Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / python / google-api-python-client / 98b51f044ede8b9b24e584ab9680e3f35f4bd469 / . / docs / dyn / ml_v1beta1.projects.jobs.html
blob: cd6f67a11626ae3442e950441758e8a27aca1cf0 [file] [log] [blame]
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]1<html><body>
2<style>
3
4body, h1, h2, h3, div, span, p, pre, a {
5 margin: 0;
6 padding: 0;
7 border: 0;
8 font-weight: inherit;
9 font-style: inherit;
10 font-size: 100%;
11 font-family: inherit;
12 vertical-align: baseline;
13}
14
15body {
16 font-size: 13px;
17 padding: 1em;
18}
19
20h1 {
21 font-size: 26px;
22 margin-bottom: 1em;
23}
24
25h2 {
26 font-size: 24px;
27 margin-bottom: 1em;
28}
29
30h3 {
31 font-size: 20px;
32 margin-bottom: 1em;
33 margin-top: 1em;
34}
35
36pre, code {
37 line-height: 1.5;
38 font-family: Monaco, 'DejaVu Sans Mono', 'Bitstream Vera Sans Mono', 'Lucida Console', monospace;
39}
40
41pre {
42 margin-top: 0.5em;
43}
44
45h1, h2, h3, p {
46 font-family: Arial, sans serif;
47}
48
49h1, h2, h3 {
50 border-bottom: solid #CCC 1px;
51}
52
53.toc_element {
54 margin-top: 0.5em;
55}
56
57.firstline {
58 margin-left: 2 em;
59}
60
61.method {
62 margin-top: 1em;
63 border: solid 1px #CCC;
64 padding: 1em;
65 background: #EEE;
66}
67
68.details {
69 font-weight: bold;
70 font-size: 14px;
71}
72
73</style>
74
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]75<h1><a href="ml_v1beta1.html">Google Cloud Machine Learning Engine</a> . <a href="ml_v1beta1.projects.html">projects</a> . <a href="ml_v1beta1.projects.jobs.html">jobs</a></h1>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]76<h2>Instance Methods</h2>
77<p class="toc_element">
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]78 <code><a href="#cancel">cancel(name, body, x__xgafv=None)</a></code></p>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]79<p class="firstline">Cancels a running job.</p>
80<p class="toc_element">
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]81 <code><a href="#create">create(parent, body, x__xgafv=None)</a></code></p>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]82<p class="firstline">Creates a training or a batch prediction job.</p>
83<p class="toc_element">
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]84 <code><a href="#get">get(name, x__xgafv=None)</a></code></p>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]85<p class="firstline">Describes a job.</p>
86<p class="toc_element">
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]87 <code><a href="#list">list(parent, pageSize=None, filter=None, pageToken=None, x__xgafv=None)</a></code></p>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]88<p class="firstline">Lists the jobs in the project.</p>
89<p class="toc_element">
90 <code><a href="#list_next">list_next(previous_request, previous_response)</a></code></p>
91<p class="firstline">Retrieves the next page of results.</p>
92<h3>Method Details</h3>
93<div class="method">
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]94 <code class="details" id="cancel">cancel(name, body, x__xgafv=None)</code>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]95 <pre>Cancels a running job.
96
97Args:
98 name: string, Required. The name of the job to cancel.
99
100Authorization: requires `Editor` role on the parent project. (required)
101 body: object, The request body. (required)
102 The object takes the form of:
103
104{ # Request message for the CancelJob method.
105 }
106
107 x__xgafv: string, V1 error format.
108 Allowed values
109 1 - v1 error format
110 2 - v2 error format
111
112Returns:
113 An object of the form:
114
115 { # A generic empty message that you can re-use to avoid defining duplicated
116 # empty messages in your APIs. A typical example is to use it as the request
117 # or the response type of an API method. For instance:
118 #
119 # service Foo {
120 # rpc Bar(google.protobuf.Empty) returns (google.protobuf.Empty);
121 # }
122 #
123 # The JSON representation for `Empty` is empty JSON object `{}`.
124 }</pre>
125</div>
126
127<div class="method">
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]128 <code class="details" id="create">create(parent, body, x__xgafv=None)</code>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]129 <pre>Creates a training or a batch prediction job.
130
131Args:
132 parent: string, Required. The project name.
133
134Authorization: requires `Editor` role on the specified project. (required)
135 body: object, The request body. (required)
136 The object takes the form of:
137
138{ # Represents a training or prediction job.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]139 "trainingOutput": { # Represents results of a training job. Output only. # The current training job result.
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]140 "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully.
141 # Only set for hyperparameter tuning jobs.
142 "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]143 "trials": [ # Results for individual Hyperparameter trials.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]144 # Only set for hyperparameter tuning jobs.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]145 { # Represents the result of a single hyperparameter tuning trial from a
146 # training job. The TrainingOutput object that is returned on successful
147 # completion of a training job with hyperparameter tuning includes a list
148 # of HyperparameterOutput objects, one for each successful trial.
149 "hyperparameters": { # The hyperparameters given to this trial.
150 "a_key": "A String",
151 },
152 "trialId": "A String", # The trial id for these results.
153 "allMetrics": [ # All recorded object metrics for this trial.
154 { # An observed value of a metric.
155 "trainingStep": "A String", # The global training step for this metric.
156 "objectiveValue": 3.14, # The objective value at this training step.
157 },
158 ],
159 "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial.
160 "trainingStep": "A String", # The global training step for this metric.
161 "objectiveValue": 3.14, # The objective value at this training step.
162 },
163 },
164 ],
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]165 "consumedMLUnits": 3.14, # The amount of ML units consumed by the job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]166 },
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]167 "trainingInput": { # Represents input parameters for a training job. # Input parameters to create a training job.
168 "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training
169 # job's worker nodes.
170 #
171 # The supported values are the same as those described in the entry for
172 # `masterType`.
173 #
174 # This value must be present when `scaleTier` is set to `CUSTOM` and
175 # `workerCount` is greater than zero.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]176 "runtimeVersion": "A String", # Optional. The Google Cloud ML runtime version to use for training. If not
177 # set, Google Cloud ML will choose the latest stable version.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]178 "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers
179 # and parameter servers.
180 "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training
181 # job's master worker.
182 #
183 # The following types are supported:
184 #
185 # <dl>
186 # <dt>standard</dt>
187 # <dd>
188 # A basic machine configuration suitable for training simple models with
189 # small to moderate datasets.
190 # </dd>
191 # <dt>large_model</dt>
192 # <dd>
193 # A machine with a lot of memory, specially suited for parameter servers
194 # when your model is large (having many hidden layers or layers with very
195 # large numbers of nodes).
196 # </dd>
197 # <dt>complex_model_s</dt>
198 # <dd>
199 # A machine suitable for the master and workers of the cluster when your
200 # model requires more computation than the standard machine can handle
201 # satisfactorily.
202 # </dd>
203 # <dt>complex_model_m</dt>
204 # <dd>
205 # A machine with roughly twice the number of cores and roughly double the
206 # memory of <code suppresswarning="true">complex_model_s</code>.
207 # </dd>
208 # <dt>complex_model_l</dt>
209 # <dd>
210 # A machine with roughly twice the number of cores and roughly double the
211 # memory of <code suppresswarning="true">complex_model_m</code>.
212 # </dd>
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]213 # <dt>standard_gpu</dt>
214 # <dd>
215 # A machine equivalent to <code suppresswarning="true">standard</code> that
216 # also includes a
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]217 # <a href="/ml-engine/docs/how-tos/using-gpus">
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]218 # GPU that you can use in your trainer</a>.
219 # </dd>
220 # <dt>complex_model_m_gpu</dt>
221 # <dd>
222 # A machine equivalent to
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]223 # <code suppresswarning="true">complex_model_m</code> that also includes
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]224 # four GPUs.
225 # </dd>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]226 # </dl>
227 #
228 # You must set this value when `scaleTier` is set to `CUSTOM`.
229 "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune.
230 "maxTrials": 42, # Optional. How many training trials should be attempted to optimize
231 # the specified hyperparameters.
232 #
233 # Defaults to one.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]234 "hyperparameterMetricTag": "A String", # Optional. The Tensorflow summary tag name to use for optimizing trials. For
235 # current versions of Tensorflow, this tag name should exactly match what is
236 # shown in Tensorboard, including all scopes. For versions of Tensorflow
237 # prior to 0.12, this should be only the tag passed to tf.Summary.
238 # By default, "training/hptuning/metric" will be used.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]239 "params": [ # Required. The set of parameters to tune.
240 { # Represents a single hyperparameter to optimize.
241 "maxValue": 3.14, # Required if typeis `DOUBLE` or `INTEGER`. This field
242 # should be unset if type is `CATEGORICAL`. This value should be integers if
243 # type is `INTEGER`.
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]244 "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories.
245 "A String",
246 ],
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]247 "discreteValues": [ # Required if type is `DISCRETE`.
248 # A list of feasible points.
249 # The list should be in strictly increasing order. For instance, this
250 # parameter might have possible settings of 1.5, 2.5, and 4.0. This list
251 # should not contain more than 1,000 values.
252 3.14,
253 ],
254 "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in
255 # a HyperparameterSpec message. E.g., "learning_rate".
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]256 "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field
257 # should be unset if type is `CATEGORICAL`. This value should be integers if
258 # type is INTEGER.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]259 "type": "A String", # Required. The type of the parameter.
260 "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube.
261 # Leave unset for categorical parameters.
262 # Some kind of scaling is strongly recommended for real or integral
263 # parameters (e.g., `UNIT_LINEAR_SCALE`).
264 },
265 ],
266 "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently.
267 # You can reduce the time it takes to perform hyperparameter tuning by adding
268 # trials in parallel. However, each trail only benefits from the information
269 # gained in completed trials. That means that a trial does not get access to
270 # the results of trials running at the same time, which could reduce the
271 # quality of the overall optimization.
272 #
273 # Each trial will use the same scale tier and machine types.
274 #
275 # Defaults to one.
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]276 "goal": "A String", # Required. The type of goal to use for tuning. Available types are
277 # `MAXIMIZE` and `MINIMIZE`.
278 #
279 # Defaults to `MAXIMIZE`.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]280 },
281 "region": "A String", # Required. The Google Compute Engine region to run the training job in.
282 "args": [ # Optional. Command line arguments to pass to the program.
283 "A String",
284 ],
285 "pythonModule": "A String", # Required. The Python module name to run after installing the packages.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]286 "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs
287 # and other data needed for training. This path is passed to your TensorFlow
288 # program as the 'job_dir' command-line argument. The benefit of specifying
289 # this field is that Cloud ML validates the path for use in training.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]290 "packageUris": [ # Required. The Google Cloud Storage location of the packages with
291 # the training program and any additional dependencies.
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]292 # The maximum number of package URIs is 100.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]293 "A String",
294 ],
295 "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each
296 # replica in the cluster will be of the type specified in `worker_type`.
297 #
298 # This value can only be used when `scale_tier` is set to `CUSTOM`. If you
299 # set this value, you must also set `worker_type`.
300 "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training
301 # job's parameter server.
302 #
303 # The supported values are the same as those described in the entry for
304 # `master_type`.
305 #
306 # This value must be present when `scaleTier` is set to `CUSTOM` and
307 # `parameter_server_count` is greater than zero.
308 "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training
309 # job. Each replica in the cluster will be of the type specified in
310 # `parameter_server_type`.
311 #
312 # This value can only be used when `scale_tier` is set to `CUSTOM`.If you
313 # set this value, you must also set `parameter_server_type`.
314 },
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]315 "startTime": "A String", # Output only. When the job processing was started.
316 "errorMessage": "A String", # Output only. The details of a failure or a cancellation.
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]317 "jobId": "A String", # Required. The user-specified id of the job.
318 "state": "A String", # Output only. The detailed state of a job.
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]319 "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job.
320 "modelName": "A String", # Use this field if you want to use the default version for the specified
321 # model. The string must use the following format:
322 #
323 # `"projects/<var>[YOUR_PROJECT]</var>/models/<var>[YOUR_MODEL]</var>"`
324 "runtimeVersion": "A String", # Optional. The Google Cloud ML runtime version to use for this batch
325 # prediction. If not set, Google Cloud ML will pick the runtime version used
326 # during the CreateVersion request for this model version, or choose the
327 # latest stable version when model version information is not available
328 # such as when the model is specified by uri.
329 "inputPaths": [ # Required. The Google Cloud Storage location of the input data files.
330 # May contain wildcards.
331 "A String",
332 ],
333 "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing.
334 # Defaults to 10 if not specified.
335 "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for
336 # the model to use.
337 "outputPath": "A String", # Required. The output Google Cloud Storage location.
338 "dataFormat": "A String", # Required. The format of the input data files.
339 "versionName": "A String", # Use this field if you want to specify a version of the model to use. The
340 # string is formatted the same way as `model_version`, with the addition
341 # of the version information:
342 #
343 # `"projects/<var>[YOUR_PROJECT]</var>/models/<var>YOUR_MODEL/versions/<var>[YOUR_VERSION]</var>"`
344 "region": "A String", # Required. The Google Compute Engine region to run the prediction job in.
345 },
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]346 "endTime": "A String", # Output only. When the job processing was completed.
347 "predictionOutput": { # Represents results of a prediction job. # The current prediction job result.
348 "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]349 "nodeHours": 3.14, # Node hours used by the batch prediction job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]350 "predictionCount": "A String", # The number of generated predictions.
351 "errorCount": "A String", # The number of data instances which resulted in errors.
352 },
353 "createTime": "A String", # Output only. When the job was created.
354 }
355
356 x__xgafv: string, V1 error format.
357 Allowed values
358 1 - v1 error format
359 2 - v2 error format
360
361Returns:
362 An object of the form:
363
364 { # Represents a training or prediction job.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]365 "trainingOutput": { # Represents results of a training job. Output only. # The current training job result.
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]366 "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully.
367 # Only set for hyperparameter tuning jobs.
368 "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]369 "trials": [ # Results for individual Hyperparameter trials.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]370 # Only set for hyperparameter tuning jobs.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]371 { # Represents the result of a single hyperparameter tuning trial from a
372 # training job. The TrainingOutput object that is returned on successful
373 # completion of a training job with hyperparameter tuning includes a list
374 # of HyperparameterOutput objects, one for each successful trial.
375 "hyperparameters": { # The hyperparameters given to this trial.
376 "a_key": "A String",
377 },
378 "trialId": "A String", # The trial id for these results.
379 "allMetrics": [ # All recorded object metrics for this trial.
380 { # An observed value of a metric.
381 "trainingStep": "A String", # The global training step for this metric.
382 "objectiveValue": 3.14, # The objective value at this training step.
383 },
384 ],
385 "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial.
386 "trainingStep": "A String", # The global training step for this metric.
387 "objectiveValue": 3.14, # The objective value at this training step.
388 },
389 },
390 ],
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]391 "consumedMLUnits": 3.14, # The amount of ML units consumed by the job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]392 },
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]393 "trainingInput": { # Represents input parameters for a training job. # Input parameters to create a training job.
394 "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training
395 # job's worker nodes.
396 #
397 # The supported values are the same as those described in the entry for
398 # `masterType`.
399 #
400 # This value must be present when `scaleTier` is set to `CUSTOM` and
401 # `workerCount` is greater than zero.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]402 "runtimeVersion": "A String", # Optional. The Google Cloud ML runtime version to use for training. If not
403 # set, Google Cloud ML will choose the latest stable version.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]404 "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers
405 # and parameter servers.
406 "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training
407 # job's master worker.
408 #
409 # The following types are supported:
410 #
411 # <dl>
412 # <dt>standard</dt>
413 # <dd>
414 # A basic machine configuration suitable for training simple models with
415 # small to moderate datasets.
416 # </dd>
417 # <dt>large_model</dt>
418 # <dd>
419 # A machine with a lot of memory, specially suited for parameter servers
420 # when your model is large (having many hidden layers or layers with very
421 # large numbers of nodes).
422 # </dd>
423 # <dt>complex_model_s</dt>
424 # <dd>
425 # A machine suitable for the master and workers of the cluster when your
426 # model requires more computation than the standard machine can handle
427 # satisfactorily.
428 # </dd>
429 # <dt>complex_model_m</dt>
430 # <dd>
431 # A machine with roughly twice the number of cores and roughly double the
432 # memory of <code suppresswarning="true">complex_model_s</code>.
433 # </dd>
434 # <dt>complex_model_l</dt>
435 # <dd>
436 # A machine with roughly twice the number of cores and roughly double the
437 # memory of <code suppresswarning="true">complex_model_m</code>.
438 # </dd>
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]439 # <dt>standard_gpu</dt>
440 # <dd>
441 # A machine equivalent to <code suppresswarning="true">standard</code> that
442 # also includes a
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]443 # <a href="/ml-engine/docs/how-tos/using-gpus">
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]444 # GPU that you can use in your trainer</a>.
445 # </dd>
446 # <dt>complex_model_m_gpu</dt>
447 # <dd>
448 # A machine equivalent to
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]449 # <code suppresswarning="true">complex_model_m</code> that also includes
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]450 # four GPUs.
451 # </dd>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]452 # </dl>
453 #
454 # You must set this value when `scaleTier` is set to `CUSTOM`.
455 "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune.
456 "maxTrials": 42, # Optional. How many training trials should be attempted to optimize
457 # the specified hyperparameters.
458 #
459 # Defaults to one.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]460 "hyperparameterMetricTag": "A String", # Optional. The Tensorflow summary tag name to use for optimizing trials. For
461 # current versions of Tensorflow, this tag name should exactly match what is
462 # shown in Tensorboard, including all scopes. For versions of Tensorflow
463 # prior to 0.12, this should be only the tag passed to tf.Summary.
464 # By default, "training/hptuning/metric" will be used.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]465 "params": [ # Required. The set of parameters to tune.
466 { # Represents a single hyperparameter to optimize.
467 "maxValue": 3.14, # Required if typeis `DOUBLE` or `INTEGER`. This field
468 # should be unset if type is `CATEGORICAL`. This value should be integers if
469 # type is `INTEGER`.
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]470 "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories.
471 "A String",
472 ],
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]473 "discreteValues": [ # Required if type is `DISCRETE`.
474 # A list of feasible points.
475 # The list should be in strictly increasing order. For instance, this
476 # parameter might have possible settings of 1.5, 2.5, and 4.0. This list
477 # should not contain more than 1,000 values.
478 3.14,
479 ],
480 "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in
481 # a HyperparameterSpec message. E.g., "learning_rate".
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]482 "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field
483 # should be unset if type is `CATEGORICAL`. This value should be integers if
484 # type is INTEGER.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]485 "type": "A String", # Required. The type of the parameter.
486 "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube.
487 # Leave unset for categorical parameters.
488 # Some kind of scaling is strongly recommended for real or integral
489 # parameters (e.g., `UNIT_LINEAR_SCALE`).
490 },
491 ],
492 "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently.
493 # You can reduce the time it takes to perform hyperparameter tuning by adding
494 # trials in parallel. However, each trail only benefits from the information
495 # gained in completed trials. That means that a trial does not get access to
496 # the results of trials running at the same time, which could reduce the
497 # quality of the overall optimization.
498 #
499 # Each trial will use the same scale tier and machine types.
500 #
501 # Defaults to one.
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]502 "goal": "A String", # Required. The type of goal to use for tuning. Available types are
503 # `MAXIMIZE` and `MINIMIZE`.
504 #
505 # Defaults to `MAXIMIZE`.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]506 },
507 "region": "A String", # Required. The Google Compute Engine region to run the training job in.
508 "args": [ # Optional. Command line arguments to pass to the program.
509 "A String",
510 ],
511 "pythonModule": "A String", # Required. The Python module name to run after installing the packages.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]512 "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs
513 # and other data needed for training. This path is passed to your TensorFlow
514 # program as the 'job_dir' command-line argument. The benefit of specifying
515 # this field is that Cloud ML validates the path for use in training.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]516 "packageUris": [ # Required. The Google Cloud Storage location of the packages with
517 # the training program and any additional dependencies.
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]518 # The maximum number of package URIs is 100.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]519 "A String",
520 ],
521 "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each
522 # replica in the cluster will be of the type specified in `worker_type`.
523 #
524 # This value can only be used when `scale_tier` is set to `CUSTOM`. If you
525 # set this value, you must also set `worker_type`.
526 "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training
527 # job's parameter server.
528 #
529 # The supported values are the same as those described in the entry for
530 # `master_type`.
531 #
532 # This value must be present when `scaleTier` is set to `CUSTOM` and
533 # `parameter_server_count` is greater than zero.
534 "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training
535 # job. Each replica in the cluster will be of the type specified in
536 # `parameter_server_type`.
537 #
538 # This value can only be used when `scale_tier` is set to `CUSTOM`.If you
539 # set this value, you must also set `parameter_server_type`.
540 },
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]541 "startTime": "A String", # Output only. When the job processing was started.
542 "errorMessage": "A String", # Output only. The details of a failure or a cancellation.
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]543 "jobId": "A String", # Required. The user-specified id of the job.
544 "state": "A String", # Output only. The detailed state of a job.
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]545 "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job.
546 "modelName": "A String", # Use this field if you want to use the default version for the specified
547 # model. The string must use the following format:
548 #
549 # `"projects/<var>[YOUR_PROJECT]</var>/models/<var>[YOUR_MODEL]</var>"`
550 "runtimeVersion": "A String", # Optional. The Google Cloud ML runtime version to use for this batch
551 # prediction. If not set, Google Cloud ML will pick the runtime version used
552 # during the CreateVersion request for this model version, or choose the
553 # latest stable version when model version information is not available
554 # such as when the model is specified by uri.
555 "inputPaths": [ # Required. The Google Cloud Storage location of the input data files.
556 # May contain wildcards.
557 "A String",
558 ],
559 "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing.
560 # Defaults to 10 if not specified.
561 "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for
562 # the model to use.
563 "outputPath": "A String", # Required. The output Google Cloud Storage location.
564 "dataFormat": "A String", # Required. The format of the input data files.
565 "versionName": "A String", # Use this field if you want to specify a version of the model to use. The
566 # string is formatted the same way as `model_version`, with the addition
567 # of the version information:
568 #
569 # `"projects/<var>[YOUR_PROJECT]</var>/models/<var>YOUR_MODEL/versions/<var>[YOUR_VERSION]</var>"`
570 "region": "A String", # Required. The Google Compute Engine region to run the prediction job in.
571 },
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]572 "endTime": "A String", # Output only. When the job processing was completed.
573 "predictionOutput": { # Represents results of a prediction job. # The current prediction job result.
574 "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]575 "nodeHours": 3.14, # Node hours used by the batch prediction job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]576 "predictionCount": "A String", # The number of generated predictions.
577 "errorCount": "A String", # The number of data instances which resulted in errors.
578 },
579 "createTime": "A String", # Output only. When the job was created.
580 }</pre>
581</div>
582
583<div class="method">
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]584 <code class="details" id="get">get(name, x__xgafv=None)</code>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]585 <pre>Describes a job.
586
587Args:
588 name: string, Required. The name of the job to get the description of.
589
590Authorization: requires `Viewer` role on the parent project. (required)
591 x__xgafv: string, V1 error format.
592 Allowed values
593 1 - v1 error format
594 2 - v2 error format
595
596Returns:
597 An object of the form:
598
599 { # Represents a training or prediction job.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]600 "trainingOutput": { # Represents results of a training job. Output only. # The current training job result.
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]601 "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully.
602 # Only set for hyperparameter tuning jobs.
603 "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]604 "trials": [ # Results for individual Hyperparameter trials.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]605 # Only set for hyperparameter tuning jobs.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]606 { # Represents the result of a single hyperparameter tuning trial from a
607 # training job. The TrainingOutput object that is returned on successful
608 # completion of a training job with hyperparameter tuning includes a list
609 # of HyperparameterOutput objects, one for each successful trial.
610 "hyperparameters": { # The hyperparameters given to this trial.
611 "a_key": "A String",
612 },
613 "trialId": "A String", # The trial id for these results.
614 "allMetrics": [ # All recorded object metrics for this trial.
615 { # An observed value of a metric.
616 "trainingStep": "A String", # The global training step for this metric.
617 "objectiveValue": 3.14, # The objective value at this training step.
618 },
619 ],
620 "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial.
621 "trainingStep": "A String", # The global training step for this metric.
622 "objectiveValue": 3.14, # The objective value at this training step.
623 },
624 },
625 ],
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]626 "consumedMLUnits": 3.14, # The amount of ML units consumed by the job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]627 },
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]628 "trainingInput": { # Represents input parameters for a training job. # Input parameters to create a training job.
629 "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training
630 # job's worker nodes.
631 #
632 # The supported values are the same as those described in the entry for
633 # `masterType`.
634 #
635 # This value must be present when `scaleTier` is set to `CUSTOM` and
636 # `workerCount` is greater than zero.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]637 "runtimeVersion": "A String", # Optional. The Google Cloud ML runtime version to use for training. If not
638 # set, Google Cloud ML will choose the latest stable version.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]639 "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers
640 # and parameter servers.
641 "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training
642 # job's master worker.
643 #
644 # The following types are supported:
645 #
646 # <dl>
647 # <dt>standard</dt>
648 # <dd>
649 # A basic machine configuration suitable for training simple models with
650 # small to moderate datasets.
651 # </dd>
652 # <dt>large_model</dt>
653 # <dd>
654 # A machine with a lot of memory, specially suited for parameter servers
655 # when your model is large (having many hidden layers or layers with very
656 # large numbers of nodes).
657 # </dd>
658 # <dt>complex_model_s</dt>
659 # <dd>
660 # A machine suitable for the master and workers of the cluster when your
661 # model requires more computation than the standard machine can handle
662 # satisfactorily.
663 # </dd>
664 # <dt>complex_model_m</dt>
665 # <dd>
666 # A machine with roughly twice the number of cores and roughly double the
667 # memory of <code suppresswarning="true">complex_model_s</code>.
668 # </dd>
669 # <dt>complex_model_l</dt>
670 # <dd>
671 # A machine with roughly twice the number of cores and roughly double the
672 # memory of <code suppresswarning="true">complex_model_m</code>.
673 # </dd>
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]674 # <dt>standard_gpu</dt>
675 # <dd>
676 # A machine equivalent to <code suppresswarning="true">standard</code> that
677 # also includes a
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]678 # <a href="/ml-engine/docs/how-tos/using-gpus">
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]679 # GPU that you can use in your trainer</a>.
680 # </dd>
681 # <dt>complex_model_m_gpu</dt>
682 # <dd>
683 # A machine equivalent to
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]684 # <code suppresswarning="true">complex_model_m</code> that also includes
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]685 # four GPUs.
686 # </dd>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]687 # </dl>
688 #
689 # You must set this value when `scaleTier` is set to `CUSTOM`.
690 "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune.
691 "maxTrials": 42, # Optional. How many training trials should be attempted to optimize
692 # the specified hyperparameters.
693 #
694 # Defaults to one.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]695 "hyperparameterMetricTag": "A String", # Optional. The Tensorflow summary tag name to use for optimizing trials. For
696 # current versions of Tensorflow, this tag name should exactly match what is
697 # shown in Tensorboard, including all scopes. For versions of Tensorflow
698 # prior to 0.12, this should be only the tag passed to tf.Summary.
699 # By default, "training/hptuning/metric" will be used.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]700 "params": [ # Required. The set of parameters to tune.
701 { # Represents a single hyperparameter to optimize.
702 "maxValue": 3.14, # Required if typeis `DOUBLE` or `INTEGER`. This field
703 # should be unset if type is `CATEGORICAL`. This value should be integers if
704 # type is `INTEGER`.
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]705 "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories.
706 "A String",
707 ],
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]708 "discreteValues": [ # Required if type is `DISCRETE`.
709 # A list of feasible points.
710 # The list should be in strictly increasing order. For instance, this
711 # parameter might have possible settings of 1.5, 2.5, and 4.0. This list
712 # should not contain more than 1,000 values.
713 3.14,
714 ],
715 "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in
716 # a HyperparameterSpec message. E.g., "learning_rate".
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]717 "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field
718 # should be unset if type is `CATEGORICAL`. This value should be integers if
719 # type is INTEGER.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]720 "type": "A String", # Required. The type of the parameter.
721 "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube.
722 # Leave unset for categorical parameters.
723 # Some kind of scaling is strongly recommended for real or integral
724 # parameters (e.g., `UNIT_LINEAR_SCALE`).
725 },
726 ],
727 "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently.
728 # You can reduce the time it takes to perform hyperparameter tuning by adding
729 # trials in parallel. However, each trail only benefits from the information
730 # gained in completed trials. That means that a trial does not get access to
731 # the results of trials running at the same time, which could reduce the
732 # quality of the overall optimization.
733 #
734 # Each trial will use the same scale tier and machine types.
735 #
736 # Defaults to one.
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]737 "goal": "A String", # Required. The type of goal to use for tuning. Available types are
738 # `MAXIMIZE` and `MINIMIZE`.
739 #
740 # Defaults to `MAXIMIZE`.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]741 },
742 "region": "A String", # Required. The Google Compute Engine region to run the training job in.
743 "args": [ # Optional. Command line arguments to pass to the program.
744 "A String",
745 ],
746 "pythonModule": "A String", # Required. The Python module name to run after installing the packages.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]747 "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs
748 # and other data needed for training. This path is passed to your TensorFlow
749 # program as the 'job_dir' command-line argument. The benefit of specifying
750 # this field is that Cloud ML validates the path for use in training.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]751 "packageUris": [ # Required. The Google Cloud Storage location of the packages with
752 # the training program and any additional dependencies.
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]753 # The maximum number of package URIs is 100.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]754 "A String",
755 ],
756 "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each
757 # replica in the cluster will be of the type specified in `worker_type`.
758 #
759 # This value can only be used when `scale_tier` is set to `CUSTOM`. If you
760 # set this value, you must also set `worker_type`.
761 "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training
762 # job's parameter server.
763 #
764 # The supported values are the same as those described in the entry for
765 # `master_type`.
766 #
767 # This value must be present when `scaleTier` is set to `CUSTOM` and
768 # `parameter_server_count` is greater than zero.
769 "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training
770 # job. Each replica in the cluster will be of the type specified in
771 # `parameter_server_type`.
772 #
773 # This value can only be used when `scale_tier` is set to `CUSTOM`.If you
774 # set this value, you must also set `parameter_server_type`.
775 },
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]776 "startTime": "A String", # Output only. When the job processing was started.
777 "errorMessage": "A String", # Output only. The details of a failure or a cancellation.
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]778 "jobId": "A String", # Required. The user-specified id of the job.
779 "state": "A String", # Output only. The detailed state of a job.
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]780 "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job.
781 "modelName": "A String", # Use this field if you want to use the default version for the specified
782 # model. The string must use the following format:
783 #
784 # `"projects/<var>[YOUR_PROJECT]</var>/models/<var>[YOUR_MODEL]</var>"`
785 "runtimeVersion": "A String", # Optional. The Google Cloud ML runtime version to use for this batch
786 # prediction. If not set, Google Cloud ML will pick the runtime version used
787 # during the CreateVersion request for this model version, or choose the
788 # latest stable version when model version information is not available
789 # such as when the model is specified by uri.
790 "inputPaths": [ # Required. The Google Cloud Storage location of the input data files.
791 # May contain wildcards.
792 "A String",
793 ],
794 "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing.
795 # Defaults to 10 if not specified.
796 "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for
797 # the model to use.
798 "outputPath": "A String", # Required. The output Google Cloud Storage location.
799 "dataFormat": "A String", # Required. The format of the input data files.
800 "versionName": "A String", # Use this field if you want to specify a version of the model to use. The
801 # string is formatted the same way as `model_version`, with the addition
802 # of the version information:
803 #
804 # `"projects/<var>[YOUR_PROJECT]</var>/models/<var>YOUR_MODEL/versions/<var>[YOUR_VERSION]</var>"`
805 "region": "A String", # Required. The Google Compute Engine region to run the prediction job in.
806 },
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]807 "endTime": "A String", # Output only. When the job processing was completed.
808 "predictionOutput": { # Represents results of a prediction job. # The current prediction job result.
809 "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]810 "nodeHours": 3.14, # Node hours used by the batch prediction job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]811 "predictionCount": "A String", # The number of generated predictions.
812 "errorCount": "A String", # The number of data instances which resulted in errors.
813 },
814 "createTime": "A String", # Output only. When the job was created.
815 }</pre>
816</div>
817
818<div class="method">
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]819 <code class="details" id="list">list(parent, pageSize=None, filter=None, pageToken=None, x__xgafv=None)</code>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]820 <pre>Lists the jobs in the project.
821
822Args:
823 parent: string, Required. The name of the project for which to list jobs.
824
825Authorization: requires `Viewer` role on the specified project. (required)
826 pageSize: integer, Optional. The number of jobs to retrieve per "page" of results. If there
827are more remaining results than this number, the response message will
828contain a valid value in the `next_page_token` field.
829
830The default value is 20, and the maximum page size is 100.
831 filter: string, Optional. Specifies the subset of jobs to retrieve.
832 pageToken: string, Optional. A page token to request the next page of results.
833
834You get the token from the `next_page_token` field of the response from
835the previous call.
836 x__xgafv: string, V1 error format.
837 Allowed values
838 1 - v1 error format
839 2 - v2 error format
840
841Returns:
842 An object of the form:
843
844 { # Response message for the ListJobs method.
845 "nextPageToken": "A String", # Optional. Pass this token as the `page_token` field of the request for a
846 # subsequent call.
847 "jobs": [ # The list of jobs.
848 { # Represents a training or prediction job.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]849 "trainingOutput": { # Represents results of a training job. Output only. # The current training job result.
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]850 "completedTrialCount": "A String", # The number of hyperparameter tuning trials that completed successfully.
851 # Only set for hyperparameter tuning jobs.
852 "isHyperparameterTuningJob": True or False, # Whether this job is a hyperparameter tuning job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]853 "trials": [ # Results for individual Hyperparameter trials.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]854 # Only set for hyperparameter tuning jobs.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]855 { # Represents the result of a single hyperparameter tuning trial from a
856 # training job. The TrainingOutput object that is returned on successful
857 # completion of a training job with hyperparameter tuning includes a list
858 # of HyperparameterOutput objects, one for each successful trial.
859 "hyperparameters": { # The hyperparameters given to this trial.
860 "a_key": "A String",
861 },
862 "trialId": "A String", # The trial id for these results.
863 "allMetrics": [ # All recorded object metrics for this trial.
864 { # An observed value of a metric.
865 "trainingStep": "A String", # The global training step for this metric.
866 "objectiveValue": 3.14, # The objective value at this training step.
867 },
868 ],
869 "finalMetric": { # An observed value of a metric. # The final objective metric seen for this trial.
870 "trainingStep": "A String", # The global training step for this metric.
871 "objectiveValue": 3.14, # The objective value at this training step.
872 },
873 },
874 ],
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]875 "consumedMLUnits": 3.14, # The amount of ML units consumed by the job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]876 },
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]877 "trainingInput": { # Represents input parameters for a training job. # Input parameters to create a training job.
878 "workerType": "A String", # Optional. Specifies the type of virtual machine to use for your training
879 # job's worker nodes.
880 #
881 # The supported values are the same as those described in the entry for
882 # `masterType`.
883 #
884 # This value must be present when `scaleTier` is set to `CUSTOM` and
885 # `workerCount` is greater than zero.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]886 "runtimeVersion": "A String", # Optional. The Google Cloud ML runtime version to use for training. If not
887 # set, Google Cloud ML will choose the latest stable version.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]888 "scaleTier": "A String", # Required. Specifies the machine types, the number of replicas for workers
889 # and parameter servers.
890 "masterType": "A String", # Optional. Specifies the type of virtual machine to use for your training
891 # job's master worker.
892 #
893 # The following types are supported:
894 #
895 # <dl>
896 # <dt>standard</dt>
897 # <dd>
898 # A basic machine configuration suitable for training simple models with
899 # small to moderate datasets.
900 # </dd>
901 # <dt>large_model</dt>
902 # <dd>
903 # A machine with a lot of memory, specially suited for parameter servers
904 # when your model is large (having many hidden layers or layers with very
905 # large numbers of nodes).
906 # </dd>
907 # <dt>complex_model_s</dt>
908 # <dd>
909 # A machine suitable for the master and workers of the cluster when your
910 # model requires more computation than the standard machine can handle
911 # satisfactorily.
912 # </dd>
913 # <dt>complex_model_m</dt>
914 # <dd>
915 # A machine with roughly twice the number of cores and roughly double the
916 # memory of <code suppresswarning="true">complex_model_s</code>.
917 # </dd>
918 # <dt>complex_model_l</dt>
919 # <dd>
920 # A machine with roughly twice the number of cores and roughly double the
921 # memory of <code suppresswarning="true">complex_model_m</code>.
922 # </dd>
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]923 # <dt>standard_gpu</dt>
924 # <dd>
925 # A machine equivalent to <code suppresswarning="true">standard</code> that
926 # also includes a
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]927 # <a href="/ml-engine/docs/how-tos/using-gpus">
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]928 # GPU that you can use in your trainer</a>.
929 # </dd>
930 # <dt>complex_model_m_gpu</dt>
931 # <dd>
932 # A machine equivalent to
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]933 # <code suppresswarning="true">complex_model_m</code> that also includes
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]934 # four GPUs.
935 # </dd>
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]936 # </dl>
937 #
938 # You must set this value when `scaleTier` is set to `CUSTOM`.
939 "hyperparameters": { # Represents a set of hyperparameters to optimize. # Optional. The set of Hyperparameters to tune.
940 "maxTrials": 42, # Optional. How many training trials should be attempted to optimize
941 # the specified hyperparameters.
942 #
943 # Defaults to one.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]944 "hyperparameterMetricTag": "A String", # Optional. The Tensorflow summary tag name to use for optimizing trials. For
945 # current versions of Tensorflow, this tag name should exactly match what is
946 # shown in Tensorboard, including all scopes. For versions of Tensorflow
947 # prior to 0.12, this should be only the tag passed to tf.Summary.
948 # By default, "training/hptuning/metric" will be used.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]949 "params": [ # Required. The set of parameters to tune.
950 { # Represents a single hyperparameter to optimize.
951 "maxValue": 3.14, # Required if typeis `DOUBLE` or `INTEGER`. This field
952 # should be unset if type is `CATEGORICAL`. This value should be integers if
953 # type is `INTEGER`.
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]954 "categoricalValues": [ # Required if type is `CATEGORICAL`. The list of possible categories.
955 "A String",
956 ],
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]957 "discreteValues": [ # Required if type is `DISCRETE`.
958 # A list of feasible points.
959 # The list should be in strictly increasing order. For instance, this
960 # parameter might have possible settings of 1.5, 2.5, and 4.0. This list
961 # should not contain more than 1,000 values.
962 3.14,
963 ],
964 "parameterName": "A String", # Required. The parameter name must be unique amongst all ParameterConfigs in
965 # a HyperparameterSpec message. E.g., "learning_rate".
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]966 "minValue": 3.14, # Required if type is `DOUBLE` or `INTEGER`. This field
967 # should be unset if type is `CATEGORICAL`. This value should be integers if
968 # type is INTEGER.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]969 "type": "A String", # Required. The type of the parameter.
970 "scaleType": "A String", # Optional. How the parameter should be scaled to the hypercube.
971 # Leave unset for categorical parameters.
972 # Some kind of scaling is strongly recommended for real or integral
973 # parameters (e.g., `UNIT_LINEAR_SCALE`).
974 },
975 ],
976 "maxParallelTrials": 42, # Optional. The number of training trials to run concurrently.
977 # You can reduce the time it takes to perform hyperparameter tuning by adding
978 # trials in parallel. However, each trail only benefits from the information
979 # gained in completed trials. That means that a trial does not get access to
980 # the results of trials running at the same time, which could reduce the
981 # quality of the overall optimization.
982 #
983 # Each trial will use the same scale tier and machine types.
984 #
985 # Defaults to one.
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]986 "goal": "A String", # Required. The type of goal to use for tuning. Available types are
987 # `MAXIMIZE` and `MINIMIZE`.
988 #
989 # Defaults to `MAXIMIZE`.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]990 },
991 "region": "A String", # Required. The Google Compute Engine region to run the training job in.
992 "args": [ # Optional. Command line arguments to pass to the program.
993 "A String",
994 ],
995 "pythonModule": "A String", # Required. The Python module name to run after installing the packages.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]996 "jobDir": "A String", # Optional. A Google Cloud Storage path in which to store training outputs
997 # and other data needed for training. This path is passed to your TensorFlow
998 # program as the 'job_dir' command-line argument. The benefit of specifying
999 # this field is that Cloud ML validates the path for use in training.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]1000 "packageUris": [ # Required. The Google Cloud Storage location of the packages with
1001 # the training program and any additional dependencies.
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]1002 # The maximum number of package URIs is 100.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]1003 "A String",
1004 ],
1005 "workerCount": "A String", # Optional. The number of worker replicas to use for the training job. Each
1006 # replica in the cluster will be of the type specified in `worker_type`.
1007 #
1008 # This value can only be used when `scale_tier` is set to `CUSTOM`. If you
1009 # set this value, you must also set `worker_type`.
1010 "parameterServerType": "A String", # Optional. Specifies the type of virtual machine to use for your training
1011 # job's parameter server.
1012 #
1013 # The supported values are the same as those described in the entry for
1014 # `master_type`.
1015 #
1016 # This value must be present when `scaleTier` is set to `CUSTOM` and
1017 # `parameter_server_count` is greater than zero.
1018 "parameterServerCount": "A String", # Optional. The number of parameter server replicas to use for the training
1019 # job. Each replica in the cluster will be of the type specified in
1020 # `parameter_server_type`.
1021 #
1022 # This value can only be used when `scale_tier` is set to `CUSTOM`.If you
1023 # set this value, you must also set `parameter_server_type`.
1024 },
Sai Cheemalapati4ba8c232017-06-06 18:46:08 -0400[diff] [blame]1025 "startTime": "A String", # Output only. When the job processing was started.
1026 "errorMessage": "A String", # Output only. The details of a failure or a cancellation.
Thomas Coffee2f245372017-03-27 10:39:26 -0700[diff] [blame]1027 "jobId": "A String", # Required. The user-specified id of the job.
1028 "state": "A String", # Output only. The detailed state of a job.
Sai Cheemalapatie833b792017-03-24 15:06:46 -0700[diff] [blame]1029 "predictionInput": { # Represents input parameters for a prediction job. # Input parameters to create a prediction job.
1030 "modelName": "A String", # Use this field if you want to use the default version for the specified
1031 # model. The string must use the following format:
1032 #
1033 # `"projects/<var>[YOUR_PROJECT]</var>/models/<var>[YOUR_MODEL]</var>"`
1034 "runtimeVersion": "A String", # Optional. The Google Cloud ML runtime version to use for this batch
1035 # prediction. If not set, Google Cloud ML will pick the runtime version used
1036 # during the CreateVersion request for this model version, or choose the
1037 # latest stable version when model version information is not available
1038 # such as when the model is specified by uri.
1039 "inputPaths": [ # Required. The Google Cloud Storage location of the input data files.
1040 # May contain wildcards.
1041 "A String",
1042 ],
1043 "maxWorkerCount": "A String", # Optional. The maximum number of workers to be used for parallel processing.
1044 # Defaults to 10 if not specified.
1045 "uri": "A String", # Use this field if you want to specify a Google Cloud Storage path for
1046 # the model to use.
1047 "outputPath": "A String", # Required. The output Google Cloud Storage location.
1048 "dataFormat": "A String", # Required. The format of the input data files.
1049 "versionName": "A String", # Use this field if you want to specify a version of the model to use. The
1050 # string is formatted the same way as `model_version`, with the addition
1051 # of the version information:
1052 #
1053 # `"projects/<var>[YOUR_PROJECT]</var>/models/<var>YOUR_MODEL/versions/<var>[YOUR_VERSION]</var>"`
1054 "region": "A String", # Required. The Google Compute Engine region to run the prediction job in.
1055 },
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]1056 "endTime": "A String", # Output only. When the job processing was completed.
1057 "predictionOutput": { # Represents results of a prediction job. # The current prediction job result.
1058 "outputPath": "A String", # The output Google Cloud Storage location provided at the job creation time.
Sai Cheemalapatic30d2b52017-03-13 12:12:03 -0400[diff] [blame]1059 "nodeHours": 3.14, # Node hours used by the batch prediction job.
Sai Cheemalapatiea3a5e12016-10-12 14:05:53 -0700[diff] [blame]1060 "predictionCount": "A String", # The number of generated predictions.
1061 "errorCount": "A String", # The number of data instances which resulted in errors.
1062 },
1063 "createTime": "A String", # Output only. When the job was created.
1064 },
1065 ],
1066 }</pre>
1067</div>
1068
1069<div class="method">
1070 <code class="details" id="list_next">list_next(previous_request, previous_response)</code>
1071 <pre>Retrieves the next page of results.
1072
1073Args:
1074 previous_request: The request for the previous page. (required)
1075 previous_response: The response from the request for the previous page. (required)
1076
1077Returns:
1078 A request object that you can call 'execute()' on to request the next
1079 page. Returns None if there are no more items in the collection.
1080 </pre>
1081</div>
1082
1083</body></html>