Task: Task

In MindFoundry/optaas-r-client: Mind Foundry OPTaaS R Client

Description

Allows you to access Task attributes and perform all Task-related functions.

Usage

Task

Format

An object of class R6ClassGenerator of length 24.

Examples

## Not run: 
task$id  # The unique ID of the Task
task$json  # The full JSON representation of the Task in OPTaaS

# To run your task:
# First define a scoring function whose arguments are identical to your parameter names.
# The function should return a score value 
# Then to run the task for n iterations and store the best result:
best_result <- task$run(scoring_function, n)
# Or to run for at most n iterations, but stop when the score is X or better:
best_result <- task$run(scoring_function, n, X)

# Or if you prefer to do things manually:
configuration <- task$generate_configuration()
result <- Result$new(configuration=configuration, score=score, ...)
next_configuration <- task$record_result(result)

task$get_best_result()  # Result with the best score (for single-objective tasks)
task$get_pareto_set()  # Set of Pareto front results (for multi-objective tasks)

# To use batching (i.e. parallel score evaluation):
configurations <- task$generate_configurations(number_of_workers)
results <- your_parallel_scoring_function(configurations)
next_configurations <- task$record_results(results)

# Other functions:
task$add_user_defined_configuration()  # Warm-start the optimization with pre-calculated results
task$get_results()  # Get all recorded results
task$get_surrogate_predictions()  # Get predicted scores for specific configurations
task$complete()  # Complete the task (no more configurations or results can be added)
task$resume()  # Resume a completed task
task$delete()  # Delete the task

## End(Not run)

MindFoundry/optaas-r-client documentation built on May 17, 2019, 7:32 p.m.