6.2.8. Output and logging

For the generation of checkpoint files, see Generate checkpoints. The results directory is a shared key, documented here

Most of the output files (the contents of training_set_results/best, training_set_results/running_loss.txt, etc., are updated every LoggingInterval%General evaluations.

The training set is logged at evaluation numbers that are multiples of LoggingInterval%General, or when the loss function reaches a new lowest value, or when the validation is evaluated (see below).

If there is a validation set, then

  • The validation set is evaluated approximately every EvaluateEvery evaluations (configured in the validation set’s DataSet block),

  • The value for EvaluateEvery is adjusted to be a multiple of LoggingInterval%General (the validation set is not evaluated if it is not logged),

  • The validation set is not necessarily evaluated/logged for evaluation numbers that are exact multiples of LoggingInterval%General. It uses the evaluation-number-per-optimizer (when there are multiple optimizers), and even then only an approximate multiple is used.

  • Whenever the validation is logged, so is the training set.

LoggingInterval
Type:

Block

Description:

Number of function evaluations between every log to file.

Flush
Type:

Integer

Default value:

10

Description:

The number of function evaluations between flushes of the output streams to disk. If Flush = General, then files will be flushed after every logged call.

General
Type:

Integer

Default value:

10

Description:

The number of function evaluations between writes to stdout, running_loss.txt, running_stats.txt and all latest/ and best/ files. This is also the interval with which the validation set will be evaluated and logged.

History
Type:

Integer

Default value:

500

Description:

The number of function evaluations between saves to the history directory with copies of latest/.

Parameters
Type:

Integer

Default value:

500

Description:

The number of function evaluations between writes to running_active_parameters.txt.

GlompoLogging
Type:

Bool

Default value:

Yes

GUI name:

Print manager logging messages:

Description:

Include status and progress information from the optimization manager in the printstreams.

SplitPrintstreams
Type:

Bool

Default value:

Yes

GUI name:

Split optimizer printstreams:

Description:

Split print statements from each optimizer to separate files.

PrintStatusInterval
Type:

Float

Default value:

600.0

Unit:

s

Description:

Number of seconds between printing of a status summary.

GlompoSummaryFiles
Type:

Multiple Choice

Default value:

None

Options:

[None, 1, 2, 3, 4]

GUI name:

GloMPO summary files:

Description:

Indicates what GloMPO-style outputs you would like saved to disk. Higher values also save all lower level information. Available options: • None: Nothing is saved. • 1: YAML file with summary info about the optimization settings, performance and the result. • 2: PNG file showing the trajectories of the optimizers. • 3: HDF5 file containing iteration history for each optimizer. • 4: More detailed HDF5 log including the residual results for each optimizer, data set and iteration.

PLAMSWorkingDirectory
Type:

String

Default value:

Description:

Path to PLAMS working directory to temporarily hold Job results files.