aws-sdk-sagemaker 1.216.0 → 1.217.0

Sign up to get free protection for your applications and to get access to all the features.
checksums.yaml CHANGED
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  SHA256:
3
- metadata.gz: ca75f03989ceecfdd619b9896e6af3dbb0554dd81ea98c3f0f6fac9be41493e2
4
- data.tar.gz: 8c9201413b4e893def08f750d9b561ac560fd77186c1a851fbb54e221c367886
3
+ metadata.gz: 44e7e1f63bcf45a9e3c3c59aa6cc3ff1c72d5dcd343e8fd05a63fe85a95375e6
4
+ data.tar.gz: 5b10df777ead8d3feb98e0b95951a94ca581285b2d9cdddbbd70248f863b6990
5
5
  SHA512:
6
- metadata.gz: ad24efe1b58d1d0a14a2c7d2e4511b0556674c998b105b84e5e786e81a4666110f9e9d21da5c48c6832c811345f51c10c0416a415aebffa90bfcff2313052d14
7
- data.tar.gz: 2a6978535604755b73ab331ac9ee44bde9f9e8bb28d26da37e4faaf457a6e7fd3261d886f96d9738848cda3af540c062e553ddd6ebf898acbd7fe9f462a0c0f8
6
+ metadata.gz: 9393949f989c2a943f2c9a616655e3cc40cc36e6c7568cb5cd513378887862dde9f253cddecd89ec670527fac0953a8e0da6a527eca42f4300e43a292a948286
7
+ data.tar.gz: f40cd05d91ebe24815c853ed1043b09f4274a08ed4c23980cea7cd53b35f703af2531b660ac78e5b9d10c2a76143b95f20759516d024df7cda3214d9c0ea2643
data/CHANGELOG.md CHANGED
@@ -1,6 +1,11 @@
1
1
  Unreleased Changes
2
2
  ------------------
3
3
 
4
+ 1.217.0 (2023-10-26)
5
+ ------------------
6
+
7
+ * Feature - Amazon Sagemaker Autopilot now supports Text Generation jobs.
8
+
4
9
  1.216.0 (2023-10-12)
5
10
  ------------------
6
11
 
data/VERSION CHANGED
@@ -1 +1 @@
1
- 1.216.0
1
+ 1.217.0
@@ -1221,8 +1221,8 @@ module Aws::SageMaker
1221
1221
  #
1222
1222
  # `CreateAutoMLJobV2` can manage tabular problem types identical to
1223
1223
  # those of its previous version `CreateAutoMLJob`, as well as
1224
- # time-series forecasting, and non-tabular problem types such as image
1225
- # or text classification.
1224
+ # time-series forecasting, non-tabular problem types such as image or
1225
+ # text classification, and text generation (LLMs fine-tuning).
1226
1226
  #
1227
1227
  # Find guidelines about how to migrate a `CreateAutoMLJob` to
1228
1228
  # `CreateAutoMLJobV2` in [Migrate a CreateAutoMLJob to
@@ -1401,8 +1401,8 @@ module Aws::SageMaker
1401
1401
  #
1402
1402
  # `CreateAutoMLJobV2` can manage tabular problem types identical to
1403
1403
  # those of its previous version `CreateAutoMLJob`, as well as
1404
- # time-series forecasting, and non-tabular problem types such as image
1405
- # or text classification.
1404
+ # time-series forecasting, non-tabular problem types such as image or
1405
+ # text classification, and text generation (LLMs fine-tuning).
1406
1406
  #
1407
1407
  # Find guidelines about how to migrate a `CreateAutoMLJob` to
1408
1408
  # `CreateAutoMLJobV2` in [Migrate a CreateAutoMLJob to
@@ -1444,6 +1444,8 @@ module Aws::SageMaker
1444
1444
  #
1445
1445
  # * For time-series forecasting: `S3Prefix`.
1446
1446
  #
1447
+ # * For text generation (LLMs fine-tuning): `S3Prefix`.
1448
+ #
1447
1449
  #
1448
1450
  #
1449
1451
  # [1]: https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_CreateAutoMLJob.html#sagemaker-CreateAutoMLJob-request-InputDataConfig
@@ -1479,16 +1481,27 @@ module Aws::SageMaker
1479
1481
  # type. For the list of default values per problem type, see
1480
1482
  # [AutoMLJobObjective][1].
1481
1483
  #
1482
- # <note markdown="1"> For tabular problem types, you must either provide both the
1483
- # `AutoMLJobObjective` and indicate the type of supervised learning
1484
- # problem in `AutoMLProblemTypeConfig` (`TabularJobConfig.ProblemType`),
1485
- # or none at all.
1484
+ # <note markdown="1"> * For tabular problem types: You must either provide both the
1485
+ # `AutoMLJobObjective` and indicate the type of supervised learning
1486
+ # problem in `AutoMLProblemTypeConfig`
1487
+ # (`TabularJobConfig.ProblemType`), or none at all.
1488
+ #
1489
+ # * For text generation problem types (LLMs fine-tuning): Fine-tuning
1490
+ # language models in Autopilot does not require setting the
1491
+ # `AutoMLJobObjective` field. Autopilot fine-tunes LLMs without
1492
+ # requiring multiple candidates to be trained and evaluated. Instead,
1493
+ # using your dataset, Autopilot directly fine-tunes your target model
1494
+ # to enhance a default objective metric, the cross-entropy loss. After
1495
+ # fine-tuning a language model, you can evaluate the quality of its
1496
+ # generated text using different metrics. For a list of the available
1497
+ # metrics, see [Metrics for fine-tuning LLMs in Autopilot][2].
1486
1498
  #
1487
1499
  # </note>
1488
1500
  #
1489
1501
  #
1490
1502
  #
1491
1503
  # [1]: https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_AutoMLJobObjective.html
1504
+ # [2]: https://docs.aws.amazon.com/sagemaker/latest/dg/llms-finetuning-models.html
1492
1505
  #
1493
1506
  # @option params [Types::ModelDeployConfig] :model_deploy_config
1494
1507
  # Specifies how to generate the endpoint name for an automatic one-click
@@ -1602,6 +1615,14 @@ module Aws::SageMaker
1602
1615
  # },
1603
1616
  # ],
1604
1617
  # },
1618
+ # text_generation_job_config: {
1619
+ # completion_criteria: {
1620
+ # max_candidates: 1,
1621
+ # max_runtime_per_training_job_in_seconds: 1,
1622
+ # max_auto_ml_job_runtime_in_seconds: 1,
1623
+ # },
1624
+ # base_model_name: "BaseModelName",
1625
+ # },
1605
1626
  # },
1606
1627
  # role_arn: "RoleArn", # required
1607
1628
  # tags: [
@@ -10314,7 +10335,7 @@ module Aws::SageMaker
10314
10335
  # resp.best_candidate.candidate_properties.candidate_metrics[0].metric_name #=> String, one of "Accuracy", "MSE", "F1", "F1macro", "AUC", "RMSE", "MAE", "R2", "BalancedAccuracy", "Precision", "PrecisionMacro", "Recall", "RecallMacro", "MAPE", "MASE", "WAPE", "AverageWeightedQuantileLoss"
10315
10336
  # resp.best_candidate.candidate_properties.candidate_metrics[0].value #=> Float
10316
10337
  # resp.best_candidate.candidate_properties.candidate_metrics[0].set #=> String, one of "Train", "Validation", "Test"
10317
- # resp.best_candidate.candidate_properties.candidate_metrics[0].standard_metric_name #=> String, one of "Accuracy", "MSE", "F1", "F1macro", "AUC", "RMSE", "MAE", "R2", "BalancedAccuracy", "Precision", "PrecisionMacro", "Recall", "RecallMacro", "LogLoss", "InferenceLatency", "MAPE", "MASE", "WAPE", "AverageWeightedQuantileLoss"
10338
+ # resp.best_candidate.candidate_properties.candidate_metrics[0].standard_metric_name #=> String, one of "Accuracy", "MSE", "F1", "F1macro", "AUC", "RMSE", "MAE", "R2", "BalancedAccuracy", "Precision", "PrecisionMacro", "Recall", "RecallMacro", "LogLoss", "InferenceLatency", "MAPE", "MASE", "WAPE", "AverageWeightedQuantileLoss", "Rouge1", "Rouge2", "RougeL", "RougeLSum", "Perplexity", "ValidationLoss", "TrainingLoss"
10318
10339
  # resp.best_candidate.inference_container_definitions #=> Hash
10319
10340
  # resp.best_candidate.inference_container_definitions["AutoMLProcessingUnit"] #=> Array
10320
10341
  # resp.best_candidate.inference_container_definitions["AutoMLProcessingUnit"][0].image #=> String
@@ -10440,6 +10461,10 @@ module Aws::SageMaker
10440
10461
  # resp.auto_ml_problem_type_config.time_series_forecasting_job_config.time_series_config.grouping_attribute_names[0] #=> String
10441
10462
  # resp.auto_ml_problem_type_config.time_series_forecasting_job_config.holiday_config #=> Array
10442
10463
  # resp.auto_ml_problem_type_config.time_series_forecasting_job_config.holiday_config[0].country_code #=> String
10464
+ # resp.auto_ml_problem_type_config.text_generation_job_config.completion_criteria.max_candidates #=> Integer
10465
+ # resp.auto_ml_problem_type_config.text_generation_job_config.completion_criteria.max_runtime_per_training_job_in_seconds #=> Integer
10466
+ # resp.auto_ml_problem_type_config.text_generation_job_config.completion_criteria.max_auto_ml_job_runtime_in_seconds #=> Integer
10467
+ # resp.auto_ml_problem_type_config.text_generation_job_config.base_model_name #=> String
10443
10468
  # resp.creation_time #=> Time
10444
10469
  # resp.end_time #=> Time
10445
10470
  # resp.last_modified_time #=> Time
@@ -10473,7 +10498,7 @@ module Aws::SageMaker
10473
10498
  # resp.best_candidate.candidate_properties.candidate_metrics[0].metric_name #=> String, one of "Accuracy", "MSE", "F1", "F1macro", "AUC", "RMSE", "MAE", "R2", "BalancedAccuracy", "Precision", "PrecisionMacro", "Recall", "RecallMacro", "MAPE", "MASE", "WAPE", "AverageWeightedQuantileLoss"
10474
10499
  # resp.best_candidate.candidate_properties.candidate_metrics[0].value #=> Float
10475
10500
  # resp.best_candidate.candidate_properties.candidate_metrics[0].set #=> String, one of "Train", "Validation", "Test"
10476
- # resp.best_candidate.candidate_properties.candidate_metrics[0].standard_metric_name #=> String, one of "Accuracy", "MSE", "F1", "F1macro", "AUC", "RMSE", "MAE", "R2", "BalancedAccuracy", "Precision", "PrecisionMacro", "Recall", "RecallMacro", "LogLoss", "InferenceLatency", "MAPE", "MASE", "WAPE", "AverageWeightedQuantileLoss"
10501
+ # resp.best_candidate.candidate_properties.candidate_metrics[0].standard_metric_name #=> String, one of "Accuracy", "MSE", "F1", "F1macro", "AUC", "RMSE", "MAE", "R2", "BalancedAccuracy", "Precision", "PrecisionMacro", "Recall", "RecallMacro", "LogLoss", "InferenceLatency", "MAPE", "MASE", "WAPE", "AverageWeightedQuantileLoss", "Rouge1", "Rouge2", "RougeL", "RougeLSum", "Perplexity", "ValidationLoss", "TrainingLoss"
10477
10502
  # resp.best_candidate.inference_container_definitions #=> Hash
10478
10503
  # resp.best_candidate.inference_container_definitions["AutoMLProcessingUnit"] #=> Array
10479
10504
  # resp.best_candidate.inference_container_definitions["AutoMLProcessingUnit"][0].image #=> String
@@ -10499,7 +10524,8 @@ module Aws::SageMaker
10499
10524
  # resp.resolved_attributes.completion_criteria.max_runtime_per_training_job_in_seconds #=> Integer
10500
10525
  # resp.resolved_attributes.completion_criteria.max_auto_ml_job_runtime_in_seconds #=> Integer
10501
10526
  # resp.resolved_attributes.auto_ml_problem_type_resolved_attributes.tabular_resolved_attributes.problem_type #=> String, one of "BinaryClassification", "MulticlassClassification", "Regression"
10502
- # resp.auto_ml_problem_type_config_name #=> String, one of "ImageClassification", "TextClassification", "Tabular", "TimeSeriesForecasting"
10527
+ # resp.resolved_attributes.auto_ml_problem_type_resolved_attributes.text_generation_resolved_attributes.base_model_name #=> String
10528
+ # resp.auto_ml_problem_type_config_name #=> String, one of "ImageClassification", "TextClassification", "Tabular", "TimeSeriesForecasting", "TextGeneration"
10503
10529
  #
10504
10530
  # @see http://docs.aws.amazon.com/goto/WebAPI/sagemaker-2017-07-24/DescribeAutoMLJobV2 AWS API Documentation
10505
10531
  #
@@ -15888,7 +15914,7 @@ module Aws::SageMaker
15888
15914
  # resp.candidates[0].candidate_properties.candidate_metrics[0].metric_name #=> String, one of "Accuracy", "MSE", "F1", "F1macro", "AUC", "RMSE", "MAE", "R2", "BalancedAccuracy", "Precision", "PrecisionMacro", "Recall", "RecallMacro", "MAPE", "MASE", "WAPE", "AverageWeightedQuantileLoss"
15889
15915
  # resp.candidates[0].candidate_properties.candidate_metrics[0].value #=> Float
15890
15916
  # resp.candidates[0].candidate_properties.candidate_metrics[0].set #=> String, one of "Train", "Validation", "Test"
15891
- # resp.candidates[0].candidate_properties.candidate_metrics[0].standard_metric_name #=> String, one of "Accuracy", "MSE", "F1", "F1macro", "AUC", "RMSE", "MAE", "R2", "BalancedAccuracy", "Precision", "PrecisionMacro", "Recall", "RecallMacro", "LogLoss", "InferenceLatency", "MAPE", "MASE", "WAPE", "AverageWeightedQuantileLoss"
15917
+ # resp.candidates[0].candidate_properties.candidate_metrics[0].standard_metric_name #=> String, one of "Accuracy", "MSE", "F1", "F1macro", "AUC", "RMSE", "MAE", "R2", "BalancedAccuracy", "Precision", "PrecisionMacro", "Recall", "RecallMacro", "LogLoss", "InferenceLatency", "MAPE", "MASE", "WAPE", "AverageWeightedQuantileLoss", "Rouge1", "Rouge2", "RougeL", "RougeLSum", "Perplexity", "ValidationLoss", "TrainingLoss"
15892
15918
  # resp.candidates[0].inference_container_definitions #=> Hash
15893
15919
  # resp.candidates[0].inference_container_definitions["AutoMLProcessingUnit"] #=> Array
15894
15920
  # resp.candidates[0].inference_container_definitions["AutoMLProcessingUnit"][0].image #=> String
@@ -24267,7 +24293,7 @@ module Aws::SageMaker
24267
24293
  params: params,
24268
24294
  config: config)
24269
24295
  context[:gem_name] = 'aws-sdk-sagemaker'
24270
- context[:gem_version] = '1.216.0'
24296
+ context[:gem_version] = '1.217.0'
24271
24297
  Seahorse::Client::Request.new(handlers, context)
24272
24298
  end
24273
24299
 
@@ -157,6 +157,7 @@ module Aws::SageMaker
157
157
  AutotuneMode = Shapes::StringShape.new(name: 'AutotuneMode')
158
158
  AwsManagedHumanLoopRequestSource = Shapes::StringShape.new(name: 'AwsManagedHumanLoopRequestSource')
159
159
  BacktestResultsLocation = Shapes::StringShape.new(name: 'BacktestResultsLocation')
160
+ BaseModelName = Shapes::StringShape.new(name: 'BaseModelName')
160
161
  BatchDataCaptureConfig = Shapes::StructureShape.new(name: 'BatchDataCaptureConfig')
161
162
  BatchDescribeModelPackageError = Shapes::StructureShape.new(name: 'BatchDescribeModelPackageError')
162
163
  BatchDescribeModelPackageErrorMap = Shapes::MapShape.new(name: 'BatchDescribeModelPackageErrorMap')
@@ -1893,6 +1894,8 @@ module Aws::SageMaker
1893
1894
  TenthFractionsOfACent = Shapes::IntegerShape.new(name: 'TenthFractionsOfACent')
1894
1895
  TerminationWaitInSeconds = Shapes::IntegerShape.new(name: 'TerminationWaitInSeconds')
1895
1896
  TextClassificationJobConfig = Shapes::StructureShape.new(name: 'TextClassificationJobConfig')
1897
+ TextGenerationJobConfig = Shapes::StructureShape.new(name: 'TextGenerationJobConfig')
1898
+ TextGenerationResolvedAttributes = Shapes::StructureShape.new(name: 'TextGenerationResolvedAttributes')
1896
1899
  ThingName = Shapes::StringShape.new(name: 'ThingName')
1897
1900
  TimeSeriesConfig = Shapes::StructureShape.new(name: 'TimeSeriesConfig')
1898
1901
  TimeSeriesForecastingJobConfig = Shapes::StructureShape.new(name: 'TimeSeriesForecastingJobConfig')
@@ -2427,17 +2430,21 @@ module Aws::SageMaker
2427
2430
  AutoMLProblemTypeConfig.add_member(:text_classification_job_config, Shapes::ShapeRef.new(shape: TextClassificationJobConfig, location_name: "TextClassificationJobConfig"))
2428
2431
  AutoMLProblemTypeConfig.add_member(:tabular_job_config, Shapes::ShapeRef.new(shape: TabularJobConfig, location_name: "TabularJobConfig"))
2429
2432
  AutoMLProblemTypeConfig.add_member(:time_series_forecasting_job_config, Shapes::ShapeRef.new(shape: TimeSeriesForecastingJobConfig, location_name: "TimeSeriesForecastingJobConfig"))
2433
+ AutoMLProblemTypeConfig.add_member(:text_generation_job_config, Shapes::ShapeRef.new(shape: TextGenerationJobConfig, location_name: "TextGenerationJobConfig"))
2430
2434
  AutoMLProblemTypeConfig.add_member(:unknown, Shapes::ShapeRef.new(shape: nil, location_name: 'unknown'))
2431
2435
  AutoMLProblemTypeConfig.add_member_subclass(:image_classification_job_config, Types::AutoMLProblemTypeConfig::ImageClassificationJobConfig)
2432
2436
  AutoMLProblemTypeConfig.add_member_subclass(:text_classification_job_config, Types::AutoMLProblemTypeConfig::TextClassificationJobConfig)
2433
2437
  AutoMLProblemTypeConfig.add_member_subclass(:tabular_job_config, Types::AutoMLProblemTypeConfig::TabularJobConfig)
2434
2438
  AutoMLProblemTypeConfig.add_member_subclass(:time_series_forecasting_job_config, Types::AutoMLProblemTypeConfig::TimeSeriesForecastingJobConfig)
2439
+ AutoMLProblemTypeConfig.add_member_subclass(:text_generation_job_config, Types::AutoMLProblemTypeConfig::TextGenerationJobConfig)
2435
2440
  AutoMLProblemTypeConfig.add_member_subclass(:unknown, Types::AutoMLProblemTypeConfig::Unknown)
2436
2441
  AutoMLProblemTypeConfig.struct_class = Types::AutoMLProblemTypeConfig
2437
2442
 
2438
2443
  AutoMLProblemTypeResolvedAttributes.add_member(:tabular_resolved_attributes, Shapes::ShapeRef.new(shape: TabularResolvedAttributes, location_name: "TabularResolvedAttributes"))
2444
+ AutoMLProblemTypeResolvedAttributes.add_member(:text_generation_resolved_attributes, Shapes::ShapeRef.new(shape: TextGenerationResolvedAttributes, location_name: "TextGenerationResolvedAttributes"))
2439
2445
  AutoMLProblemTypeResolvedAttributes.add_member(:unknown, Shapes::ShapeRef.new(shape: nil, location_name: 'unknown'))
2440
2446
  AutoMLProblemTypeResolvedAttributes.add_member_subclass(:tabular_resolved_attributes, Types::AutoMLProblemTypeResolvedAttributes::TabularResolvedAttributes)
2447
+ AutoMLProblemTypeResolvedAttributes.add_member_subclass(:text_generation_resolved_attributes, Types::AutoMLProblemTypeResolvedAttributes::TextGenerationResolvedAttributes)
2441
2448
  AutoMLProblemTypeResolvedAttributes.add_member_subclass(:unknown, Types::AutoMLProblemTypeResolvedAttributes::Unknown)
2442
2449
  AutoMLProblemTypeResolvedAttributes.struct_class = Types::AutoMLProblemTypeResolvedAttributes
2443
2450
 
@@ -8529,6 +8536,13 @@ module Aws::SageMaker
8529
8536
  TextClassificationJobConfig.add_member(:target_label_column, Shapes::ShapeRef.new(shape: TargetLabelColumn, required: true, location_name: "TargetLabelColumn"))
8530
8537
  TextClassificationJobConfig.struct_class = Types::TextClassificationJobConfig
8531
8538
 
8539
+ TextGenerationJobConfig.add_member(:completion_criteria, Shapes::ShapeRef.new(shape: AutoMLJobCompletionCriteria, location_name: "CompletionCriteria"))
8540
+ TextGenerationJobConfig.add_member(:base_model_name, Shapes::ShapeRef.new(shape: BaseModelName, location_name: "BaseModelName"))
8541
+ TextGenerationJobConfig.struct_class = Types::TextGenerationJobConfig
8542
+
8543
+ TextGenerationResolvedAttributes.add_member(:base_model_name, Shapes::ShapeRef.new(shape: BaseModelName, location_name: "BaseModelName"))
8544
+ TextGenerationResolvedAttributes.struct_class = Types::TextGenerationResolvedAttributes
8545
+
8532
8546
  TimeSeriesConfig.add_member(:target_attribute_name, Shapes::ShapeRef.new(shape: TargetAttributeName, required: true, location_name: "TargetAttributeName"))
8533
8547
  TimeSeriesConfig.add_member(:timestamp_attribute_name, Shapes::ShapeRef.new(shape: TimestampAttributeName, required: true, location_name: "TimestampAttributeName"))
8534
8548
  TimeSeriesConfig.add_member(:item_identifier_attribute_name, Shapes::ShapeRef.new(shape: ItemIdentifierAttributeName, required: true, location_name: "ItemIdentifierAttributeName"))
@@ -32,11 +32,11 @@ module Aws::SageMaker
32
32
  raise ArgumentError, "FIPS and DualStack are enabled, but this partition does not support one or both"
33
33
  end
34
34
  if Aws::Endpoints::Matchers.boolean_equals?(use_fips, true)
35
- if Aws::Endpoints::Matchers.boolean_equals?(true, Aws::Endpoints::Matchers.attr(partition_result, "supportsFIPS"))
36
- if Aws::Endpoints::Matchers.string_equals?("aws", Aws::Endpoints::Matchers.attr(partition_result, "name"))
35
+ if Aws::Endpoints::Matchers.boolean_equals?(Aws::Endpoints::Matchers.attr(partition_result, "supportsFIPS"), true)
36
+ if Aws::Endpoints::Matchers.string_equals?(Aws::Endpoints::Matchers.attr(partition_result, "name"), "aws")
37
37
  return Aws::Endpoints::Endpoint.new(url: "https://api-fips.sagemaker.#{region}.amazonaws.com", headers: {}, properties: {})
38
38
  end
39
- if Aws::Endpoints::Matchers.string_equals?("aws-us-gov", Aws::Endpoints::Matchers.attr(partition_result, "name"))
39
+ if Aws::Endpoints::Matchers.string_equals?(Aws::Endpoints::Matchers.attr(partition_result, "name"), "aws-us-gov")
40
40
  return Aws::Endpoints::Endpoint.new(url: "https://api-fips.sagemaker.#{region}.amazonaws.com", headers: {}, properties: {})
41
41
  end
42
42
  return Aws::Endpoints::Endpoint.new(url: "https://api.sagemaker-fips.#{region}.#{partition_result['dnsSuffix']}", headers: {}, properties: {})
@@ -2140,6 +2140,10 @@ module Aws::SageMaker
2140
2140
  # * For time-series forecasting: `text/csv;header=present` or
2141
2141
  # `x-application/vnd.amazon+parquet`. The default value is
2142
2142
  # `text/csv;header=present`.
2143
+ #
2144
+ # * For text generation (LLMs fine-tuning): `text/csv;header=present`
2145
+ # or `x-application/vnd.amazon+parquet`. The default value is
2146
+ # `text/csv;header=present`.
2143
2147
  # @return [String]
2144
2148
  #
2145
2149
  # @!attribute [rw] compression_type
@@ -2170,9 +2174,9 @@ module Aws::SageMaker
2170
2174
  # @!attribute [rw] max_candidates
2171
2175
  # The maximum number of times a training job is allowed to run.
2172
2176
  #
2173
- # For text and image classification, as well as time-series
2174
- # forecasting problem types, the supported value is 1. For tabular
2175
- # problem types, the maximum value is 750.
2177
+ # For text and image classification, time-series forecasting, as well
2178
+ # as text generation (LLMs fine-tuning) problem types, the supported
2179
+ # value is 1. For tabular problem types, the maximum value is 750.
2176
2180
  # @return [Integer]
2177
2181
  #
2178
2182
  # @!attribute [rw] max_runtime_per_training_job_in_seconds
@@ -2275,7 +2279,8 @@ module Aws::SageMaker
2275
2279
  include Aws::Structure
2276
2280
  end
2277
2281
 
2278
- # Specifies a metric to minimize or maximize as the objective of a job.
2282
+ # Specifies a metric to minimize or maximize as the objective of an
2283
+ # AutoML job.
2279
2284
  #
2280
2285
  # @!attribute [rw] metric_name
2281
2286
  # The name of the objective metric used to measure the predictive
@@ -2284,28 +2289,70 @@ module Aws::SageMaker
2284
2289
  # on the feedback provided by the objective metric when evaluating the
2285
2290
  # model on the validation dataset.
2286
2291
  #
2287
- # For the list of all available metrics supported by Autopilot, see
2288
- # [Autopilot metrics][1].
2289
- #
2290
- # If you do not specify a metric explicitly, the default behavior is
2291
- # to automatically use:
2292
+ # The list of available metrics supported by Autopilot and the default
2293
+ # metric applied when you do not specify a metric name explicitly
2294
+ # depend on the problem type.
2292
2295
  #
2293
2296
  # * For tabular problem types:
2294
2297
  #
2295
- # * Regression: `MSE`.
2298
+ # * List of available metrics:
2299
+ #
2300
+ # * Regression: `InferenceLatency`, `MAE`, `MSE`, `R2`, `RMSE`
2301
+ #
2302
+ # * Binary classification: `Accuracy`, `AUC`, `BalancedAccuracy`,
2303
+ # `F1`, `InferenceLatency`, `LogLoss`, `Precision`, `Recall`
2304
+ #
2305
+ # * Multiclass classification: `Accuracy`, `BalancedAccuracy`,
2306
+ # `F1macro`, `InferenceLatency`, `LogLoss`, `PrecisionMacro`,
2307
+ # `RecallMacro`
2308
+ #
2309
+ # For a description of each metric, see [Autopilot metrics for
2310
+ # classification and regression][1].
2311
+ #
2312
+ # * Default objective metrics:
2313
+ #
2314
+ # * Regression: `MSE`.
2296
2315
  #
2297
- # * Binary classification: `F1`.
2316
+ # * Binary classification: `F1`.
2298
2317
  #
2299
- # * Multiclass classification: `Accuracy`.
2318
+ # * Multiclass classification: `Accuracy`.
2300
2319
  #
2301
- # * For image or text classification problem types: `Accuracy`
2320
+ # * For image or text classification problem types:
2321
+ #
2322
+ # * List of available metrics: `Accuracy`
2323
+ #
2324
+ # For a description of each metric, see [Autopilot metrics for
2325
+ # text and image classification][2].
2326
+ #
2327
+ # * Default objective metrics: `Accuracy`
2302
2328
  #
2303
2329
  # * For time-series forecasting problem types:
2304
- # `AverageWeightedQuantileLoss`
2330
+ #
2331
+ # * List of available metrics: `RMSE`, `wQL`, `Average wQL`, `MASE`,
2332
+ # `MAPE`, `WAPE`
2333
+ #
2334
+ # For a description of each metric, see [Autopilot metrics for
2335
+ # time-series forecasting][3].
2336
+ #
2337
+ # * Default objective metrics: `AverageWeightedQuantileLoss`
2338
+ #
2339
+ # * For text generation problem types (LLMs fine-tuning): Fine-tuning
2340
+ # language models in Autopilot does not require setting the
2341
+ # `AutoMLJobObjective` field. Autopilot fine-tunes LLMs without
2342
+ # requiring multiple candidates to be trained and evaluated.
2343
+ # Instead, using your dataset, Autopilot directly fine-tunes your
2344
+ # target model to enhance a default objective metric, the
2345
+ # cross-entropy loss. After fine-tuning a language model, you can
2346
+ # evaluate the quality of its generated text using different
2347
+ # metrics. For a list of the available metrics, see [Metrics for
2348
+ # fine-tuning LLMs in Autopilot][4].
2305
2349
  #
2306
2350
  #
2307
2351
  #
2308
2352
  # [1]: https://docs.aws.amazon.com/sagemaker/latest/dg/autopilot-metrics-validation.html#autopilot-metrics
2353
+ # [2]: https://docs.aws.amazon.com/sagemaker/latest/dg/text-classification-data-format-and-metric.html
2354
+ # [3]: https://docs.aws.amazon.com/sagemaker/latest/dg/timeseries-objective-metric.html
2355
+ # [4]: https://docs.aws.amazon.com/sagemaker/latest/dg/llms-finetuning-models.html
2309
2356
  # @return [String]
2310
2357
  #
2311
2358
  # @see http://docs.aws.amazon.com/goto/WebAPI/sagemaker-2017-07-24/AutoMLJobObjective AWS API Documentation
@@ -2437,15 +2484,31 @@ module Aws::SageMaker
2437
2484
  # @return [Types::TextClassificationJobConfig]
2438
2485
  #
2439
2486
  # @!attribute [rw] tabular_job_config
2440
- # Settings used to configure an AutoML job V2 for a tabular problem
2487
+ # Settings used to configure an AutoML job V2 for the tabular problem
2441
2488
  # type (regression, classification).
2442
2489
  # @return [Types::TabularJobConfig]
2443
2490
  #
2444
2491
  # @!attribute [rw] time_series_forecasting_job_config
2445
- # Settings used to configure an AutoML job V2 for a time-series
2492
+ # Settings used to configure an AutoML job V2 for the time-series
2446
2493
  # forecasting problem type.
2447
2494
  # @return [Types::TimeSeriesForecastingJobConfig]
2448
2495
  #
2496
+ # @!attribute [rw] text_generation_job_config
2497
+ # Settings used to configure an AutoML job V2 for the text generation
2498
+ # (LLMs fine-tuning) problem type.
2499
+ #
2500
+ # <note markdown="1"> The text generation models that support fine-tuning in Autopilot are
2501
+ # currently accessible exclusively in regions supported by Canvas.
2502
+ # Refer to the documentation of Canvas for the [full list of its
2503
+ # supported Regions][1].
2504
+ #
2505
+ # </note>
2506
+ #
2507
+ #
2508
+ #
2509
+ # [1]: https://docs.aws.amazon.com/sagemaker/latest/dg/canvas.html
2510
+ # @return [Types::TextGenerationJobConfig]
2511
+ #
2449
2512
  # @see http://docs.aws.amazon.com/goto/WebAPI/sagemaker-2017-07-24/AutoMLProblemTypeConfig AWS API Documentation
2450
2513
  #
2451
2514
  class AutoMLProblemTypeConfig < Struct.new(
@@ -2453,6 +2516,7 @@ module Aws::SageMaker
2453
2516
  :text_classification_job_config,
2454
2517
  :tabular_job_config,
2455
2518
  :time_series_forecasting_job_config,
2519
+ :text_generation_job_config,
2456
2520
  :unknown)
2457
2521
  SENSITIVE = []
2458
2522
  include Aws::Structure
@@ -2462,36 +2526,43 @@ module Aws::SageMaker
2462
2526
  class TextClassificationJobConfig < AutoMLProblemTypeConfig; end
2463
2527
  class TabularJobConfig < AutoMLProblemTypeConfig; end
2464
2528
  class TimeSeriesForecastingJobConfig < AutoMLProblemTypeConfig; end
2529
+ class TextGenerationJobConfig < AutoMLProblemTypeConfig; end
2465
2530
  class Unknown < AutoMLProblemTypeConfig; end
2466
2531
  end
2467
2532
 
2468
- # The resolved attributes specific to the problem type of an AutoML job
2469
- # V2.
2533
+ # Stores resolved attributes specific to the problem type of an AutoML
2534
+ # job V2.
2470
2535
  #
2471
2536
  # @note AutoMLProblemTypeResolvedAttributes is a union - when returned from an API call exactly one value will be set and the returned type will be a subclass of AutoMLProblemTypeResolvedAttributes corresponding to the set member.
2472
2537
  #
2473
2538
  # @!attribute [rw] tabular_resolved_attributes
2474
- # Defines the resolved attributes for the `TABULAR` problem type.
2539
+ # The resolved attributes for the tabular problem type.
2475
2540
  # @return [Types::TabularResolvedAttributes]
2476
2541
  #
2542
+ # @!attribute [rw] text_generation_resolved_attributes
2543
+ # The resolved attributes for the text generation problem type.
2544
+ # @return [Types::TextGenerationResolvedAttributes]
2545
+ #
2477
2546
  # @see http://docs.aws.amazon.com/goto/WebAPI/sagemaker-2017-07-24/AutoMLProblemTypeResolvedAttributes AWS API Documentation
2478
2547
  #
2479
2548
  class AutoMLProblemTypeResolvedAttributes < Struct.new(
2480
2549
  :tabular_resolved_attributes,
2550
+ :text_generation_resolved_attributes,
2481
2551
  :unknown)
2482
2552
  SENSITIVE = []
2483
2553
  include Aws::Structure
2484
2554
  include Aws::Structure::Union
2485
2555
 
2486
2556
  class TabularResolvedAttributes < AutoMLProblemTypeResolvedAttributes; end
2557
+ class TextGenerationResolvedAttributes < AutoMLProblemTypeResolvedAttributes; end
2487
2558
  class Unknown < AutoMLProblemTypeResolvedAttributes; end
2488
2559
  end
2489
2560
 
2490
2561
  # The resolved attributes used to configure an AutoML job V2.
2491
2562
  #
2492
2563
  # @!attribute [rw] auto_ml_job_objective
2493
- # Specifies a metric to minimize or maximize as the objective of a
2494
- # job.
2564
+ # Specifies a metric to minimize or maximize as the objective of an
2565
+ # AutoML job.
2495
2566
  # @return [Types::AutoMLJobObjective]
2496
2567
  #
2497
2568
  # @!attribute [rw] completion_criteria
@@ -4842,6 +4913,8 @@ module Aws::SageMaker
4842
4913
  #
4843
4914
  # * For time-series forecasting: `S3Prefix`.
4844
4915
  #
4916
+ # * For text generation (LLMs fine-tuning): `S3Prefix`.
4917
+ #
4845
4918
  #
4846
4919
  #
4847
4920
  # [1]: https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_CreateAutoMLJob.html#sagemaker-CreateAutoMLJob-request-InputDataConfig
@@ -4883,16 +4956,28 @@ module Aws::SageMaker
4883
4956
  # problem type. For the list of default values per problem type, see
4884
4957
  # [AutoMLJobObjective][1].
4885
4958
  #
4886
- # <note markdown="1"> For tabular problem types, you must either provide both the
4887
- # `AutoMLJobObjective` and indicate the type of supervised learning
4888
- # problem in `AutoMLProblemTypeConfig`
4889
- # (`TabularJobConfig.ProblemType`), or none at all.
4959
+ # <note markdown="1"> * For tabular problem types: You must either provide both the
4960
+ # `AutoMLJobObjective` and indicate the type of supervised learning
4961
+ # problem in `AutoMLProblemTypeConfig`
4962
+ # (`TabularJobConfig.ProblemType`), or none at all.
4963
+ #
4964
+ # * For text generation problem types (LLMs fine-tuning): Fine-tuning
4965
+ # language models in Autopilot does not require setting the
4966
+ # `AutoMLJobObjective` field. Autopilot fine-tunes LLMs without
4967
+ # requiring multiple candidates to be trained and evaluated.
4968
+ # Instead, using your dataset, Autopilot directly fine-tunes your
4969
+ # target model to enhance a default objective metric, the
4970
+ # cross-entropy loss. After fine-tuning a language model, you can
4971
+ # evaluate the quality of its generated text using different
4972
+ # metrics. For a list of the available metrics, see [Metrics for
4973
+ # fine-tuning LLMs in Autopilot][2].
4890
4974
  #
4891
4975
  # </note>
4892
4976
  #
4893
4977
  #
4894
4978
  #
4895
4979
  # [1]: https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_AutoMLJobObjective.html
4980
+ # [2]: https://docs.aws.amazon.com/sagemaker/latest/dg/llms-finetuning-models.html
4896
4981
  # @return [Types::AutoMLJobObjective]
4897
4982
  #
4898
4983
  # @!attribute [rw] model_deploy_config
@@ -20226,7 +20311,13 @@ module Aws::SageMaker
20226
20311
  # Hyperparameter tuning uses the value of this metric to evaluate the
20227
20312
  # training jobs it launches, and returns the training job that results
20228
20313
  # in either the highest or lowest value for this metric, depending on
20229
- # the value you specify for the `Type` parameter.
20314
+ # the value you specify for the `Type` parameter. If you want to
20315
+ # define a custom objective metric, see [Define metrics and
20316
+ # environment variables][1].
20317
+ #
20318
+ #
20319
+ #
20320
+ # [1]: https://docs.aws.amazon.com/sagemaker/latest/dg/automatic-model-tuning-define-metrics-variables.html
20230
20321
  # @return [Types::HyperParameterTuningJobObjective]
20231
20322
  #
20232
20323
  # @!attribute [rw] hyper_parameter_ranges
@@ -20695,7 +20786,13 @@ module Aws::SageMaker
20695
20786
  # Hyperparameter tuning uses the value of this metric to evaluate the
20696
20787
  # training jobs it launches, and returns the training job that results
20697
20788
  # in either the highest or lowest value for this metric, depending on
20698
- # the value you specify for the `Type` parameter.
20789
+ # the value you specify for the `Type` parameter. If you want to define
20790
+ # a custom objective metric, see [Define metrics and environment
20791
+ # variables][1].
20792
+ #
20793
+ #
20794
+ #
20795
+ # [1]: https://docs.aws.amazon.com/sagemaker/latest/dg/automatic-model-tuning-define-metrics-variables.html
20699
20796
  #
20700
20797
  # @!attribute [rw] type
20701
20798
  # Whether to minimize or maximize the objective metric.
@@ -21311,8 +21408,8 @@ module Aws::SageMaker
21311
21408
  include Aws::Structure
21312
21409
  end
21313
21410
 
21314
- # Stores the configuration information for the image classification
21315
- # problem of an AutoML job V2.
21411
+ # The collection of settings used by an AutoML job V2 for the image
21412
+ # classification problem type.
21316
21413
  #
21317
21414
  # @!attribute [rw] completion_criteria
21318
21415
  # How long a job is allowed to run, or how many candidates a job is
@@ -35238,8 +35335,8 @@ module Aws::SageMaker
35238
35335
  # The resolved attributes.
35239
35336
  #
35240
35337
  # @!attribute [rw] auto_ml_job_objective
35241
- # Specifies a metric to minimize or maximize as the objective of a
35242
- # job.
35338
+ # Specifies a metric to minimize or maximize as the objective of an
35339
+ # AutoML job.
35243
35340
  # @return [Types::AutoMLJobObjective]
35244
35341
  #
35245
35342
  # @!attribute [rw] problem_type
@@ -37479,7 +37576,7 @@ module Aws::SageMaker
37479
37576
  include Aws::Structure
37480
37577
  end
37481
37578
 
37482
- # The collection of settings used by an AutoML job V2 for the `TABULAR`
37579
+ # The collection of settings used by an AutoML job V2 for the tabular
37483
37580
  # problem type.
37484
37581
  #
37485
37582
  # @!attribute [rw] candidate_generation_config
@@ -37620,7 +37717,7 @@ module Aws::SageMaker
37620
37717
  include Aws::Structure
37621
37718
  end
37622
37719
 
37623
- # The resolved attributes specific to the `TABULAR` problem type.
37720
+ # The resolved attributes specific to the tabular problem type.
37624
37721
  #
37625
37722
  # @!attribute [rw] problem_type
37626
37723
  # The type of supervised learning problem available for the model
@@ -37794,8 +37891,8 @@ module Aws::SageMaker
37794
37891
  include Aws::Structure
37795
37892
  end
37796
37893
 
37797
- # Stores the configuration information for the text classification
37798
- # problem of an AutoML job V2.
37894
+ # The collection of settings used by an AutoML job V2 for the text
37895
+ # classification problem type.
37799
37896
  #
37800
37897
  # @!attribute [rw] completion_criteria
37801
37898
  # How long a job is allowed to run, or how many candidates a job is
@@ -37822,6 +37919,60 @@ module Aws::SageMaker
37822
37919
  include Aws::Structure
37823
37920
  end
37824
37921
 
37922
+ # The collection of settings used by an AutoML job V2 for the text
37923
+ # generation problem type.
37924
+ #
37925
+ # <note markdown="1"> The text generation models that support fine-tuning in Autopilot are
37926
+ # currently accessible exclusively in regions supported by Canvas. Refer
37927
+ # to the documentation of Canvas for the [full list of its supported
37928
+ # Regions][1].
37929
+ #
37930
+ # </note>
37931
+ #
37932
+ #
37933
+ #
37934
+ # [1]: https://docs.aws.amazon.com/sagemaker/latest/dg/canvas.html
37935
+ #
37936
+ # @!attribute [rw] completion_criteria
37937
+ # How long a job is allowed to run, or how many candidates a job is
37938
+ # allowed to generate.
37939
+ # @return [Types::AutoMLJobCompletionCriteria]
37940
+ #
37941
+ # @!attribute [rw] base_model_name
37942
+ # The name of the base model to fine-tune. Autopilot supports
37943
+ # fine-tuning a variety of large language models. For information on
37944
+ # the list of supported models, see [Text generation models supporting
37945
+ # fine-tuning in Autopilot][1]. If no `BaseModelName` is provided, the
37946
+ # default model used is Falcon-7B-Instruct.
37947
+ #
37948
+ #
37949
+ #
37950
+ # [1]: https://docs.aws.amazon.com/sagemaker/src/AWSIronmanApiDoc/build/server-root/sagemaker/latest/dg/llms-finetuning-models.html#llms-finetuning-supported-llms
37951
+ # @return [String]
37952
+ #
37953
+ # @see http://docs.aws.amazon.com/goto/WebAPI/sagemaker-2017-07-24/TextGenerationJobConfig AWS API Documentation
37954
+ #
37955
+ class TextGenerationJobConfig < Struct.new(
37956
+ :completion_criteria,
37957
+ :base_model_name)
37958
+ SENSITIVE = []
37959
+ include Aws::Structure
37960
+ end
37961
+
37962
+ # The resolved attributes specific to the text generation problem type.
37963
+ #
37964
+ # @!attribute [rw] base_model_name
37965
+ # The name of the base model to fine-tune.
37966
+ # @return [String]
37967
+ #
37968
+ # @see http://docs.aws.amazon.com/goto/WebAPI/sagemaker-2017-07-24/TextGenerationResolvedAttributes AWS API Documentation
37969
+ #
37970
+ class TextGenerationResolvedAttributes < Struct.new(
37971
+ :base_model_name)
37972
+ SENSITIVE = []
37973
+ include Aws::Structure
37974
+ end
37975
+
37825
37976
  # The collection of components that defines the time-series.
37826
37977
  #
37827
37978
  # @!attribute [rw] target_attribute_name
@@ -53,6 +53,6 @@ require_relative 'aws-sdk-sagemaker/customizations'
53
53
  # @!group service
54
54
  module Aws::SageMaker
55
55
 
56
- GEM_VERSION = '1.216.0'
56
+ GEM_VERSION = '1.217.0'
57
57
 
58
58
  end
metadata CHANGED
@@ -1,14 +1,14 @@
1
1
  --- !ruby/object:Gem::Specification
2
2
  name: aws-sdk-sagemaker
3
3
  version: !ruby/object:Gem::Version
4
- version: 1.216.0
4
+ version: 1.217.0
5
5
  platform: ruby
6
6
  authors:
7
7
  - Amazon Web Services
8
8
  autorequire:
9
9
  bindir: bin
10
10
  cert_chain: []
11
- date: 2023-10-12 00:00:00.000000000 Z
11
+ date: 2023-10-26 00:00:00.000000000 Z
12
12
  dependencies:
13
13
  - !ruby/object:Gem::Dependency
14
14
  name: aws-sdk-core