6.1. General input¶

6.1.1. Task¶

The Task keyword specifies which type of program to run:

Task

Type:

Multiple Choice

Default value:

Optimization

Options:

[Optimization, GenerateReference, SinglePoint, Sensitivity, MachineLearning]

Description:

Task to run. Available options: •MachineLearning: Optimization for machine learning models. •Optimization: Global optimization powered by GloMPO •Generate Reference: Run jobs with reference engine to get reference values •Single Point: Evaluate the current configuration of jobs, training data, and parameters •Sensitivity: Measure the sensitivity of the loss function to each of the active parameters

See more information about

• Optimization

• GenerateReference

• SinglePoint

• Sensitivity

6.1.2. Job Collection Key¶

JobCollection

Type:

String

Default value:

job_collection.yaml

Description:

Path to JobCollection YAML file.

6.1.3. Engine Collection Key¶

If you specify an EngineCollection in the input file, it will override the engine collection specified inside the job_collection.yaml file.

EngineCollection

Type:

String

Default value:

job_collection_engines.yaml

Description:

Path to (optional) JobCollection Engines YAML file.

6.1.4. ParameterInterface Key¶

ParameterInterface

Type:

String

Default value:

parameter_interface.yaml

Description:

Path to parameter interface YAML file.

6.1.5. DataSet Key¶

The training set, validation set, and any other data sets are specified with the DataSet recurring block.

DataSet
   BatchSize integer
   EvaluateEvery integer
   LossFunction [mae | rmse | sse | sae]
   MaxJobs integer
   MaxJobsShuffle Yes/No
   Name string
   Path string
   UsePipe Yes/No
End

DataSet

Type:

Block

Recurring:

True

Description:

Configuration settings for each data set in the optimization.

BatchSize

Type:

Integer

Default value:

0

Description:

Number of data set entries to be evaluated per epoch. Default 0 means all entries.

EvaluateEvery

Type:

Integer

Default value:

1

Description:

This data set is evaluated every n evaluations of the training set. This will always be set to 1 for the training set. For other data sets it will be adjusted to the closest multiple of LoggingInterval%General, i.e., you cannot evaluate an extra data set more frequently than you log it.

LossFunction

Type:

Multiple Choice

Default value:

sse

Options:

[mae, rmse, sse, sae]

Description:

Loss function used to quantify the error between model and reference values. This becomes the minimization task. Available options: • mae: Mean absolute error • rmse: Root mean squared error • sse: Sum of squared errors • sae: Sum of absolute errors

MaxJobs

Type:

Integer

Default value:

0

Description:

Limit each evaluation to a subset of n jobs. Default 0 meaning all jobs are used.

MaxJobsShuffle

Type:

Bool

Default value:

No

Description:

Use a different job subset every for every evaluation.

Name

Type:

String

Default value:

Description:

Unique data set identifier. The first occurrence of DataSet will always be called training_set. The second will always be called validation_set. These cannot be overwritten. Later occurrences will default to data_set_xx where xx starts at 03 and increments from there. This field can be used to customize the latter names.

Path

Type:

String

Description:

Path to DataSet YAML file.

UsePipe

Type:

Bool

Default value:

Yes

Description:

Use AMS Pipe for suitable jobs to speed-up evaluation.

6.1.6. Results Directory¶

This key specifies the name of the directory into which the results will be saved. ParAMS will attempt to create this directory. If it already exists, ParAMS will automatically increment the directory name. For example, if results already exists ParAMS will attempt results.001, if results.001 exists it will attempt results.002 and so on.

$AMSBIN/params --replace

$AMSBIN/params -o path/to/results

ResultsDirectory

Type:

String

Default value:

results

GUI name:

Working directory:

Description:

Directory in which output files will be created.