameters to tune. |
| 938 | { # Represents a single hyperparameter to optimize. |
| 939 | "type": "A String", # Required. The type of the parameter. |
| 940 | "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. |
| 941 | "A String", |
| 942 | ], |
| 943 | "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in |
| 944 | # a HyperparameterSpec message. E.g., "learning_rate". |
| 945 | "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 946 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 947 | # type is INTEGER. |
| 948 | "discreteValues": [ # Required if type is `DISCRETE`. |
| 949 | # A list of feasible points. |
| 950 | # The list should be in strictly increasing order. For instance, this |
| 951 | # parameter might have possible settings of 1.5, 2.5, and 4.0. This list |
| 952 | # should not contain more than 1,000 values. |
| 953 | 3.14, |
| 954 | ], |
| 955 | "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. |
| 956 | # Leave unset for categorical parameters. |
| 957 | # Some kind of scaling is strongly recommended for real or integral |
| 958 | # parameters (e.g., `UNIT_LINEAR_SCALE`). |
| 959 | "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 960 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 961 | # type is `INTEGER`. |
| 962 | }, |
| 963 | ], |
| 964 | "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial |
| 965 | # early stopping. |
| 966 | "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to |
| 967 | # continue with. The job id will be used to find the corresponding vizier |
| 968 | # study guid and resume the study. |
| 969 | "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. |
| 970 | # You can reduce the time it takes to perform hyperparameter tuning by adding |
| 971 | # trials in parallel. However, each trail only benefits from the information |
| 972 | # gained in completed trials. That means that a trial does not get access to |
| 973 | # the results of trials running at the same time, which could reduce the |
| 974 | # quality of the overall optimization. |
| 975 | # |
| 976 | # Each trial will use the same scale tier and machine types. |
| 977 | # |
| 978 | # Defaults to one. |
| 979 | "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing |
| 980 | # the hyperparameter tuning job. You can specify this field to override the |
| 981 | # default failing criteria for AI Platform hyperparameter tuning jobs. |
| 982 | # |
| 983 | # Defaults to zero, which means the service decides when a hyperparameter |
| 984 | # job should fail. |
| 985 | "goal": "A String", # Required. The type of goal to use for tuning. Available types are |
| 986 | # `MAXIMIZE` and `MINIMIZE`. |
| 987 | # |
| 988 | # Defaults to `MAXIMIZE`. |
| 989 | "maxTrials": 42, # Optional. How many training trials should be attempted to optimize |
| 990 | # the specified hyperparameters. |
| 991 | # |
| 992 | # Defaults to one. |
| 993 | "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter |
| 994 | # tuning job. |
| 995 | # Uses the default AI Platform hyperparameter tuning |
| 996 | # algorithm if unspecified. |
| 997 | }, |
| 998 | "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. |
| 999 | # |
| 1000 | # You should only set `workerConfig.acceleratorConfig` if `workerType` is set |
| 1001 | # to a Compute Engine machine type. [Learn about restrictions on accelerator |
| 1002 | # configurations for |
| 1003 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1004 | # |
| 1005 | # Set `workerConfig.imageUri` only if you build a custom image for your |
| 1006 | # worker. If `workerConfig.imageUri` has not been set, AI Platform uses |
| 1007 | # the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 1008 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1009 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 1010 | # the one used in the custom container. This field is required if the replica |
| 1011 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 1012 | # this field. This must be a [runtime version that currently supports |
| 1013 | # training with |
| 1014 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 1015 | # |
| 1016 | # Note that the version of TensorFlow included in a runtime version may |
| 1017 | # differ from the numbering of the runtime version itself, because it may |
| 1018 | # have a different [patch |
| 1019 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 1020 | # In this field, you must specify the runtime version (TensorFlow minor |
| 1021 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 1022 | # specify `1.x`. |
| 1023 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 1024 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 1025 | # If not provided, the docker image's ENTRYPOINT is used. |
| 1026 | # It cannot be set if custom container image is |
| 1027 | # not provided. |
| 1028 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1029 | # both cannot be set at the same time. |
| 1030 | "A String", |
| 1031 | ], |
| 1032 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 1033 | # Registry. Learn more about [configuring custom |
| 1034 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1035 | "containerArgs": [ # Arguments to the entrypoint command. |
| 1036 | # The following rules apply for container_command and container_args: |
| 1037 | # - If you do not supply command or args: |
| 1038 | # The defaults defined in the Docker image are used. |
| 1039 | # - If you supply a command but no args: |
| 1040 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 1041 | # are ignored. Your command is run without any arguments. |
| 1042 | # - If you supply only args: |
| 1043 | # The default Entrypoint defined in the Docker image is run with the args |
| 1044 | # that you supplied. |
| 1045 | # - If you supply a command and args: |
| 1046 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 1047 | # are ignored. Your command is run with your args. |
| 1048 | # It cannot be set if custom container image is |
| 1049 | # not provided. |
| 1050 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1051 | # both cannot be set at the same time. |
| 1052 | "A String", |
| 1053 | ], |
| 1054 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 1055 | # [Learn about restrictions on accelerator configurations for |
| 1056 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1057 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 1058 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 1059 | # [accelerators for online |
| 1060 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 1061 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 1062 | "type": "A String", # The type of accelerator to use. |
| 1063 | }, |
| 1064 | }, |
| 1065 | "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training |
| 1066 | # job. Each replica in the cluster will be of the type specified in |
| 1067 | # `parameter_server_type`. |
| 1068 | # |
| 1069 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 1070 | # set this value, you must also set `parameter_server_type`. |
| 1071 | # |
| 1072 | # The default value is zero. |
| 1073 | "packageUris": [ # Required. The Google Cloud Storage location of the packages with |
| 1074 | # the training program and any additional dependencies. |
| 1075 | # The maximum number of package URIs is 100. |
| 1076 | "A String", |
| 1077 | ], |
| 1078 | "evaluatorCount": "A String", # Optional. The number of evaluator replicas to use for the training job. |
| 1079 | # Each replica in the cluster will be of the type specified in |
| 1080 | # `evaluator_type`. |
| 1081 | # |
| 1082 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 1083 | # set this value, you must also set `evaluator_type`. |
| 1084 | # |
| 1085 | # The default value is zero. |
| 1086 | "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 1087 | # job's master worker. You must specify this field when `scaleTier` is set to |
| 1088 | # `CUSTOM`. |
| 1089 | # |
| 1090 | # You can use certain Compute Engine machine types directly in this field. |
| 1091 | # The following types are supported: |
| 1092 | # |
| 1093 | # - `n1-standard-4` |
| 1094 | # - `n1-standard-8` |
| 1095 | # - `n1-standard-16` |
| 1096 | # - `n1-standard-32` |
| 1097 | # - `n1-standard-64` |
| 1098 | # - `n1-standard-96` |
| 1099 | # - `n1-highmem-2` |
| 1100 | # - `n1-highmem-4` |
| 1101 | # - `n1-highmem-8` |
| 1102 | # - `n1-highmem-16` |
| 1103 | # - `n1-highmem-32` |
| 1104 | # - `n1-highmem-64` |
| 1105 | # - `n1-highmem-96` |
| 1106 | # - `n1-highcpu-16` |
| 1107 | # - `n1-highcpu-32` |
| 1108 | # - `n1-highcpu-64` |
| 1109 | # - `n1-highcpu-96` |
| 1110 | # |
| 1111 | # Learn more about [using Compute Engine machine |
| 1112 | # types](/ml-engine/docs/machine-types#compute-engine-machine-types). |
| 1113 | # |
| 1114 | # Alternatively, you can use the following legacy machine types: |
| 1115 | # |
| 1116 | # - `standard` |
| 1117 | # - `large_model` |
| 1118 | # - `complex_model_s` |
| 1119 | # - `complex_model_m` |
| 1120 | # - `complex_model_l` |
| 1121 | # - `standard_gpu` |
| 1122 | # - `complex_model_m_gpu` |
| 1123 | # - `complex_model_l_gpu` |
| 1124 | # - `standard_p100` |
| 1125 | # - `complex_model_m_p100` |
| 1126 | # - `standard_v100` |
| 1127 | # - `large_model_v100` |
| 1128 | # - `complex_model_m_v100` |
| 1129 | # - `complex_model_l_v100` |
| 1130 | # |
| 1131 | # Learn more about [using legacy machine |
| 1132 | # types](/ml-engine/docs/machine-types#legacy-machine-types). |
| 1133 | # |
| 1134 | # Finally, if you want to use a TPU for training, specify `cloud_tpu` in this |
| 1135 | # field. Learn more about the [special configuration options for training |
| 1136 | # with |
| 1137 | # TPUs](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| 1138 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. You must |
| 1139 | # either specify this field or specify `masterConfig.imageUri`. |
| 1140 | # |
| 1141 | # For more information, see the [runtime version |
| 1142 | # list](/ai-platform/training/docs/runtime-version-list) and learn [how to |
| 1143 | # manage runtime versions](/ai-platform/training/docs/versioning). |
| 1144 | "evaluatorType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 1145 | # job's evaluator nodes. |
| 1146 | # |
| 1147 | # The supported values are the same as those described in the entry for |
| 1148 | # `masterType`. |
| 1149 | # |
| 1150 | # This value must be consistent with the category of machine type that |
| 1151 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 1152 | # types or both must be legacy machine types. |
| 1153 | # |
| 1154 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 1155 | # `evaluatorCount` is greater than zero. |
| 1156 | "region": "A String", # Required. The region to run the training job in. See the [available |
| 1157 | # regions](/ai-platform/training/docs/regions) for AI Platform Training. |
| 1158 | "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 1159 | # job's worker nodes. |
| 1160 | # |
| 1161 | # The supported values are the same as those described in the entry for |
| 1162 | # `masterType`. |
| 1163 | # |
| 1164 | # This value must be consistent with the category of machine type that |
| 1165 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 1166 | # types or both must be legacy machine types. |
| 1167 | # |
| 1168 | # If you use `cloud_tpu` for this value, see special instructions for |
| 1169 | # [configuring a custom TPU |
| 1170 | # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| 1171 | # |
| 1172 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 1173 | # `workerCount` is greater than zero. |
| 1174 | "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 1175 | # job's parameter server. |
| 1176 | # |
| 1177 | # The supported values are the same as those described in the entry for |
| 1178 | # `master_type`. |
| 1179 | # |
| 1180 | # This value must be consistent with the category of machine type that |
| 1181 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 1182 | # types or both must be legacy machine types. |
| 1183 | # |
| 1184 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 1185 | # `parameter_server_count` is greater than zero. |
| 1186 | "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. |
| 1187 | # |
| 1188 | # You should only set `masterConfig.acceleratorConfig` if `masterType` is set |
| 1189 | # to a Compute Engine machine type. Learn about [restrictions on accelerator |
| 1190 | # configurations for |
| 1191 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1192 | # |
| 1193 | # Set `masterConfig.imageUri` only if you build a custom image. Only one of |
| 1194 | # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more |
| 1195 | # about [configuring custom |
| 1196 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1197 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 1198 | # the one used in the custom container. This field is required if the replica |
| 1199 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 1200 | # this field. This must be a [runtime version that currently supports |
| 1201 | # training with |
| 1202 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 1203 | # |
| 1204 | # Note that the version of TensorFlow included in a runtime version may |
| 1205 | # differ from the numbering of the runtime version itself, because it may |
| 1206 | # have a different [patch |
| 1207 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 1208 | # In this field, you must specify the runtime version (TensorFlow minor |
| 1209 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 1210 | # specify `1.x`. |
| 1211 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 1212 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 1213 | # If not provided, the docker image's ENTRYPOINT is used. |
| 1214 | # It cannot be set if custom container image is |
| 1215 | # not provided. |
| 1216 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1217 | # both cannot be set at the same time. |
| 1218 | "A String", |
| 1219 | ], |
| 1220 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 1221 | # Registry. Learn more about [configuring custom |
| 1222 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1223 | "containerArgs": [ # Arguments to the entrypoint command. |
| 1224 | # The following rules apply for container_command and container_args: |
| 1225 | # - If you do not supply command or args: |
| 1226 | # The defaults defined in the Docker image are used. |
| 1227 | # - If you supply a command but no args: |
| 1228 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 1229 | # are ignored. Your command is run without any arguments. |
| 1230 | # - If you supply only args: |
| 1231 | # The default Entrypoint defined in the Docker image is run with the args |
| 1232 | # that you supplied. |
| 1233 | # - If you supply a command and args: |
| 1234 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 1235 | # are ignored. Your command is run with your args. |
| 1236 | # It cannot be set if custom container image is |
| 1237 | # not provided. |
| 1238 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1239 | # both cannot be set at the same time. |
| 1240 | "A String", |
| 1241 | ], |
| 1242 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 1243 | # [Learn about restrictions on accelerator configurations for |
| 1244 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1245 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 1246 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 1247 | # [accelerators for online |
| 1248 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 1249 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 1250 | "type": "A String", # The type of accelerator to use. |
| 1251 | }, |
| 1252 | }, |
| 1253 | "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers |
| 1254 | # and parameter servers. |
| 1255 | "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs |
| 1256 | # and other data needed for training. This path is passed to your TensorFlow |
| 1257 | # program as the '--job-dir' command-line argument. The benefit of specifying |
| 1258 | # this field is that Cloud ML validates the path for use in training. |
| 1259 | "pythonVersion": "A String", # Optional. The version of Python used in training. You must either specify |
| 1260 | # this field or specify `masterConfig.imageUri`. |
| 1261 | # |
| 1262 | # The following Python versions are available: |
| 1263 | # |
| 1264 | # * Python '3.7' is available when `runtime_version` is set to '1.15' or |
| 1265 | # later. |
| 1266 | # * Python '3.5' is available when `runtime_version` is set to a version |
| 1267 | # from '1.4' to '1.14'. |
| 1268 | # * Python '2.7' is available when `runtime_version` is set to '1.15' or |
| 1269 | # earlier. |
| 1270 | # |
| 1271 | # Read more about the Python versions available for [each runtime |
| 1272 | # version](/ml-engine/docs/runtime-version-list). |
| 1273 | "network": "A String", # Optional. The full name of the Google Compute Engine |
| 1274 | # [network](/compute/docs/networks-and-firewalls#networks) to which the Job |
| 1275 | # is peered. For example, projects/12345/global/networks/myVPC. Format is of |
| 1276 | # the form projects/{project}/global/networks/{network}. Where {project} is a |
| 1277 | # project number, as in '12345', and {network} is network name.". |
| 1278 | # |
| 1279 | # Private services access must already be configured for the network. If left |
| 1280 | # unspecified, the Job is not peered with any network. Learn more - |
| 1281 | # Connecting Job to user network over private |
| 1282 | # IP. |
| 1283 | "scheduling": { # All parameters related to scheduling of training jobs. # Optional. Scheduling options for a training job. |
| 1284 | "maxWaitTime": "A String", |
| 1285 | "maxRunningTime": "A String", # Optional. The maximum job running time, expressed in seconds. The field can |
| 1286 | # contain up to nine fractional digits, terminated by `s`. If not specified, |
| 1287 | # this field defaults to `604800s` (seven days). |
| 1288 | # |
| 1289 | # If the training job is still running after this duration, AI Platform |
| 1290 | # Training cancels it. |
| 1291 | # |
| 1292 | # For example, if you want to ensure your job runs for no more than 2 hours, |
| 1293 | # set this field to `7200s` (2 hours * 60 minutes / hour * 60 seconds / |
| 1294 | # minute). |
| 1295 | # |
| 1296 | # If you submit your training job using the `gcloud` tool, you can [provide |
| 1297 | # this field in a `config.yaml` |
| 1298 | # file](/ai-platform/training/docs/training-jobs#formatting_your_configuration_parameters). |
| 1299 | # For example: |
| 1300 | # |
| 1301 | # ```yaml |
| 1302 | # trainingInput: |
| 1303 | # ... |
| 1304 | # scheduling: |
| 1305 | # maxRunningTime: 7200s |
| 1306 | # ... |
| 1307 | # ``` |
| 1308 | }, |
| 1309 | "evaluatorConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for evaluators. |
| 1310 | # |
| 1311 | # You should only set `evaluatorConfig.acceleratorConfig` if |
| 1312 | # `evaluatorType` is set to a Compute Engine machine type. [Learn |
| 1313 | # about restrictions on accelerator configurations for |
| 1314 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1315 | # |
| 1316 | # Set `evaluatorConfig.imageUri` only if you build a custom image for |
| 1317 | # your evaluator. If `evaluatorConfig.imageUri` has not been |
| 1318 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 1319 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1320 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 1321 | # the one used in the custom container. This field is required if the replica |
| 1322 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 1323 | # this field. This must be a [runtime version that currently supports |
| 1324 | # training with |
| 1325 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 1326 | # |
| 1327 | # Note that the version of TensorFlow included in a runtime version may |
| 1328 | # differ from the numbering of the runtime version itself, because it may |
| 1329 | # have a different [patch |
| 1330 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 1331 | # In this field, you must specify the runtime version (TensorFlow minor |
| 1332 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 1333 | # specify `1.x`. |
| 1334 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 1335 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 1336 | # If not provided, the docker image's ENTRYPOINT is used. |
| 1337 | # It cannot be set if custom container image is |
| 1338 | # not provided. |
| 1339 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1340 | # both cannot be set at the same time. |
| 1341 | "A String", |
| 1342 | ], |
| 1343 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 1344 | # Registry. Learn more about [configuring custom |
| 1345 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1346 | "containerArgs": [ # Arguments to the entrypoint command. |
| 1347 | # The following rules apply for container_command and container_args: |
| 1348 | # - If you do not supply command or args: |
| 1349 | # The defaults defined in the Docker image are used. |
| 1350 | # - If you supply a command but no args: |
| 1351 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 1352 | # are ignored. Your command is run without any arguments. |
| 1353 | # - If you supply only args: |
| 1354 | # The default Entrypoint defined in the Docker image is run with the args |
| 1355 | # that you supplied. |
| 1356 | # - If you supply a command and args: |
| 1357 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 1358 | # are ignored. Your command is run with your args. |
| 1359 | # It cannot be set if custom container image is |
| 1360 | # not provided. |
| 1361 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1362 | # both cannot be set at the same time. |
| 1363 | "A String", |
| 1364 | ], |
| 1365 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 1366 | # [Learn about restrictions on accelerator configurations for |
| 1367 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1368 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 1369 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 1370 | # [accelerators for online |
| 1371 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 1372 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 1373 | "type": "A String", # The type of accelerator to use. |
| 1374 | }, |
| 1375 | }, |
| 1376 | "useChiefInTfConfig": True or False, # Optional. Use `chief` instead of `master` in the `TF_CONFIG` environment |
| 1377 | # variable when training with a custom container. Defaults to `false`. [Learn |
| 1378 | # more about this |
| 1379 | # field.](/ai-platform/training/docs/distributed-training-details#chief-versus-master) |
| 1380 | # |
| 1381 | # This field has no effect for training jobs that don't use a custom |
| 1382 | # container. |
| 1383 | "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each |
| 1384 | # replica in the cluster will be of the type specified in `worker_type`. |
| 1385 | # |
| 1386 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 1387 | # set this value, you must also set `worker_type`. |
| 1388 | # |
| 1389 | # The default value is zero. |
| 1390 | "pythonModule": "A String", # Required. The Python module name to run after installing the packages. |
| 1391 | "args": [ # Optional. Command-line arguments passed to the training application when it |
| 1392 | # starts. If your job uses a custom container, then the arguments are passed |
| 1393 | # to the container's <a class="external" target="_blank" |
| 1394 | # href="https://docs.docker.com/engine/reference/builder/#entrypoint"> |
| 1395 | # `ENTRYPOINT`</a> command. |
| 1396 | "A String", |
| 1397 | ], |
| 1398 | }, |
| 1399 | "state": "A String", # Output only. The detailed state of a job. |
| 1400 | "jobId": "A String", # Required. The user-specified id of the job. |
| 1401 | "endTime": "A String", # Output only. When the job processing was completed. |
| 1402 | "startTime": "A String", # Output only. When the job processing was started. |
| 1403 | "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. |
| 1404 | "errorCount": "A String", # The number of data instances which resulted in errors. |
| 1405 | "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. |
| 1406 | "nodeHours": 3.14, # Node hours used by the batch prediction job. |
| 1407 | "predictionCount": "A String", # The number of generated predictions. |
| 1408 | }, |
| 1409 | "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. |
| 1410 | "trials": [ # Results for individual Hyperparameter trials. |
| 1411 | # Only set for hyperparameter tuning jobs. |
| 1412 | { # Represents the result of a single hyperparameter tuning trial from a |
| 1413 | # training job. The TrainingOutput object that is returned on successful |
| 1414 | # completion of a training job with hyperparameter tuning includes a list |
| 1415 | # of HyperparameterOutput objects, one for each successful trial. |
| 1416 | "allMetrics": [ # All recorded object metrics for this trial. This field is not currently |
| 1417 | # populated. |
| 1418 | { # An observed value of a metric. |
| 1419 | "objectiveValue": 3.14, # The objective value at this training step. |
| 1420 | "trainingStep": "A String", # The global training step for this metric. |
| 1421 | }, |
| 1422 | ], |
| 1423 | "hyperparameters": { # The hyperparameters given to this trial. |
| 1424 | "a_key": "A String", |
| 1425 | }, |
| 1426 | "trialId": "A String", # The trial id for these results. |
| 1427 | "endTime": "A String", # Output only. End time for the trial. |
| 1428 | "isTrialStoppedEarly": True or False, # True if the trial is stopped early. |
| 1429 | "startTime": "A String", # Output only. Start time for the trial. |
| 1430 | "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. |
| 1431 | "objectiveValue": 3.14, # The objective value at this training step. |
| 1432 | "trainingStep": "A String", # The global training step for this metric. |
| 1433 | }, |
| 1434 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 1435 | # Only set for trials of built-in algorithms jobs that have succeeded. |
| 1436 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 1437 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 1438 | # saves the trained model. Only set for successful jobs that don't use |
| 1439 | # hyperparameter tuning. |
| 1440 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 1441 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 1442 | # trained. |
| 1443 | }, |
| 1444 | "state": "A String", # Output only. The detailed state of the trial. |
| 1445 | }, |
| 1446 | ], |
| 1447 | "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning |
| 1448 | # trials. See |
| 1449 | # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) |
| 1450 | # for more information. Only set for hyperparameter tuning jobs. |
| 1451 | "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. |
| 1452 | # Only set for hyperparameter tuning jobs. |
| 1453 | "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. |
| 1454 | "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. |
| 1455 | "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. |
| 1456 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 1457 | # Only set for built-in algorithms jobs. |
| 1458 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 1459 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 1460 | # saves the trained model. Only set for successful jobs that don't use |
| 1461 | # hyperparameter tuning. |
| 1462 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 1463 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 1464 | # trained. |
| 1465 | }, |
| 1466 | }, |
| 1467 | "createTime": "A String", # Output only. When the job was created. |
| 1468 | "labels": { # Optional. One or more labels that you can add, to organize your jobs. |
| 1469 | # Each label is a key-value pair, where both the key and the value are |
| 1470 | # arbitrary strings that you supply. |
| 1471 | # For more information, see the documentation on |
| 1472 | # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. |
| 1473 | "a_key": "A String", |
| 1474 | }, |
| 1475 | "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. |
| 1476 | "outputPath": "A String", # Required. The output Google Cloud Storage location. |
| 1477 | "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. |
| 1478 | "dataFormat": "A String", # Required. The format of the input data files. |
| 1479 | "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. |
| 1480 | # The service will buffer batch_size number of records in memory before |
| 1481 | # invoking one Tensorflow prediction call internally. So take the record |
| 1482 | # size and memory available into consideration when setting this parameter. |
| 1483 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch |
| 1484 | # prediction. If not set, AI Platform will pick the runtime version used |
| 1485 | # during the CreateVersion request for this model version, or choose the |
| 1486 | # latest stable version when model version information is not available |
| 1487 | # such as when the model is specified by uri. |
| 1488 | "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain |
| 1489 | # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. |
| 1490 | "A String", |
| 1491 | ], |
| 1492 | "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. |
| 1493 | # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> |
| 1494 | # for AI Platform services. |
| 1495 | "versionName": "A String", # Use this field if you want to specify a version of the model to use. The |
| 1496 | # string is formatted the same way as `model_version`, with the addition |
| 1497 | # of the version information: |
| 1498 | # |
| 1499 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` |
| 1500 | "modelName": "A String", # Use this field if you want to use the default version for the specified |
| 1501 | # model. The string must use the following format: |
| 1502 | # |
| 1503 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` |
| 1504 | "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for |
| 1505 | # the model to use. |
| 1506 | "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. |
| 1507 | # Defaults to 10 if not specified. |
| 1508 | "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for |
| 1509 | # this job. Please refer to |
| 1510 | # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) |
| 1511 | # for information about how to use signatures. |
| 1512 | # |
| 1513 | # Defaults to |
| 1514 | # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) |
| 1515 | # , which is "serving_default". |
| 1516 | }, |
| 1517 | "errorMessage": "A String", # Output only. The details of a failure or a cancellation. |
| 1518 | }</pre> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1519 | </div> |
| 1520 | |
| 1521 | <div class="method"> |
| 1522 | <code class="details" id="get">get(name, x__xgafv=None)</code> |
| 1523 | <pre>Describes a job. |
| 1524 | |
| 1525 | Args: |
| 1526 | name: string, Required. The name of the job to get the description of. (required) |
| 1527 | x__xgafv: string, V1 error format. |
| 1528 | Allowed values |
| 1529 | 1 - v1 error format |
| 1530 | 2 - v2 error format |
| 1531 | |
| 1532 | Returns: |
| 1533 | An object of the form: |
| 1534 | |
| 1535 | { # Represents a training or prediction job. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1536 | "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help |
| 1537 | # prevent simultaneous updates of a job from overwriting each other. |
| 1538 | # It is strongly suggested that systems make use of the `etag` in the |
| 1539 | # read-modify-write cycle to perform job updates in order to avoid race |
| 1540 | # conditions: An `etag` is returned in the response to `GetJob`, and |
| 1541 | # systems are expected to put that etag in the request to `UpdateJob` to |
| 1542 | # ensure that their change will be applied to the same version of the job. |
| 1543 | "trainingInput": { # Represents input parameters for a training job. When using the gcloud command # Input parameters to create a training job. |
| 1544 | # to submit your training job, you can specify the input parameters as |
| 1545 | # command-line arguments and/or in a YAML configuration file referenced from |
| 1546 | # the --config command-line argument. For details, see the guide to [submitting |
| 1547 | # a training job](/ai-platform/training/docs/training-jobs). |
| 1548 | "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. |
| 1549 | # |
| 1550 | # You should only set `parameterServerConfig.acceleratorConfig` if |
| 1551 | # `parameterServerType` is set to a Compute Engine machine type. [Learn |
| 1552 | # about restrictions on accelerator configurations for |
| 1553 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1554 | # |
| 1555 | # Set `parameterServerConfig.imageUri` only if you build a custom image for |
| 1556 | # your parameter server. If `parameterServerConfig.imageUri` has not been |
| 1557 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 1558 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1559 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 1560 | # the one used in the custom container. This field is required if the replica |
| 1561 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 1562 | # this field. This must be a [runtime version that currently supports |
| 1563 | # training with |
| 1564 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 1565 | # |
| 1566 | # Note that the version of TensorFlow included in a runtime version may |
| 1567 | # differ from the numbering of the runtime version itself, because it may |
| 1568 | # have a different [patch |
| 1569 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 1570 | # In this field, you must specify the runtime version (TensorFlow minor |
| 1571 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 1572 | # specify `1.x`. |
| 1573 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 1574 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 1575 | # If not provided, the docker image's ENTRYPOINT is used. |
| 1576 | # It cannot be set if custom container image is |
| 1577 | # not provided. |
| 1578 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1579 | # both cannot be set at the same time. |
| 1580 | "A String", |
| 1581 | ], |
| 1582 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 1583 | # Registry. Learn more about [configuring custom |
| 1584 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1585 | "containerArgs": [ # Arguments to the entrypoint command. |
| 1586 | # The following rules apply for container_command and container_args: |
| 1587 | # - If you do not supply command or args: |
| 1588 | # The defaults defined in the Docker image are used. |
| 1589 | # - If you supply a command but no args: |
| 1590 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 1591 | # are ignored. Your command is run without any arguments. |
| 1592 | # - If you supply only args: |
| 1593 | # The default Entrypoint defined in the Docker image is run with the args |
| 1594 | # that you supplied. |
| 1595 | # - If you supply a command and args: |
| 1596 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 1597 | # are ignored. Your command is run with your args. |
| 1598 | # It cannot be set if custom container image is |
| 1599 | # not provided. |
| 1600 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1601 | # both cannot be set at the same time. |
| 1602 | "A String", |
| 1603 | ], |
| 1604 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 1605 | # [Learn about restrictions on accelerator configurations for |
| 1606 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1607 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 1608 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 1609 | # [accelerators for online |
| 1610 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 1611 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 1612 | "type": "A String", # The type of accelerator to use. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1613 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1614 | }, |
| 1615 | "encryptionConfig": { # Represents a custom encryption key configuration that can be applied to # Optional. Options for using customer-managed encryption keys (CMEK) to |
| 1616 | # protect resources created by a training job, instead of using Google's |
| 1617 | # default encryption. If this is set, then all resources created by the |
| 1618 | # training job will be encrypted with the customer-managed encryption key |
| 1619 | # that you specify. |
| 1620 | # |
| 1621 | # [Learn how and when to use CMEK with AI Platform |
| 1622 | # Training](/ai-platform/training/docs/cmek). |
| 1623 | # a resource. |
| 1624 | "kmsKeyName": "A String", # The Cloud KMS resource identifier of the customer-managed encryption key |
| 1625 | # used to protect a resource, such as a training job. It has the following |
| 1626 | # format: |
| 1627 | # `projects/{PROJECT_ID}/locations/{REGION}/keyRings/{KEY_RING_NAME}/cryptoKeys/{KEY_NAME}` |
| 1628 | }, |
| 1629 | "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. |
| 1630 | "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For |
| 1631 | # current versions of TensorFlow, this tag name should exactly match what is |
| 1632 | # shown in TensorBoard, including all scopes. For versions of TensorFlow |
| 1633 | # prior to 0.12, this should be only the tag passed to tf.Summary. |
| 1634 | # By default, "training/hptuning/metric" will be used. |
| 1635 | "params": [ # Required. The set of parameters to tune. |
| 1636 | { # Represents a single hyperparameter to optimize. |
| 1637 | "type": "A String", # Required. The type of the parameter. |
| 1638 | "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. |
| 1639 | "A String", |
| 1640 | ], |
| 1641 | "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in |
| 1642 | # a HyperparameterSpec message. E.g., "learning_rate". |
| 1643 | "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 1644 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 1645 | # type is INTEGER. |
| 1646 | "discreteValues": [ # Required if type is `DISCRETE`. |
| 1647 | # A list of feasible points. |
| 1648 | # The list should be in strictly increasing order. For instance, this |
| 1649 | # parameter might have possible settings of 1.5, 2.5, and 4.0. This list |
| 1650 | # should not contain more than 1,000 values. |
| 1651 | 3.14, |
| 1652 | ], |
| 1653 | "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. |
| 1654 | # Leave unset for categorical parameters. |
| 1655 | # Some kind of scaling is strongly recommended for real or integral |
| 1656 | # parameters (e.g., `UNIT_LINEAR_SCALE`). |
| 1657 | "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 1658 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 1659 | # type is `INTEGER`. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1660 | }, |
| 1661 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1662 | "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial |
| 1663 | # early stopping. |
| 1664 | "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to |
| 1665 | # continue with. The job id will be used to find the corresponding vizier |
| 1666 | # study guid and resume the study. |
| 1667 | "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. |
| 1668 | # You can reduce the time it takes to perform hyperparameter tuning by adding |
| 1669 | # trials in parallel. However, each trail only benefits from the information |
| 1670 | # gained in completed trials. That means that a trial does not get access to |
| 1671 | # the results of trials running at the same time, which could reduce the |
| 1672 | # quality of the overall optimization. |
| 1673 | # |
| 1674 | # Each trial will use the same scale tier and machine types. |
| 1675 | # |
| 1676 | # Defaults to one. |
| 1677 | "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing |
| 1678 | # the hyperparameter tuning job. You can specify this field to override the |
| 1679 | # default failing criteria for AI Platform hyperparameter tuning jobs. |
| 1680 | # |
| 1681 | # Defaults to zero, which means the service decides when a hyperparameter |
| 1682 | # job should fail. |
| 1683 | "goal": "A String", # Required. The type of goal to use for tuning. Available types are |
| 1684 | # `MAXIMIZE` and `MINIMIZE`. |
| 1685 | # |
| 1686 | # Defaults to `MAXIMIZE`. |
| 1687 | "maxTrials": 42, # Optional. How many training trials should be attempted to optimize |
| 1688 | # the specified hyperparameters. |
| 1689 | # |
| 1690 | # Defaults to one. |
| 1691 | "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter |
| 1692 | # tuning job. |
| 1693 | # Uses the default AI Platform hyperparameter tuning |
| 1694 | # algorithm if unspecified. |
| 1695 | }, |
| 1696 | "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. |
| 1697 | # |
| 1698 | # You should only set `workerConfig.acceleratorConfig` if `workerType` is set |
| 1699 | # to a Compute Engine machine type. [Learn about restrictions on accelerator |
| 1700 | # configurations for |
| 1701 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1702 | # |
| 1703 | # Set `workerConfig.imageUri` only if you build a custom image for your |
| 1704 | # worker. If `workerConfig.imageUri` has not been set, AI Platform uses |
| 1705 | # the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 1706 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1707 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 1708 | # the one used in the custom container. This field is required if the replica |
| 1709 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 1710 | # this field. This must be a [runtime version that currently supports |
| 1711 | # training with |
| 1712 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 1713 | # |
| 1714 | # Note that the version of TensorFlow included in a runtime version may |
| 1715 | # differ from the numbering of the runtime version itself, because it may |
| 1716 | # have a different [patch |
| 1717 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 1718 | # In this field, you must specify the runtime version (TensorFlow minor |
| 1719 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 1720 | # specify `1.x`. |
| 1721 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 1722 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 1723 | # If not provided, the docker image's ENTRYPOINT is used. |
| 1724 | # It cannot be set if custom container image is |
| 1725 | # not provided. |
| 1726 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1727 | # both cannot be set at the same time. |
| 1728 | "A String", |
| 1729 | ], |
| 1730 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 1731 | # Registry. Learn more about [configuring custom |
| 1732 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1733 | "containerArgs": [ # Arguments to the entrypoint command. |
| 1734 | # The following rules apply for container_command and container_args: |
| 1735 | # - If you do not supply command or args: |
| 1736 | # The defaults defined in the Docker image are used. |
| 1737 | # - If you supply a command but no args: |
| 1738 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 1739 | # are ignored. Your command is run without any arguments. |
| 1740 | # - If you supply only args: |
| 1741 | # The default Entrypoint defined in the Docker image is run with the args |
| 1742 | # that you supplied. |
| 1743 | # - If you supply a command and args: |
| 1744 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 1745 | # are ignored. Your command is run with your args. |
| 1746 | # It cannot be set if custom container image is |
| 1747 | # not provided. |
| 1748 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1749 | # both cannot be set at the same time. |
| 1750 | "A String", |
| 1751 | ], |
| 1752 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 1753 | # [Learn about restrictions on accelerator configurations for |
| 1754 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1755 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 1756 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 1757 | # [accelerators for online |
| 1758 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 1759 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 1760 | "type": "A String", # The type of accelerator to use. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 1761 | }, |
| 1762 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1763 | "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training |
| 1764 | # job. Each replica in the cluster will be of the type specified in |
| 1765 | # `parameter_server_type`. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1766 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1767 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 1768 | # set this value, you must also set `parameter_server_type`. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1769 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1770 | # The default value is zero. |
| 1771 | "packageUris": [ # Required. The Google Cloud Storage location of the packages with |
| 1772 | # the training program and any additional dependencies. |
| 1773 | # The maximum number of package URIs is 100. |
| 1774 | "A String", |
| 1775 | ], |
| 1776 | "evaluatorCount": "A String", # Optional. The number of evaluator replicas to use for the training job. |
| 1777 | # Each replica in the cluster will be of the type specified in |
| 1778 | # `evaluator_type`. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1779 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1780 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 1781 | # set this value, you must also set `evaluator_type`. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1782 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1783 | # The default value is zero. |
| 1784 | "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 1785 | # job's master worker. You must specify this field when `scaleTier` is set to |
| 1786 | # `CUSTOM`. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1787 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1788 | # You can use certain Compute Engine machine types directly in this field. |
| 1789 | # The following types are supported: |
| 1790 | # |
| 1791 | # - `n1-standard-4` |
| 1792 | # - `n1-standard-8` |
| 1793 | # - `n1-standard-16` |
| 1794 | # - `n1-standard-32` |
| 1795 | # - `n1-standard-64` |
| 1796 | # - `n1-standard-96` |
| 1797 | # - `n1-highmem-2` |
| 1798 | # - `n1-highmem-4` |
| 1799 | # - `n1-highmem-8` |
| 1800 | # - `n1-highmem-16` |
| 1801 | # - `n1-highmem-32` |
| 1802 | # - `n1-highmem-64` |
| 1803 | # - `n1-highmem-96` |
| 1804 | # - `n1-highcpu-16` |
| 1805 | # - `n1-highcpu-32` |
| 1806 | # - `n1-highcpu-64` |
| 1807 | # - `n1-highcpu-96` |
| 1808 | # |
| 1809 | # Learn more about [using Compute Engine machine |
| 1810 | # types](/ml-engine/docs/machine-types#compute-engine-machine-types). |
| 1811 | # |
| 1812 | # Alternatively, you can use the following legacy machine types: |
| 1813 | # |
| 1814 | # - `standard` |
| 1815 | # - `large_model` |
| 1816 | # - `complex_model_s` |
| 1817 | # - `complex_model_m` |
| 1818 | # - `complex_model_l` |
| 1819 | # - `standard_gpu` |
| 1820 | # - `complex_model_m_gpu` |
| 1821 | # - `complex_model_l_gpu` |
| 1822 | # - `standard_p100` |
| 1823 | # - `complex_model_m_p100` |
| 1824 | # - `standard_v100` |
| 1825 | # - `large_model_v100` |
| 1826 | # - `complex_model_m_v100` |
| 1827 | # - `complex_model_l_v100` |
| 1828 | # |
| 1829 | # Learn more about [using legacy machine |
| 1830 | # types](/ml-engine/docs/machine-types#legacy-machine-types). |
| 1831 | # |
| 1832 | # Finally, if you want to use a TPU for training, specify `cloud_tpu` in this |
| 1833 | # field. Learn more about the [special configuration options for training |
| 1834 | # with |
| 1835 | # TPUs](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| 1836 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. You must |
| 1837 | # either specify this field or specify `masterConfig.imageUri`. |
| 1838 | # |
| 1839 | # For more information, see the [runtime version |
| 1840 | # list](/ai-platform/training/docs/runtime-version-list) and learn [how to |
| 1841 | # manage runtime versions](/ai-platform/training/docs/versioning). |
| 1842 | "evaluatorType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 1843 | # job's evaluator nodes. |
| 1844 | # |
| 1845 | # The supported values are the same as those described in the entry for |
| 1846 | # `masterType`. |
| 1847 | # |
| 1848 | # This value must be consistent with the category of machine type that |
| 1849 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 1850 | # types or both must be legacy machine types. |
| 1851 | # |
| 1852 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 1853 | # `evaluatorCount` is greater than zero. |
| 1854 | "region": "A String", # Required. The region to run the training job in. See the [available |
| 1855 | # regions](/ai-platform/training/docs/regions) for AI Platform Training. |
| 1856 | "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 1857 | # job's worker nodes. |
| 1858 | # |
| 1859 | # The supported values are the same as those described in the entry for |
| 1860 | # `masterType`. |
| 1861 | # |
| 1862 | # This value must be consistent with the category of machine type that |
| 1863 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 1864 | # types or both must be legacy machine types. |
| 1865 | # |
| 1866 | # If you use `cloud_tpu` for this value, see special instructions for |
| 1867 | # [configuring a custom TPU |
| 1868 | # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| 1869 | # |
| 1870 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 1871 | # `workerCount` is greater than zero. |
| 1872 | "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 1873 | # job's parameter server. |
| 1874 | # |
| 1875 | # The supported values are the same as those described in the entry for |
| 1876 | # `master_type`. |
| 1877 | # |
| 1878 | # This value must be consistent with the category of machine type that |
| 1879 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 1880 | # types or both must be legacy machine types. |
| 1881 | # |
| 1882 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 1883 | # `parameter_server_count` is greater than zero. |
| 1884 | "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. |
| 1885 | # |
| 1886 | # You should only set `masterConfig.acceleratorConfig` if `masterType` is set |
| 1887 | # to a Compute Engine machine type. Learn about [restrictions on accelerator |
| 1888 | # configurations for |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1889 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 1890 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 1891 | # Set `masterConfig.imageUri` only if you build a custom image. Only one of |
| 1892 | # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more |
| 1893 | # about [configuring custom |
| 1894 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1895 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 1896 | # the one used in the custom container. This field is required if the replica |
| 1897 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 1898 | # this field. This must be a [runtime version that currently supports |
| 1899 | # training with |
| 1900 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 1901 | # |
| 1902 | # Note that the version of TensorFlow included in a runtime version may |
| 1903 | # differ from the numbering of the runtime version itself, because it may |
| 1904 | # have a different [patch |
| 1905 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 1906 | # In this field, you must specify the runtime version (TensorFlow minor |
| 1907 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 1908 | # specify `1.x`. |
| 1909 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 1910 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 1911 | # If not provided, the docker image's ENTRYPOINT is used. |
| 1912 | # It cannot be set if custom container image is |
| 1913 | # not provided. |
| 1914 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1915 | # both cannot be set at the same time. |
| 1916 | "A String", |
| 1917 | ], |
| 1918 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 1919 | # Registry. Learn more about [configuring custom |
| 1920 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 1921 | "containerArgs": [ # Arguments to the entrypoint command. |
| 1922 | # The following rules apply for container_command and container_args: |
| 1923 | # - If you do not supply command or args: |
| 1924 | # The defaults defined in the Docker image are used. |
| 1925 | # - If you supply a command but no args: |
| 1926 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 1927 | # are ignored. Your command is run without any arguments. |
| 1928 | # - If you supply only args: |
| 1929 | # The default Entrypoint defined in the Docker image is run with the args |
| 1930 | # that you supplied. |
| 1931 | # - If you supply a command and args: |
| 1932 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 1933 | # are ignored. Your command is run with your args. |
| 1934 | # It cannot be set if custom container image is |
| 1935 | # not provided. |
| 1936 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 1937 | # both cannot be set at the same time. |
| 1938 | "A String", |
| 1939 | ], |
| 1940 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 1941 | # [Learn about restrictions on accelerator configurations for |
| 1942 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 1943 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 1944 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 1945 | # [accelerators for online |
| 1946 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 1947 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 1948 | "type": "A String", # The type of accelerator to use. |
| 1949 | }, |
| 1950 | }, |
| 1951 | "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers |
| 1952 | # and parameter servers. |
| 1953 | "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs |
| 1954 | # and other data needed for training. This path is passed to your TensorFlow |
| 1955 | # program as the '--job-dir' command-line argument. The benefit of specifying |
| 1956 | # this field is that Cloud ML validates the path for use in training. |
| 1957 | "pythonVersion": "A String", # Optional. The version of Python used in training. You must either specify |
| 1958 | # this field or specify `masterConfig.imageUri`. |
| 1959 | # |
| 1960 | # The following Python versions are available: |
| 1961 | # |
| 1962 | # * Python '3.7' is available when `runtime_version` is set to '1.15' or |
| 1963 | # later. |
| 1964 | # * Python '3.5' is available when `runtime_version` is set to a version |
| 1965 | # from '1.4' to '1.14'. |
| 1966 | # * Python '2.7' is available when `runtime_version` is set to '1.15' or |
| 1967 | # earlier. |
| 1968 | # |
| 1969 | # Read more about the Python versions available for [each runtime |
| 1970 | # version](/ml-engine/docs/runtime-version-list). |
| 1971 | "network": "A String", # Optional. The full name of the Google Compute Engine |
| 1972 | # [network](/compute/docs/networks-and-firewalls#networks) to which the Job |
| 1973 | # is peered. For example, projects/12345/global/networks/myVPC. Format is of |
| 1974 | # the form projects/{project}/global/networks/{network}. Where {project} is a |
| 1975 | # project number, as in '12345', and {network} is network name.". |
| 1976 | # |
| 1977 | # Private services access must already be configured for the network. If left |
| 1978 | # unspecified, the Job is not peered with any network. Learn more - |
| 1979 | # Connecting Job to user network over private |
| 1980 | # IP. |
| 1981 | "scheduling": { # All parameters related to scheduling of training jobs. # Optional. Scheduling options for a training job. |
| 1982 | "maxWaitTime": "A String", |
| 1983 | "maxRunningTime": "A String", # Optional. The maximum job running time, expressed in seconds. The field can |
| 1984 | # contain up to nine fractional digits, terminated by `s`. If not specified, |
| 1985 | # this field defaults to `604800s` (seven days). |
| 1986 | # |
| 1987 | # If the training job is still running after this duration, AI Platform |
| 1988 | # Training cancels it. |
| 1989 | # |
| 1990 | # For example, if you want to ensure your job runs for no more than 2 hours, |
| 1991 | # set this field to `7200s` (2 hours * 60 minutes / hour * 60 seconds / |
| 1992 | # minute). |
| 1993 | # |
| 1994 | # If you submit your training job using the `gcloud` tool, you can [provide |
| 1995 | # this field in a `config.yaml` |
| 1996 | # file](/ai-platform/training/docs/training-jobs#formatting_your_configuration_parameters). |
| 1997 | # For example: |
| 1998 | # |
| 1999 | # ```yaml |
| 2000 | # trainingInput: |
| 2001 | # ... |
| 2002 | # scheduling: |
| 2003 | # maxRunningTime: 7200s |
| 2004 | # ... |
| 2005 | # ``` |
| 2006 | }, |
| 2007 | "evaluatorConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for evaluators. |
| 2008 | # |
| 2009 | # You should only set `evaluatorConfig.acceleratorConfig` if |
| 2010 | # `evaluatorType` is set to a Compute Engine machine type. [Learn |
| 2011 | # about restrictions on accelerator configurations for |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2012 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2013 | # |
| 2014 | # Set `evaluatorConfig.imageUri` only if you build a custom image for |
| 2015 | # your evaluator. If `evaluatorConfig.imageUri` has not been |
| 2016 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2017 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2018 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 2019 | # the one used in the custom container. This field is required if the replica |
| 2020 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 2021 | # this field. This must be a [runtime version that currently supports |
| 2022 | # training with |
| 2023 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 2024 | # |
| 2025 | # Note that the version of TensorFlow included in a runtime version may |
| 2026 | # differ from the numbering of the runtime version itself, because it may |
| 2027 | # have a different [patch |
| 2028 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 2029 | # In this field, you must specify the runtime version (TensorFlow minor |
| 2030 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 2031 | # specify `1.x`. |
| 2032 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 2033 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 2034 | # If not provided, the docker image's ENTRYPOINT is used. |
| 2035 | # It cannot be set if custom container image is |
| 2036 | # not provided. |
| 2037 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 2038 | # both cannot be set at the same time. |
| 2039 | "A String", |
| 2040 | ], |
| 2041 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 2042 | # Registry. Learn more about [configuring custom |
| 2043 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 2044 | "containerArgs": [ # Arguments to the entrypoint command. |
| 2045 | # The following rules apply for container_command and container_args: |
| 2046 | # - If you do not supply command or args: |
| 2047 | # The defaults defined in the Docker image are used. |
| 2048 | # - If you supply a command but no args: |
| 2049 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 2050 | # are ignored. Your command is run without any arguments. |
| 2051 | # - If you supply only args: |
| 2052 | # The default Entrypoint defined in the Docker image is run with the args |
| 2053 | # that you supplied. |
| 2054 | # - If you supply a command and args: |
| 2055 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 2056 | # are ignored. Your command is run with your args. |
| 2057 | # It cannot be set if custom container image is |
| 2058 | # not provided. |
| 2059 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 2060 | # both cannot be set at the same time. |
| 2061 | "A String", |
| 2062 | ], |
| 2063 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 2064 | # [Learn about restrictions on accelerator configurations for |
| 2065 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 2066 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 2067 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 2068 | # [accelerators for online |
| 2069 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 2070 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 2071 | "type": "A String", # The type of accelerator to use. |
| 2072 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2073 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2074 | "useChiefInTfConfig": True or False, # Optional. Use `chief` instead of `master` in the `TF_CONFIG` environment |
| 2075 | # variable when training with a custom container. Defaults to `false`. [Learn |
| 2076 | # more about this |
| 2077 | # field.](/ai-platform/training/docs/distributed-training-details#chief-versus-master) |
| 2078 | # |
| 2079 | # This field has no effect for training jobs that don't use a custom |
| 2080 | # container. |
| 2081 | "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each |
| 2082 | # replica in the cluster will be of the type specified in `worker_type`. |
| 2083 | # |
| 2084 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 2085 | # set this value, you must also set `worker_type`. |
| 2086 | # |
| 2087 | # The default value is zero. |
| 2088 | "pythonModule": "A String", # Required. The Python module name to run after installing the packages. |
| 2089 | "args": [ # Optional. Command-line arguments passed to the training application when it |
| 2090 | # starts. If your job uses a custom container, then the arguments are passed |
| 2091 | # to the container's <a class="external" target="_blank" |
| 2092 | # href="https://docs.docker.com/engine/reference/builder/#entrypoint"> |
| 2093 | # `ENTRYPOINT`</a> command. |
| 2094 | "A String", |
| 2095 | ], |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2096 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2097 | "state": "A String", # Output only. The detailed state of a job. |
| 2098 | "jobId": "A String", # Required. The user-specified id of the job. |
| 2099 | "endTime": "A String", # Output only. When the job processing was completed. |
| 2100 | "startTime": "A String", # Output only. When the job processing was started. |
| 2101 | "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. |
| 2102 | "errorCount": "A String", # The number of data instances which resulted in errors. |
| 2103 | "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. |
| 2104 | "nodeHours": 3.14, # Node hours used by the batch prediction job. |
| 2105 | "predictionCount": "A String", # The number of generated predictions. |
| 2106 | }, |
| 2107 | "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. |
| 2108 | "trials": [ # Results for individual Hyperparameter trials. |
| 2109 | # Only set for hyperparameter tuning jobs. |
| 2110 | { # Represents the result of a single hyperparameter tuning trial from a |
| 2111 | # training job. The TrainingOutput object that is returned on successful |
| 2112 | # completion of a training job with hyperparameter tuning includes a list |
| 2113 | # of HyperparameterOutput objects, one for each successful trial. |
| 2114 | "allMetrics": [ # All recorded object metrics for this trial. This field is not currently |
| 2115 | # populated. |
| 2116 | { # An observed value of a metric. |
| 2117 | "objectiveValue": 3.14, # The objective value at this training step. |
| 2118 | "trainingStep": "A String", # The global training step for this metric. |
| 2119 | }, |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2120 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2121 | "hyperparameters": { # The hyperparameters given to this trial. |
| 2122 | "a_key": "A String", |
| 2123 | }, |
| 2124 | "trialId": "A String", # The trial id for these results. |
| 2125 | "endTime": "A String", # Output only. End time for the trial. |
| 2126 | "isTrialStoppedEarly": True or False, # True if the trial is stopped early. |
| 2127 | "startTime": "A String", # Output only. Start time for the trial. |
| 2128 | "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. |
| 2129 | "objectiveValue": 3.14, # The objective value at this training step. |
| 2130 | "trainingStep": "A String", # The global training step for this metric. |
| 2131 | }, |
| 2132 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 2133 | # Only set for trials of built-in algorithms jobs that have succeeded. |
| 2134 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 2135 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 2136 | # saves the trained model. Only set for successful jobs that don't use |
| 2137 | # hyperparameter tuning. |
| 2138 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 2139 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 2140 | # trained. |
| 2141 | }, |
| 2142 | "state": "A String", # Output only. The detailed state of the trial. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2143 | }, |
| 2144 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2145 | "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning |
| 2146 | # trials. See |
| 2147 | # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) |
| 2148 | # for more information. Only set for hyperparameter tuning jobs. |
| 2149 | "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. |
| 2150 | # Only set for hyperparameter tuning jobs. |
| 2151 | "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. |
| 2152 | "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. |
| 2153 | "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. |
| 2154 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 2155 | # Only set for built-in algorithms jobs. |
| 2156 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 2157 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 2158 | # saves the trained model. Only set for successful jobs that don't use |
| 2159 | # hyperparameter tuning. |
| 2160 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 2161 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 2162 | # trained. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2163 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2164 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2165 | "createTime": "A String", # Output only. When the job was created. |
| 2166 | "labels": { # Optional. One or more labels that you can add, to organize your jobs. |
| 2167 | # Each label is a key-value pair, where both the key and the value are |
| 2168 | # arbitrary strings that you supply. |
| 2169 | # For more information, see the documentation on |
| 2170 | # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. |
| 2171 | "a_key": "A String", |
| 2172 | }, |
| 2173 | "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. |
| 2174 | "outputPath": "A String", # Required. The output Google Cloud Storage location. |
| 2175 | "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. |
| 2176 | "dataFormat": "A String", # Required. The format of the input data files. |
| 2177 | "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. |
| 2178 | # The service will buffer batch_size number of records in memory before |
| 2179 | # invoking one Tensorflow prediction call internally. So take the record |
| 2180 | # size and memory available into consideration when setting this parameter. |
| 2181 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch |
| 2182 | # prediction. If not set, AI Platform will pick the runtime version used |
| 2183 | # during the CreateVersion request for this model version, or choose the |
| 2184 | # latest stable version when model version information is not available |
| 2185 | # such as when the model is specified by uri. |
| 2186 | "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain |
| 2187 | # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. |
| 2188 | "A String", |
| 2189 | ], |
| 2190 | "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. |
| 2191 | # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> |
| 2192 | # for AI Platform services. |
| 2193 | "versionName": "A String", # Use this field if you want to specify a version of the model to use. The |
| 2194 | # string is formatted the same way as `model_version`, with the addition |
| 2195 | # of the version information: |
| 2196 | # |
| 2197 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` |
| 2198 | "modelName": "A String", # Use this field if you want to use the default version for the specified |
| 2199 | # model. The string must use the following format: |
| 2200 | # |
| 2201 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` |
| 2202 | "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for |
| 2203 | # the model to use. |
| 2204 | "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. |
| 2205 | # Defaults to 10 if not specified. |
| 2206 | "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for |
| 2207 | # this job. Please refer to |
| 2208 | # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) |
| 2209 | # for information about how to use signatures. |
| 2210 | # |
| 2211 | # Defaults to |
| 2212 | # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) |
| 2213 | # , which is "serving_default". |
| 2214 | }, |
| 2215 | "errorMessage": "A String", # Output only. The details of a failure or a cancellation. |
| 2216 | }</pre> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 2217 | </div> |
| 2218 | |
| 2219 | <div class="method"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2220 | <code class="details" id="getIamPolicy">getIamPolicy(resource, options_requestedPolicyVersion=None, x__xgafv=None)</code> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2221 | <pre>Gets the access control policy for a resource. |
| 2222 | Returns an empty policy if the resource exists and does not have a policy |
| 2223 | set. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 2224 | |
| 2225 | Args: |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2226 | resource: string, REQUIRED: The resource for which the policy is being requested. |
| 2227 | See the operation documentation for the appropriate value for this field. (required) |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2228 | options_requestedPolicyVersion: integer, Optional. The policy format version to be returned. |
| 2229 | |
| 2230 | Valid values are 0, 1, and 3. Requests specifying an invalid value will be |
| 2231 | rejected. |
| 2232 | |
| 2233 | Requests for policies with any conditional bindings must specify version 3. |
| 2234 | Policies without any conditional bindings may specify any valid value or |
| 2235 | leave the field unset. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2236 | |
| 2237 | To learn which resources support conditions in their IAM policies, see the |
| 2238 | [IAM |
| 2239 | documentation](https://cloud.google.com/iam/help/conditions/resource-policies). |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 2240 | x__xgafv: string, V1 error format. |
| 2241 | Allowed values |
| 2242 | 1 - v1 error format |
| 2243 | 2 - v2 error format |
| 2244 | |
| 2245 | Returns: |
| 2246 | An object of the form: |
| 2247 | |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2248 | { # An Identity and Access Management (IAM) policy, which specifies access |
| 2249 | # controls for Google Cloud resources. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2250 | # |
| 2251 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2252 | # A `Policy` is a collection of `bindings`. A `binding` binds one or more |
| 2253 | # `members` to a single `role`. Members can be user accounts, service accounts, |
| 2254 | # Google groups, and domains (such as G Suite). A `role` is a named list of |
| 2255 | # permissions; each `role` can be an IAM predefined role or a user-created |
| 2256 | # custom role. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2257 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2258 | # For some types of Google Cloud resources, a `binding` can also specify a |
| 2259 | # `condition`, which is a logical expression that allows access to a resource |
| 2260 | # only if the expression evaluates to `true`. A condition can add constraints |
| 2261 | # based on attributes of the request, the resource, or both. To learn which |
| 2262 | # resources support conditions in their IAM policies, see the |
| 2263 | # [IAM documentation](https://cloud.google.com/iam/help/conditions/resource-policies). |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2264 | # |
| 2265 | # **JSON example:** |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2266 | # |
| 2267 | # { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2268 | # "bindings": [ |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2269 | # { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2270 | # "role": "roles/resourcemanager.organizationAdmin", |
| 2271 | # "members": [ |
| 2272 | # "user:mike@example.com", |
| 2273 | # "group:admins@example.com", |
| 2274 | # "domain:google.com", |
| 2275 | # "serviceAccount:my-project-id@appspot.gserviceaccount.com" |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2276 | # ] |
| 2277 | # }, |
| 2278 | # { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2279 | # "role": "roles/resourcemanager.organizationViewer", |
| 2280 | # "members": [ |
| 2281 | # "user:eve@example.com" |
| 2282 | # ], |
| 2283 | # "condition": { |
| 2284 | # "title": "expirable access", |
| 2285 | # "description": "Does not grant access after Sep 2020", |
| 2286 | # "expression": "request.time < timestamp('2020-10-01T00:00:00.000Z')", |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2287 | # } |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2288 | # } |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2289 | # ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2290 | # "etag": "BwWWja0YfJA=", |
| 2291 | # "version": 3 |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2292 | # } |
| 2293 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2294 | # **YAML example:** |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2295 | # |
| 2296 | # bindings: |
| 2297 | # - members: |
| 2298 | # - user:mike@example.com |
| 2299 | # - group:admins@example.com |
| 2300 | # - domain:google.com |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2301 | # - serviceAccount:my-project-id@appspot.gserviceaccount.com |
| 2302 | # role: roles/resourcemanager.organizationAdmin |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2303 | # - members: |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2304 | # - user:eve@example.com |
| 2305 | # role: roles/resourcemanager.organizationViewer |
| 2306 | # condition: |
| 2307 | # title: expirable access |
| 2308 | # description: Does not grant access after Sep 2020 |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2309 | # expression: request.time < timestamp('2020-10-01T00:00:00.000Z') |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2310 | # - etag: BwWWja0YfJA= |
| 2311 | # - version: 3 |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2312 | # |
| 2313 | # For a description of IAM and its features, see the |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2314 | # [IAM documentation](https://cloud.google.com/iam/docs/). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2315 | "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help |
| 2316 | # prevent simultaneous updates of a policy from overwriting each other. |
| 2317 | # It is strongly suggested that systems make use of the `etag` in the |
| 2318 | # read-modify-write cycle to perform policy updates in order to avoid race |
| 2319 | # conditions: An `etag` is returned in the response to `getIamPolicy`, and |
| 2320 | # systems are expected to put that etag in the request to `setIamPolicy` to |
| 2321 | # ensure that their change will be applied to the same version of the policy. |
| 2322 | # |
| 2323 | # **Important:** If you use IAM Conditions, you must include the `etag` field |
| 2324 | # whenever you call `setIamPolicy`. If you omit this field, then IAM allows |
| 2325 | # you to overwrite a version `3` policy with a version `1` policy, and all of |
| 2326 | # the conditions in the version `3` policy are lost. |
| 2327 | "version": 42, # Specifies the format of the policy. |
| 2328 | # |
| 2329 | # Valid values are `0`, `1`, and `3`. Requests that specify an invalid value |
| 2330 | # are rejected. |
| 2331 | # |
| 2332 | # Any operation that affects conditional role bindings must specify version |
| 2333 | # `3`. This requirement applies to the following operations: |
| 2334 | # |
| 2335 | # * Getting a policy that includes a conditional role binding |
| 2336 | # * Adding a conditional role binding to a policy |
| 2337 | # * Changing a conditional role binding in a policy |
| 2338 | # * Removing any role binding, with or without a condition, from a policy |
| 2339 | # that includes conditions |
| 2340 | # |
| 2341 | # **Important:** If you use IAM Conditions, you must include the `etag` field |
| 2342 | # whenever you call `setIamPolicy`. If you omit this field, then IAM allows |
| 2343 | # you to overwrite a version `3` policy with a version `1` policy, and all of |
| 2344 | # the conditions in the version `3` policy are lost. |
| 2345 | # |
| 2346 | # If a policy does not include any conditions, operations on that policy may |
| 2347 | # specify any valid version or leave the field unset. |
| 2348 | # |
| 2349 | # To learn which resources support conditions in their IAM policies, see the |
| 2350 | # [IAM documentation](https://cloud.google.com/iam/help/conditions/resource-policies). |
| 2351 | "auditConfigs": [ # Specifies cloud audit logging configuration for this policy. |
| 2352 | { # Specifies the audit configuration for a service. |
| 2353 | # The configuration determines which permission types are logged, and what |
| 2354 | # identities, if any, are exempted from logging. |
| 2355 | # An AuditConfig must have one or more AuditLogConfigs. |
| 2356 | # |
| 2357 | # If there are AuditConfigs for both `allServices` and a specific service, |
| 2358 | # the union of the two AuditConfigs is used for that service: the log_types |
| 2359 | # specified in each AuditConfig are enabled, and the exempted_members in each |
| 2360 | # AuditLogConfig are exempted. |
| 2361 | # |
| 2362 | # Example Policy with multiple AuditConfigs: |
| 2363 | # |
| 2364 | # { |
| 2365 | # "audit_configs": [ |
| 2366 | # { |
| 2367 | # "service": "allServices" |
| 2368 | # "audit_log_configs": [ |
| 2369 | # { |
| 2370 | # "log_type": "DATA_READ", |
| 2371 | # "exempted_members": [ |
| 2372 | # "user:jose@example.com" |
| 2373 | # ] |
| 2374 | # }, |
| 2375 | # { |
| 2376 | # "log_type": "DATA_WRITE", |
| 2377 | # }, |
| 2378 | # { |
| 2379 | # "log_type": "ADMIN_READ", |
| 2380 | # } |
| 2381 | # ] |
| 2382 | # }, |
| 2383 | # { |
| 2384 | # "service": "sampleservice.googleapis.com" |
| 2385 | # "audit_log_configs": [ |
| 2386 | # { |
| 2387 | # "log_type": "DATA_READ", |
| 2388 | # }, |
| 2389 | # { |
| 2390 | # "log_type": "DATA_WRITE", |
| 2391 | # "exempted_members": [ |
| 2392 | # "user:aliya@example.com" |
| 2393 | # ] |
| 2394 | # } |
| 2395 | # ] |
| 2396 | # } |
| 2397 | # ] |
| 2398 | # } |
| 2399 | # |
| 2400 | # For sampleservice, this policy enables DATA_READ, DATA_WRITE and ADMIN_READ |
| 2401 | # logging. It also exempts jose@example.com from DATA_READ logging, and |
| 2402 | # aliya@example.com from DATA_WRITE logging. |
| 2403 | "service": "A String", # Specifies a service that will be enabled for audit logging. |
| 2404 | # For example, `storage.googleapis.com`, `cloudsql.googleapis.com`. |
| 2405 | # `allServices` is a special value that covers all services. |
| 2406 | "auditLogConfigs": [ # The configuration for logging of each type of permission. |
| 2407 | { # Provides the configuration for logging a type of permissions. |
| 2408 | # Example: |
| 2409 | # |
| 2410 | # { |
| 2411 | # "audit_log_configs": [ |
| 2412 | # { |
| 2413 | # "log_type": "DATA_READ", |
| 2414 | # "exempted_members": [ |
| 2415 | # "user:jose@example.com" |
| 2416 | # ] |
| 2417 | # }, |
| 2418 | # { |
| 2419 | # "log_type": "DATA_WRITE", |
| 2420 | # } |
| 2421 | # ] |
| 2422 | # } |
| 2423 | # |
| 2424 | # This enables 'DATA_READ' and 'DATA_WRITE' logging, while exempting |
| 2425 | # jose@example.com from DATA_READ logging. |
| 2426 | "exemptedMembers": [ # Specifies the identities that do not cause logging for this type of |
| 2427 | # permission. |
| 2428 | # Follows the same format of Binding.members. |
| 2429 | "A String", |
| 2430 | ], |
| 2431 | "logType": "A String", # The log type that this config enables. |
| 2432 | }, |
| 2433 | ], |
| 2434 | }, |
| 2435 | ], |
| 2436 | "bindings": [ # Associates a list of `members` to a `role`. Optionally, may specify a |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2437 | # `condition` that determines how and when the `bindings` are applied. Each |
| 2438 | # of the `bindings` must contain at least one member. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2439 | { # Associates `members` with a `role`. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2440 | "condition": { # Represents a textual expression in the Common Expression Language (CEL) # The condition that is associated with this binding. |
| 2441 | # |
| 2442 | # If the condition evaluates to `true`, then this binding applies to the |
| 2443 | # current request. |
| 2444 | # |
| 2445 | # If the condition evaluates to `false`, then this binding does not apply to |
| 2446 | # the current request. However, a different role binding might grant the same |
| 2447 | # role to one or more of the members in this binding. |
| 2448 | # |
| 2449 | # To learn which resources support conditions in their IAM policies, see the |
| 2450 | # [IAM |
| 2451 | # documentation](https://cloud.google.com/iam/help/conditions/resource-policies). |
| 2452 | # syntax. CEL is a C-like expression language. The syntax and semantics of CEL |
| 2453 | # are documented at https://github.com/google/cel-spec. |
| 2454 | # |
| 2455 | # Example (Comparison): |
| 2456 | # |
| 2457 | # title: "Summary size limit" |
| 2458 | # description: "Determines if a summary is less than 100 chars" |
| 2459 | # expression: "document.summary.size() < 100" |
| 2460 | # |
| 2461 | # Example (Equality): |
| 2462 | # |
| 2463 | # title: "Requestor is owner" |
| 2464 | # description: "Determines if requestor is the document owner" |
| 2465 | # expression: "document.owner == request.auth.claims.email" |
| 2466 | # |
| 2467 | # Example (Logic): |
| 2468 | # |
| 2469 | # title: "Public documents" |
| 2470 | # description: "Determine whether the document should be publicly visible" |
| 2471 | # expression: "document.type != 'private' && document.type != 'internal'" |
| 2472 | # |
| 2473 | # Example (Data Manipulation): |
| 2474 | # |
| 2475 | # title: "Notification string" |
| 2476 | # description: "Create a notification string with a timestamp." |
| 2477 | # expression: "'New message received at ' + string(document.create_time)" |
| 2478 | # |
| 2479 | # The exact variables and functions that may be referenced within an expression |
| 2480 | # are determined by the service that evaluates it. See the service |
| 2481 | # documentation for additional information. |
| 2482 | "title": "A String", # Optional. Title for the expression, i.e. a short string describing |
| 2483 | # its purpose. This can be used e.g. in UIs which allow to enter the |
| 2484 | # expression. |
| 2485 | "location": "A String", # Optional. String indicating the location of the expression for error |
| 2486 | # reporting, e.g. a file name and a position in the file. |
| 2487 | "description": "A String", # Optional. Description of the expression. This is a longer text which |
| 2488 | # describes the expression, e.g. when hovered over it in a UI. |
| 2489 | "expression": "A String", # Textual representation of an expression in Common Expression Language |
| 2490 | # syntax. |
| 2491 | }, |
| 2492 | "members": [ # Specifies the identities requesting access for a Cloud Platform resource. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2493 | # `members` can have the following values: |
| 2494 | # |
| 2495 | # * `allUsers`: A special identifier that represents anyone who is |
| 2496 | # on the internet; with or without a Google account. |
| 2497 | # |
| 2498 | # * `allAuthenticatedUsers`: A special identifier that represents anyone |
| 2499 | # who is authenticated with a Google account or a service account. |
| 2500 | # |
| 2501 | # * `user:{emailid}`: An email address that represents a specific Google |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2502 | # account. For example, `alice@example.com` . |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2503 | # |
| 2504 | # |
| 2505 | # * `serviceAccount:{emailid}`: An email address that represents a service |
| 2506 | # account. For example, `my-other-app@appspot.gserviceaccount.com`. |
| 2507 | # |
| 2508 | # * `group:{emailid}`: An email address that represents a Google group. |
| 2509 | # For example, `admins@example.com`. |
| 2510 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2511 | # * `deleted:user:{emailid}?uid={uniqueid}`: An email address (plus unique |
| 2512 | # identifier) representing a user that has been recently deleted. For |
| 2513 | # example, `alice@example.com?uid=123456789012345678901`. If the user is |
| 2514 | # recovered, this value reverts to `user:{emailid}` and the recovered user |
| 2515 | # retains the role in the binding. |
| 2516 | # |
| 2517 | # * `deleted:serviceAccount:{emailid}?uid={uniqueid}`: An email address (plus |
| 2518 | # unique identifier) representing a service account that has been recently |
| 2519 | # deleted. For example, |
| 2520 | # `my-other-app@appspot.gserviceaccount.com?uid=123456789012345678901`. |
| 2521 | # If the service account is undeleted, this value reverts to |
| 2522 | # `serviceAccount:{emailid}` and the undeleted service account retains the |
| 2523 | # role in the binding. |
| 2524 | # |
| 2525 | # * `deleted:group:{emailid}?uid={uniqueid}`: An email address (plus unique |
| 2526 | # identifier) representing a Google group that has been recently |
| 2527 | # deleted. For example, `admins@example.com?uid=123456789012345678901`. If |
| 2528 | # the group is recovered, this value reverts to `group:{emailid}` and the |
| 2529 | # recovered group retains the role in the binding. |
| 2530 | # |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2531 | # |
| 2532 | # * `domain:{domain}`: The G Suite domain (primary) that represents all the |
| 2533 | # users of that domain. For example, `google.com` or `example.com`. |
| 2534 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2535 | "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2536 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2537 | "role": "A String", # Role that is assigned to `members`. |
| 2538 | # For example, `roles/viewer`, `roles/editor`, or `roles/owner`. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2539 | }, |
| 2540 | ], |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2541 | }</pre> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 2542 | </div> |
| 2543 | |
| 2544 | <div class="method"> |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2545 | <code class="details" id="list">list(parent, pageToken=None, pageSize=None, filter=None, x__xgafv=None)</code> |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 2546 | <pre>Lists the jobs in the project. |
| 2547 | |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2548 | If there are no jobs that match the request parameters, the list |
| 2549 | request returns an empty response body: {}. |
| 2550 | |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 2551 | Args: |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2552 | parent: string, Required. The name of the project for which to list jobs. (required) |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 2553 | pageToken: string, Optional. A page token to request the next page of results. |
| 2554 | |
| 2555 | You get the token from the `next_page_token` field of the response from |
| 2556 | the previous call. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2557 | pageSize: integer, Optional. The number of jobs to retrieve per "page" of results. If there |
| 2558 | are more remaining results than this number, the response message will |
| 2559 | contain a valid value in the `next_page_token` field. |
| 2560 | |
| 2561 | The default value is 20, and the maximum page size is 100. |
| 2562 | filter: string, Optional. Specifies the subset of jobs to retrieve. |
| 2563 | You can filter on the value of one or more attributes of the job object. |
| 2564 | For example, retrieve jobs with a job identifier that starts with 'census': |
| 2565 | <p><code>gcloud ai-platform jobs list --filter='jobId:census*'</code> |
| 2566 | <p>List all failed jobs with names that start with 'rnn': |
| 2567 | <p><code>gcloud ai-platform jobs list --filter='jobId:rnn* |
| 2568 | AND state:FAILED'</code> |
| 2569 | <p>For more examples, see the guide to |
| 2570 | <a href="/ml-engine/docs/tensorflow/monitor-training">monitoring jobs</a>. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 2571 | x__xgafv: string, V1 error format. |
| 2572 | Allowed values |
| 2573 | 1 - v1 error format |
| 2574 | 2 - v2 error format |
| 2575 | |
| 2576 | Returns: |
| 2577 | An object of the form: |
| 2578 | |
| 2579 | { # Response message for the ListJobs method. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2580 | "jobs": [ # The list of jobs. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 2581 | { # Represents a training or prediction job. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2582 | "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help |
| 2583 | # prevent simultaneous updates of a job from overwriting each other. |
| 2584 | # It is strongly suggested that systems make use of the `etag` in the |
| 2585 | # read-modify-write cycle to perform job updates in order to avoid race |
| 2586 | # conditions: An `etag` is returned in the response to `GetJob`, and |
| 2587 | # systems are expected to put that etag in the request to `UpdateJob` to |
| 2588 | # ensure that their change will be applied to the same version of the job. |
| 2589 | "trainingInput": { # Represents input parameters for a training job. When using the gcloud command # Input parameters to create a training job. |
| 2590 | # to submit your training job, you can specify the input parameters as |
| 2591 | # command-line arguments and/or in a YAML configuration file referenced from |
| 2592 | # the --config command-line argument. For details, see the guide to [submitting |
| 2593 | # a training job](/ai-platform/training/docs/training-jobs). |
| 2594 | "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. |
| 2595 | # |
| 2596 | # You should only set `parameterServerConfig.acceleratorConfig` if |
| 2597 | # `parameterServerType` is set to a Compute Engine machine type. [Learn |
| 2598 | # about restrictions on accelerator configurations for |
| 2599 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 2600 | # |
| 2601 | # Set `parameterServerConfig.imageUri` only if you build a custom image for |
| 2602 | # your parameter server. If `parameterServerConfig.imageUri` has not been |
| 2603 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 2604 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 2605 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 2606 | # the one used in the custom container. This field is required if the replica |
| 2607 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 2608 | # this field. This must be a [runtime version that currently supports |
| 2609 | # training with |
| 2610 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 2611 | # |
| 2612 | # Note that the version of TensorFlow included in a runtime version may |
| 2613 | # differ from the numbering of the runtime version itself, because it may |
| 2614 | # have a different [patch |
| 2615 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 2616 | # In this field, you must specify the runtime version (TensorFlow minor |
| 2617 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 2618 | # specify `1.x`. |
| 2619 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 2620 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 2621 | # If not provided, the docker image's ENTRYPOINT is used. |
| 2622 | # It cannot be set if custom container image is |
| 2623 | # not provided. |
| 2624 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 2625 | # both cannot be set at the same time. |
| 2626 | "A String", |
| 2627 | ], |
| 2628 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 2629 | # Registry. Learn more about [configuring custom |
| 2630 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 2631 | "containerArgs": [ # Arguments to the entrypoint command. |
| 2632 | # The following rules apply for container_command and container_args: |
| 2633 | # - If you do not supply command or args: |
| 2634 | # The defaults defined in the Docker image are used. |
| 2635 | # - If you supply a command but no args: |
| 2636 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 2637 | # are ignored. Your command is run without any arguments. |
| 2638 | # - If you supply only args: |
| 2639 | # The default Entrypoint defined in the Docker image is run with the args |
| 2640 | # that you supplied. |
| 2641 | # - If you supply a command and args: |
| 2642 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 2643 | # are ignored. Your command is run with your args. |
| 2644 | # It cannot be set if custom container image is |
| 2645 | # not provided. |
| 2646 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 2647 | # both cannot be set at the same time. |
| 2648 | "A String", |
| 2649 | ], |
| 2650 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 2651 | # [Learn about restrictions on accelerator configurations for |
| 2652 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 2653 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 2654 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 2655 | # [accelerators for online |
| 2656 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 2657 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 2658 | "type": "A String", # The type of accelerator to use. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2659 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2660 | }, |
| 2661 | "encryptionConfig": { # Represents a custom encryption key configuration that can be applied to # Optional. Options for using customer-managed encryption keys (CMEK) to |
| 2662 | # protect resources created by a training job, instead of using Google's |
| 2663 | # default encryption. If this is set, then all resources created by the |
| 2664 | # training job will be encrypted with the customer-managed encryption key |
| 2665 | # that you specify. |
| 2666 | # |
| 2667 | # [Learn how and when to use CMEK with AI Platform |
| 2668 | # Training](/ai-platform/training/docs/cmek). |
| 2669 | # a resource. |
| 2670 | "kmsKeyName": "A String", # The Cloud KMS resource identifier of the customer-managed encryption key |
| 2671 | # used to protect a resource, such as a training job. It has the following |
| 2672 | # format: |
| 2673 | # `projects/{PROJECT_ID}/locations/{REGION}/keyRings/{KEY_RING_NAME}/cryptoKeys/{KEY_NAME}` |
| 2674 | }, |
| 2675 | "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. |
| 2676 | "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For |
| 2677 | # current versions of TensorFlow, this tag name should exactly match what is |
| 2678 | # shown in TensorBoard, including all scopes. For versions of TensorFlow |
| 2679 | # prior to 0.12, this should be only the tag passed to tf.Summary. |
| 2680 | # By default, "training/hptuning/metric" will be used. |
| 2681 | "params": [ # Required. The set of parameters to tune. |
| 2682 | { # Represents a single hyperparameter to optimize. |
| 2683 | "type": "A String", # Required. The type of the parameter. |
| 2684 | "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. |
| 2685 | "A String", |
| 2686 | ], |
| 2687 | "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in |
| 2688 | # a HyperparameterSpec message. E.g., "learning_rate". |
| 2689 | "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 2690 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 2691 | # type is INTEGER. |
| 2692 | "discreteValues": [ # Required if type is `DISCRETE`. |
| 2693 | # A list of feasible points. |
| 2694 | # The list should be in strictly increasing order. For instance, this |
| 2695 | # parameter might have possible settings of 1.5, 2.5, and 4.0. This list |
| 2696 | # should not contain more than 1,000 values. |
| 2697 | 3.14, |
| 2698 | ], |
| 2699 | "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. |
| 2700 | # Leave unset for categorical parameters. |
| 2701 | # Some kind of scaling is strongly recommended for real or integral |
| 2702 | # parameters (e.g., `UNIT_LINEAR_SCALE`). |
| 2703 | "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 2704 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 2705 | # type is `INTEGER`. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 2706 | }, |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2707 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2708 | "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial |
| 2709 | # early stopping. |
| 2710 | "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to |
| 2711 | # continue with. The job id will be used to find the corresponding vizier |
| 2712 | # study guid and resume the study. |
| 2713 | "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. |
| 2714 | # You can reduce the time it takes to perform hyperparameter tuning by adding |
| 2715 | # trials in parallel. However, each trail only benefits from the information |
| 2716 | # gained in completed trials. That means that a trial does not get access to |
| 2717 | # the results of trials running at the same time, which could reduce the |
| 2718 | # quality of the overall optimization. |
| 2719 | # |
| 2720 | # Each trial will use the same scale tier and machine types. |
| 2721 | # |
| 2722 | # Defaults to one. |
| 2723 | "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing |
| 2724 | # the hyperparameter tuning job. You can specify this field to override the |
| 2725 | # default failing criteria for AI Platform hyperparameter tuning jobs. |
| 2726 | # |
| 2727 | # Defaults to zero, which means the service decides when a hyperparameter |
| 2728 | # job should fail. |
| 2729 | "goal": "A String", # Required. The type of goal to use for tuning. Available types are |
| 2730 | # `MAXIMIZE` and `MINIMIZE`. |
| 2731 | # |
| 2732 | # Defaults to `MAXIMIZE`. |
| 2733 | "maxTrials": 42, # Optional. How many training trials should be attempted to optimize |
| 2734 | # the specified hyperparameters. |
| 2735 | # |
| 2736 | # Defaults to one. |
| 2737 | "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter |
| 2738 | # tuning job. |
| 2739 | # Uses the default AI Platform hyperparameter tuning |
| 2740 | # algorithm if unspecified. |
| 2741 | }, |
| 2742 | "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. |
| 2743 | # |
| 2744 | # You should only set `workerConfig.acceleratorConfig` if `workerType` is set |
| 2745 | # to a Compute Engine machine type. [Learn about restrictions on accelerator |
| 2746 | # configurations for |
| 2747 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 2748 | # |
| 2749 | # Set `workerConfig.imageUri` only if you build a custom image for your |
| 2750 | # worker. If `workerConfig.imageUri` has not been set, AI Platform uses |
| 2751 | # the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 2752 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 2753 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 2754 | # the one used in the custom container. This field is required if the replica |
| 2755 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 2756 | # this field. This must be a [runtime version that currently supports |
| 2757 | # training with |
| 2758 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 2759 | # |
| 2760 | # Note that the version of TensorFlow included in a runtime version may |
| 2761 | # differ from the numbering of the runtime version itself, because it may |
| 2762 | # have a different [patch |
| 2763 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 2764 | # In this field, you must specify the runtime version (TensorFlow minor |
| 2765 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 2766 | # specify `1.x`. |
| 2767 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 2768 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 2769 | # If not provided, the docker image's ENTRYPOINT is used. |
| 2770 | # It cannot be set if custom container image is |
| 2771 | # not provided. |
| 2772 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 2773 | # both cannot be set at the same time. |
| 2774 | "A String", |
| 2775 | ], |
| 2776 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 2777 | # Registry. Learn more about [configuring custom |
| 2778 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 2779 | "containerArgs": [ # Arguments to the entrypoint command. |
| 2780 | # The following rules apply for container_command and container_args: |
| 2781 | # - If you do not supply command or args: |
| 2782 | # The defaults defined in the Docker image are used. |
| 2783 | # - If you supply a command but no args: |
| 2784 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 2785 | # are ignored. Your command is run without any arguments. |
| 2786 | # - If you supply only args: |
| 2787 | # The default Entrypoint defined in the Docker image is run with the args |
| 2788 | # that you supplied. |
| 2789 | # - If you supply a command and args: |
| 2790 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 2791 | # are ignored. Your command is run with your args. |
| 2792 | # It cannot be set if custom container image is |
| 2793 | # not provided. |
| 2794 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 2795 | # both cannot be set at the same time. |
| 2796 | "A String", |
| 2797 | ], |
| 2798 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 2799 | # [Learn about restrictions on accelerator configurations for |
| 2800 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 2801 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 2802 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 2803 | # [accelerators for online |
| 2804 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 2805 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 2806 | "type": "A String", # The type of accelerator to use. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 2807 | }, |
| 2808 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2809 | "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training |
| 2810 | # job. Each replica in the cluster will be of the type specified in |
| 2811 | # `parameter_server_type`. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2812 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2813 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 2814 | # set this value, you must also set `parameter_server_type`. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2815 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2816 | # The default value is zero. |
| 2817 | "packageUris": [ # Required. The Google Cloud Storage location of the packages with |
| 2818 | # the training program and any additional dependencies. |
| 2819 | # The maximum number of package URIs is 100. |
| 2820 | "A String", |
| 2821 | ], |
| 2822 | "evaluatorCount": "A String", # Optional. The number of evaluator replicas to use for the training job. |
| 2823 | # Each replica in the cluster will be of the type specified in |
| 2824 | # `evaluator_type`. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2825 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2826 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 2827 | # set this value, you must also set `evaluator_type`. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2828 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2829 | # The default value is zero. |
| 2830 | "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 2831 | # job's master worker. You must specify this field when `scaleTier` is set to |
| 2832 | # `CUSTOM`. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2833 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2834 | # You can use certain Compute Engine machine types directly in this field. |
| 2835 | # The following types are supported: |
| 2836 | # |
| 2837 | # - `n1-standard-4` |
| 2838 | # - `n1-standard-8` |
| 2839 | # - `n1-standard-16` |
| 2840 | # - `n1-standard-32` |
| 2841 | # - `n1-standard-64` |
| 2842 | # - `n1-standard-96` |
| 2843 | # - `n1-highmem-2` |
| 2844 | # - `n1-highmem-4` |
| 2845 | # - `n1-highmem-8` |
| 2846 | # - `n1-highmem-16` |
| 2847 | # - `n1-highmem-32` |
| 2848 | # - `n1-highmem-64` |
| 2849 | # - `n1-highmem-96` |
| 2850 | # - `n1-highcpu-16` |
| 2851 | # - `n1-highcpu-32` |
| 2852 | # - `n1-highcpu-64` |
| 2853 | # - `n1-highcpu-96` |
| 2854 | # |
| 2855 | # Learn more about [using Compute Engine machine |
| 2856 | # types](/ml-engine/docs/machine-types#compute-engine-machine-types). |
| 2857 | # |
| 2858 | # Alternatively, you can use the following legacy machine types: |
| 2859 | # |
| 2860 | # - `standard` |
| 2861 | # - `large_model` |
| 2862 | # - `complex_model_s` |
| 2863 | # - `complex_model_m` |
| 2864 | # - `complex_model_l` |
| 2865 | # - `standard_gpu` |
| 2866 | # - `complex_model_m_gpu` |
| 2867 | # - `complex_model_l_gpu` |
| 2868 | # - `standard_p100` |
| 2869 | # - `complex_model_m_p100` |
| 2870 | # - `standard_v100` |
| 2871 | # - `large_model_v100` |
| 2872 | # - `complex_model_m_v100` |
| 2873 | # - `complex_model_l_v100` |
| 2874 | # |
| 2875 | # Learn more about [using legacy machine |
| 2876 | # types](/ml-engine/docs/machine-types#legacy-machine-types). |
| 2877 | # |
| 2878 | # Finally, if you want to use a TPU for training, specify `cloud_tpu` in this |
| 2879 | # field. Learn more about the [special configuration options for training |
| 2880 | # with |
| 2881 | # TPUs](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| 2882 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. You must |
| 2883 | # either specify this field or specify `masterConfig.imageUri`. |
| 2884 | # |
| 2885 | # For more information, see the [runtime version |
| 2886 | # list](/ai-platform/training/docs/runtime-version-list) and learn [how to |
| 2887 | # manage runtime versions](/ai-platform/training/docs/versioning). |
| 2888 | "evaluatorType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 2889 | # job's evaluator nodes. |
| 2890 | # |
| 2891 | # The supported values are the same as those described in the entry for |
| 2892 | # `masterType`. |
| 2893 | # |
| 2894 | # This value must be consistent with the category of machine type that |
| 2895 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 2896 | # types or both must be legacy machine types. |
| 2897 | # |
| 2898 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 2899 | # `evaluatorCount` is greater than zero. |
| 2900 | "region": "A String", # Required. The region to run the training job in. See the [available |
| 2901 | # regions](/ai-platform/training/docs/regions) for AI Platform Training. |
| 2902 | "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 2903 | # job's worker nodes. |
| 2904 | # |
| 2905 | # The supported values are the same as those described in the entry for |
| 2906 | # `masterType`. |
| 2907 | # |
| 2908 | # This value must be consistent with the category of machine type that |
| 2909 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 2910 | # types or both must be legacy machine types. |
| 2911 | # |
| 2912 | # If you use `cloud_tpu` for this value, see special instructions for |
| 2913 | # [configuring a custom TPU |
| 2914 | # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| 2915 | # |
| 2916 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 2917 | # `workerCount` is greater than zero. |
| 2918 | "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 2919 | # job's parameter server. |
| 2920 | # |
| 2921 | # The supported values are the same as those described in the entry for |
| 2922 | # `master_type`. |
| 2923 | # |
| 2924 | # This value must be consistent with the category of machine type that |
| 2925 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 2926 | # types or both must be legacy machine types. |
| 2927 | # |
| 2928 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 2929 | # `parameter_server_count` is greater than zero. |
| 2930 | "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. |
| 2931 | # |
| 2932 | # You should only set `masterConfig.acceleratorConfig` if `masterType` is set |
| 2933 | # to a Compute Engine machine type. Learn about [restrictions on accelerator |
| 2934 | # configurations for |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2935 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 2936 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 2937 | # Set `masterConfig.imageUri` only if you build a custom image. Only one of |
| 2938 | # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more |
| 2939 | # about [configuring custom |
| 2940 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 2941 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 2942 | # the one used in the custom container. This field is required if the replica |
| 2943 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 2944 | # this field. This must be a [runtime version that currently supports |
| 2945 | # training with |
| 2946 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 2947 | # |
| 2948 | # Note that the version of TensorFlow included in a runtime version may |
| 2949 | # differ from the numbering of the runtime version itself, because it may |
| 2950 | # have a different [patch |
| 2951 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 2952 | # In this field, you must specify the runtime version (TensorFlow minor |
| 2953 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 2954 | # specify `1.x`. |
| 2955 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 2956 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 2957 | # If not provided, the docker image's ENTRYPOINT is used. |
| 2958 | # It cannot be set if custom container image is |
| 2959 | # not provided. |
| 2960 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 2961 | # both cannot be set at the same time. |
| 2962 | "A String", |
| 2963 | ], |
| 2964 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 2965 | # Registry. Learn more about [configuring custom |
| 2966 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 2967 | "containerArgs": [ # Arguments to the entrypoint command. |
| 2968 | # The following rules apply for container_command and container_args: |
| 2969 | # - If you do not supply command or args: |
| 2970 | # The defaults defined in the Docker image are used. |
| 2971 | # - If you supply a command but no args: |
| 2972 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 2973 | # are ignored. Your command is run without any arguments. |
| 2974 | # - If you supply only args: |
| 2975 | # The default Entrypoint defined in the Docker image is run with the args |
| 2976 | # that you supplied. |
| 2977 | # - If you supply a command and args: |
| 2978 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 2979 | # are ignored. Your command is run with your args. |
| 2980 | # It cannot be set if custom container image is |
| 2981 | # not provided. |
| 2982 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 2983 | # both cannot be set at the same time. |
| 2984 | "A String", |
| 2985 | ], |
| 2986 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 2987 | # [Learn about restrictions on accelerator configurations for |
| 2988 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 2989 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 2990 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 2991 | # [accelerators for online |
| 2992 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 2993 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 2994 | "type": "A String", # The type of accelerator to use. |
| 2995 | }, |
| 2996 | }, |
| 2997 | "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers |
| 2998 | # and parameter servers. |
| 2999 | "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs |
| 3000 | # and other data needed for training. This path is passed to your TensorFlow |
| 3001 | # program as the '--job-dir' command-line argument. The benefit of specifying |
| 3002 | # this field is that Cloud ML validates the path for use in training. |
| 3003 | "pythonVersion": "A String", # Optional. The version of Python used in training. You must either specify |
| 3004 | # this field or specify `masterConfig.imageUri`. |
| 3005 | # |
| 3006 | # The following Python versions are available: |
| 3007 | # |
| 3008 | # * Python '3.7' is available when `runtime_version` is set to '1.15' or |
| 3009 | # later. |
| 3010 | # * Python '3.5' is available when `runtime_version` is set to a version |
| 3011 | # from '1.4' to '1.14'. |
| 3012 | # * Python '2.7' is available when `runtime_version` is set to '1.15' or |
| 3013 | # earlier. |
| 3014 | # |
| 3015 | # Read more about the Python versions available for [each runtime |
| 3016 | # version](/ml-engine/docs/runtime-version-list). |
| 3017 | "network": "A String", # Optional. The full name of the Google Compute Engine |
| 3018 | # [network](/compute/docs/networks-and-firewalls#networks) to which the Job |
| 3019 | # is peered. For example, projects/12345/global/networks/myVPC. Format is of |
| 3020 | # the form projects/{project}/global/networks/{network}. Where {project} is a |
| 3021 | # project number, as in '12345', and {network} is network name.". |
| 3022 | # |
| 3023 | # Private services access must already be configured for the network. If left |
| 3024 | # unspecified, the Job is not peered with any network. Learn more - |
| 3025 | # Connecting Job to user network over private |
| 3026 | # IP. |
| 3027 | "scheduling": { # All parameters related to scheduling of training jobs. # Optional. Scheduling options for a training job. |
| 3028 | "maxWaitTime": "A String", |
| 3029 | "maxRunningTime": "A String", # Optional. The maximum job running time, expressed in seconds. The field can |
| 3030 | # contain up to nine fractional digits, terminated by `s`. If not specified, |
| 3031 | # this field defaults to `604800s` (seven days). |
| 3032 | # |
| 3033 | # If the training job is still running after this duration, AI Platform |
| 3034 | # Training cancels it. |
| 3035 | # |
| 3036 | # For example, if you want to ensure your job runs for no more than 2 hours, |
| 3037 | # set this field to `7200s` (2 hours * 60 minutes / hour * 60 seconds / |
| 3038 | # minute). |
| 3039 | # |
| 3040 | # If you submit your training job using the `gcloud` tool, you can [provide |
| 3041 | # this field in a `config.yaml` |
| 3042 | # file](/ai-platform/training/docs/training-jobs#formatting_your_configuration_parameters). |
| 3043 | # For example: |
| 3044 | # |
| 3045 | # ```yaml |
| 3046 | # trainingInput: |
| 3047 | # ... |
| 3048 | # scheduling: |
| 3049 | # maxRunningTime: 7200s |
| 3050 | # ... |
| 3051 | # ``` |
| 3052 | }, |
| 3053 | "evaluatorConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for evaluators. |
| 3054 | # |
| 3055 | # You should only set `evaluatorConfig.acceleratorConfig` if |
| 3056 | # `evaluatorType` is set to a Compute Engine machine type. [Learn |
| 3057 | # about restrictions on accelerator configurations for |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3058 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3059 | # |
| 3060 | # Set `evaluatorConfig.imageUri` only if you build a custom image for |
| 3061 | # your evaluator. If `evaluatorConfig.imageUri` has not been |
| 3062 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3063 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3064 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 3065 | # the one used in the custom container. This field is required if the replica |
| 3066 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 3067 | # this field. This must be a [runtime version that currently supports |
| 3068 | # training with |
| 3069 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 3070 | # |
| 3071 | # Note that the version of TensorFlow included in a runtime version may |
| 3072 | # differ from the numbering of the runtime version itself, because it may |
| 3073 | # have a different [patch |
| 3074 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 3075 | # In this field, you must specify the runtime version (TensorFlow minor |
| 3076 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 3077 | # specify `1.x`. |
| 3078 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 3079 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 3080 | # If not provided, the docker image's ENTRYPOINT is used. |
| 3081 | # It cannot be set if custom container image is |
| 3082 | # not provided. |
| 3083 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 3084 | # both cannot be set at the same time. |
| 3085 | "A String", |
| 3086 | ], |
| 3087 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 3088 | # Registry. Learn more about [configuring custom |
| 3089 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 3090 | "containerArgs": [ # Arguments to the entrypoint command. |
| 3091 | # The following rules apply for container_command and container_args: |
| 3092 | # - If you do not supply command or args: |
| 3093 | # The defaults defined in the Docker image are used. |
| 3094 | # - If you supply a command but no args: |
| 3095 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 3096 | # are ignored. Your command is run without any arguments. |
| 3097 | # - If you supply only args: |
| 3098 | # The default Entrypoint defined in the Docker image is run with the args |
| 3099 | # that you supplied. |
| 3100 | # - If you supply a command and args: |
| 3101 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 3102 | # are ignored. Your command is run with your args. |
| 3103 | # It cannot be set if custom container image is |
| 3104 | # not provided. |
| 3105 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 3106 | # both cannot be set at the same time. |
| 3107 | "A String", |
| 3108 | ], |
| 3109 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 3110 | # [Learn about restrictions on accelerator configurations for |
| 3111 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 3112 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 3113 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 3114 | # [accelerators for online |
| 3115 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 3116 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 3117 | "type": "A String", # The type of accelerator to use. |
| 3118 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3119 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3120 | "useChiefInTfConfig": True or False, # Optional. Use `chief` instead of `master` in the `TF_CONFIG` environment |
| 3121 | # variable when training with a custom container. Defaults to `false`. [Learn |
| 3122 | # more about this |
| 3123 | # field.](/ai-platform/training/docs/distributed-training-details#chief-versus-master) |
| 3124 | # |
| 3125 | # This field has no effect for training jobs that don't use a custom |
| 3126 | # container. |
| 3127 | "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each |
| 3128 | # replica in the cluster will be of the type specified in `worker_type`. |
| 3129 | # |
| 3130 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 3131 | # set this value, you must also set `worker_type`. |
| 3132 | # |
| 3133 | # The default value is zero. |
| 3134 | "pythonModule": "A String", # Required. The Python module name to run after installing the packages. |
| 3135 | "args": [ # Optional. Command-line arguments passed to the training application when it |
| 3136 | # starts. If your job uses a custom container, then the arguments are passed |
| 3137 | # to the container's <a class="external" target="_blank" |
| 3138 | # href="https://docs.docker.com/engine/reference/builder/#entrypoint"> |
| 3139 | # `ENTRYPOINT`</a> command. |
| 3140 | "A String", |
| 3141 | ], |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3142 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3143 | "state": "A String", # Output only. The detailed state of a job. |
| 3144 | "jobId": "A String", # Required. The user-specified id of the job. |
| 3145 | "endTime": "A String", # Output only. When the job processing was completed. |
| 3146 | "startTime": "A String", # Output only. When the job processing was started. |
| 3147 | "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. |
| 3148 | "errorCount": "A String", # The number of data instances which resulted in errors. |
| 3149 | "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. |
| 3150 | "nodeHours": 3.14, # Node hours used by the batch prediction job. |
| 3151 | "predictionCount": "A String", # The number of generated predictions. |
| 3152 | }, |
| 3153 | "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. |
| 3154 | "trials": [ # Results for individual Hyperparameter trials. |
| 3155 | # Only set for hyperparameter tuning jobs. |
| 3156 | { # Represents the result of a single hyperparameter tuning trial from a |
| 3157 | # training job. The TrainingOutput object that is returned on successful |
| 3158 | # completion of a training job with hyperparameter tuning includes a list |
| 3159 | # of HyperparameterOutput objects, one for each successful trial. |
| 3160 | "allMetrics": [ # All recorded object metrics for this trial. This field is not currently |
| 3161 | # populated. |
| 3162 | { # An observed value of a metric. |
| 3163 | "objectiveValue": 3.14, # The objective value at this training step. |
| 3164 | "trainingStep": "A String", # The global training step for this metric. |
| 3165 | }, |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3166 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3167 | "hyperparameters": { # The hyperparameters given to this trial. |
| 3168 | "a_key": "A String", |
| 3169 | }, |
| 3170 | "trialId": "A String", # The trial id for these results. |
| 3171 | "endTime": "A String", # Output only. End time for the trial. |
| 3172 | "isTrialStoppedEarly": True or False, # True if the trial is stopped early. |
| 3173 | "startTime": "A String", # Output only. Start time for the trial. |
| 3174 | "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. |
| 3175 | "objectiveValue": 3.14, # The objective value at this training step. |
| 3176 | "trainingStep": "A String", # The global training step for this metric. |
| 3177 | }, |
| 3178 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 3179 | # Only set for trials of built-in algorithms jobs that have succeeded. |
| 3180 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 3181 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 3182 | # saves the trained model. Only set for successful jobs that don't use |
| 3183 | # hyperparameter tuning. |
| 3184 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 3185 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 3186 | # trained. |
| 3187 | }, |
| 3188 | "state": "A String", # Output only. The detailed state of the trial. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 3189 | }, |
| 3190 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3191 | "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning |
| 3192 | # trials. See |
| 3193 | # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) |
| 3194 | # for more information. Only set for hyperparameter tuning jobs. |
| 3195 | "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. |
| 3196 | # Only set for hyperparameter tuning jobs. |
| 3197 | "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. |
| 3198 | "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. |
| 3199 | "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. |
| 3200 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 3201 | # Only set for built-in algorithms jobs. |
| 3202 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 3203 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 3204 | # saves the trained model. Only set for successful jobs that don't use |
| 3205 | # hyperparameter tuning. |
| 3206 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 3207 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 3208 | # trained. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3209 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3210 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3211 | "createTime": "A String", # Output only. When the job was created. |
| 3212 | "labels": { # Optional. One or more labels that you can add, to organize your jobs. |
| 3213 | # Each label is a key-value pair, where both the key and the value are |
| 3214 | # arbitrary strings that you supply. |
| 3215 | # For more information, see the documentation on |
| 3216 | # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. |
| 3217 | "a_key": "A String", |
| 3218 | }, |
| 3219 | "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. |
| 3220 | "outputPath": "A String", # Required. The output Google Cloud Storage location. |
| 3221 | "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. |
| 3222 | "dataFormat": "A String", # Required. The format of the input data files. |
| 3223 | "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. |
| 3224 | # The service will buffer batch_size number of records in memory before |
| 3225 | # invoking one Tensorflow prediction call internally. So take the record |
| 3226 | # size and memory available into consideration when setting this parameter. |
| 3227 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch |
| 3228 | # prediction. If not set, AI Platform will pick the runtime version used |
| 3229 | # during the CreateVersion request for this model version, or choose the |
| 3230 | # latest stable version when model version information is not available |
| 3231 | # such as when the model is specified by uri. |
| 3232 | "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain |
| 3233 | # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. |
| 3234 | "A String", |
| 3235 | ], |
| 3236 | "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. |
| 3237 | # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> |
| 3238 | # for AI Platform services. |
| 3239 | "versionName": "A String", # Use this field if you want to specify a version of the model to use. The |
| 3240 | # string is formatted the same way as `model_version`, with the addition |
| 3241 | # of the version information: |
| 3242 | # |
| 3243 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` |
| 3244 | "modelName": "A String", # Use this field if you want to use the default version for the specified |
| 3245 | # model. The string must use the following format: |
| 3246 | # |
| 3247 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` |
| 3248 | "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for |
| 3249 | # the model to use. |
| 3250 | "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. |
| 3251 | # Defaults to 10 if not specified. |
| 3252 | "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for |
| 3253 | # this job. Please refer to |
| 3254 | # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) |
| 3255 | # for information about how to use signatures. |
| 3256 | # |
| 3257 | # Defaults to |
| 3258 | # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) |
| 3259 | # , which is "serving_default". |
| 3260 | }, |
| 3261 | "errorMessage": "A String", # Output only. The details of a failure or a cancellation. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 3262 | }, |
| 3263 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3264 | "nextPageToken": "A String", # Optional. Pass this token as the `page_token` field of the request for a |
| 3265 | # subsequent call. |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 3266 | }</pre> |
| 3267 | </div> |
| 3268 | |
| 3269 | <div class="method"> |
| 3270 | <code class="details" id="list_next">list_next(previous_request, previous_response)</code> |
| 3271 | <pre>Retrieves the next page of results. |
| 3272 | |
| 3273 | Args: |
| 3274 | previous_request: The request for the previous page. (required) |
| 3275 | previous_response: The response from the request for the previous page. (required) |
| 3276 | |
| 3277 | Returns: |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3278 | A request object that you can call 'execute()' on to request the next |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 3279 | page. Returns None if there are no more items in the collection. |
| 3280 | </pre> |
| 3281 | </div> |
| 3282 | |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3283 | <div class="method"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3284 | <code class="details" id="patch">patch(name, body=None, updateMask=None, x__xgafv=None)</code> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3285 | <pre>Updates a specific job resource. |
| 3286 | |
| 3287 | Currently the only supported fields to update are `labels`. |
| 3288 | |
| 3289 | Args: |
| 3290 | name: string, Required. The job name. (required) |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3291 | body: object, The request body. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3292 | The object takes the form of: |
| 3293 | |
| 3294 | { # Represents a training or prediction job. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3295 | "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3296 | # prevent simultaneous updates of a job from overwriting each other. |
| 3297 | # It is strongly suggested that systems make use of the `etag` in the |
| 3298 | # read-modify-write cycle to perform job updates in order to avoid race |
| 3299 | # conditions: An `etag` is returned in the response to `GetJob`, and |
| 3300 | # systems are expected to put that etag in the request to `UpdateJob` to |
| 3301 | # ensure that their change will be applied to the same version of the job. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3302 | "trainingInput": { # Represents input parameters for a training job. When using the gcloud command # Input parameters to create a training job. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3303 | # to submit your training job, you can specify the input parameters as |
| 3304 | # command-line arguments and/or in a YAML configuration file referenced from |
| 3305 | # the --config command-line argument. For details, see the guide to [submitting |
| 3306 | # a training job](/ai-platform/training/docs/training-jobs). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3307 | "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. |
| 3308 | # |
| 3309 | # You should only set `parameterServerConfig.acceleratorConfig` if |
| 3310 | # `parameterServerType` is set to a Compute Engine machine type. [Learn |
| 3311 | # about restrictions on accelerator configurations for |
| 3312 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 3313 | # |
| 3314 | # Set `parameterServerConfig.imageUri` only if you build a custom image for |
| 3315 | # your parameter server. If `parameterServerConfig.imageUri` has not been |
| 3316 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 3317 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 3318 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 3319 | # the one used in the custom container. This field is required if the replica |
| 3320 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 3321 | # this field. This must be a [runtime version that currently supports |
| 3322 | # training with |
| 3323 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 3324 | # |
| 3325 | # Note that the version of TensorFlow included in a runtime version may |
| 3326 | # differ from the numbering of the runtime version itself, because it may |
| 3327 | # have a different [patch |
| 3328 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 3329 | # In this field, you must specify the runtime version (TensorFlow minor |
| 3330 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 3331 | # specify `1.x`. |
| 3332 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 3333 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 3334 | # If not provided, the docker image's ENTRYPOINT is used. |
| 3335 | # It cannot be set if custom container image is |
| 3336 | # not provided. |
| 3337 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 3338 | # both cannot be set at the same time. |
| 3339 | "A String", |
| 3340 | ], |
| 3341 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 3342 | # Registry. Learn more about [configuring custom |
| 3343 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 3344 | "containerArgs": [ # Arguments to the entrypoint command. |
| 3345 | # The following rules apply for container_command and container_args: |
| 3346 | # - If you do not supply command or args: |
| 3347 | # The defaults defined in the Docker image are used. |
| 3348 | # - If you supply a command but no args: |
| 3349 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 3350 | # are ignored. Your command is run without any arguments. |
| 3351 | # - If you supply only args: |
| 3352 | # The default Entrypoint defined in the Docker image is run with the args |
| 3353 | # that you supplied. |
| 3354 | # - If you supply a command and args: |
| 3355 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 3356 | # are ignored. Your command is run with your args. |
| 3357 | # It cannot be set if custom container image is |
| 3358 | # not provided. |
| 3359 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 3360 | # both cannot be set at the same time. |
| 3361 | "A String", |
| 3362 | ], |
| 3363 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 3364 | # [Learn about restrictions on accelerator configurations for |
| 3365 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 3366 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 3367 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 3368 | # [accelerators for online |
| 3369 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 3370 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 3371 | "type": "A String", # The type of accelerator to use. |
| 3372 | }, |
| 3373 | }, |
| 3374 | "encryptionConfig": { # Represents a custom encryption key configuration that can be applied to # Optional. Options for using customer-managed encryption keys (CMEK) to |
| 3375 | # protect resources created by a training job, instead of using Google's |
| 3376 | # default encryption. If this is set, then all resources created by the |
| 3377 | # training job will be encrypted with the customer-managed encryption key |
| 3378 | # that you specify. |
| 3379 | # |
| 3380 | # [Learn how and when to use CMEK with AI Platform |
| 3381 | # Training](/ai-platform/training/docs/cmek). |
| 3382 | # a resource. |
| 3383 | "kmsKeyName": "A String", # The Cloud KMS resource identifier of the customer-managed encryption key |
| 3384 | # used to protect a resource, such as a training job. It has the following |
| 3385 | # format: |
| 3386 | # `projects/{PROJECT_ID}/locations/{REGION}/keyRings/{KEY_RING_NAME}/cryptoKeys/{KEY_NAME}` |
| 3387 | }, |
| 3388 | "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. |
| 3389 | "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For |
| 3390 | # current versions of TensorFlow, this tag name should exactly match what is |
| 3391 | # shown in TensorBoard, including all scopes. For versions of TensorFlow |
| 3392 | # prior to 0.12, this should be only the tag passed to tf.Summary. |
| 3393 | # By default, "training/hptuning/metric" will be used. |
| 3394 | "params": [ # Required. The set of parameters to tune. |
| 3395 | { # Represents a single hyperparameter to optimize. |
| 3396 | "type": "A String", # Required. The type of the parameter. |
| 3397 | "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. |
| 3398 | "A String", |
| 3399 | ], |
| 3400 | "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in |
| 3401 | # a HyperparameterSpec message. E.g., "learning_rate". |
| 3402 | "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 3403 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 3404 | # type is INTEGER. |
| 3405 | "discreteValues": [ # Required if type is `DISCRETE`. |
| 3406 | # A list of feasible points. |
| 3407 | # The list should be in strictly increasing order. For instance, this |
| 3408 | # parameter might have possible settings of 1.5, 2.5, and 4.0. This list |
| 3409 | # should not contain more than 1,000 values. |
| 3410 | 3.14, |
| 3411 | ], |
| 3412 | "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. |
| 3413 | # Leave unset for categorical parameters. |
| 3414 | # Some kind of scaling is strongly recommended for real or integral |
| 3415 | # parameters (e.g., `UNIT_LINEAR_SCALE`). |
| 3416 | "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 3417 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 3418 | # type is `INTEGER`. |
| 3419 | }, |
| 3420 | ], |
| 3421 | "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial |
| 3422 | # early stopping. |
| 3423 | "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to |
| 3424 | # continue with. The job id will be used to find the corresponding vizier |
| 3425 | # study guid and resume the study. |
| 3426 | "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. |
| 3427 | # You can reduce the time it takes to perform hyperparameter tuning by adding |
| 3428 | # trials in parallel. However, each trail only benefits from the information |
| 3429 | # gained in completed trials. That means that a trial does not get access to |
| 3430 | # the results of trials running at the same time, which could reduce the |
| 3431 | # quality of the overall optimization. |
| 3432 | # |
| 3433 | # Each trial will use the same scale tier and machine types. |
| 3434 | # |
| 3435 | # Defaults to one. |
| 3436 | "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing |
| 3437 | # the hyperparameter tuning job. You can specify this field to override the |
| 3438 | # default failing criteria for AI Platform hyperparameter tuning jobs. |
| 3439 | # |
| 3440 | # Defaults to zero, which means the service decides when a hyperparameter |
| 3441 | # job should fail. |
| 3442 | "goal": "A String", # Required. The type of goal to use for tuning. Available types are |
| 3443 | # `MAXIMIZE` and `MINIMIZE`. |
| 3444 | # |
| 3445 | # Defaults to `MAXIMIZE`. |
| 3446 | "maxTrials": 42, # Optional. How many training trials should be attempted to optimize |
| 3447 | # the specified hyperparameters. |
| 3448 | # |
| 3449 | # Defaults to one. |
| 3450 | "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter |
| 3451 | # tuning job. |
| 3452 | # Uses the default AI Platform hyperparameter tuning |
| 3453 | # algorithm if unspecified. |
| 3454 | }, |
| 3455 | "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. |
| 3456 | # |
| 3457 | # You should only set `workerConfig.acceleratorConfig` if `workerType` is set |
| 3458 | # to a Compute Engine machine type. [Learn about restrictions on accelerator |
| 3459 | # configurations for |
| 3460 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 3461 | # |
| 3462 | # Set `workerConfig.imageUri` only if you build a custom image for your |
| 3463 | # worker. If `workerConfig.imageUri` has not been set, AI Platform uses |
| 3464 | # the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 3465 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 3466 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 3467 | # the one used in the custom container. This field is required if the replica |
| 3468 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 3469 | # this field. This must be a [runtime version that currently supports |
| 3470 | # training with |
| 3471 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 3472 | # |
| 3473 | # Note that the version of TensorFlow included in a runtime version may |
| 3474 | # differ from the numbering of the runtime version itself, because it may |
| 3475 | # have a different [patch |
| 3476 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 3477 | # In this field, you must specify the runtime version (TensorFlow minor |
| 3478 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 3479 | # specify `1.x`. |
| 3480 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 3481 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 3482 | # If not provided, the docker image's ENTRYPOINT is used. |
| 3483 | # It cannot be set if custom container image is |
| 3484 | # not provided. |
| 3485 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 3486 | # both cannot be set at the same time. |
| 3487 | "A String", |
| 3488 | ], |
| 3489 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 3490 | # Registry. Learn more about [configuring custom |
| 3491 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 3492 | "containerArgs": [ # Arguments to the entrypoint command. |
| 3493 | # The following rules apply for container_command and container_args: |
| 3494 | # - If you do not supply command or args: |
| 3495 | # The defaults defined in the Docker image are used. |
| 3496 | # - If you supply a command but no args: |
| 3497 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 3498 | # are ignored. Your command is run without any arguments. |
| 3499 | # - If you supply only args: |
| 3500 | # The default Entrypoint defined in the Docker image is run with the args |
| 3501 | # that you supplied. |
| 3502 | # - If you supply a command and args: |
| 3503 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 3504 | # are ignored. Your command is run with your args. |
| 3505 | # It cannot be set if custom container image is |
| 3506 | # not provided. |
| 3507 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 3508 | # both cannot be set at the same time. |
| 3509 | "A String", |
| 3510 | ], |
| 3511 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 3512 | # [Learn about restrictions on accelerator configurations for |
| 3513 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 3514 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 3515 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 3516 | # [accelerators for online |
| 3517 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 3518 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 3519 | "type": "A String", # The type of accelerator to use. |
| 3520 | }, |
| 3521 | }, |
| 3522 | "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training |
| 3523 | # job. Each replica in the cluster will be of the type specified in |
| 3524 | # `parameter_server_type`. |
| 3525 | # |
| 3526 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 3527 | # set this value, you must also set `parameter_server_type`. |
| 3528 | # |
| 3529 | # The default value is zero. |
| 3530 | "packageUris": [ # Required. The Google Cloud Storage location of the packages with |
| 3531 | # the training program and any additional dependencies. |
| 3532 | # The maximum number of package URIs is 100. |
| 3533 | "A String", |
| 3534 | ], |
| 3535 | "evaluatorCount": "A String", # Optional. The number of evaluator replicas to use for the training job. |
| 3536 | # Each replica in the cluster will be of the type specified in |
| 3537 | # `evaluator_type`. |
| 3538 | # |
| 3539 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 3540 | # set this value, you must also set `evaluator_type`. |
| 3541 | # |
| 3542 | # The default value is zero. |
| 3543 | "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 3544 | # job's master worker. You must specify this field when `scaleTier` is set to |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3545 | # `CUSTOM`. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3546 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3547 | # You can use certain Compute Engine machine types directly in this field. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3548 | # The following types are supported: |
| 3549 | # |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3550 | # - `n1-standard-4` |
| 3551 | # - `n1-standard-8` |
| 3552 | # - `n1-standard-16` |
| 3553 | # - `n1-standard-32` |
| 3554 | # - `n1-standard-64` |
| 3555 | # - `n1-standard-96` |
| 3556 | # - `n1-highmem-2` |
| 3557 | # - `n1-highmem-4` |
| 3558 | # - `n1-highmem-8` |
| 3559 | # - `n1-highmem-16` |
| 3560 | # - `n1-highmem-32` |
| 3561 | # - `n1-highmem-64` |
| 3562 | # - `n1-highmem-96` |
| 3563 | # - `n1-highcpu-16` |
| 3564 | # - `n1-highcpu-32` |
| 3565 | # - `n1-highcpu-64` |
| 3566 | # - `n1-highcpu-96` |
| 3567 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3568 | # Learn more about [using Compute Engine machine |
| 3569 | # types](/ml-engine/docs/machine-types#compute-engine-machine-types). |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3570 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3571 | # Alternatively, you can use the following legacy machine types: |
| 3572 | # |
| 3573 | # - `standard` |
| 3574 | # - `large_model` |
| 3575 | # - `complex_model_s` |
| 3576 | # - `complex_model_m` |
| 3577 | # - `complex_model_l` |
| 3578 | # - `standard_gpu` |
| 3579 | # - `complex_model_m_gpu` |
| 3580 | # - `complex_model_l_gpu` |
| 3581 | # - `standard_p100` |
| 3582 | # - `complex_model_m_p100` |
| 3583 | # - `standard_v100` |
| 3584 | # - `large_model_v100` |
| 3585 | # - `complex_model_m_v100` |
| 3586 | # - `complex_model_l_v100` |
| 3587 | # |
| 3588 | # Learn more about [using legacy machine |
| 3589 | # types](/ml-engine/docs/machine-types#legacy-machine-types). |
| 3590 | # |
| 3591 | # Finally, if you want to use a TPU for training, specify `cloud_tpu` in this |
| 3592 | # field. Learn more about the [special configuration options for training |
| 3593 | # with |
| 3594 | # TPUs](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3595 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. You must |
| 3596 | # either specify this field or specify `masterConfig.imageUri`. |
| 3597 | # |
| 3598 | # For more information, see the [runtime version |
| 3599 | # list](/ai-platform/training/docs/runtime-version-list) and learn [how to |
| 3600 | # manage runtime versions](/ai-platform/training/docs/versioning). |
| 3601 | "evaluatorType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 3602 | # job's evaluator nodes. |
| 3603 | # |
| 3604 | # The supported values are the same as those described in the entry for |
| 3605 | # `masterType`. |
| 3606 | # |
| 3607 | # This value must be consistent with the category of machine type that |
| 3608 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 3609 | # types or both must be legacy machine types. |
| 3610 | # |
| 3611 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 3612 | # `evaluatorCount` is greater than zero. |
| 3613 | "region": "A String", # Required. The region to run the training job in. See the [available |
| 3614 | # regions](/ai-platform/training/docs/regions) for AI Platform Training. |
| 3615 | "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 3616 | # job's worker nodes. |
| 3617 | # |
| 3618 | # The supported values are the same as those described in the entry for |
| 3619 | # `masterType`. |
| 3620 | # |
| 3621 | # This value must be consistent with the category of machine type that |
| 3622 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 3623 | # types or both must be legacy machine types. |
| 3624 | # |
| 3625 | # If you use `cloud_tpu` for this value, see special instructions for |
| 3626 | # [configuring a custom TPU |
| 3627 | # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| 3628 | # |
| 3629 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 3630 | # `workerCount` is greater than zero. |
| 3631 | "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 3632 | # job's parameter server. |
| 3633 | # |
| 3634 | # The supported values are the same as those described in the entry for |
| 3635 | # `master_type`. |
| 3636 | # |
| 3637 | # This value must be consistent with the category of machine type that |
| 3638 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 3639 | # types or both must be legacy machine types. |
| 3640 | # |
| 3641 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 3642 | # `parameter_server_count` is greater than zero. |
| 3643 | "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. |
| 3644 | # |
| 3645 | # You should only set `masterConfig.acceleratorConfig` if `masterType` is set |
| 3646 | # to a Compute Engine machine type. Learn about [restrictions on accelerator |
| 3647 | # configurations for |
| 3648 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 3649 | # |
| 3650 | # Set `masterConfig.imageUri` only if you build a custom image. Only one of |
| 3651 | # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more |
| 3652 | # about [configuring custom |
| 3653 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 3654 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 3655 | # the one used in the custom container. This field is required if the replica |
| 3656 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 3657 | # this field. This must be a [runtime version that currently supports |
| 3658 | # training with |
| 3659 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 3660 | # |
| 3661 | # Note that the version of TensorFlow included in a runtime version may |
| 3662 | # differ from the numbering of the runtime version itself, because it may |
| 3663 | # have a different [patch |
| 3664 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 3665 | # In this field, you must specify the runtime version (TensorFlow minor |
| 3666 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 3667 | # specify `1.x`. |
| 3668 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 3669 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 3670 | # If not provided, the docker image's ENTRYPOINT is used. |
| 3671 | # It cannot be set if custom container image is |
| 3672 | # not provided. |
| 3673 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 3674 | # both cannot be set at the same time. |
| 3675 | "A String", |
| 3676 | ], |
| 3677 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 3678 | # Registry. Learn more about [configuring custom |
| 3679 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 3680 | "containerArgs": [ # Arguments to the entrypoint command. |
| 3681 | # The following rules apply for container_command and container_args: |
| 3682 | # - If you do not supply command or args: |
| 3683 | # The defaults defined in the Docker image are used. |
| 3684 | # - If you supply a command but no args: |
| 3685 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 3686 | # are ignored. Your command is run without any arguments. |
| 3687 | # - If you supply only args: |
| 3688 | # The default Entrypoint defined in the Docker image is run with the args |
| 3689 | # that you supplied. |
| 3690 | # - If you supply a command and args: |
| 3691 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 3692 | # are ignored. Your command is run with your args. |
| 3693 | # It cannot be set if custom container image is |
| 3694 | # not provided. |
| 3695 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 3696 | # both cannot be set at the same time. |
| 3697 | "A String", |
| 3698 | ], |
| 3699 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 3700 | # [Learn about restrictions on accelerator configurations for |
| 3701 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 3702 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 3703 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 3704 | # [accelerators for online |
| 3705 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 3706 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 3707 | "type": "A String", # The type of accelerator to use. |
| 3708 | }, |
| 3709 | }, |
| 3710 | "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers |
| 3711 | # and parameter servers. |
| 3712 | "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3713 | # and other data needed for training. This path is passed to your TensorFlow |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3714 | # program as the '--job-dir' command-line argument. The benefit of specifying |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3715 | # this field is that Cloud ML validates the path for use in training. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3716 | "pythonVersion": "A String", # Optional. The version of Python used in training. You must either specify |
| 3717 | # this field or specify `masterConfig.imageUri`. |
| 3718 | # |
| 3719 | # The following Python versions are available: |
| 3720 | # |
| 3721 | # * Python '3.7' is available when `runtime_version` is set to '1.15' or |
| 3722 | # later. |
| 3723 | # * Python '3.5' is available when `runtime_version` is set to a version |
| 3724 | # from '1.4' to '1.14'. |
| 3725 | # * Python '2.7' is available when `runtime_version` is set to '1.15' or |
| 3726 | # earlier. |
| 3727 | # |
| 3728 | # Read more about the Python versions available for [each runtime |
| 3729 | # version](/ml-engine/docs/runtime-version-list). |
| 3730 | "network": "A String", # Optional. The full name of the Google Compute Engine |
| 3731 | # [network](/compute/docs/networks-and-firewalls#networks) to which the Job |
| 3732 | # is peered. For example, projects/12345/global/networks/myVPC. Format is of |
| 3733 | # the form projects/{project}/global/networks/{network}. Where {project} is a |
| 3734 | # project number, as in '12345', and {network} is network name.". |
| 3735 | # |
| 3736 | # Private services access must already be configured for the network. If left |
| 3737 | # unspecified, the Job is not peered with any network. Learn more - |
| 3738 | # Connecting Job to user network over private |
| 3739 | # IP. |
| 3740 | "scheduling": { # All parameters related to scheduling of training jobs. # Optional. Scheduling options for a training job. |
| 3741 | "maxWaitTime": "A String", |
| 3742 | "maxRunningTime": "A String", # Optional. The maximum job running time, expressed in seconds. The field can |
| 3743 | # contain up to nine fractional digits, terminated by `s`. If not specified, |
| 3744 | # this field defaults to `604800s` (seven days). |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3745 | # |
| 3746 | # If the training job is still running after this duration, AI Platform |
| 3747 | # Training cancels it. |
| 3748 | # |
| 3749 | # For example, if you want to ensure your job runs for no more than 2 hours, |
| 3750 | # set this field to `7200s` (2 hours * 60 minutes / hour * 60 seconds / |
| 3751 | # minute). |
| 3752 | # |
| 3753 | # If you submit your training job using the `gcloud` tool, you can [provide |
| 3754 | # this field in a `config.yaml` |
| 3755 | # file](/ai-platform/training/docs/training-jobs#formatting_your_configuration_parameters). |
| 3756 | # For example: |
| 3757 | # |
| 3758 | # ```yaml |
| 3759 | # trainingInput: |
| 3760 | # ... |
| 3761 | # scheduling: |
| 3762 | # maxRunningTime: 7200s |
| 3763 | # ... |
| 3764 | # ``` |
| 3765 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3766 | "evaluatorConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for evaluators. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3767 | # |
| 3768 | # You should only set `evaluatorConfig.acceleratorConfig` if |
| 3769 | # `evaluatorType` is set to a Compute Engine machine type. [Learn |
| 3770 | # about restrictions on accelerator configurations for |
| 3771 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 3772 | # |
| 3773 | # Set `evaluatorConfig.imageUri` only if you build a custom image for |
| 3774 | # your evaluator. If `evaluatorConfig.imageUri` has not been |
| 3775 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 3776 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3777 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3778 | # the one used in the custom container. This field is required if the replica |
| 3779 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 3780 | # this field. This must be a [runtime version that currently supports |
| 3781 | # training with |
| 3782 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 3783 | # |
| 3784 | # Note that the version of TensorFlow included in a runtime version may |
| 3785 | # differ from the numbering of the runtime version itself, because it may |
| 3786 | # have a different [patch |
| 3787 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 3788 | # In this field, you must specify the runtime version (TensorFlow minor |
| 3789 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 3790 | # specify `1.x`. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3791 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 3792 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 3793 | # If not provided, the docker image's ENTRYPOINT is used. |
| 3794 | # It cannot be set if custom container image is |
| 3795 | # not provided. |
| 3796 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 3797 | # both cannot be set at the same time. |
| 3798 | "A String", |
| 3799 | ], |
| 3800 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 3801 | # Registry. Learn more about [configuring custom |
| 3802 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 3803 | "containerArgs": [ # Arguments to the entrypoint command. |
| 3804 | # The following rules apply for container_command and container_args: |
| 3805 | # - If you do not supply command or args: |
| 3806 | # The defaults defined in the Docker image are used. |
| 3807 | # - If you supply a command but no args: |
| 3808 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 3809 | # are ignored. Your command is run without any arguments. |
| 3810 | # - If you supply only args: |
| 3811 | # The default Entrypoint defined in the Docker image is run with the args |
| 3812 | # that you supplied. |
| 3813 | # - If you supply a command and args: |
| 3814 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 3815 | # are ignored. Your command is run with your args. |
| 3816 | # It cannot be set if custom container image is |
| 3817 | # not provided. |
| 3818 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 3819 | # both cannot be set at the same time. |
| 3820 | "A String", |
| 3821 | ], |
| 3822 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3823 | # [Learn about restrictions on accelerator configurations for |
| 3824 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 3825 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 3826 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 3827 | # [accelerators for online |
| 3828 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3829 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 3830 | "type": "A String", # The type of accelerator to use. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3831 | }, |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3832 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3833 | "useChiefInTfConfig": True or False, # Optional. Use `chief` instead of `master` in the `TF_CONFIG` environment |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3834 | # variable when training with a custom container. Defaults to `false`. [Learn |
| 3835 | # more about this |
| 3836 | # field.](/ai-platform/training/docs/distributed-training-details#chief-versus-master) |
| 3837 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3838 | # This field has no effect for training jobs that don't use a custom |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3839 | # container. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3840 | "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3841 | # replica in the cluster will be of the type specified in `worker_type`. |
| 3842 | # |
| 3843 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 3844 | # set this value, you must also set `worker_type`. |
| 3845 | # |
| 3846 | # The default value is zero. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3847 | "pythonModule": "A String", # Required. The Python module name to run after installing the packages. |
| 3848 | "args": [ # Optional. Command-line arguments passed to the training application when it |
| 3849 | # starts. If your job uses a custom container, then the arguments are passed |
| 3850 | # to the container's <a class="external" target="_blank" |
| 3851 | # href="https://docs.docker.com/engine/reference/builder/#entrypoint"> |
| 3852 | # `ENTRYPOINT`</a> command. |
| 3853 | "A String", |
| 3854 | ], |
| 3855 | }, |
| 3856 | "state": "A String", # Output only. The detailed state of a job. |
| 3857 | "jobId": "A String", # Required. The user-specified id of the job. |
| 3858 | "endTime": "A String", # Output only. When the job processing was completed. |
| 3859 | "startTime": "A String", # Output only. When the job processing was started. |
| 3860 | "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. |
| 3861 | "errorCount": "A String", # The number of data instances which resulted in errors. |
| 3862 | "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. |
| 3863 | "nodeHours": 3.14, # Node hours used by the batch prediction job. |
| 3864 | "predictionCount": "A String", # The number of generated predictions. |
| 3865 | }, |
| 3866 | "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. |
| 3867 | "trials": [ # Results for individual Hyperparameter trials. |
| 3868 | # Only set for hyperparameter tuning jobs. |
| 3869 | { # Represents the result of a single hyperparameter tuning trial from a |
| 3870 | # training job. The TrainingOutput object that is returned on successful |
| 3871 | # completion of a training job with hyperparameter tuning includes a list |
| 3872 | # of HyperparameterOutput objects, one for each successful trial. |
| 3873 | "allMetrics": [ # All recorded object metrics for this trial. This field is not currently |
| 3874 | # populated. |
| 3875 | { # An observed value of a metric. |
| 3876 | "objectiveValue": 3.14, # The objective value at this training step. |
| 3877 | "trainingStep": "A String", # The global training step for this metric. |
| 3878 | }, |
| 3879 | ], |
| 3880 | "hyperparameters": { # The hyperparameters given to this trial. |
| 3881 | "a_key": "A String", |
| 3882 | }, |
| 3883 | "trialId": "A String", # The trial id for these results. |
| 3884 | "endTime": "A String", # Output only. End time for the trial. |
| 3885 | "isTrialStoppedEarly": True or False, # True if the trial is stopped early. |
| 3886 | "startTime": "A String", # Output only. Start time for the trial. |
| 3887 | "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. |
| 3888 | "objectiveValue": 3.14, # The objective value at this training step. |
| 3889 | "trainingStep": "A String", # The global training step for this metric. |
| 3890 | }, |
| 3891 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 3892 | # Only set for trials of built-in algorithms jobs that have succeeded. |
| 3893 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 3894 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 3895 | # saves the trained model. Only set for successful jobs that don't use |
| 3896 | # hyperparameter tuning. |
| 3897 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 3898 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 3899 | # trained. |
| 3900 | }, |
| 3901 | "state": "A String", # Output only. The detailed state of the trial. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3902 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3903 | ], |
| 3904 | "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning |
| 3905 | # trials. See |
| 3906 | # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) |
| 3907 | # for more information. Only set for hyperparameter tuning jobs. |
| 3908 | "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. |
| 3909 | # Only set for hyperparameter tuning jobs. |
| 3910 | "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. |
| 3911 | "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. |
| 3912 | "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. |
| 3913 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 3914 | # Only set for built-in algorithms jobs. |
| 3915 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 3916 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 3917 | # saves the trained model. Only set for successful jobs that don't use |
| 3918 | # hyperparameter tuning. |
| 3919 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 3920 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 3921 | # trained. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 3922 | }, |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3923 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3924 | "createTime": "A String", # Output only. When the job was created. |
| 3925 | "labels": { # Optional. One or more labels that you can add, to organize your jobs. |
| 3926 | # Each label is a key-value pair, where both the key and the value are |
| 3927 | # arbitrary strings that you supply. |
| 3928 | # For more information, see the documentation on |
| 3929 | # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. |
| 3930 | "a_key": "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 3931 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 3932 | "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. |
| 3933 | "outputPath": "A String", # Required. The output Google Cloud Storage location. |
| 3934 | "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. |
| 3935 | "dataFormat": "A String", # Required. The format of the input data files. |
| 3936 | "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. |
| 3937 | # The service will buffer batch_size number of records in memory before |
| 3938 | # invoking one Tensorflow prediction call internally. So take the record |
| 3939 | # size and memory available into consideration when setting this parameter. |
| 3940 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch |
| 3941 | # prediction. If not set, AI Platform will pick the runtime version used |
| 3942 | # during the CreateVersion request for this model version, or choose the |
| 3943 | # latest stable version when model version information is not available |
| 3944 | # such as when the model is specified by uri. |
| 3945 | "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain |
| 3946 | # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. |
| 3947 | "A String", |
| 3948 | ], |
| 3949 | "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. |
| 3950 | # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> |
| 3951 | # for AI Platform services. |
| 3952 | "versionName": "A String", # Use this field if you want to specify a version of the model to use. The |
| 3953 | # string is formatted the same way as `model_version`, with the addition |
| 3954 | # of the version information: |
| 3955 | # |
| 3956 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` |
| 3957 | "modelName": "A String", # Use this field if you want to use the default version for the specified |
| 3958 | # model. The string must use the following format: |
| 3959 | # |
| 3960 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` |
| 3961 | "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for |
| 3962 | # the model to use. |
| 3963 | "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. |
| 3964 | # Defaults to 10 if not specified. |
| 3965 | "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for |
| 3966 | # this job. Please refer to |
| 3967 | # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) |
| 3968 | # for information about how to use signatures. |
| 3969 | # |
| 3970 | # Defaults to |
| 3971 | # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) |
| 3972 | # , which is "serving_default". |
| 3973 | }, |
| 3974 | "errorMessage": "A String", # Output only. The details of a failure or a cancellation. |
| 3975 | } |
| 3976 | |
| 3977 | updateMask: string, Required. Specifies the path, relative to `Job`, of the field to update. |
| 3978 | To adopt etag mechanism, include `etag` field in the mask, and include the |
| 3979 | `etag` value in your job resource. |
| 3980 | |
| 3981 | For example, to change the labels of a job, the `update_mask` parameter |
| 3982 | would be specified as `labels`, `etag`, and the |
| 3983 | `PATCH` request body would specify the new value, as follows: |
| 3984 | { |
| 3985 | "labels": { |
| 3986 | "owner": "Google", |
| 3987 | "color": "Blue" |
| 3988 | } |
| 3989 | "etag": "33a64df551425fcc55e4d42a148795d9f25f89d4" |
| 3990 | } |
| 3991 | If `etag` matches the one on the server, the labels of the job will be |
| 3992 | replaced with the given ones, and the server end `etag` will be |
| 3993 | recalculated. |
| 3994 | |
| 3995 | Currently the only supported update masks are `labels` and `etag`. |
| 3996 | x__xgafv: string, V1 error format. |
| 3997 | Allowed values |
| 3998 | 1 - v1 error format |
| 3999 | 2 - v2 error format |
| 4000 | |
| 4001 | Returns: |
| 4002 | An object of the form: |
| 4003 | |
| 4004 | { # Represents a training or prediction job. |
| 4005 | "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help |
| 4006 | # prevent simultaneous updates of a job from overwriting each other. |
| 4007 | # It is strongly suggested that systems make use of the `etag` in the |
| 4008 | # read-modify-write cycle to perform job updates in order to avoid race |
| 4009 | # conditions: An `etag` is returned in the response to `GetJob`, and |
| 4010 | # systems are expected to put that etag in the request to `UpdateJob` to |
| 4011 | # ensure that their change will be applied to the same version of the job. |
| 4012 | "trainingInput": { # Represents input parameters for a training job. When using the gcloud command # Input parameters to create a training job. |
| 4013 | # to submit your training job, you can specify the input parameters as |
| 4014 | # command-line arguments and/or in a YAML configuration file referenced from |
| 4015 | # the --config command-line argument. For details, see the guide to [submitting |
| 4016 | # a training job](/ai-platform/training/docs/training-jobs). |
| 4017 | "parameterServerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for parameter servers. |
| 4018 | # |
| 4019 | # You should only set `parameterServerConfig.acceleratorConfig` if |
| 4020 | # `parameterServerType` is set to a Compute Engine machine type. [Learn |
| 4021 | # about restrictions on accelerator configurations for |
| 4022 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 4023 | # |
| 4024 | # Set `parameterServerConfig.imageUri` only if you build a custom image for |
| 4025 | # your parameter server. If `parameterServerConfig.imageUri` has not been |
| 4026 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 4027 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 4028 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 4029 | # the one used in the custom container. This field is required if the replica |
| 4030 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 4031 | # this field. This must be a [runtime version that currently supports |
| 4032 | # training with |
| 4033 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 4034 | # |
| 4035 | # Note that the version of TensorFlow included in a runtime version may |
| 4036 | # differ from the numbering of the runtime version itself, because it may |
| 4037 | # have a different [patch |
| 4038 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 4039 | # In this field, you must specify the runtime version (TensorFlow minor |
| 4040 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 4041 | # specify `1.x`. |
| 4042 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 4043 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 4044 | # If not provided, the docker image's ENTRYPOINT is used. |
| 4045 | # It cannot be set if custom container image is |
| 4046 | # not provided. |
| 4047 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 4048 | # both cannot be set at the same time. |
| 4049 | "A String", |
| 4050 | ], |
| 4051 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 4052 | # Registry. Learn more about [configuring custom |
| 4053 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 4054 | "containerArgs": [ # Arguments to the entrypoint command. |
| 4055 | # The following rules apply for container_command and container_args: |
| 4056 | # - If you do not supply command or args: |
| 4057 | # The defaults defined in the Docker image are used. |
| 4058 | # - If you supply a command but no args: |
| 4059 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 4060 | # are ignored. Your command is run without any arguments. |
| 4061 | # - If you supply only args: |
| 4062 | # The default Entrypoint defined in the Docker image is run with the args |
| 4063 | # that you supplied. |
| 4064 | # - If you supply a command and args: |
| 4065 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 4066 | # are ignored. Your command is run with your args. |
| 4067 | # It cannot be set if custom container image is |
| 4068 | # not provided. |
| 4069 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 4070 | # both cannot be set at the same time. |
| 4071 | "A String", |
| 4072 | ], |
| 4073 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 4074 | # [Learn about restrictions on accelerator configurations for |
| 4075 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 4076 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 4077 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 4078 | # [accelerators for online |
| 4079 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 4080 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 4081 | "type": "A String", # The type of accelerator to use. |
| 4082 | }, |
| 4083 | }, |
| 4084 | "encryptionConfig": { # Represents a custom encryption key configuration that can be applied to # Optional. Options for using customer-managed encryption keys (CMEK) to |
| 4085 | # protect resources created by a training job, instead of using Google's |
| 4086 | # default encryption. If this is set, then all resources created by the |
| 4087 | # training job will be encrypted with the customer-managed encryption key |
| 4088 | # that you specify. |
| 4089 | # |
| 4090 | # [Learn how and when to use CMEK with AI Platform |
| 4091 | # Training](/ai-platform/training/docs/cmek). |
| 4092 | # a resource. |
| 4093 | "kmsKeyName": "A String", # The Cloud KMS resource identifier of the customer-managed encryption key |
| 4094 | # used to protect a resource, such as a training job. It has the following |
| 4095 | # format: |
| 4096 | # `projects/{PROJECT_ID}/locations/{REGION}/keyRings/{KEY_RING_NAME}/cryptoKeys/{KEY_NAME}` |
| 4097 | }, |
| 4098 | "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune. |
| 4099 | "hyperparameterMetricTag": "A String", # Optional. The TensorFlow summary tag name to use for optimizing trials. For |
| 4100 | # current versions of TensorFlow, this tag name should exactly match what is |
| 4101 | # shown in TensorBoard, including all scopes. For versions of TensorFlow |
| 4102 | # prior to 0.12, this should be only the tag passed to tf.Summary. |
| 4103 | # By default, "training/hptuning/metric" will be used. |
| 4104 | "params": [ # Required. The set of parameters to tune. |
| 4105 | { # Represents a single hyperparameter to optimize. |
| 4106 | "type": "A String", # Required. The type of the parameter. |
| 4107 | "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories. |
| 4108 | "A String", |
| 4109 | ], |
| 4110 | "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in |
| 4111 | # a HyperparameterSpec message. E.g., "learning_rate". |
| 4112 | "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 4113 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 4114 | # type is INTEGER. |
| 4115 | "discreteValues": [ # Required if type is `DISCRETE`. |
| 4116 | # A list of feasible points. |
| 4117 | # The list should be in strictly increasing order. For instance, this |
| 4118 | # parameter might have possible settings of 1.5, 2.5, and 4.0. This list |
| 4119 | # should not contain more than 1,000 values. |
| 4120 | 3.14, |
| 4121 | ], |
| 4122 | "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube. |
| 4123 | # Leave unset for categorical parameters. |
| 4124 | # Some kind of scaling is strongly recommended for real or integral |
| 4125 | # parameters (e.g., `UNIT_LINEAR_SCALE`). |
| 4126 | "maxValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field |
| 4127 | # should be unset if type is `CATEGORICAL`. This value should be integers if |
| 4128 | # type is `INTEGER`. |
| 4129 | }, |
| 4130 | ], |
| 4131 | "enableTrialEarlyStopping": True or False, # Optional. Indicates if the hyperparameter tuning job enables auto trial |
| 4132 | # early stopping. |
| 4133 | "resumePreviousJobId": "A String", # Optional. The prior hyperparameter tuning job id that users hope to |
| 4134 | # continue with. The job id will be used to find the corresponding vizier |
| 4135 | # study guid and resume the study. |
| 4136 | "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently. |
| 4137 | # You can reduce the time it takes to perform hyperparameter tuning by adding |
| 4138 | # trials in parallel. However, each trail only benefits from the information |
| 4139 | # gained in completed trials. That means that a trial does not get access to |
| 4140 | # the results of trials running at the same time, which could reduce the |
| 4141 | # quality of the overall optimization. |
| 4142 | # |
| 4143 | # Each trial will use the same scale tier and machine types. |
| 4144 | # |
| 4145 | # Defaults to one. |
| 4146 | "maxFailedTrials": 42, # Optional. The number of failed trials that need to be seen before failing |
| 4147 | # the hyperparameter tuning job. You can specify this field to override the |
| 4148 | # default failing criteria for AI Platform hyperparameter tuning jobs. |
| 4149 | # |
| 4150 | # Defaults to zero, which means the service decides when a hyperparameter |
| 4151 | # job should fail. |
| 4152 | "goal": "A String", # Required. The type of goal to use for tuning. Available types are |
| 4153 | # `MAXIMIZE` and `MINIMIZE`. |
| 4154 | # |
| 4155 | # Defaults to `MAXIMIZE`. |
| 4156 | "maxTrials": 42, # Optional. How many training trials should be attempted to optimize |
| 4157 | # the specified hyperparameters. |
| 4158 | # |
| 4159 | # Defaults to one. |
| 4160 | "algorithm": "A String", # Optional. The search algorithm specified for the hyperparameter |
| 4161 | # tuning job. |
| 4162 | # Uses the default AI Platform hyperparameter tuning |
| 4163 | # algorithm if unspecified. |
| 4164 | }, |
| 4165 | "workerConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for workers. |
| 4166 | # |
| 4167 | # You should only set `workerConfig.acceleratorConfig` if `workerType` is set |
| 4168 | # to a Compute Engine machine type. [Learn about restrictions on accelerator |
| 4169 | # configurations for |
| 4170 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 4171 | # |
| 4172 | # Set `workerConfig.imageUri` only if you build a custom image for your |
| 4173 | # worker. If `workerConfig.imageUri` has not been set, AI Platform uses |
| 4174 | # the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 4175 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 4176 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 4177 | # the one used in the custom container. This field is required if the replica |
| 4178 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 4179 | # this field. This must be a [runtime version that currently supports |
| 4180 | # training with |
| 4181 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 4182 | # |
| 4183 | # Note that the version of TensorFlow included in a runtime version may |
| 4184 | # differ from the numbering of the runtime version itself, because it may |
| 4185 | # have a different [patch |
| 4186 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 4187 | # In this field, you must specify the runtime version (TensorFlow minor |
| 4188 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 4189 | # specify `1.x`. |
| 4190 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 4191 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 4192 | # If not provided, the docker image's ENTRYPOINT is used. |
| 4193 | # It cannot be set if custom container image is |
| 4194 | # not provided. |
| 4195 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 4196 | # both cannot be set at the same time. |
| 4197 | "A String", |
| 4198 | ], |
| 4199 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 4200 | # Registry. Learn more about [configuring custom |
| 4201 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 4202 | "containerArgs": [ # Arguments to the entrypoint command. |
| 4203 | # The following rules apply for container_command and container_args: |
| 4204 | # - If you do not supply command or args: |
| 4205 | # The defaults defined in the Docker image are used. |
| 4206 | # - If you supply a command but no args: |
| 4207 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 4208 | # are ignored. Your command is run without any arguments. |
| 4209 | # - If you supply only args: |
| 4210 | # The default Entrypoint defined in the Docker image is run with the args |
| 4211 | # that you supplied. |
| 4212 | # - If you supply a command and args: |
| 4213 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 4214 | # are ignored. Your command is run with your args. |
| 4215 | # It cannot be set if custom container image is |
| 4216 | # not provided. |
| 4217 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 4218 | # both cannot be set at the same time. |
| 4219 | "A String", |
| 4220 | ], |
| 4221 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 4222 | # [Learn about restrictions on accelerator configurations for |
| 4223 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 4224 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 4225 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 4226 | # [accelerators for online |
| 4227 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 4228 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 4229 | "type": "A String", # The type of accelerator to use. |
| 4230 | }, |
| 4231 | }, |
| 4232 | "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training |
| 4233 | # job. Each replica in the cluster will be of the type specified in |
| 4234 | # `parameter_server_type`. |
| 4235 | # |
| 4236 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 4237 | # set this value, you must also set `parameter_server_type`. |
| 4238 | # |
| 4239 | # The default value is zero. |
| 4240 | "packageUris": [ # Required. The Google Cloud Storage location of the packages with |
| 4241 | # the training program and any additional dependencies. |
| 4242 | # The maximum number of package URIs is 100. |
| 4243 | "A String", |
| 4244 | ], |
| 4245 | "evaluatorCount": "A String", # Optional. The number of evaluator replicas to use for the training job. |
| 4246 | # Each replica in the cluster will be of the type specified in |
| 4247 | # `evaluator_type`. |
| 4248 | # |
| 4249 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 4250 | # set this value, you must also set `evaluator_type`. |
| 4251 | # |
| 4252 | # The default value is zero. |
| 4253 | "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 4254 | # job's master worker. You must specify this field when `scaleTier` is set to |
| 4255 | # `CUSTOM`. |
| 4256 | # |
| 4257 | # You can use certain Compute Engine machine types directly in this field. |
| 4258 | # The following types are supported: |
| 4259 | # |
| 4260 | # - `n1-standard-4` |
| 4261 | # - `n1-standard-8` |
| 4262 | # - `n1-standard-16` |
| 4263 | # - `n1-standard-32` |
| 4264 | # - `n1-standard-64` |
| 4265 | # - `n1-standard-96` |
| 4266 | # - `n1-highmem-2` |
| 4267 | # - `n1-highmem-4` |
| 4268 | # - `n1-highmem-8` |
| 4269 | # - `n1-highmem-16` |
| 4270 | # - `n1-highmem-32` |
| 4271 | # - `n1-highmem-64` |
| 4272 | # - `n1-highmem-96` |
| 4273 | # - `n1-highcpu-16` |
| 4274 | # - `n1-highcpu-32` |
| 4275 | # - `n1-highcpu-64` |
| 4276 | # - `n1-highcpu-96` |
| 4277 | # |
| 4278 | # Learn more about [using Compute Engine machine |
| 4279 | # types](/ml-engine/docs/machine-types#compute-engine-machine-types). |
| 4280 | # |
| 4281 | # Alternatively, you can use the following legacy machine types: |
| 4282 | # |
| 4283 | # - `standard` |
| 4284 | # - `large_model` |
| 4285 | # - `complex_model_s` |
| 4286 | # - `complex_model_m` |
| 4287 | # - `complex_model_l` |
| 4288 | # - `standard_gpu` |
| 4289 | # - `complex_model_m_gpu` |
| 4290 | # - `complex_model_l_gpu` |
| 4291 | # - `standard_p100` |
| 4292 | # - `complex_model_m_p100` |
| 4293 | # - `standard_v100` |
| 4294 | # - `large_model_v100` |
| 4295 | # - `complex_model_m_v100` |
| 4296 | # - `complex_model_l_v100` |
| 4297 | # |
| 4298 | # Learn more about [using legacy machine |
| 4299 | # types](/ml-engine/docs/machine-types#legacy-machine-types). |
| 4300 | # |
| 4301 | # Finally, if you want to use a TPU for training, specify `cloud_tpu` in this |
| 4302 | # field. Learn more about the [special configuration options for training |
| 4303 | # with |
| 4304 | # TPUs](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| 4305 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for training. You must |
| 4306 | # either specify this field or specify `masterConfig.imageUri`. |
| 4307 | # |
| 4308 | # For more information, see the [runtime version |
| 4309 | # list](/ai-platform/training/docs/runtime-version-list) and learn [how to |
| 4310 | # manage runtime versions](/ai-platform/training/docs/versioning). |
| 4311 | "evaluatorType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 4312 | # job's evaluator nodes. |
| 4313 | # |
| 4314 | # The supported values are the same as those described in the entry for |
| 4315 | # `masterType`. |
| 4316 | # |
| 4317 | # This value must be consistent with the category of machine type that |
| 4318 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 4319 | # types or both must be legacy machine types. |
| 4320 | # |
| 4321 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 4322 | # `evaluatorCount` is greater than zero. |
| 4323 | "region": "A String", # Required. The region to run the training job in. See the [available |
| 4324 | # regions](/ai-platform/training/docs/regions) for AI Platform Training. |
| 4325 | "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 4326 | # job's worker nodes. |
| 4327 | # |
| 4328 | # The supported values are the same as those described in the entry for |
| 4329 | # `masterType`. |
| 4330 | # |
| 4331 | # This value must be consistent with the category of machine type that |
| 4332 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 4333 | # types or both must be legacy machine types. |
| 4334 | # |
| 4335 | # If you use `cloud_tpu` for this value, see special instructions for |
| 4336 | # [configuring a custom TPU |
| 4337 | # machine](/ml-engine/docs/tensorflow/using-tpus#configuring_a_custom_tpu_machine). |
| 4338 | # |
| 4339 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 4340 | # `workerCount` is greater than zero. |
| 4341 | "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training |
| 4342 | # job's parameter server. |
| 4343 | # |
| 4344 | # The supported values are the same as those described in the entry for |
| 4345 | # `master_type`. |
| 4346 | # |
| 4347 | # This value must be consistent with the category of machine type that |
| 4348 | # `masterType` uses. In other words, both must be Compute Engine machine |
| 4349 | # types or both must be legacy machine types. |
| 4350 | # |
| 4351 | # This value must be present when `scaleTier` is set to `CUSTOM` and |
| 4352 | # `parameter_server_count` is greater than zero. |
| 4353 | "masterConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for your master worker. |
| 4354 | # |
| 4355 | # You should only set `masterConfig.acceleratorConfig` if `masterType` is set |
| 4356 | # to a Compute Engine machine type. Learn about [restrictions on accelerator |
| 4357 | # configurations for |
| 4358 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 4359 | # |
| 4360 | # Set `masterConfig.imageUri` only if you build a custom image. Only one of |
| 4361 | # `masterConfig.imageUri` and `runtimeVersion` should be set. Learn more |
| 4362 | # about [configuring custom |
| 4363 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 4364 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 4365 | # the one used in the custom container. This field is required if the replica |
| 4366 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 4367 | # this field. This must be a [runtime version that currently supports |
| 4368 | # training with |
| 4369 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 4370 | # |
| 4371 | # Note that the version of TensorFlow included in a runtime version may |
| 4372 | # differ from the numbering of the runtime version itself, because it may |
| 4373 | # have a different [patch |
| 4374 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 4375 | # In this field, you must specify the runtime version (TensorFlow minor |
| 4376 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 4377 | # specify `1.x`. |
| 4378 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 4379 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 4380 | # If not provided, the docker image's ENTRYPOINT is used. |
| 4381 | # It cannot be set if custom container image is |
| 4382 | # not provided. |
| 4383 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 4384 | # both cannot be set at the same time. |
| 4385 | "A String", |
| 4386 | ], |
| 4387 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 4388 | # Registry. Learn more about [configuring custom |
| 4389 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 4390 | "containerArgs": [ # Arguments to the entrypoint command. |
| 4391 | # The following rules apply for container_command and container_args: |
| 4392 | # - If you do not supply command or args: |
| 4393 | # The defaults defined in the Docker image are used. |
| 4394 | # - If you supply a command but no args: |
| 4395 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 4396 | # are ignored. Your command is run without any arguments. |
| 4397 | # - If you supply only args: |
| 4398 | # The default Entrypoint defined in the Docker image is run with the args |
| 4399 | # that you supplied. |
| 4400 | # - If you supply a command and args: |
| 4401 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 4402 | # are ignored. Your command is run with your args. |
| 4403 | # It cannot be set if custom container image is |
| 4404 | # not provided. |
| 4405 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 4406 | # both cannot be set at the same time. |
| 4407 | "A String", |
| 4408 | ], |
| 4409 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 4410 | # [Learn about restrictions on accelerator configurations for |
| 4411 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 4412 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 4413 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 4414 | # [accelerators for online |
| 4415 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 4416 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 4417 | "type": "A String", # The type of accelerator to use. |
| 4418 | }, |
| 4419 | }, |
| 4420 | "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers |
| 4421 | # and parameter servers. |
| 4422 | "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs |
| 4423 | # and other data needed for training. This path is passed to your TensorFlow |
| 4424 | # program as the '--job-dir' command-line argument. The benefit of specifying |
| 4425 | # this field is that Cloud ML validates the path for use in training. |
| 4426 | "pythonVersion": "A String", # Optional. The version of Python used in training. You must either specify |
| 4427 | # this field or specify `masterConfig.imageUri`. |
| 4428 | # |
| 4429 | # The following Python versions are available: |
| 4430 | # |
| 4431 | # * Python '3.7' is available when `runtime_version` is set to '1.15' or |
| 4432 | # later. |
| 4433 | # * Python '3.5' is available when `runtime_version` is set to a version |
| 4434 | # from '1.4' to '1.14'. |
| 4435 | # * Python '2.7' is available when `runtime_version` is set to '1.15' or |
| 4436 | # earlier. |
| 4437 | # |
| 4438 | # Read more about the Python versions available for [each runtime |
| 4439 | # version](/ml-engine/docs/runtime-version-list). |
| 4440 | "network": "A String", # Optional. The full name of the Google Compute Engine |
| 4441 | # [network](/compute/docs/networks-and-firewalls#networks) to which the Job |
| 4442 | # is peered. For example, projects/12345/global/networks/myVPC. Format is of |
| 4443 | # the form projects/{project}/global/networks/{network}. Where {project} is a |
| 4444 | # project number, as in '12345', and {network} is network name.". |
| 4445 | # |
| 4446 | # Private services access must already be configured for the network. If left |
| 4447 | # unspecified, the Job is not peered with any network. Learn more - |
| 4448 | # Connecting Job to user network over private |
| 4449 | # IP. |
| 4450 | "scheduling": { # All parameters related to scheduling of training jobs. # Optional. Scheduling options for a training job. |
| 4451 | "maxWaitTime": "A String", |
| 4452 | "maxRunningTime": "A String", # Optional. The maximum job running time, expressed in seconds. The field can |
| 4453 | # contain up to nine fractional digits, terminated by `s`. If not specified, |
| 4454 | # this field defaults to `604800s` (seven days). |
| 4455 | # |
| 4456 | # If the training job is still running after this duration, AI Platform |
| 4457 | # Training cancels it. |
| 4458 | # |
| 4459 | # For example, if you want to ensure your job runs for no more than 2 hours, |
| 4460 | # set this field to `7200s` (2 hours * 60 minutes / hour * 60 seconds / |
| 4461 | # minute). |
| 4462 | # |
| 4463 | # If you submit your training job using the `gcloud` tool, you can [provide |
| 4464 | # this field in a `config.yaml` |
| 4465 | # file](/ai-platform/training/docs/training-jobs#formatting_your_configuration_parameters). |
| 4466 | # For example: |
| 4467 | # |
| 4468 | # ```yaml |
| 4469 | # trainingInput: |
| 4470 | # ... |
| 4471 | # scheduling: |
| 4472 | # maxRunningTime: 7200s |
| 4473 | # ... |
| 4474 | # ``` |
| 4475 | }, |
| 4476 | "evaluatorConfig": { # Represents the configuration for a replica in a cluster. # Optional. The configuration for evaluators. |
| 4477 | # |
| 4478 | # You should only set `evaluatorConfig.acceleratorConfig` if |
| 4479 | # `evaluatorType` is set to a Compute Engine machine type. [Learn |
| 4480 | # about restrictions on accelerator configurations for |
| 4481 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 4482 | # |
| 4483 | # Set `evaluatorConfig.imageUri` only if you build a custom image for |
| 4484 | # your evaluator. If `evaluatorConfig.imageUri` has not been |
| 4485 | # set, AI Platform uses the value of `masterConfig.imageUri`. Learn more about [configuring custom |
| 4486 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 4487 | "tpuTfVersion": "A String", # The AI Platform runtime version that includes a TensorFlow version matching |
| 4488 | # the one used in the custom container. This field is required if the replica |
| 4489 | # is a TPU worker that uses a custom container. Otherwise, do not specify |
| 4490 | # this field. This must be a [runtime version that currently supports |
| 4491 | # training with |
| 4492 | # TPUs](/ml-engine/docs/tensorflow/runtime-version-list#tpu-support). |
| 4493 | # |
| 4494 | # Note that the version of TensorFlow included in a runtime version may |
| 4495 | # differ from the numbering of the runtime version itself, because it may |
| 4496 | # have a different [patch |
| 4497 | # version](https://www.tensorflow.org/guide/version_compat#semantic_versioning_20). |
| 4498 | # In this field, you must specify the runtime version (TensorFlow minor |
| 4499 | # version). For example, if your custom container runs TensorFlow `1.x.y`, |
| 4500 | # specify `1.x`. |
| 4501 | "containerCommand": [ # The command with which the replica's custom container is run. |
| 4502 | # If provided, it will override default ENTRYPOINT of the docker image. |
| 4503 | # If not provided, the docker image's ENTRYPOINT is used. |
| 4504 | # It cannot be set if custom container image is |
| 4505 | # not provided. |
| 4506 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 4507 | # both cannot be set at the same time. |
| 4508 | "A String", |
| 4509 | ], |
| 4510 | "imageUri": "A String", # The Docker image to run on the replica. This image must be in Container |
| 4511 | # Registry. Learn more about [configuring custom |
| 4512 | # containers](/ai-platform/training/docs/distributed-training-containers). |
| 4513 | "containerArgs": [ # Arguments to the entrypoint command. |
| 4514 | # The following rules apply for container_command and container_args: |
| 4515 | # - If you do not supply command or args: |
| 4516 | # The defaults defined in the Docker image are used. |
| 4517 | # - If you supply a command but no args: |
| 4518 | # The default EntryPoint and the default Cmd defined in the Docker image |
| 4519 | # are ignored. Your command is run without any arguments. |
| 4520 | # - If you supply only args: |
| 4521 | # The default Entrypoint defined in the Docker image is run with the args |
| 4522 | # that you supplied. |
| 4523 | # - If you supply a command and args: |
| 4524 | # The default Entrypoint and the default Cmd defined in the Docker image |
| 4525 | # are ignored. Your command is run with your args. |
| 4526 | # It cannot be set if custom container image is |
| 4527 | # not provided. |
| 4528 | # Note that this field and [TrainingInput.args] are mutually exclusive, i.e., |
| 4529 | # both cannot be set at the same time. |
| 4530 | "A String", |
| 4531 | ], |
| 4532 | "acceleratorConfig": { # Represents a hardware accelerator request config. # Represents the type and number of accelerators used by the replica. |
| 4533 | # [Learn about restrictions on accelerator configurations for |
| 4534 | # training.](/ai-platform/training/docs/using-gpus#compute-engine-machine-types-with-gpu) |
| 4535 | # Note that the AcceleratorConfig can be used in both Jobs and Versions. |
| 4536 | # Learn more about [accelerators for training](/ml-engine/docs/using-gpus) and |
| 4537 | # [accelerators for online |
| 4538 | # prediction](/ml-engine/docs/machine-types-online-prediction#gpus). |
| 4539 | "count": "A String", # The number of accelerators to attach to each machine running the job. |
| 4540 | "type": "A String", # The type of accelerator to use. |
| 4541 | }, |
| 4542 | }, |
| 4543 | "useChiefInTfConfig": True or False, # Optional. Use `chief` instead of `master` in the `TF_CONFIG` environment |
| 4544 | # variable when training with a custom container. Defaults to `false`. [Learn |
| 4545 | # more about this |
| 4546 | # field.](/ai-platform/training/docs/distributed-training-details#chief-versus-master) |
| 4547 | # |
| 4548 | # This field has no effect for training jobs that don't use a custom |
| 4549 | # container. |
| 4550 | "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each |
| 4551 | # replica in the cluster will be of the type specified in `worker_type`. |
| 4552 | # |
| 4553 | # This value can only be used when `scale_tier` is set to `CUSTOM`. If you |
| 4554 | # set this value, you must also set `worker_type`. |
| 4555 | # |
| 4556 | # The default value is zero. |
| 4557 | "pythonModule": "A String", # Required. The Python module name to run after installing the packages. |
| 4558 | "args": [ # Optional. Command-line arguments passed to the training application when it |
| 4559 | # starts. If your job uses a custom container, then the arguments are passed |
| 4560 | # to the container's <a class="external" target="_blank" |
| 4561 | # href="https://docs.docker.com/engine/reference/builder/#entrypoint"> |
| 4562 | # `ENTRYPOINT`</a> command. |
| 4563 | "A String", |
| 4564 | ], |
| 4565 | }, |
| 4566 | "state": "A String", # Output only. The detailed state of a job. |
| 4567 | "jobId": "A String", # Required. The user-specified id of the job. |
| 4568 | "endTime": "A String", # Output only. When the job processing was completed. |
| 4569 | "startTime": "A String", # Output only. When the job processing was started. |
| 4570 | "predictionOutput": { # Represents results of a prediction job. # The current prediction job result. |
| 4571 | "errorCount": "A String", # The number of data instances which resulted in errors. |
| 4572 | "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time. |
| 4573 | "nodeHours": 3.14, # Node hours used by the batch prediction job. |
| 4574 | "predictionCount": "A String", # The number of generated predictions. |
| 4575 | }, |
| 4576 | "trainingOutput": { # Represents results of a training job. Output only. # The current training job result. |
| 4577 | "trials": [ # Results for individual Hyperparameter trials. |
| 4578 | # Only set for hyperparameter tuning jobs. |
| 4579 | { # Represents the result of a single hyperparameter tuning trial from a |
| 4580 | # training job. The TrainingOutput object that is returned on successful |
| 4581 | # completion of a training job with hyperparameter tuning includes a list |
| 4582 | # of HyperparameterOutput objects, one for each successful trial. |
| 4583 | "allMetrics": [ # All recorded object metrics for this trial. This field is not currently |
| 4584 | # populated. |
| 4585 | { # An observed value of a metric. |
| 4586 | "objectiveValue": 3.14, # The objective value at this training step. |
| 4587 | "trainingStep": "A String", # The global training step for this metric. |
| 4588 | }, |
| 4589 | ], |
| 4590 | "hyperparameters": { # The hyperparameters given to this trial. |
| 4591 | "a_key": "A String", |
| 4592 | }, |
| 4593 | "trialId": "A String", # The trial id for these results. |
| 4594 | "endTime": "A String", # Output only. End time for the trial. |
| 4595 | "isTrialStoppedEarly": True or False, # True if the trial is stopped early. |
| 4596 | "startTime": "A String", # Output only. Start time for the trial. |
| 4597 | "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial. |
| 4598 | "objectiveValue": 3.14, # The objective value at this training step. |
| 4599 | "trainingStep": "A String", # The global training step for this metric. |
| 4600 | }, |
| 4601 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 4602 | # Only set for trials of built-in algorithms jobs that have succeeded. |
| 4603 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 4604 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 4605 | # saves the trained model. Only set for successful jobs that don't use |
| 4606 | # hyperparameter tuning. |
| 4607 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 4608 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 4609 | # trained. |
| 4610 | }, |
| 4611 | "state": "A String", # Output only. The detailed state of the trial. |
| 4612 | }, |
| 4613 | ], |
| 4614 | "hyperparameterMetricTag": "A String", # The TensorFlow summary tag name used for optimizing hyperparameter tuning |
| 4615 | # trials. See |
| 4616 | # [`HyperparameterSpec.hyperparameterMetricTag`](#HyperparameterSpec.FIELDS.hyperparameter_metric_tag) |
| 4617 | # for more information. Only set for hyperparameter tuning jobs. |
| 4618 | "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully. |
| 4619 | # Only set for hyperparameter tuning jobs. |
| 4620 | "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job. |
| 4621 | "consumedMLUnits": 3.14, # The amount of ML units consumed by the job. |
| 4622 | "isBuiltInAlgorithmJob": True or False, # Whether this job is a built-in Algorithm job. |
| 4623 | "builtInAlgorithmOutput": { # Represents output related to a built-in algorithm Job. # Details related to built-in algorithms jobs. |
| 4624 | # Only set for built-in algorithms jobs. |
| 4625 | "framework": "A String", # Framework on which the built-in algorithm was trained. |
| 4626 | "modelPath": "A String", # The Cloud Storage path to the `model/` directory where the training job |
| 4627 | # saves the trained model. Only set for successful jobs that don't use |
| 4628 | # hyperparameter tuning. |
| 4629 | "pythonVersion": "A String", # Python version on which the built-in algorithm was trained. |
| 4630 | "runtimeVersion": "A String", # AI Platform runtime version on which the built-in algorithm was |
| 4631 | # trained. |
| 4632 | }, |
| 4633 | }, |
| 4634 | "createTime": "A String", # Output only. When the job was created. |
| 4635 | "labels": { # Optional. One or more labels that you can add, to organize your jobs. |
| 4636 | # Each label is a key-value pair, where both the key and the value are |
| 4637 | # arbitrary strings that you supply. |
| 4638 | # For more information, see the documentation on |
| 4639 | # <a href="/ml-engine/docs/tensorflow/resource-labels">using labels</a>. |
| 4640 | "a_key": "A String", |
| 4641 | }, |
| 4642 | "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job. |
| 4643 | "outputPath": "A String", # Required. The output Google Cloud Storage location. |
| 4644 | "outputDataFormat": "A String", # Optional. Format of the output data files, defaults to JSON. |
| 4645 | "dataFormat": "A String", # Required. The format of the input data files. |
| 4646 | "batchSize": "A String", # Optional. Number of records per batch, defaults to 64. |
| 4647 | # The service will buffer batch_size number of records in memory before |
| 4648 | # invoking one Tensorflow prediction call internally. So take the record |
| 4649 | # size and memory available into consideration when setting this parameter. |
| 4650 | "runtimeVersion": "A String", # Optional. The AI Platform runtime version to use for this batch |
| 4651 | # prediction. If not set, AI Platform will pick the runtime version used |
| 4652 | # during the CreateVersion request for this model version, or choose the |
| 4653 | # latest stable version when model version information is not available |
| 4654 | # such as when the model is specified by uri. |
| 4655 | "inputPaths": [ # Required. The Cloud Storage location of the input data files. May contain |
| 4656 | # <a href="/storage/docs/gsutil/addlhelp/WildcardNames">wildcards</a>. |
| 4657 | "A String", |
| 4658 | ], |
| 4659 | "region": "A String", # Required. The Google Compute Engine region to run the prediction job in. |
| 4660 | # See the <a href="/ml-engine/docs/tensorflow/regions">available regions</a> |
| 4661 | # for AI Platform services. |
| 4662 | "versionName": "A String", # Use this field if you want to specify a version of the model to use. The |
| 4663 | # string is formatted the same way as `model_version`, with the addition |
| 4664 | # of the version information: |
| 4665 | # |
| 4666 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL/versions/YOUR_VERSION"` |
| 4667 | "modelName": "A String", # Use this field if you want to use the default version for the specified |
| 4668 | # model. The string must use the following format: |
| 4669 | # |
| 4670 | # `"projects/YOUR_PROJECT/models/YOUR_MODEL"` |
| 4671 | "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for |
| 4672 | # the model to use. |
| 4673 | "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing. |
| 4674 | # Defaults to 10 if not specified. |
| 4675 | "signatureName": "A String", # Optional. The name of the signature defined in the SavedModel to use for |
| 4676 | # this job. Please refer to |
| 4677 | # [SavedModel](https://tensorflow.github.io/serving/serving_basic.html) |
| 4678 | # for information about how to use signatures. |
| 4679 | # |
| 4680 | # Defaults to |
| 4681 | # [DEFAULT_SERVING_SIGNATURE_DEF_KEY](https://www.tensorflow.org/api_docs/python/tf/saved_model/signature_constants) |
| 4682 | # , which is "serving_default". |
| 4683 | }, |
| 4684 | "errorMessage": "A String", # Output only. The details of a failure or a cancellation. |
| 4685 | }</pre> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4686 | </div> |
| 4687 | |
| 4688 | <div class="method"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4689 | <code class="details" id="setIamPolicy">setIamPolicy(resource, body=None, x__xgafv=None)</code> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4690 | <pre>Sets the access control policy on the specified resource. Replaces any |
| 4691 | existing policy. |
| 4692 | |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4693 | Can return `NOT_FOUND`, `INVALID_ARGUMENT`, and `PERMISSION_DENIED` errors. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4694 | |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4695 | Args: |
| 4696 | resource: string, REQUIRED: The resource for which the policy is being specified. |
| 4697 | See the operation documentation for the appropriate value for this field. (required) |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4698 | body: object, The request body. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4699 | The object takes the form of: |
| 4700 | |
| 4701 | { # Request message for `SetIamPolicy` method. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4702 | "policy": { # An Identity and Access Management (IAM) policy, which specifies access # REQUIRED: The complete policy to be applied to the `resource`. The size of |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4703 | # the policy is limited to a few 10s of KB. An empty policy is a |
| 4704 | # valid policy but certain Cloud Platform services (such as Projects) |
| 4705 | # might reject them. |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4706 | # controls for Google Cloud resources. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4707 | # |
| 4708 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4709 | # A `Policy` is a collection of `bindings`. A `binding` binds one or more |
| 4710 | # `members` to a single `role`. Members can be user accounts, service accounts, |
| 4711 | # Google groups, and domains (such as G Suite). A `role` is a named list of |
| 4712 | # permissions; each `role` can be an IAM predefined role or a user-created |
| 4713 | # custom role. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4714 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4715 | # For some types of Google Cloud resources, a `binding` can also specify a |
| 4716 | # `condition`, which is a logical expression that allows access to a resource |
| 4717 | # only if the expression evaluates to `true`. A condition can add constraints |
| 4718 | # based on attributes of the request, the resource, or both. To learn which |
| 4719 | # resources support conditions in their IAM policies, see the |
| 4720 | # [IAM documentation](https://cloud.google.com/iam/help/conditions/resource-policies). |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4721 | # |
| 4722 | # **JSON example:** |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4723 | # |
| 4724 | # { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4725 | # "bindings": [ |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4726 | # { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4727 | # "role": "roles/resourcemanager.organizationAdmin", |
| 4728 | # "members": [ |
| 4729 | # "user:mike@example.com", |
| 4730 | # "group:admins@example.com", |
| 4731 | # "domain:google.com", |
| 4732 | # "serviceAccount:my-project-id@appspot.gserviceaccount.com" |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4733 | # ] |
| 4734 | # }, |
| 4735 | # { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4736 | # "role": "roles/resourcemanager.organizationViewer", |
| 4737 | # "members": [ |
| 4738 | # "user:eve@example.com" |
| 4739 | # ], |
| 4740 | # "condition": { |
| 4741 | # "title": "expirable access", |
| 4742 | # "description": "Does not grant access after Sep 2020", |
| 4743 | # "expression": "request.time < timestamp('2020-10-01T00:00:00.000Z')", |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4744 | # } |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4745 | # } |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4746 | # ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4747 | # "etag": "BwWWja0YfJA=", |
| 4748 | # "version": 3 |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4749 | # } |
| 4750 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4751 | # **YAML example:** |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4752 | # |
| 4753 | # bindings: |
| 4754 | # - members: |
| 4755 | # - user:mike@example.com |
| 4756 | # - group:admins@example.com |
| 4757 | # - domain:google.com |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4758 | # - serviceAccount:my-project-id@appspot.gserviceaccount.com |
| 4759 | # role: roles/resourcemanager.organizationAdmin |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4760 | # - members: |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4761 | # - user:eve@example.com |
| 4762 | # role: roles/resourcemanager.organizationViewer |
| 4763 | # condition: |
| 4764 | # title: expirable access |
| 4765 | # description: Does not grant access after Sep 2020 |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4766 | # expression: request.time < timestamp('2020-10-01T00:00:00.000Z') |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4767 | # - etag: BwWWja0YfJA= |
| 4768 | # - version: 3 |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4769 | # |
| 4770 | # For a description of IAM and its features, see the |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4771 | # [IAM documentation](https://cloud.google.com/iam/docs/). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4772 | "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help |
| 4773 | # prevent simultaneous updates of a policy from overwriting each other. |
| 4774 | # It is strongly suggested that systems make use of the `etag` in the |
| 4775 | # read-modify-write cycle to perform policy updates in order to avoid race |
| 4776 | # conditions: An `etag` is returned in the response to `getIamPolicy`, and |
| 4777 | # systems are expected to put that etag in the request to `setIamPolicy` to |
| 4778 | # ensure that their change will be applied to the same version of the policy. |
| 4779 | # |
| 4780 | # **Important:** If you use IAM Conditions, you must include the `etag` field |
| 4781 | # whenever you call `setIamPolicy`. If you omit this field, then IAM allows |
| 4782 | # you to overwrite a version `3` policy with a version `1` policy, and all of |
| 4783 | # the conditions in the version `3` policy are lost. |
| 4784 | "version": 42, # Specifies the format of the policy. |
| 4785 | # |
| 4786 | # Valid values are `0`, `1`, and `3`. Requests that specify an invalid value |
| 4787 | # are rejected. |
| 4788 | # |
| 4789 | # Any operation that affects conditional role bindings must specify version |
| 4790 | # `3`. This requirement applies to the following operations: |
| 4791 | # |
| 4792 | # * Getting a policy that includes a conditional role binding |
| 4793 | # * Adding a conditional role binding to a policy |
| 4794 | # * Changing a conditional role binding in a policy |
| 4795 | # * Removing any role binding, with or without a condition, from a policy |
| 4796 | # that includes conditions |
| 4797 | # |
| 4798 | # **Important:** If you use IAM Conditions, you must include the `etag` field |
| 4799 | # whenever you call `setIamPolicy`. If you omit this field, then IAM allows |
| 4800 | # you to overwrite a version `3` policy with a version `1` policy, and all of |
| 4801 | # the conditions in the version `3` policy are lost. |
| 4802 | # |
| 4803 | # If a policy does not include any conditions, operations on that policy may |
| 4804 | # specify any valid version or leave the field unset. |
| 4805 | # |
| 4806 | # To learn which resources support conditions in their IAM policies, see the |
| 4807 | # [IAM documentation](https://cloud.google.com/iam/help/conditions/resource-policies). |
| 4808 | "auditConfigs": [ # Specifies cloud audit logging configuration for this policy. |
| 4809 | { # Specifies the audit configuration for a service. |
| 4810 | # The configuration determines which permission types are logged, and what |
| 4811 | # identities, if any, are exempted from logging. |
| 4812 | # An AuditConfig must have one or more AuditLogConfigs. |
| 4813 | # |
| 4814 | # If there are AuditConfigs for both `allServices` and a specific service, |
| 4815 | # the union of the two AuditConfigs is used for that service: the log_types |
| 4816 | # specified in each AuditConfig are enabled, and the exempted_members in each |
| 4817 | # AuditLogConfig are exempted. |
| 4818 | # |
| 4819 | # Example Policy with multiple AuditConfigs: |
| 4820 | # |
| 4821 | # { |
| 4822 | # "audit_configs": [ |
| 4823 | # { |
| 4824 | # "service": "allServices" |
| 4825 | # "audit_log_configs": [ |
| 4826 | # { |
| 4827 | # "log_type": "DATA_READ", |
| 4828 | # "exempted_members": [ |
| 4829 | # "user:jose@example.com" |
| 4830 | # ] |
| 4831 | # }, |
| 4832 | # { |
| 4833 | # "log_type": "DATA_WRITE", |
| 4834 | # }, |
| 4835 | # { |
| 4836 | # "log_type": "ADMIN_READ", |
| 4837 | # } |
| 4838 | # ] |
| 4839 | # }, |
| 4840 | # { |
| 4841 | # "service": "sampleservice.googleapis.com" |
| 4842 | # "audit_log_configs": [ |
| 4843 | # { |
| 4844 | # "log_type": "DATA_READ", |
| 4845 | # }, |
| 4846 | # { |
| 4847 | # "log_type": "DATA_WRITE", |
| 4848 | # "exempted_members": [ |
| 4849 | # "user:aliya@example.com" |
| 4850 | # ] |
| 4851 | # } |
| 4852 | # ] |
| 4853 | # } |
| 4854 | # ] |
| 4855 | # } |
| 4856 | # |
| 4857 | # For sampleservice, this policy enables DATA_READ, DATA_WRITE and ADMIN_READ |
| 4858 | # logging. It also exempts jose@example.com from DATA_READ logging, and |
| 4859 | # aliya@example.com from DATA_WRITE logging. |
| 4860 | "service": "A String", # Specifies a service that will be enabled for audit logging. |
| 4861 | # For example, `storage.googleapis.com`, `cloudsql.googleapis.com`. |
| 4862 | # `allServices` is a special value that covers all services. |
| 4863 | "auditLogConfigs": [ # The configuration for logging of each type of permission. |
| 4864 | { # Provides the configuration for logging a type of permissions. |
| 4865 | # Example: |
| 4866 | # |
| 4867 | # { |
| 4868 | # "audit_log_configs": [ |
| 4869 | # { |
| 4870 | # "log_type": "DATA_READ", |
| 4871 | # "exempted_members": [ |
| 4872 | # "user:jose@example.com" |
| 4873 | # ] |
| 4874 | # }, |
| 4875 | # { |
| 4876 | # "log_type": "DATA_WRITE", |
| 4877 | # } |
| 4878 | # ] |
| 4879 | # } |
| 4880 | # |
| 4881 | # This enables 'DATA_READ' and 'DATA_WRITE' logging, while exempting |
| 4882 | # jose@example.com from DATA_READ logging. |
| 4883 | "exemptedMembers": [ # Specifies the identities that do not cause logging for this type of |
| 4884 | # permission. |
| 4885 | # Follows the same format of Binding.members. |
| 4886 | "A String", |
| 4887 | ], |
| 4888 | "logType": "A String", # The log type that this config enables. |
| 4889 | }, |
| 4890 | ], |
| 4891 | }, |
| 4892 | ], |
| 4893 | "bindings": [ # Associates a list of `members` to a `role`. Optionally, may specify a |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4894 | # `condition` that determines how and when the `bindings` are applied. Each |
| 4895 | # of the `bindings` must contain at least one member. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4896 | { # Associates `members` with a `role`. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4897 | "condition": { # Represents a textual expression in the Common Expression Language (CEL) # The condition that is associated with this binding. |
| 4898 | # |
| 4899 | # If the condition evaluates to `true`, then this binding applies to the |
| 4900 | # current request. |
| 4901 | # |
| 4902 | # If the condition evaluates to `false`, then this binding does not apply to |
| 4903 | # the current request. However, a different role binding might grant the same |
| 4904 | # role to one or more of the members in this binding. |
| 4905 | # |
| 4906 | # To learn which resources support conditions in their IAM policies, see the |
| 4907 | # [IAM |
| 4908 | # documentation](https://cloud.google.com/iam/help/conditions/resource-policies). |
| 4909 | # syntax. CEL is a C-like expression language. The syntax and semantics of CEL |
| 4910 | # are documented at https://github.com/google/cel-spec. |
| 4911 | # |
| 4912 | # Example (Comparison): |
| 4913 | # |
| 4914 | # title: "Summary size limit" |
| 4915 | # description: "Determines if a summary is less than 100 chars" |
| 4916 | # expression: "document.summary.size() < 100" |
| 4917 | # |
| 4918 | # Example (Equality): |
| 4919 | # |
| 4920 | # title: "Requestor is owner" |
| 4921 | # description: "Determines if requestor is the document owner" |
| 4922 | # expression: "document.owner == request.auth.claims.email" |
| 4923 | # |
| 4924 | # Example (Logic): |
| 4925 | # |
| 4926 | # title: "Public documents" |
| 4927 | # description: "Determine whether the document should be publicly visible" |
| 4928 | # expression: "document.type != 'private' && document.type != 'internal'" |
| 4929 | # |
| 4930 | # Example (Data Manipulation): |
| 4931 | # |
| 4932 | # title: "Notification string" |
| 4933 | # description: "Create a notification string with a timestamp." |
| 4934 | # expression: "'New message received at ' + string(document.create_time)" |
| 4935 | # |
| 4936 | # The exact variables and functions that may be referenced within an expression |
| 4937 | # are determined by the service that evaluates it. See the service |
| 4938 | # documentation for additional information. |
| 4939 | "title": "A String", # Optional. Title for the expression, i.e. a short string describing |
| 4940 | # its purpose. This can be used e.g. in UIs which allow to enter the |
| 4941 | # expression. |
| 4942 | "location": "A String", # Optional. String indicating the location of the expression for error |
| 4943 | # reporting, e.g. a file name and a position in the file. |
| 4944 | "description": "A String", # Optional. Description of the expression. This is a longer text which |
| 4945 | # describes the expression, e.g. when hovered over it in a UI. |
| 4946 | "expression": "A String", # Textual representation of an expression in Common Expression Language |
| 4947 | # syntax. |
| 4948 | }, |
| 4949 | "members": [ # Specifies the identities requesting access for a Cloud Platform resource. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4950 | # `members` can have the following values: |
| 4951 | # |
| 4952 | # * `allUsers`: A special identifier that represents anyone who is |
| 4953 | # on the internet; with or without a Google account. |
| 4954 | # |
| 4955 | # * `allAuthenticatedUsers`: A special identifier that represents anyone |
| 4956 | # who is authenticated with a Google account or a service account. |
| 4957 | # |
| 4958 | # * `user:{emailid}`: An email address that represents a specific Google |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4959 | # account. For example, `alice@example.com` . |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4960 | # |
| 4961 | # |
| 4962 | # * `serviceAccount:{emailid}`: An email address that represents a service |
| 4963 | # account. For example, `my-other-app@appspot.gserviceaccount.com`. |
| 4964 | # |
| 4965 | # * `group:{emailid}`: An email address that represents a Google group. |
| 4966 | # For example, `admins@example.com`. |
| 4967 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 4968 | # * `deleted:user:{emailid}?uid={uniqueid}`: An email address (plus unique |
| 4969 | # identifier) representing a user that has been recently deleted. For |
| 4970 | # example, `alice@example.com?uid=123456789012345678901`. If the user is |
| 4971 | # recovered, this value reverts to `user:{emailid}` and the recovered user |
| 4972 | # retains the role in the binding. |
| 4973 | # |
| 4974 | # * `deleted:serviceAccount:{emailid}?uid={uniqueid}`: An email address (plus |
| 4975 | # unique identifier) representing a service account that has been recently |
| 4976 | # deleted. For example, |
| 4977 | # `my-other-app@appspot.gserviceaccount.com?uid=123456789012345678901`. |
| 4978 | # If the service account is undeleted, this value reverts to |
| 4979 | # `serviceAccount:{emailid}` and the undeleted service account retains the |
| 4980 | # role in the binding. |
| 4981 | # |
| 4982 | # * `deleted:group:{emailid}?uid={uniqueid}`: An email address (plus unique |
| 4983 | # identifier) representing a Google group that has been recently |
| 4984 | # deleted. For example, `admins@example.com?uid=123456789012345678901`. If |
| 4985 | # the group is recovered, this value reverts to `group:{emailid}` and the |
| 4986 | # recovered group retains the role in the binding. |
| 4987 | # |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4988 | # |
| 4989 | # * `domain:{domain}`: The G Suite domain (primary) that represents all the |
| 4990 | # users of that domain. For example, `google.com` or `example.com`. |
| 4991 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4992 | "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4993 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4994 | "role": "A String", # Role that is assigned to `members`. |
| 4995 | # For example, `roles/viewer`, `roles/editor`, or `roles/owner`. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4996 | }, |
| 4997 | ], |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 4998 | }, |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 4999 | "updateMask": "A String", # OPTIONAL: A FieldMask specifying which fields of the policy to modify. Only |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5000 | # the fields in the mask will be modified. If no mask is provided, the |
| 5001 | # following default mask is used: |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5002 | # |
| 5003 | # `paths: "bindings, etag"` |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5004 | } |
| 5005 | |
| 5006 | x__xgafv: string, V1 error format. |
| 5007 | Allowed values |
| 5008 | 1 - v1 error format |
| 5009 | 2 - v2 error format |
| 5010 | |
| 5011 | Returns: |
| 5012 | An object of the form: |
| 5013 | |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5014 | { # An Identity and Access Management (IAM) policy, which specifies access |
| 5015 | # controls for Google Cloud resources. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5016 | # |
| 5017 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5018 | # A `Policy` is a collection of `bindings`. A `binding` binds one or more |
| 5019 | # `members` to a single `role`. Members can be user accounts, service accounts, |
| 5020 | # Google groups, and domains (such as G Suite). A `role` is a named list of |
| 5021 | # permissions; each `role` can be an IAM predefined role or a user-created |
| 5022 | # custom role. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5023 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5024 | # For some types of Google Cloud resources, a `binding` can also specify a |
| 5025 | # `condition`, which is a logical expression that allows access to a resource |
| 5026 | # only if the expression evaluates to `true`. A condition can add constraints |
| 5027 | # based on attributes of the request, the resource, or both. To learn which |
| 5028 | # resources support conditions in their IAM policies, see the |
| 5029 | # [IAM documentation](https://cloud.google.com/iam/help/conditions/resource-policies). |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5030 | # |
| 5031 | # **JSON example:** |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5032 | # |
| 5033 | # { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5034 | # "bindings": [ |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5035 | # { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5036 | # "role": "roles/resourcemanager.organizationAdmin", |
| 5037 | # "members": [ |
| 5038 | # "user:mike@example.com", |
| 5039 | # "group:admins@example.com", |
| 5040 | # "domain:google.com", |
| 5041 | # "serviceAccount:my-project-id@appspot.gserviceaccount.com" |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5042 | # ] |
| 5043 | # }, |
| 5044 | # { |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5045 | # "role": "roles/resourcemanager.organizationViewer", |
| 5046 | # "members": [ |
| 5047 | # "user:eve@example.com" |
| 5048 | # ], |
| 5049 | # "condition": { |
| 5050 | # "title": "expirable access", |
| 5051 | # "description": "Does not grant access after Sep 2020", |
| 5052 | # "expression": "request.time < timestamp('2020-10-01T00:00:00.000Z')", |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5053 | # } |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5054 | # } |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5055 | # ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5056 | # "etag": "BwWWja0YfJA=", |
| 5057 | # "version": 3 |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5058 | # } |
| 5059 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5060 | # **YAML example:** |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5061 | # |
| 5062 | # bindings: |
| 5063 | # - members: |
| 5064 | # - user:mike@example.com |
| 5065 | # - group:admins@example.com |
| 5066 | # - domain:google.com |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5067 | # - serviceAccount:my-project-id@appspot.gserviceaccount.com |
| 5068 | # role: roles/resourcemanager.organizationAdmin |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5069 | # - members: |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5070 | # - user:eve@example.com |
| 5071 | # role: roles/resourcemanager.organizationViewer |
| 5072 | # condition: |
| 5073 | # title: expirable access |
| 5074 | # description: Does not grant access after Sep 2020 |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5075 | # expression: request.time < timestamp('2020-10-01T00:00:00.000Z') |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5076 | # - etag: BwWWja0YfJA= |
| 5077 | # - version: 3 |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5078 | # |
| 5079 | # For a description of IAM and its features, see the |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5080 | # [IAM documentation](https://cloud.google.com/iam/docs/). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5081 | "etag": "A String", # `etag` is used for optimistic concurrency control as a way to help |
| 5082 | # prevent simultaneous updates of a policy from overwriting each other. |
| 5083 | # It is strongly suggested that systems make use of the `etag` in the |
| 5084 | # read-modify-write cycle to perform policy updates in order to avoid race |
| 5085 | # conditions: An `etag` is returned in the response to `getIamPolicy`, and |
| 5086 | # systems are expected to put that etag in the request to `setIamPolicy` to |
| 5087 | # ensure that their change will be applied to the same version of the policy. |
| 5088 | # |
| 5089 | # **Important:** If you use IAM Conditions, you must include the `etag` field |
| 5090 | # whenever you call `setIamPolicy`. If you omit this field, then IAM allows |
| 5091 | # you to overwrite a version `3` policy with a version `1` policy, and all of |
| 5092 | # the conditions in the version `3` policy are lost. |
| 5093 | "version": 42, # Specifies the format of the policy. |
| 5094 | # |
| 5095 | # Valid values are `0`, `1`, and `3`. Requests that specify an invalid value |
| 5096 | # are rejected. |
| 5097 | # |
| 5098 | # Any operation that affects conditional role bindings must specify version |
| 5099 | # `3`. This requirement applies to the following operations: |
| 5100 | # |
| 5101 | # * Getting a policy that includes a conditional role binding |
| 5102 | # * Adding a conditional role binding to a policy |
| 5103 | # * Changing a conditional role binding in a policy |
| 5104 | # * Removing any role binding, with or without a condition, from a policy |
| 5105 | # that includes conditions |
| 5106 | # |
| 5107 | # **Important:** If you use IAM Conditions, you must include the `etag` field |
| 5108 | # whenever you call `setIamPolicy`. If you omit this field, then IAM allows |
| 5109 | # you to overwrite a version `3` policy with a version `1` policy, and all of |
| 5110 | # the conditions in the version `3` policy are lost. |
| 5111 | # |
| 5112 | # If a policy does not include any conditions, operations on that policy may |
| 5113 | # specify any valid version or leave the field unset. |
| 5114 | # |
| 5115 | # To learn which resources support conditions in their IAM policies, see the |
| 5116 | # [IAM documentation](https://cloud.google.com/iam/help/conditions/resource-policies). |
| 5117 | "auditConfigs": [ # Specifies cloud audit logging configuration for this policy. |
| 5118 | { # Specifies the audit configuration for a service. |
| 5119 | # The configuration determines which permission types are logged, and what |
| 5120 | # identities, if any, are exempted from logging. |
| 5121 | # An AuditConfig must have one or more AuditLogConfigs. |
| 5122 | # |
| 5123 | # If there are AuditConfigs for both `allServices` and a specific service, |
| 5124 | # the union of the two AuditConfigs is used for that service: the log_types |
| 5125 | # specified in each AuditConfig are enabled, and the exempted_members in each |
| 5126 | # AuditLogConfig are exempted. |
| 5127 | # |
| 5128 | # Example Policy with multiple AuditConfigs: |
| 5129 | # |
| 5130 | # { |
| 5131 | # "audit_configs": [ |
| 5132 | # { |
| 5133 | # "service": "allServices" |
| 5134 | # "audit_log_configs": [ |
| 5135 | # { |
| 5136 | # "log_type": "DATA_READ", |
| 5137 | # "exempted_members": [ |
| 5138 | # "user:jose@example.com" |
| 5139 | # ] |
| 5140 | # }, |
| 5141 | # { |
| 5142 | # "log_type": "DATA_WRITE", |
| 5143 | # }, |
| 5144 | # { |
| 5145 | # "log_type": "ADMIN_READ", |
| 5146 | # } |
| 5147 | # ] |
| 5148 | # }, |
| 5149 | # { |
| 5150 | # "service": "sampleservice.googleapis.com" |
| 5151 | # "audit_log_configs": [ |
| 5152 | # { |
| 5153 | # "log_type": "DATA_READ", |
| 5154 | # }, |
| 5155 | # { |
| 5156 | # "log_type": "DATA_WRITE", |
| 5157 | # "exempted_members": [ |
| 5158 | # "user:aliya@example.com" |
| 5159 | # ] |
| 5160 | # } |
| 5161 | # ] |
| 5162 | # } |
| 5163 | # ] |
| 5164 | # } |
| 5165 | # |
| 5166 | # For sampleservice, this policy enables DATA_READ, DATA_WRITE and ADMIN_READ |
| 5167 | # logging. It also exempts jose@example.com from DATA_READ logging, and |
| 5168 | # aliya@example.com from DATA_WRITE logging. |
| 5169 | "service": "A String", # Specifies a service that will be enabled for audit logging. |
| 5170 | # For example, `storage.googleapis.com`, `cloudsql.googleapis.com`. |
| 5171 | # `allServices` is a special value that covers all services. |
| 5172 | "auditLogConfigs": [ # The configuration for logging of each type of permission. |
| 5173 | { # Provides the configuration for logging a type of permissions. |
| 5174 | # Example: |
| 5175 | # |
| 5176 | # { |
| 5177 | # "audit_log_configs": [ |
| 5178 | # { |
| 5179 | # "log_type": "DATA_READ", |
| 5180 | # "exempted_members": [ |
| 5181 | # "user:jose@example.com" |
| 5182 | # ] |
| 5183 | # }, |
| 5184 | # { |
| 5185 | # "log_type": "DATA_WRITE", |
| 5186 | # } |
| 5187 | # ] |
| 5188 | # } |
| 5189 | # |
| 5190 | # This enables 'DATA_READ' and 'DATA_WRITE' logging, while exempting |
| 5191 | # jose@example.com from DATA_READ logging. |
| 5192 | "exemptedMembers": [ # Specifies the identities that do not cause logging for this type of |
| 5193 | # permission. |
| 5194 | # Follows the same format of Binding.members. |
| 5195 | "A String", |
| 5196 | ], |
| 5197 | "logType": "A String", # The log type that this config enables. |
| 5198 | }, |
| 5199 | ], |
| 5200 | }, |
| 5201 | ], |
| 5202 | "bindings": [ # Associates a list of `members` to a `role`. Optionally, may specify a |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5203 | # `condition` that determines how and when the `bindings` are applied. Each |
| 5204 | # of the `bindings` must contain at least one member. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5205 | { # Associates `members` with a `role`. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5206 | "condition": { # Represents a textual expression in the Common Expression Language (CEL) # The condition that is associated with this binding. |
| 5207 | # |
| 5208 | # If the condition evaluates to `true`, then this binding applies to the |
| 5209 | # current request. |
| 5210 | # |
| 5211 | # If the condition evaluates to `false`, then this binding does not apply to |
| 5212 | # the current request. However, a different role binding might grant the same |
| 5213 | # role to one or more of the members in this binding. |
| 5214 | # |
| 5215 | # To learn which resources support conditions in their IAM policies, see the |
| 5216 | # [IAM |
| 5217 | # documentation](https://cloud.google.com/iam/help/conditions/resource-policies). |
| 5218 | # syntax. CEL is a C-like expression language. The syntax and semantics of CEL |
| 5219 | # are documented at https://github.com/google/cel-spec. |
| 5220 | # |
| 5221 | # Example (Comparison): |
| 5222 | # |
| 5223 | # title: "Summary size limit" |
| 5224 | # description: "Determines if a summary is less than 100 chars" |
| 5225 | # expression: "document.summary.size() < 100" |
| 5226 | # |
| 5227 | # Example (Equality): |
| 5228 | # |
| 5229 | # title: "Requestor is owner" |
| 5230 | # description: "Determines if requestor is the document owner" |
| 5231 | # expression: "document.owner == request.auth.claims.email" |
| 5232 | # |
| 5233 | # Example (Logic): |
| 5234 | # |
| 5235 | # title: "Public documents" |
| 5236 | # description: "Determine whether the document should be publicly visible" |
| 5237 | # expression: "document.type != 'private' && document.type != 'internal'" |
| 5238 | # |
| 5239 | # Example (Data Manipulation): |
| 5240 | # |
| 5241 | # title: "Notification string" |
| 5242 | # description: "Create a notification string with a timestamp." |
| 5243 | # expression: "'New message received at ' + string(document.create_time)" |
| 5244 | # |
| 5245 | # The exact variables and functions that may be referenced within an expression |
| 5246 | # are determined by the service that evaluates it. See the service |
| 5247 | # documentation for additional information. |
| 5248 | "title": "A String", # Optional. Title for the expression, i.e. a short string describing |
| 5249 | # its purpose. This can be used e.g. in UIs which allow to enter the |
| 5250 | # expression. |
| 5251 | "location": "A String", # Optional. String indicating the location of the expression for error |
| 5252 | # reporting, e.g. a file name and a position in the file. |
| 5253 | "description": "A String", # Optional. Description of the expression. This is a longer text which |
| 5254 | # describes the expression, e.g. when hovered over it in a UI. |
| 5255 | "expression": "A String", # Textual representation of an expression in Common Expression Language |
| 5256 | # syntax. |
| 5257 | }, |
| 5258 | "members": [ # Specifies the identities requesting access for a Cloud Platform resource. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5259 | # `members` can have the following values: |
| 5260 | # |
| 5261 | # * `allUsers`: A special identifier that represents anyone who is |
| 5262 | # on the internet; with or without a Google account. |
| 5263 | # |
| 5264 | # * `allAuthenticatedUsers`: A special identifier that represents anyone |
| 5265 | # who is authenticated with a Google account or a service account. |
| 5266 | # |
| 5267 | # * `user:{emailid}`: An email address that represents a specific Google |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5268 | # account. For example, `alice@example.com` . |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5269 | # |
| 5270 | # |
| 5271 | # * `serviceAccount:{emailid}`: An email address that represents a service |
| 5272 | # account. For example, `my-other-app@appspot.gserviceaccount.com`. |
| 5273 | # |
| 5274 | # * `group:{emailid}`: An email address that represents a Google group. |
| 5275 | # For example, `admins@example.com`. |
| 5276 | # |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5277 | # * `deleted:user:{emailid}?uid={uniqueid}`: An email address (plus unique |
| 5278 | # identifier) representing a user that has been recently deleted. For |
| 5279 | # example, `alice@example.com?uid=123456789012345678901`. If the user is |
| 5280 | # recovered, this value reverts to `user:{emailid}` and the recovered user |
| 5281 | # retains the role in the binding. |
| 5282 | # |
| 5283 | # * `deleted:serviceAccount:{emailid}?uid={uniqueid}`: An email address (plus |
| 5284 | # unique identifier) representing a service account that has been recently |
| 5285 | # deleted. For example, |
| 5286 | # `my-other-app@appspot.gserviceaccount.com?uid=123456789012345678901`. |
| 5287 | # If the service account is undeleted, this value reverts to |
| 5288 | # `serviceAccount:{emailid}` and the undeleted service account retains the |
| 5289 | # role in the binding. |
| 5290 | # |
| 5291 | # * `deleted:group:{emailid}?uid={uniqueid}`: An email address (plus unique |
| 5292 | # identifier) representing a Google group that has been recently |
| 5293 | # deleted. For example, `admins@example.com?uid=123456789012345678901`. If |
| 5294 | # the group is recovered, this value reverts to `group:{emailid}` and the |
| 5295 | # recovered group retains the role in the binding. |
| 5296 | # |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5297 | # |
| 5298 | # * `domain:{domain}`: The G Suite domain (primary) that represents all the |
| 5299 | # users of that domain. For example, `google.com` or `example.com`. |
| 5300 | # |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5301 | "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5302 | ], |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5303 | "role": "A String", # Role that is assigned to `members`. |
| 5304 | # For example, `roles/viewer`, `roles/editor`, or `roles/owner`. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5305 | }, |
| 5306 | ], |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5307 | }</pre> |
| 5308 | </div> |
| 5309 | |
| 5310 | <div class="method"> |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5311 | <code class="details" id="testIamPermissions">testIamPermissions(resource, body=None, x__xgafv=None)</code> |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5312 | <pre>Returns permissions that a caller has on the specified resource. |
| 5313 | If the resource does not exist, this will return an empty set of |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5314 | permissions, not a `NOT_FOUND` error. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5315 | |
| 5316 | Note: This operation is designed to be used for building permission-aware |
| 5317 | UIs and command-line tools, not for authorization checking. This operation |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5318 | may "fail open" without warning. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5319 | |
| 5320 | Args: |
| 5321 | resource: string, REQUIRED: The resource for which the policy detail is being requested. |
| 5322 | See the operation documentation for the appropriate value for this field. (required) |
| Dan O'Meara | dd49464 | 2020-05-01 07:42:23 -0700 | [diff] [blame] | 5323 | body: object, The request body. |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5324 | The object takes the form of: |
| 5325 | |
| 5326 | { # Request message for `TestIamPermissions` method. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5327 | "permissions": [ # The set of permissions to check for the `resource`. Permissions with |
| 5328 | # wildcards (such as '*' or 'storage.*') are not allowed. For more |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5329 | # information see |
| 5330 | # [IAM Overview](https://cloud.google.com/iam/docs/overview#permissions). |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5331 | "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5332 | ], |
| 5333 | } |
| 5334 | |
| 5335 | x__xgafv: string, V1 error format. |
| 5336 | Allowed values |
| 5337 | 1 - v1 error format |
| 5338 | 2 - v2 error format |
| 5339 | |
| 5340 | Returns: |
| 5341 | An object of the form: |
| 5342 | |
| 5343 | { # Response message for `TestIamPermissions` method. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5344 | "permissions": [ # A subset of `TestPermissionsRequest.permissions` that the caller is |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5345 | # allowed. |
| Bu Sun Kim | 6502091 | 2020-05-20 12:08:20 -0700 | [diff] [blame] | 5346 | "A String", |
| Bu Sun Kim | 715bd7f | 2019-06-14 16:50:42 -0700 | [diff] [blame] | 5347 | ], |
| 5348 | }</pre> |
| 5349 | </div> |
| 5350 | |
| Sai Cheemalapati | c30d2b5 | 2017-03-13 12:12:03 -0400 | [diff] [blame] | 5351 | </body></html> |